linguistic structures - Professor Binkert's Webpage - Oakland ...

LINGUISTIC STRUCTURES
LECTURE NOTES AND WORKBOOK
Peter J. Binkert
Oakland University
Rochester, Michigan

Copyright © The Langtech Corporation 2012
The Langtech Corporation
643 Hazy View Lane
Milford, Michigan 48381-2159
All rights reserved, including those of translation into foreign languages. Except for the quotation
of short passages for the purposes of criticism and review, no part of this publication may be
reproduced or transmitted in any form or by any means, electronic or mechanical, including
photocopy, recording, or any information storage and retrieval system, without prior permission in
writing from the publisher.
Printed in Milford, Michigan.

PREFACE
This book is not a traditional introductory textbook for linguistics. Although many of the traditional
topics are discussed in the following chapters, the book does not aim at a comprehensive summary
of the discipline. Rather, the intent is to present to the student a unified theory of human language
which demonstrates specifically that the structure of human language is not arbitrary and that it
derives directly from genetically determined faculties of human beings. The essential nature of
human language is not a matter of choice, convention, or whim. Cultural diversity among different
societies and linguistic diversity among different languages and dialects reflect superficial variations
on this basic, biologically determined structure.
The central theme of this text is quite simple: human language reflects the capacities and limitations
of human beings. There are three facts supporting this theme. First, all normal human beings,
regardless of the languages they speak and the cultures they represent, have the same basic
biological makeup. Second, all children learn whatever language they are exposed to in whatever
cultural setting; children are not predestined to learn specific languages. Third, language acquisition
in all children proceeds in a uniform and predictable fashion despite widely varying environmental
conditions.
These facts clearly indicate that the structure of all languages must be based on the nature of human
beings. This is not an original idea; indeed, much of the current work in linguistics in all theoretical
frameworks proceeds from this position. For example, it is basic to Chomsky’s concept of linguistic
universals.
Where this text differs from most introductory texts is in the manner in which the various
subdivisions within linguistics are described. At every point possible, the development of our theme
will be bolstered by specific arguments that have reference to human biology and psychology. All
components of language will be related to each other, rather than presented as discrete units, so that
elements of linguistic structure, change, and variation are integrated. Our discussion of different
languages and different cultures will show how linguistic divergence is constrained by linguistic
universals.
Generally, linguistics departments on university campuses are grouped with the social sciences
(psychology, sociology, anthropology, etc.) or humanities (philosophy, music, art, literature, etc.).
In either of these contexts, linguistics is in a distinctive position to make contributions to the history
of ideas. Linguistic argumentation, that is, the justification of grammars, is a highly developed
methodology that can be used to make predictions about the nature of man and mind. Traditional
justification of theoretical models in the natural and physical sciences derives from experimentation.
The techniques of justification in biology and physics are familiar to every student. In linguistics
also, it is possible to formulate hypotheses of considerable rigor and subject them to scientific
scrutiny that leads directly to their verification or refutation. This book aims to show the
introductory student how this is possible.
More than anything else, this text is designed to disabuse readers of the many misconceptions that
surround the study of language. These fallacies include, among other things, the idea that different

dialects, different cultural patterns, and even different languages reflect different levels of human
competence, as well as the prevalent idea that grammatical structure is haphazard and unjustifiable.
Although it may be difficult to believe at this point, the study of grammar can lead to provocative
and interesting ideas about the nature of human beings and the origin of cultural diversity.
Needless to say, one cannot attain such a high level of generality about language, or anything else,
without attention to detail. Therefore, in the following chapters, students will be introduced to some
of the technical vocabulary of linguistics. The aim is not to memorize facts per se, although mastery
of some details is essential before application can begin; rather, it is to show how rigorous
investigation can lead to meaningful generalizations. Lists of memorized facts rarely stay in
people’s minds for very long, but genuine comprehension of issues does. Moreover, such
comprehension can have a significant impact on one’s life. In this regard, this text aims to satisfy
some of the major objectives of general education, which include, among other things, helping
students understand and master basic techniques for the analysis and synthesis of ideas. Such
techniques involve the ability to gather, organize, and interpret data, to separate what is significant
and interesting from what is irrelevant and trivial, and to formulate hypotheses of real explanatory
power. Effective thinking occurs when one is able to uncover the essential nature of any given
problem and to propose reasonable solutions consistent with available resources.
ii

TABLE OF CONTENTS
Introduction: Linguistics and the Job Market ........................................1
Chapter One: Fundamentals of Linguistics ........................................23
1.1 Grammatical Characterization and Grammatical Realization ...............23
1.2 The Study of Grammar ............................................28
1.3 Speech and Writing ...............................................29
1.4 Linguistic Argumentation ..........................................30
1.5 Universal Grammar (UG) ..........................................32
1.6 Levels of Adequacy ...............................................38
1.7 Language Variation ...............................................43
1.8 Language and Culture .............................................44
1.9 Summary .......................................................47
Exercises for Chapter One ................................................50
Chapter Two: Phonetics and Phonology ...........................................55
2.1 Phonetics .......................................................55
2.1.1 Vowels ...................................................55
Figure One: the Vocal Apparatus ..........................................56
Figure Two: English Vowels .............................................57
Figure Three: Distinctive Features for English Vowels .........................58
2.1.2 Consonants ................................................59
Figure Four: English Consonants, Liquids, and Glides .........................60
2.1.3 Liquids and Glides ..........................................62
Figure Five: the Major Phonological Categories ..............................62
Figure Six: Distinctive Features for English Consonants, Liquids & Glides .........64
2.1.4 Review of Issues ...........................................65
2.1.5 Suprasegmentals ...........................................67
2.1.6 Syllables ..................................................67
2.2 Phonology ......................................................69
Figure Seven: Phonological and Phonetic Representations ......................75
2.3 Phonotactics .....................................................76
2.4 Morphology .....................................................77
2.5 The Biological Basis of Phonology ...................................79
2.5.1 First Set of Observations .....................................81
2.5.2 Second Set of Observations ...................................83
2.5.3 Third Set of Observations ....................................85
2.6 The Phonological and Lexical Components of a Grammar .................85
2.7 English Spelling ..................................................87
2.8 Summary of Rule Formalism........................................89
2.9 Notes on Syllables ................................................89

Summary of English Sounds and Spelling ....................................90
Exercises for Chapter Two ...............................................92
Appendix A: Answers to Transcription Exercises .....................103
Appendix B: Phonological Rules ..................................108
Appendix C: Slash–dash Notation .................................111
Appendix D: Phonology Problem ..................................112
Appendix E: Morphology Problem .................................113
Appendix F: English Vowel Shift..................................114
Appendix G: English Morphology .................................115
Chapter Three: Syntax .......................................................117
3.1 Linguistic Argumentation Revisited .................................117
3.1.1 Linguistic Argumentation: Sentence Types ......................117
3.1.2 Linguistic Argumentation: Parts of Speech ......................120
3.1.3 Linguistic Argumentation: Phrases ............................121
3.1.4 Linguistic Argumentation: Strategies ..........................123
3.2 Ordering Constraints .............................................124
3.2.1 Testing Hypotheses: Reference, Omission, Placement .............125
3.2.2 Language and Memory .....................................126
3.3 Traditional and Structural Grammar .................................127
3.4 Tense and the Internal Structure of Sentences..........................133
3.5 Determiners and the Internal Structure of Noun Phrases .................137
3.5.1 Abstract Elements: Ø .......................................138
3.5.2 Phrase Structure Rules: First Proposal ..........................142
3.6 The Endocentricity Constraint......................................145
3.7 The Lexicon ....................................................147
3.8 A Note on Scientific Inquiry .......................................150
3.9 X–bar Syntax ...................................................152
3.10 Generalizing Phrase Structure: S and NP .............................157
3.11 Adjuncts (Modifiers) and Complements ..............................159
3.12 Residential Grammar (RG) ........................................170
Figure Eight: Features for English Morphosyntactic Categories .................183
3.13 Transformational Generative Grammar ...............................191
3.14 Nontransformational Generative Grammar ............................194
3.15 Command Relations in Phrase Structure ..............................195
3.16 Empty Categories ................................................197
3.16 The English Auxiliary ............................................203
3.16.1 The Facts and Generalizations ................................203
3.16.2 Problems with the Classic TG Analysis ........................206
3.16.3 Resolution ...............................................214
3.16.4 Summary of Nominals and Verbals in RG ......................217
3.16.5 Parses Illustrating the English Auxiliary ........................218
iv

Exercises for Chapter Three .............................................227
Appendix A: Answers to Exercises ..................................231
Appendix B: Latin Syntax ........................................249
Appendix C: Summary of Tree Structures ............................251
Chapter Four: Semantics......................................................255
4.1 Preliminary Remarks .............................................255
4.2 Denotation and Connotation .......................................255
4.3 Synonymy, Homonymy, and Antonymy ..............................256
4.4 Hyponymy/hypernymy and Meronymy/holonymy .....................256
4.5 Idioms ........................................................256
4.6 Ambiguity and Paraphrase.........................................257
4.7 Pragmatics .....................................................259
4.8 Thematic Relations: Case Grammar .................................260
4.9 The Meaning and Use of the Word Case ..............................263
4.10 Problems with Case Grammar ......................................265
4.11 RG Semantics ...................................................266
4.12 Positional Thematic Relations in Rg .................................273
4.13 Some Basic Feature Distinctions ....................................274
Figure Nine I: Proximal Positional Thematic Relations ([+PST, –TMP, +PRX]) . . . 277
Figure Nine II: Nonproximal Positional Thematic Relations ([+PST, –TMP, –PRX]) 277
4.14 Nonpositional Thematic Relations ...................................279
Figure Ten I: Proximal Nonpositional Thematic Relations ([–PST, +PRX]) ........280
Figure Ten II: Nonproximal Nonpositional Thematic Relations ([–PST, –PRX]) ....280
4.15 Dimensional Thematic Relations ....................................283
Figure Eleven: Classificatory Thematic Relations ............................284
4.16 Prototypes and Strength of Association ...............................286
4.17 The Perceptual Basis of Semantics ..................................292
4.18 Semantic Networks ..............................................294
4.19 Postscript ......................................................296
Exercises for Chapter Four ..............................................297
Appendix A: Summary of Thematic Relations...............................302
Chapter Five: Language, People, and Computers ...................................305
5.1 Syntax and Cognition .............................................305
5.2 Overview of Anaphora ............................................325
Exercises for Chapter Five...............................................337
v

Supplement One: The Nature of Human Language ............................339
Fundamental Empirical Problems for Linguistics .......................339
Theories of Language ............................................345
Linguistic Universals .............................................347
Types of Languages ..............................................363
The Evolution of Human Language ..................................366
Limitations on the Form of Natural Language .........................369
Supplement Two: Grammar Review ......................................371
Syntactic Categories (The Parts of Speech) ............................371
Syntactic Constructions ...........................................373
Morphology ....................................................374
Some Inflectional Categories.......................................375
Supplement Three (Part I): Typological Classification ........................376
Supplement Three (Part II): Typological Classification ........................377
Supplement Four (Part I): Historical Classification ...........................378
Supplement Four (Part II): Historical Classification ..........................379
Supplement Five (Part I): Languages of Africa and the mid East ................380
Supplement Five (Part II): American Indian Languages .......................381
Supplement Five (Part III): Indo–european Languages ........................382
Supplement Five (Part IV): Languages of East Asia & the Pacific ...............383
Supplement Five (Part V): Other Languages of Europe and Asia ................384
Supplement Six: Languages and Language Groups ...........................385
Supplement Seven: Some Alphabets .......................................387
Supplement Eight: Sample Parses from the Langtech Parser ....................389
Sample Parses Illustrating Basic Sentence Patterns .....................389
Supplement Nine: Sample Parses from the Langtech Parser .....................427
Sample Parses Illustrating Noun Phrases ..............................427
Supplement Ten: Sample Parses from the Langtech Parser .....................431
Empty Categories: Ø, [U], and [E] ..................................431
Supplement Eleven: Sample Parses from the Langtech Parser ...................434
Sample Parses Illustrating Participles ................................434
Index .....................................................................443
vi

INTRODUCTION: LINGUISTICS AND THE JOB MARKET
There is no doubt about it: the competition for jobs in today’s market is intense. According to an
Associated Press news story on April 25, 2012, half of all new graduates are either jobless or
underemployed:
“In the last year, [graduates] were more likely to be employed as waiters, waitresses,
bartenders and food-service helpers than as engineers, physicists, chemists and
mathematicians combined (100,000 versus 90,000). There were more working in
office-related jobs such as receptionist or payroll clerk than in all computer professional jobs
(163,000 versus 100,000). More also were employed as cashiers, retail clerks and customer
representatives than engineers (125,000 versus 80,000).
“According to government projections released last month, only three of the 30 occupations
with the largest projected number of job openings by 2020 will require a bachelor’s degree
or higher to fill the position – teachers, college professors and accountants. Most job openings
are in professions such as retail sales, fast food and truck driving, jobs which aren’t easily
replaced by computers.” (Recent College Graduates Finding Few Job Prospects, Hope Yen,
Associated Press, www.detroitnews.com/article/20120425)
Despite these numbers, college graduates still fare better than those without degrees according to
the Bureau of Labor Statistics (Yes, A College Education Is Worth the Costs, Rodney K. Smith, USA
Today, 2011, www.usatoday.com/news/opinion/forum/story/2011-12-06/). Unemployment rates
decrease as education increases (Bureau of Labor Statistics, 2010, www.bls.gov):
•14.9% of those without a high school diploma
•10.3% of those with a high school education
•7% of those with an associate degree
•5.4% of those with a bachelor’s degree
•2.4% of those with a professional degree
•1.9% of those with a doctoral degree.
Also, educational attainment correlates with income. Here’s the average weekly income for those
who have jobs (Bureau of Labor Statistics, 2010, www.bls.gov):
•$444 for those with less than a high school degree
•$626 for those with a high school degree
•$767 for those with an associate degree
•$1,038 for those with a bachelor’s degree
•$1,550 for those with a doctoral degree
Clearly, making the most out of a college education is crucial. Before choosing a program of study,
students should spend some time surfing the internet and googling such search items as “what
employers want,” “skills needed for the workplace,” “trends in the marketplace,” and so on. There
are many myths and traditions surrounding options like the choice of a major in college. For

2
example, students often think that it will be impossible to get a job in a specific discipline unless
they study that discipline in depth in college. They cannot, for instance, become a manager unless
they are a business major. Actually, careers often have little to do with specific majors in college.
The Employee Benefits Research Institute, reports that the average tenure for all wage and salary
workers age 25 or older was 5.2 years in 2010, compared with 5.0 years in 1983 (www.ebri.
org/publications, Vol 31, No 12, 2010). That means that in the United States for nearly three
decades workers generally stayed at the same job for about 5 years. As a result, if individuals work
for 50 years, they can expects to have about 10 different jobs during their working careers. Different
sources report slightly different statistics, but surfing the internet clearly shows that frequent job and
career change is the norm in today’s market.
Such change is in stark contrast to the situation in the past when individuals not only kept the same
job for their entire lives but also had the same job as their parents and grandparents whether they be
policemen, farmers, teachers, doctors, coal miners, etc. Nowadays, on the other hand, people can
expect to change not only the company they work for but actually the type of work they do, that is,
change their careers. Thus, one of the first things students should consider in choosing a program
of study is finding one that can provide them with the skills to be adaptable, flexible, and mobile in
a global and unpredictable economy. It is unlikely that the job a student finds after graduation will
be directly linked to the course content of a specific undergraduate major.
Problem Solving Skills Needed for the Workplace
What exactly are the skills that will prepare a student to be adaptable, flexible, and mobile? One
way to answer that question is to look at the kinds of skills that various employers look for. Again,
surfing the internet, one finds that employers across the board look for many of the same skills in
their employees. Chief among these skills is the ability to solve problems. For example, one survey
of 301 executives in Fortune 1000 companies found that 99% of the executives think that the ability
to solve problems and to think critically is either an absolutely essential or very important skill for
college and career readiness (The MetLife Survey of the American Teacher: Preparing Students for
College and Careers, 2011, p. 21).
Other studies also emphasize the importance of creative problem-solving skills. “Employers know
that in business, the chessboard changes daily. As soon as we think all is fine, the economy changes
or the competition makes a surprise move and the company’s own strategy must change,” said Mark
Stevens, author of Your Marketing Sucks (Crown Business, 2003) and CEO of MSCO, a global
marketing firm. “A person who gets locked into a set way of doing things finds it difficult or
impossible to adjust. They are a drag on the business as opposed to an asset for it. [An employee
must know] how to tackle challenges and opportunities in a way no one will find in a textbook.”
(CNN: Top 10 Reasons Employers Want to Hire You, Rachel Zupek, www.CareerBuilder.com,
2011).
A multitude of sources including magazine articles, newspapers, on-line reports, scholarly papers,
and books all suggest that critical thinking and problem solving skills are essential for success in

today’s job market. There is also general agreement on what such skills entail. Two on-line sources
define the necessary skills as follows:
(1) Quintessential Careers: What Do Employers Really Want? Top Skills and Values Employers
Seek from Job-Seekers, Randall Hansen and Katharine Hansen (www.quintcareers.com, 2012).
a. Analytical/Research Skills. Deals with your ability to assess a situation, seek multiple
perspectives, gather more information if necessary, and identify key issues that need to
be addressed. Highly analytical thinking with demonstrated talent for identifying,
scrutinizing, improving, and streamlining complex work processes.
b. Problem-Solving/Reasoning/Creativity. Involves the ability to find solutions to
problems using your creativity, reasoning, and past experiences along with the available
information and resources. An innovative problem-solver can generate workable
solutions and resolve complaints.
(2) UC Davis, Human Resources: What do Employers Want from Employees? (www.hr.
ucdavis.edu/sdps, 2012).
a. Analytical Thinking – The ability to generate and evaluate a number of alternative
solutions and to make a sound decision regarding a plan of action.
b. Researching – The ability to search for needed data and to use references to obtain
appropriate information.
c. Organizing – The ability to arrange systems and routines to streamline work and
maintain order.
It is important to note that sources like the above do not stress the need for workers to have
extensive knowledge of particular facts in order to be employable. Actually, nowadays it is rather
easy to get information about almost anything very quicky using the internet. What workers do need
instead is the ability to organize, analyze and integrate the information they find so that they can
make viable proposals to solve problems.
The Information Explosion
In addition to finding a program of study that emphasizes critical thinking and problem solving
skills, students should also consider whether the program will prepare them to deal effectively with
the nature of the modern workplace and the specific problems they will be asked to solve on the job.
First, in today’s workplace, the amount of information that the average corporate worker must digest
is enormous. Consider, for example, just the issue of emails which must be looked at, evaluated,
and often answered before work on a problem can even begin. By 2014, it is projected that the
average number of legitimate corporate emails a worker can expect to receive per day will be 65;
3

4
the average number sent will be 39 (Email Statistics Report, 2010, Sara Radicati, The Radicati
Group, Inc., April, 2010). Adding instant messaging and social networking, it becomes clear that
workers must handle a massive amount of data.
Again, different sources provide different statistics; yet, an undeniable fact is that the internet has
fostered an explosion in the amount of information that is available virtually instantaneously. “Each
year the world produces 800 MB of data per person. It would take approximately 30 feet of shelf
space to hold that amount of information in books. The amount of data produced each year would
fill 37,000 libraries the size of the Library of Congress” (How Much Information Is There? Mark
Shead, 2007; www.productivity501.com). Recent lapses from agencies like Homeland Security, the
CIA, and the FBI have shown alarmingly how easy it is for important communiques to get lost in
the avalanche of data even though such agencies employ experts in surveillance. While national
security is rarely at stake in office emails, the point is clear: to be successful, modern workers need
to be able to digest, sort, organize, and prioritize massive amounts of information efficiently and
expertly.
The Problems Facing Today’s Workers
Another factor that students should consider in selecting a program of study involves the nature of
the problems that workers face on a routine basis in the 21st century. Today’s problems are
invariably complex, multi-faceted, and subtle. It used to be the case that one could formulate a
problem into a relatively well-defined and stable statement, that one would know when a satisfactory
solution was reached, and that one could clearly evaluate the solution objectively as being good or
bad, or right or wrong. That is no longer the case for many problems facing society, government,
and businesses such as the following:
(3) a. Should colleagues be disciplined if they are caught using drugs?
b. How should we deal with crime and violence in our schools?
c. How can we make air travel safe from terrorism?
d. Should people be fired if it is discovered that they are illegal aliens?
e. How can we improve the gas mileage of family cars?
f. Should same sex partners have medical benefits like married couples?
g. How should we improve the language programs in US schools?
As these examples indicate, today’s problems are frequently ill-defined and ambiguous. For
example, in (3a), the question raises other questions like the following: How much drugs, What
kinds of drugs, Are they prescription drugs, Is alcohol included, Is it a first offense, and so on.
Often, today’s problems are associated with strong moral, social, political and professional issues.
For example, it is not possible to talk about preventing violence in schools (3b) or about making air
travel safer (3c) without also considering whether increased surveillance will destroy individual
freedoms or whether certain policies will lead to stereotyping groups on the basis of color or
ethnicity. In some instances such as (3d), there is little consensus about what the problem is, let
alone how to resolve it. Overall, today’s problems are rarely fixed and stable; they involve sets of

complex, interacting issues developing in a dynamic social context. Often, new forms of problems
emerge as a result of trying to understand and solve the original issues (Wicked Problems - Social
Messes, Tom Ritchey, Springer, 2011). For example, proposals about gas mileage (3e) invariably
raise issues about pollution, energy policies, dependence on foreign oil, the dangers of fossil fuels,
the materials from which vehicles are made, and a host of other matters. As a result, in dealing with
today’s problems, there is generally no right answer, and proposed solutions are not good or bad,
but better or worse. To be successful, modern workers must be able to juggle competing proposals
to determine which alternative satisfies the most factors or constituents in the most effective way.
Abstract Thinking
Summarizing, today’s workers need to be good problem solvers and creative thinkers; they need to
be able to organize, analyze, and interpret massive amounts of information and to make proposals
that consider all the stakeholders involved and all the possible consequences that might arise. It is
with the those considerations in mind that students should approach the courses they take at
university and the exercises and problems they are asked to complete in those courses. Frequently,
students fail to see the connection between solving a specific problem in a particular course and the
problems they will face outside of the classroom. In fact, there are strategies and procedures that
are useful in attacking all problems, and teaching students how to make such connections is an
objective of education. Students often miss those connections when they focus instead on what
appear to them to be irrelevant facts about useless issues.
The ability to form such connections is vital to success both in college and in the workplace.
Consider the following scenario. A young boy about two years old is walking with his grandfather
in the country near a dirt road when the grandfather says, “Alex, don’t go in the road. You could
get hurt.” Later that day, the family visits some relatives who live in a suburban subdivision. Alex
runs into the paved street in front of the house to get a ball. The grandfather, very upset, says to
Alex, “I told you not to go into the street.” Alex looks at his grandfather as if to say, “No, you
didn’t. You told me not to go into the road.” At the age of two, children cannot see the connection
between a dirt road in the county and a paved street in a subdivision. Their interpretation of the
world is concrete and literal. It will take a few more years before they are capable of the abstract
thinking required to make such connections.
Abstract thinking is a level of thinking that is removed from concrete, literal, here-and-now
situations. Abstract thinking assesses a variety of specific examples of phenomena and unites them
into a generalized concept that embraces all of the individual instances. Abstract thinkers are able
to put together seemingly disconnected issues and see a larger picture. As an example, consider
another scenario. Two people are applying for a job with an automobile manufacturer. The
interviewer mentions that one of the industry’s most pressing problems is gas mileage; consumers
are concerned about the cost of fuel. The interviewer asks the applicants general questions about
customer satisfaction, options available on new models, electric cars, hybrids, changes needed in
the industry, and so on. During the discussion of options, one applicant volunteers that she is
unhappy with the little “donut” spare tires that come with cars because a person cannot drive very
5

6
long with them and must rush to have the flat tire fixed; she believes that cars should come with full
spare tires the way they used to. The other applicant responds by saying that a full spare tire weighs
more than a donut tire. The increased weight will make the car heavier and, therefore, reduce gas
mileage. Who gets the job?
Practice
Students don’t always see the connections between the various exercises that they are assigned to
complete in various courses and the “real world.” But, learning how to solve problems, think
abstractly, analyze data, propose hypotheses, and argue for a particular position – all of the skills
required for today’s workplace – require exactly the kind of practice one gets working on different
exercises in different courses. Practice is essential. Even experienced people practice and train on
a daily basis so they can maintain their expertise and develop professionally. Ballet dancers spend
long hours every day at the barre practicing plies and tendus so they can become proficient. Boxers
train every day; authors write every day; scientists experiment; philosophers ponder; golfers golf.
It’s often said that it takes 10,000 hours of practice to achieve mastery in any skill. Students cannot
expect to become successful thinkers if they don’t have practice solving a variety of different
problems as often as possible. Students generally think that they should be able to solve any
problem in twenty minutes, and, if they can’t, something must be wrong with them or their teachers.
That is incorrect as the above facts about today’s problems clearly indicate. Developing skill in
problem-solving like everything else takes practice. When students don’t see the purpose of taking
a particular course to satisfy a particular general education requirement, they need to try to see the
larger picture. It is exactly that larger picture – the relationship between roads and streets, the
relationship between gas mileage and tire size – that is crucial to adaptability, to flexibility, to
mobility, and to success.
Learning by Doing
Students in the United States generally believe that teachers should supply them with facts,
illustrations and the specific means by which they can successfully complete an assignment. There
is an excellent reason for that belief: teachers in the United States generally supply students with
facts, illustrations and the specific means by which they can successfully complete an assignment.
Unfortunately, the problems that one faces in life and at work do not come with instructions for their
solution. In short, what happens in American classrooms frequently has little bearing on what
happens outside them. When evaluation measures depend on memorization and replication, there
is little motivation for creativity, imagination, and objective thought. Consequently, high school and
college graduates often have great difficulty applying what they have learned in new situations. They
have difficulty with abstract thinking. Complexity often paralyzes them because they have few tools
to break down problems into manageable parts. They have not had enough practice doing so.
Students need to have practice and experience dealing with unknowns, managing problems that have
no clear answers, and evaluating competing approaches to find the best solution for the

circumstances when no single approach is completely satisfactory. As the above discussion on
today’s problems shows, those are the skills that are needed in today’s workplace. A teacher’s
function is not to supply students with the right answers or even with the formula to arrive at the
right answers. Rather, teachers need to act more as guides, helping students figure out how to solve
a problem and discover by themselves the principles that best describe the data under investigation.
In that way, students become better able to cope with new and different problems, which will always
occur. It is counterproductive for students to complain that teachers don’t give them the right
answers or the precise way to find the right answer. After all, teachers will not be following students
around for the rest of their lives pointing out the right answers to problems that arise.
Most students pursue a college degree knowing that it will help them in the job market. They know
that college graduates have more and better job opportunities than people without college degrees.
Furthermore, there is abundant research indicating that “college graduates are healthier, contribute
more to their communities, and raise kids who are better prepared academically” (How Much Is a
College Degree Really Worth, Kim Clark, US News, 2008, usnews.com/education). Having decided
to pursue a college education and chosen a major, an immediate concern for students is finding
courses and programs (concentrations, minors, etc.) which will make them most marketable, that is,
ones which will give them practice in the specific skills employers are looking for. As a result,
students should select classes which will help them become innovative and creative thinkers who
can make proposals and solve problems on their own. In that regard, courses in linguistics are
especially appropriate. All linguistics courses focus on gathering, organizing, and analyzing data,
setting up hypotheses to account for the data, and proposing and defending the most generalized and
effective solutions. By investigating problems in a variety of languages, students of linguistics get
practice developing just those skills that employers are looking for.
A Linguistic Example: English Syntax
As an illustration, consider two approaches that might be used in teaching some basic facts of
English syntax, which is the study of the way sentences are constructed. This illustration might
appear to be absolutely irrelevant to a student’s life or to the “real world” outside of the classroom,
but we will see that it isn’t. One approach to syntax begins with definitions like those in (4) which
teachers ask their students to memorize.
(4) a. The subject of a verb is the person or thing that performs the action in the verb.
b. The object of a verb is the person or thing that receives the action in the verb.
Given (4), students are then asked to find the subjects and objects in sentences like (5).
(5) The dogs frightened the little girl.
Applying (4) to (5), students identify the dogs as the subject phrase because they are causing the
fear, and the little girl as the object phrase because she is experiencing the fear.
7

8
The approach, therefore, seems successful. The students have memorized some formulas, the
definitions in (4), applied them in exercises, and learned a tool that can be reused elsewhere. The
teachers have done their job in supplying the formula for completing the assignment.
The difficulty is that English also has sentences like (6).
(6) The little girl feared the dogs.
In (6), the dogs are still causing the fear and the little girl is still experiencing the fear so it seems,
according to (4), that the dogs is still the subject phrase and the little girl is still the object phrase.
If that is correct, then we cannot say that subjects precede verbs in English and objects follow, which
seems to be the case in most sentences. So something is wrong. Since students do not know how
or why the definitions in (4) were proposed in the first place, they generally have no idea what to
do when the definitions seem to fail as they do in sentences like (6).
Now notice what is going on here at an abstract level. We have some data, which happen to be
sentences in English. We have a hypothesis which describes the data, namely, (4). We get some
further data which the hypothesis can’t handle. Thus, we must revise the hypothesis and look for
another proposal. This is exactly what happens frequently in the workplace. Consider, for example,
another scenario from the auto industry.
Several years ago, Ford Motor Company produced a van called the Mercury Villager. This van
originally only had a sliding door on the passenger side. Marketing analysts at Ford had determined
that parents would not want to have a sliding door on the driver’s side because their children might
get out of the van on the side of traffic. The other major auto manufacturers had vans with sliding
doors on both sides, because their marketing analysts had determined that people would find it too
inconvenient going to one side of the car to deposit items like groceries and then walking to the
other side of the car to get into the driver’s seat. Having sliding doors on both sides of the van was
considered a plus. Ultimately, Ford modified its design and offered a new Villager with sliding
doors on both sides. “A sliding door on the driver’s side of the vehicle [was] the subject of popular
demand from customers” (carpartswholesale.com/cpw/mercury~villager~parts.html). At an abstract
level, this problem has the same characteristics as our grammatical problem. Ford began with data,
namely, its marketing analysis of what it seemed parents wanted in a van. Ford then designed a car
accordingly. But further data indicated that the design did not match all the needs of its customers.
So, Ford began to produce a new van which did match the additional customer information.
Returning to our grammatical example, consider now an alternative approach to describing subjects
and objects which begins by looking at the data, that is, good sentences like those in (5), (6), and (7),
as well as bad sentences like those in (8), where the asterisk means that the sentence is
ungrammatical.
(7) a. They frightened her.
b. She feared them.

(8) a. *Them frightened she.
b. *Her feared they.
Given data like the above, it is clear that words like she and they must be distinguished from words
like her and them. All native speakers of English know this fact unconsciously whether or not they
have studied English grammar in school. They know that sentences like (7) are good and those like
(8) are not; so they say (7), not (8), even though they usually cannot explain why. As a result,
distinguishing the two groups of words is a necessity for a speaker of English, not a convention or
a convenience. The distinction is part of the English language; it is not something that teachers of
grammar made up.
Since the two groups of words exist, suppose we give them each a name. Notice that there is
nothing odd about this: we have names for all kinds of groups (cars, animals, food, laws, sports,
etc.). Suppose we call words like she and they “subject pronouns” and words like her and them
“object pronouns.” With these new names, we can now succinctly state the distinction between
subject phrases and object phrases, also a necessity if one wants to be a speaker of English.
Consider (9).
(9) a. Subject phrases are specified by subject pronouns (she, they, etc.)
b. Object phrases are specified by object pronouns (her, them, etc.)
Given (9), students attempt to replace phrases in sentences like (5) and (6) with pronouns. The
result is always sentences like (7), never (8). Therefore, students, like native speakers, immediately
know what the subjects and objects are: if a phrase can be specified by a subject pronoun, then it
is the subject; if it can be specified by an object pronoun, then it is an object. In addition, students
learn that the principles in (9) are motivated by facts about the English language, specifically the
distribution of the two groups of pronouns.
The approach just illustrated gets to the heart of the matter and the result is worth repeating.
Grammatical facts are not the result of convention, whim or convenience. They are a necessity;
indeed, they are a biological necessity. The sentences of every human language are broken up into
phrases like subject and object because the human brain cannot process unstructured material very
well. Try, for example, to recall the numbers in (10) after just one reading.
(10) 1 - 4 - 9 - 2 - 2 - 0 - 0 -1 - 1 - 8 - 1 - 2
Now try to recall the same numbers in (11).
(11) 1492 - 2001 - 1812
The numbers in (11) are much easier to recall and process because they have structure. In a similar
way, the sentences of human languages must be composed of structured units like subject and object.
Unstructured sentences without phrases are incomprehensible to human brains. Just try processing
the last sentence backwards. In short, languages have grammar, because human biology demands
9

10
it. Phrases which teachers of grammar arbitrarily call “subject” and “object” would exist even if
there were no teachers around to name them and describe them.
Again, we need to take stock at this point to see what we have done. We began with a relatively
specific matter, namely, how to identify subjects and objects in English sentences. We then
determined the following: first, that subjects and objects are phrases; second, that phrases are
structures; third, that the sentences of all languages have such structures; fourth, that sentence
structure is an essential property of language; and fifth, that the properties of language are
biologically determined. In short, the study of grammar is essentially a branch of biology. The
transition here has been a bit abrupt to be wholly convincing; in the discussion below, the argument
will be justified in further detail. At this point, however, it is possible to understand the
methodology and to see what the overall objective is. Specifically, we are attempting to find the
most generalized and comprehensive way to account for the data before us. At an abstract level, that
is exactly what today’s workers must do on the job.
The two approaches mentioned above have been called deductive and inductive. In the deductive
approach, one begins with the principle (rule, theory, definition, etc.) and tries to apply it to the data.
We began with (4) and applied it to (5). In the inductive approach, one begins with the data and tries
to discover what the principle (rule, theory, definition, etc.) is. We looked at good sentences like
(7) and bad sentences like (8) and then formulated (9).
Since the principles in (9) are objective and explicit, they are verifiable. Testing them with other
data reveals that they are, in fact, more robust than the definitions in (4), which fail in many cases:
(12) a. The stewardess is cooking the meals. She is cooking them.
b. The meals are cooking. They are cooking.
(13) a. The laundress is ironing the shirts. She is ironing them.
b. The shirts iron easily. They iron easily.
(14) a. The waitress tasted the potatoes. She tasted them.
b. The potatoes tasted fine. They tasted fine.
(15) a. The girl broke the windows. She broke them.
b. The windows broke. They broke.
The inductive approach supplies students with an exercise in problem solving, critical thinking, and
objective analysis. It has the potential of uncovering important generalizations like the principles
in (9) and of helping students understand that such principles are justifiable and, in fact, inevitable
when they are driven by an empirical investigation of the data. As a result, rather than learning
something by rote, students can develop skills for life-long learning, skills that can help even when
there are no teachers available.

In the above illustration, if students focus on the terms “subject” and “object,” they might conclude
that the exercise is irrelevant to their life in the real world, that knowledge of grammatical terms will
not help them on the job. Of course, that is probably correct: very few promotions depend on
knowledge of grammar. But the criticism misses the point. The above illustration is an exercise in
problem solving, in learning to think abstractly, and in formulating robust hypotheses. The solution
to any problem begins with collecting, sorting, and organizing the data, no matter what the data
involve. That step is followed by attempting to find generalizations and connections to account for
the data, principles that will explain the data. In turn, those principles need to be tested on new data
to see if they are robust. When they fail, alternative solutions need to be tried and retested to see if
they are successful. In the end, one produces the best possible proposal. The people who produce
the best proposals are the ones who advance in their careers.
There is a TV commercial which ask the following question: “How can rice production in India
affect wheat output in the U.S, the shipping industry in Norway, and the rubber industry in South
America?” (T. Rowe Price ad; www.youtube.com/watch?v=Ae5BH5KdcI0) The underlying
message of the commercial is that companies, workers, and people in general must be able to make
such connections, because understanding connections that are important and discovering
connections that no one has noticed before are crucial in finding solutions to problems. It is,
therefore, essential that students make every effort to try to understand how the exercises they are
asked to complete in their courses do, in fact, apply to the lives.
Initially, many students find an inductive approach frustrating; they would prefer that teachers tell
them what to memorize and then test them on their recall. Rather than trying to understand why one
proposal is better than another, they just want to know which one will be on the next test. But
memorization is not learning, and knowledge is not constant. Further, people do not succeed in a
career just by knowing a lot of existing facts. We discover and uncover new phenomena every day
which require changes in existing explanations, theories, practices, and methods. Education must
emphasize the fact that there often are no right answers, only what we know at any given moment.
The purpose of education is to bring students to the point where they can continue to learn on their
own without teachers to guide them.
In short, good teachers strive to make themselves unnecessary by making their students self-reliant.
If a few courses that students take during their education attain that outcome, then students have a
lasting model for life-long learning and for greater adaptability in the workplace. For that to happen,
students need to be given exercises and problems that are new and different from what they have
practiced before. It is useful to remember that the average workers will have ten different jobs
during their working careers. Memorization of a lot of facts is not going to help students cope with
a fluid and variable job market, one that is evolving so fast that it is difficult to keep up with the
changes. The more practice students have with the greater varieties of problems, the broader their
perspectives become. In turn, this broader experience will help them be more adaptable, more
flexible, and more mobile. Although a diploma by itself will give students an edge in securing
employment, they will fare much better over time if that diploma includes courses that give them
the skills and perspectives they need to be responsive to the dynamic nature of today’s job market.
11

12
Another Linguistic Example: English Phonology
Let us consider next another linguistic example, this time from phonology, the study of the sounds
of language. Verbs in English come in a variety of tenses such as present tense (walks), past tense
(walked), and future tense (will walk). When people are asked how to form the past tense of regular
English verbs, they generally respond with a reference to written English and say something like,
“Add an ed to the end.” Such a rule cannot be part of the natural grammar of any language.
Children know how to correctly produce the past tense of regular verbs long before they learn to
read and write, let alone spell correctly. Indeed, some adults are illiterate; still, they know how to
form the past tense of regular verbs. Writing is based on convention, and learning to write is
optional. There are hundreds of languages which have never been written down; conversely, there
is no natural human language that exists only in written form. These are facts; they form part of our
corpus of data that must be accounted for. They indicate that no language has any natural
grammatical rule based on writing. Our discussion of English syntax above lead to a hypothesis that
the rules of natural grammar are based on human biology. Language is a product of the human
language apparatus, which includes the brain and the organs of speech and hearing. It is important
to investigate whether this hypothesis can be sustained in other aspects of language like phonology.
Now notice that there are actually three ways to pronounce the past tense of regular English verbs.
First, the past tense is pronounced as the sound [t] in a verb like race (They raced from the house).
Second, it is pronounced as the sound [d] in a verb like raise (They raised their hands) or raze (They
razed the building meaning ‘They demolished the building’). And third, it is pronounced as a
separate syllable with the reduced vowel [c] before [d] in a verb like rate (They rated the movies)
or raid (They raided the refrigerator). So, we have the following three possibilities which clearly
show that spelling does not determine the correct form of the past tense:
(16) a. [t] raced and also coped, hiked, laughed, ached, etc.
b. [d] raised, razed and also combed, hugged, loved, aged, etc.
c. [cd] rated, raided and also coded, hunted, loaded, aided, etc.
It is also clear from the above examples that the correct past tense for any given regular verb ([t],
[d], or [cd]) cannot be random. Young children typically try to change irregular verbs into regular
verbs, saying things like I goed there. Importantly, children make up such forms without ever
having heard them. No matter how inelegant an adult’s speech is, no adult would ever say
something like I goed there yesterday. Further, if go were a regular verb, its past tense would have
to be pronounced goed, and children’s spontaneous production of forms like goed indicates that they
know that. All this indicates clearly that children do not acquire their native language by simply
imitating the people around them; there must be more to language acquisition that imitation.
Second, when speakers coin new verbs, they always make them regular and pick the appropriate past
tense marker from among the three possibilities given in (16). Consider, for example, the verb
material girl in a sentence like Madonna has material girled her way to superstardom. Notice that
the [d] variant in example (16b) is the only possible option. This means that when the verb material
girl was first used, its past tense had to be pronounced with a [d] and not either a [t] or an [cd]. It

is worth noting here that people, including children, invent new words all the time and, when they
do, they always make them regular with the expected variations. Consider the following examples:
(17) a. I ketchuped up my French Fries. (A five-year-old child)
b. She two-footed that landing. (A TV commentator on figure skating)
c. We clearanced those dresses. (A department store sales person)
Some new past tense verbs that have recently become common include transitioned, decontented,
googled, texted, friended, etc. The essential fact is that native speakers automatically understand
all these words the first time they hear them.
If the form of regular past tenses is not random and the determining factor is not spelling, then there
is an obvious question: What is it that governs the formation of the past tense of regular verbs? To
answer that question, one must first gather the necessary data to see exactly where each of the three
variants for the past tense occur. Note that this step is exactly the step one must take on the job
when trying to solve any problem, namely, one must gather and organize the data. Doing so in the
present case yields data like the following (ignore the spelling and listen to the sounds):
(18) a. [t] occurs in verbs ending in the sounds [p] (hyped), [k] (cracked), [f] (cuffed), [s]
(kissed), etc.
b. [d] occurs in verbs ending in the sounds [b] (robbed), [g] (shrugged), [v] (moved), [z]
(cruised), etc.
c. [cd] occurs in verbs ending in the sounds [t] (granted) and [d] (guarded)
Testing and verifying the data in (18) is the next step. One looks at as many different regular verbs
as possible to see if the statements are correct. Such investigation reveals that the statements, in fact,
are correct; however, the statements are no more than a list, and a list does not explain why the
sounds are distributed as they are. In other words, why is [p] in (18a) and not (18b) or (18c)? These
are very important steps in proposing solutions to any problem: one must not only gather, organize,
and verify the data; one must explain why the data fall out the way they do if one is going to put
forth the most robust proposal about the data. Although this is a theoretical example about English
phonology, one can still discuss its effectiveness as a matter of cost. The analysis in (18) is not very
cost-effective theoretically because it reveals no generalizations. It is not a very good way to
account for the data because it is merely a list. We need to look for common features among the
items in the list to see why they are in one list and not the other. We need to look for
generalizations. To see what this means in a situation one is more likely to encounter in business,
consider the following example.
In the business world, cost is often a primary consideration. The choice of one car design over
another will often depend on its cost-effectiveness. It is most cost-effective for a manufacturer to
vary only the exterior of various models and to use the same underlying mechanical and electrical
systems, the same chassis, the same engines, etc. If every vehicle in a manufacturer’s fleet is built
in a completely different way from every other vehicle and there are no common features, costs will
13

14
increase. Thus, manufacturers look for designs that can be generalized over many different specific
vehicles. It is not an accident that many vehicles look alike.
Similarly, in designing homes in a subdivision, builders will use the same underlying floor plan and
only vary the exterior elevations because that cuts down on costs. Architects will put bathrooms on
the second floor above bathrooms on the first floor rather than on opposite sides of the house
because that design is more cost effective. If you ask the architect why the upstairs bathroom is
where it is and not somewhere else, the architect can give you the reason and that reason probably
has to do with cost.
In designing vehicles and buildings, engineers and architects will look for ways in which they can
utilize as many of the same features as they can in all their products as a cost-saving measure. They
will try to generalize. Of course, the same is true not only in the automobile and construction
industries, but other businesses as well. At an abstract level, that is exactly what we need to do with
the data in (18). We need to look for generalizations. Specifically, we need to determine why the
data in (18) fall out the way they do. What do the members of each group have in common?
Since language is a product of human biology, we can expect that the formation of the past tense is
governed by a specific set of principles which are directly related to the nature of the human vocal
apparatus, the part of human anatomy concerned with producing sounds. That expectation turns out
to be correct. To see this, consider first, the verbs race and raise. Notice that the final sound of the
verb race is [s] and the final sound of the verb raise is [z]. The difference between [s] and [z] is
technical. When speakers make the sound [s], the vocal cords, which are located in the throat and
help to produce different sounds, do not vibrate, which means that they do not produce a buzzing
sound. There is no buzzing sound when one says [sssssss], for example. On the other hand, when
speakers make the sound [z], the vocal cords do vibrate and produce a buzzing sound, which can be
heard when one says [zzzzzzz]. Most native speakers do not consciously realize this difference, but
they instinctively know when they should vibrate their vocals cords and when they shouldn’t.
Actually, speakers can feel the difference between the sounds if they place a hand on their throat
when they say [sssssss] and [zzzzzzz].
Now notice that there is no vocal cord vibration in the sound [t], but that there is vocal cord vibration
in the sound [d]. Again, speakers can actually feel the tension in their throat when they make the
sound [d], and the tension is the same when they make the sound [z]. However, there is no similar
tension in saying either [s] or [t]. Linguists call sounds which involve vibration of the vocal cords
“voiced” sounds, and those which do not involve vibration of the vocal cords “voiceless” sounds.
During the production of words, it is natural to put sounds together that have the same features, that
is, put voiced sounds together and voiceless sounds together. There are lots of English words that
end in the sounds [st] (raced, missed, passed, etc.) and lots that end in [zd] (raised, dazed, posed,
etc.). However, there are no English words that end in either the sounds [sd] or the sounds [zt],
because such words would join voiced and voiceless sounds together.
Therefore, as a result of the nature of the vocal apparatus, a verb like race, which ends in [s]
(voiceless), should have a past tense that is pronounced with a [t] (voiceless), and it does. A verb

like raise, which ends in a [z] (voiced), should have a past tense that is pronounced with a [d]
(voiced), and it does. Using the technical terms, part of the rule for forming the past tense of regular
verbs is as follows:
(19) a. The past tense is pronounced [t], which is voiceless, in regular verbs that end in
voiceless sounds (coped [pt], hiked [kt], laughed [ft], raced [st], etc.).
b. The past tense is pronounced [d], which is voiced, in regular verbs that end in voiced
sounds (rubbed [bd], hugged [gd], loved [vd], raised [zd], razed [zd], etc.).
Consider now the third variant in (16), namely, the past tense [cd] in verbs like hunted, rated,
handed, raided, etc. Notice that there are no words in English that end in the sounds [tt] or the
sounds [dd]. Be careful not to think of spelling in such cases. A verb like putt in The golfer does
not putt well ends in the sound [t], not the sounds [tt]. There is a good reason for this: English does
not allow double consonants at the end of any word or syllable. So, the variants in (19) can’t apply
when a verb already ends in [t] or [d]. In such cases, English uses the third form [cd], making the
past tense a separate pronounceable syllable. The full set of rules is as follows:
(20) a. The past tense is pronounced [t], which is voiceless, in regular verbs that end in
voiceless sounds (coped [pt], hiked [kt], laughed [ft], raced [st], etc.).
b. The past tense is pronounced [d], which is voiced, in regular verbs that end in voiced
sounds (rubbed [bd], hugged [gd], loved [vd], raised [zd], razed [zd], etc.).
c. The past tense is pronounced [cd] in regular verbs that end in a [t] or a [d] (hunted [tcd],
rated [tcd], handed [dcd], raided [dcd], etc.)
Notice that (20) is much more cost-effective theoretically than (18). Whereas (18) is merely a list
of items without any principle, (20) explains clearly why each item is in the list that it is in. It is that
kind of generalized solution that linguists are looking for. It is also the kind of generalized solution
that employers are looking for from their employees.
With generalizations such as those in (20), we can now begin to answer some fundamental questions
regarding the acquisition of such principles by children, namely, how they mange to acquire them
so quickly and why they make the kinds of mistakes they do. It is clear that the rules for forming
the correct past tense of regular verbs are known unconsciously to all native speakers of English,
including toddlers. Because the principles are based on the nature of the human vocal apparatus,
they are natural and can be acquired relatively quickly. Furthermore, the past tense of irregular
verbs – forms like went for the past tense of go – are unpredictable and, in a sense, unnatural. It
takes children a long time to master such forms. That is why three-year-old children say things like
He hurted me, which they have never heard; they haven’t yet realized that hurt is an irregular verb.
Note that, if it were a regular verb, its past tense would have to be hurted, following (20c) and
parallel to other regular past tenses like hunted, hoisted, handed, hoarded, etc.
Given the above discussion, a child exposed to English must only figure out that the English past
tense is signaled by adding a [t] or [d] type sound to the end of the verb. The distribution of the
three variants ([t], [d], and [cd]) is not something that the child must worry about. That distribution
15

16
follows largely from the nature of the human vocal apparatus, which makes the distribution of the
variants a matter of necessity, rather than free choice. In short, there is no guesswork involved in
forming the past tense of regular verbs. As a result, we can account for the fact that children acquire
knowledge of the past tense rapidly and without instruction. We also can explain why children
invent forms such as hurted, which they have never heard. The principles prevent hurtt and hurtd.
Most important, children do not need to hear the past tense of every regular verb in order to know
which of the three possible endings is appropriate; from hearing a small number of forms, they can
derive all the rest. These are some of the reasons why children acquire their native language so
rapidly, and why language development can proceed normally even in cases of neglect, abuse,
disease, poverty and other unfavorable circumstances. What makes language acquisition possible
in children is this: the rules – the grammar – of all languages are not haphazard, unpredictable and
random. Rather, they are based on the principles, capacities, and limitations of the human language
apparatus, which all normal children share. As a result, in almost all cases, children can acquire any
language as a native language without even realizing that they are doing it.
Native speakers of English understand the grammar of English completely; yet, their knowledge of
the grammar is generally unconscious. For example, it is safe to say that no one reading this
discussion was consciously aware of the phonetic variations in the past tense ([t], [d], and [cd])
unless he or she happened to have taken a course in English grammar taught by someone with
expertise in linguistics. Since the knowledge is unconscious, it follows that parents do not teach
their children the rules of grammar. Quite clearly, parents cannot teach their children a set of rules
if they are not consciously aware of what the rules are. So then, how do children acquire their native
language if no one teaches it to them? The answer is quite simple. The rules of language are based
on principles that are directly related to the nature of the human language apparatus which matures
as children acquire their native language. Children acquire the rules of their native language as a
natural part of their development because the rules, like those in (20), have a natural basis.
Acquiring a native language is like acquiring the skills and coordination for walking. Even without
specific help or prodding from parents, all normal children will eventually walk if they get the
proper nourishment necessary for the physical requirements of the task such as strength and balance.
Similarly, all normal children will eventually acquire a native language if they get the proper
linguistic nourishment, which means only that they need to be exposed to some language or
languages. That is why absence of the ability to acquire a native language as a child is so rare.
Again, as before, notice what is going on here at an abstract level. We have some data, which
happen to be about past tenses in English. We have a hypothesis which describes the data, namely,
that past tenses are formed by adding ed to the end of the verb. We observe some further data which
the hypothesis can’t handle, namely, that children know how to form past tenses before they know
how to spell. Thus, we must revise the hypothesis and look for another proposal. The revision led
to the list of alternatives in (18). We determined that lists like (18) are unsatisfactory; they are not
cost-effective because they reveal no generalizations. What is needed is a set of principles that can
explain the behavior of native speakers, in particular, the fact that they can produce and understand
an unlimited number of new past tenses and can predict what the correct form will be. Accordingly,
we revised (18) to (20) relating the data to human biology and giving a real explanation for why the
examples fall out the way they do. In other words, what we have done with this problem in

phonology is exactly what we did with the problem in syntax, and, at an abstract level, exactly what
Ford engineers did when they modified the design of the Villager.
Linguistics
It is hard to imagine spending one waking moment without language. Whether we are alone or
among other people, whether we dream or daydream, whether we write poetry, follow a recipe,
cheer for the home team, speak or sing, language is involved. All normal children acquire a native
language, no matter where they are born, what the language is or what their home life is like.
Absence of the ability to acquire language is very rare. People who are deaf have language; so do
those who are blind, mute, completely paralyzed, intellectually disabled, or emotionally disturbed.
Language can be disrupted by injury or disease, processed by machines, altered for special
occasions, and exploited for ulterior motives. Despite this extraordinary presence, versatility and
variability, every human language, whether Old English or Modern Japanese, shares universal
features which are directly related to human biology.
The goal of courses in linguistics is to formulate a theory of language that explains what language
is and how it is acquired and used. In particular, a theory of language must explain how all normal
children can master any of the world’s languages as a native language and can do so without any
specific instruction from parents or care-givers, whereas learning a language as an adult requires
intense study and training. In fact, this problem is huge since it is ultimately concerned with the
biological characteristics of human beings, specifically, with the characteristics of the human
language apparatus which includes the brain and the organs of speech and hearing. Linguistics
courses break down that huge problem into manageable subparts to find satisfactory explanations.
Since many linguistic courses deal with English, which students already know, linguistics is an
excellent discipline for practicing the skills needed for success in the workplace. Students learn how
to collect, organize, and analyze data in English and other languages, make hypotheses to account
for the data, alter hypotheses to accommodate newly discovered data, and evaluate competing
proposals to choose the most viable and robust analysis.
Because language is so pervasive and so peculiarly human, students of linguistics find careers in
many different areas. Some, such as teachers, computer scientists, and speech therapists, use
linguistics directly; others such as market analysts, editors, and advertising executives, use it
indirectly. Still others use their undergraduate major in linguistics as a springboard to careers in law,
education, business, artificial intelligence, medicine, and international relations, as well as graduate
study in linguistics and other fields. A list of career opportunities posted by the Linguistic Society
of America is included as an Addendum to this Introduction. That list and other useful information
can be found at lsadc.org. More job opportunities are available at linguistlist.org/jobs.
Quite naturally, linguistics courses are of direct benefit to language teachers, especially teachers of
ESL (English as a Second Language). A typical ESL class has students who speak different native
languages with varying levels of proficiency in both English and their native language. It is
impossible for ESL teachers to know ahead of time what kinds of questions about English grammar
17

18
the students in such mixed classes will ask. The best preparation for ESL teachers is a solid
understanding of English linguistics, that is, a linguistic description of English grammar such as the
description given in this book. This does not mean that ESL teachers should memorize the various
analyses given here. Quite the opposite is the case: a formal linguistic description is not going to
be very helpful to students learning English. What is important is that ESL teachers develop skill
in linguistic analysis, that they learn how to look at languages properly, not as chaotic and whimsical
collections of rules, but as systems rooted in human biology. In this way, ESL teachers can
understand the facts and issues behind their students’ questions and know that there is a real
explanation for those facts. As a result, they will be able to teach English and to answer students’
questions in a way that is truly meaningful.
Many linguistics courses satisfy the general education requirement in formal reasoning because they
emphasize objective methods of analysis, synthesis and evaluation. They are specifically designed
to introduce students to systematic ways of organizing, processing, and analyzing data. The focus
on all LIN courses is on formulating hypotheses explicitly and testing them rigorously. In these
courses, students are asked to solve individual problems in English and a variety of languages as a
means to developing strategies and expertise for solving any kind of problem. Overall, students
learn how to develop explanatory models that can predict the phenomena they observe, that is,
students learn how to propose solutions to problems that are rigorous enough to incorporate the ever
expanding corpus of data. As this entire discussion has indicated, the skills that students acquire in
linguistics are exactly the skills needed in today’s workplace.
Since courses in linguistics typically involve investigation of data in a variety of different languages,
they also help to give students a global, multi-cultural perspective. In today’s world, information
can be almost instantly shared around the globe, as phenomena like the Arab spring testified so
stunningly. Companies and businesses have global interests and must adapt their products and
messages to widely divergent populations. They cannot afford to be egocentric or sociocentric.
Now, more than ever before, it is important for students to understand different cultures and different
ways of looking at the world. Linguistics by its very nature provides students with that global
perspective.
We began this discussion by noting that there are many myths and traditions surrounding options
like the choice of a major in college, for example, that students often think it will be impossible to
get a job in a specific discipline unless they study that discipline in depth in college. Another myth
is that a college education ought to be quantifiable, that students ought to be able to hold up a bunch
of books upon graduation and say, “I am educated because I know everything in these books.” But
expertise in any area is not tangible; it is not something a person can weigh. Becoming a good
golfer or a prima ballerina, for example, involves practice, false starts, and bad days. A good golfer
may be under par one day and over it the next. A ballerina may perform flawlessly on one evening,
and falter on another. Of course, what makes a golfer and a ballerina very good is that they are
generally just that: very good. Over time, we can measure the success of a golfer and a ballerina
with such quantifiable records as scores and box office receipts. But those measures are the effects
of the expertise, not the expertise itself. There are thousands of excellent golfers who have never
won a tournament and thousands of proficient ballerinas who have never made it to the big stage.

The ability to solve problems, to organize and analyze data, to come up with proposals that are
innovative, creative, and robust takes insight, and insight is intangible. It cannot be weighed. The
confidence and security that lead one to perform well on the job are a consequence of practice and
training. Students should not look at their education as something that will give them a store of
enumerable facts or an arsenal of memorized information. Education is about developing intuition,
sensitivity, vision, and perspective, and those attributes are qualities that cannot be directly
measured. Moreover, it is those kinds of attributes that will help students succeed in today’s
variable and unpredictable job market.
19

20
ADDENDUM: CAREER OPPORTUNITIES IN LINGUISTICS
Monica Macaulay and Kristen Syrett (Linguistic Society of America)
Work in the computer industry: Training in linguistics can equip you to work on speech
recognition, text-to-speech synthesis, artificial intelligence, natural language processing, and
computer-mediated language learning.
Work in education: People with a background in linguistics and education can develop materials
for different populations, train teachers, design assessments, find effective ways to teach languagerelated
topics in specific communities, or use the language of a community effectively in instruction.
Many applied linguists are involved in teacher education and educational research.
Teach English as a Second Language (ESL) in the United States or abroad: If you want to teach
ESL in the US, you will probably need additional training in language pedagogy, such as credentials
in Teaching English as a Second or Other Language (TESOL). Many teaching positions abroad
require only an undergraduate degree, but at least some specialized training in the subject will make
you a much more effective teacher. Linguistics can give you a valuable cross-language perspective.
Teach at the university level: If you go on to get a graduate degree in linguistics you might teach
in departments such as Linguistics, Philosophy, Psychology, Speech/Communication Sciences,
Anthropology, English, and departments focused on specific foreign languages.
Work as a translator or interpreter: Skilled translators and interpreters are needed everywhere,
from government to hospitals to courts of law. For this line of work, a high level of proficiency in
the relevant language(s) is necessary, and additional specialized training may be required.
Teach a foreign language: Your students will benefit from your knowledge of language structure
and your ability to make certain aspects of the language especially clear. You will need to be very
proficient in the relevant language, and you may need additional training in language pedagogy.
Work on language documentation or conduct fieldwork: Some agencies and institutes seek
linguists to work with language consultants in order to document, analyze, and preserve languages
(many of which are endangered). Some organizations engage in language-related fieldwork,
conducting language surveys, establishing literacy programs, and translating documents of cultural
heritage.
Work in the publishing industry, as a technical writer, or as a journalist: The verbal skills that
linguists develop are ideal for positions in editing, publishing, and writing.
Work for a testing agency: Linguists help prepare and evaluate standardized exams and conduct
research on assessment issues.

Work with dictionaries (lexicography): The development of good dictionaries requires the help
of qualified linguistic consultants. Knowledge of phonology, morphology, historical linguistics,
dialectology, and sociolinguistics is key to becoming a lexicographer.
Become a consultant on language in professions such as law or medicine: The subfield of
forensic linguistics involves studying the language of legal texts, linguistic aspects of evidence,
issues of voice identification, and so on. Law enforcement agencies such as the FBI and police
departments, law firms, and the courts hire linguists for these purposes.
Work for an advertising company: Companies that specialize in advertising often do extensive
linguistic research on the associations that people make with particular sounds and classes of sounds
and the kind of wording that would appeal to potential consumers.
Work for the government: The federal government hires linguists for the Foreign Service, the
Federal Bureau of Intelligence (FBI), the National Security Agency (NSA), the Central Intelligence
Agency (CIA), the Department of Defense, the Department of Education, and so on. Similar
opportunities may exist at the state level.
Become an actor or train actors: Actors need training in pronunciation, intonation, and different
elements of grammar in order to sound like real speakers of a language or dialect. They may even
need to know how to make mistakes to sound like an authentic non-native speaker.
http://www.lsadc.org/info/ling-faqs-whymajor.cfm
21

CHAPTER ONE: FUNDAMENTALS OF LINGUISTICS
1.1 GRAMMATICAL CHARACTERIZATION AND GRAMMATICAL REALIZATION
The first language that children acquire during childhood is called their NATIVE LANGUAGE.*
As they mature, children are considered NATIVE SPEAKERS of that language. Knowledge of
one’s native language is quite special. It differs in many ways from the knowledge one gains
studying a language in school, for example, in Spanish class. In particular, native speakers are not
generally aware of what they know about their native language. Also, they usually cannot describe
what they know or explain how they know it. Their knowledge is unconscious. To see this,
consider the following sentences, which differ only in that (1b) contains the word that:
(1) a. Do you think the judge will tell the truth?
b. Do you think that the judge will tell the truth?
Every native speaker of English knows that these two sentences are yes/no questions, that is,
questions that can get either a positive or negative answer. Now suppose that a person does not
know who will tell the truth and wants to find out. In such a case, the person could ask a
WH–question, which is a question that begins with a WH–word like who, what, where, etc. Given
sentences like those in (1), every native speaker unconsciously knows that the results of attempting
to ask such a WH–question will be different. Sentence (2a), a WH–question related to (1a), is
perfectly GRAMMATICAL, meaning it does not violate any rules. On the other hand, sentence
(2b), a WH–question related to (1b) and containing the word that, is UNGRAMMATICAL,
meaning it violates some rule (an asterisk at the beginning of a sentence indicates that it is
ungrammatical).
(2) a. Who do you think will tell the truth?
b. *Who do you think that will tell the truth?
There is a specific principle that rules out sentences like (2b), yet native speakers typically are not
consciously aware of what that principle is. Nor can they explain why the word that can be left out
of (3a) to produce the synonymous (3b), whereas a similar deletion in (4a) produces the
ungrammatical (4b).
(3) a. The lawyers that the judges admire will tell the truth.
b. The lawyers the judges admire will tell the truth.
(4) a. The lawyers that admire the judges will tell the truth.
b. *The lawyers admire the judges will tell the truth.
_______
* Words and phrases in boldface capitals are important concepts. They are found in the index,
which indicates the page(s) on which the concepts are introduced or reviewed. There is also a
glossary of most of these concepts on Moodle in the files section under the folder HANDOUTS.

24
Again, there is a specific principle that accounts for the ungrammaticality of (4b) which native
speakers must know, at least unconsciously, because they all recognize that (4b) is not a well-formed
sentence. About the best that the average native speaker can say about (4b) is that it doesn’t sound
right. The same is true of (2b) and, in fact, all ungrammatical sentences. In general, native speakers
cannot explain why grammatical sentences are grammatical and why ungrammatical sentences are
ungrammatical. Yet they do know the difference between the two.
The principle that predicts the ungrammaticality of (4b) is highly specific. Changing admire to
know in (3) produces the pair of sentences in (5) which mean the same thing.
(5) a. The lawyers that the judges know will tell the truth.
(= The ones who will tell the truth are the lawyers that the judges know.)
b. The lawyers the judges know will tell the truth. (=5a)
On the other hand, changing admire to know in (4) produces the pair of sentences in (6) which mean
very different things.
(6) a. The lawyers that know the judges will tell the truth.
(= The ones who will tell the truth are the lawyers that know the judges.)
b. The lawyers know the judges will tell the truth. (=/6a)
Notice also that (6b) looks a lot like (4b); however, (6b) is grammatical, but (4b) is not. Further,
(5a) and (6a) can be expanded as in (7), but a similar expansion of (4a) produces (8) which is not
grammatical.
(7) a. The lawyers that the judges know will tell the truth are from Michigan.
b. The lawyers that know the judges will tell the truth are from Michigan.
(8) *The lawyers that admire the judges will tell the truth are from Michigan.
These examples are not gimmicky or atypical; thousands of pairs of grammatical/ungrammatical
sentences like them can be found in the linguistics literature. Such abundant examples emphasize
that native speakers unconsciously know a vast and detailed array of facts about their language, and
that these facts are largely hidden from introspection and analysis. It is simply incorrect to say that
the facts of language – the grammar – are well known to (even educated) speakers. This means, of
course, that there is a difference between having knowledge of a language and being able to talk
about that knowledge. Linguists refer to the unconscious knowledge that native speakers have of
their native language as their LINGUISTIC COMPETENCE.
Since native speakers do not consciously know what makes a sentence grammatical or
ungrammatical in their native language, it follows that native speakers do not teach language to their
children. While parents do correct children on occasion, it is hardly likely that all native speakers
of English have been told by their parents that sentences like (2b), (4b), and (8) are ungrammatical
and should never be uttered. If speakers required specific instruction of this type, no one would

acquire a native language because there is an infinite number of grammatical and ungrammatical
sentences in every human language. In addition, even when native speakers have expert, fluent
knowledge of grammatical principles, they can’t share this knowledge with their children because
children are unable to comprehend even the most elementary statements about grammar. Most of
one’s native language is acquired before one possesses the cognitive skills to discuss the
grammatical principles which underlie the ability. For example, kindergartners clearly know the
difference between statements and questions before their first day of school, yet none can discuss
the difference in principled terms that reveal the facts behind the difference.
In view of the above, modern linguistics is principally concerned with two broad empirical problems
which have often been referred to as the GRAMMATICAL CHARACTERIZATION
PROBLEM and the GRAMMATICAL REALIZATION PROBLEM. Grammatical
characterization entails describing the linguistic competence of native speakers, that is, discovering
and generalizing the grammatical principles that constitute their unconscious knowledge of their
native language. In grammatical characterization, a linguist describes what the principles are which
determine the grammaticality of examples like (1) through (8).
Grammatical realization, on the other hand, entails accounting for native speakers’ LINGUISTIC
PERFORMANCE, that is, their acquisition and use of their unconscious knowledge. In
grammatical realization, a linguist describes how such principles become part of the linguistic
competence of native speakers, in short, how children achieve mastery of their native language.
Linguists must address both the grammatical characterization problem and the grammatical
realization problem in formulating a theory of human language. Clearly, there are huge obstacles
to doing so. Fundamentally, the linguistic competence of humans is a property of the human mind,
and studying the nature of the human mind is as complex a task as studying the nature of the
universe. The grammatical information that native speakers have unconsciously internalized about
their native language is vast, detailed, and often extraordinarily subtle.
One cannot emphasize enough the difference between the unconscious knowledge that native
speakers have of their native language and the knowledge they learn about their language in school.
For example, in school, children learn that English contains many words called homonyms which
are spelled differently but pronounced the same, such as two, to, and too, all of which are
pronounced [tu]. While this is something children consciously learn in school, it is clear that they
already know the difference between these words even when they don’t know how to spell them.
Consider the following sentences, remembering that children do not know how to read when they
first encounter such sentences in speech:
(9) a. I have to clean shirts.
b. I have two clean shirts.
When people read the sentences in (9), they can obviously see the difference between to and two.
But even without reading those sentences or having any knowledge of English spelling or
homonyms, it is clear that native speakers, including children, know the difference between to and
25

26
two. Thus, they know that they can say (10a) instead of (9a) contracting have and to into “hafta”,
but they cannot say (10b) instead of (9b).
(10) a. I hafta clean shirts. (where hafta = have to)
b. *I hafta clean shirts. (where hafta = have two)
Similarly, in elementary school, children learn to recognize and talk about the distinction between
nouns and verbs. They learn that school is a noun and eat is a verb in sentences like the following:
(11) a. I am going to school. (school is a noun)
b. I am going to eat. (eat is a verb)
Unconsciously, they already know that school is a noun and eat is a verb before the teacher ever
brings the matter up, because they will give an answer like (12b), but never one like (13b).
(12) a. Question: Where are you going to?
b. Answer: School.
(13) a. Question: Where are you going to?
b. Answer: *Eat.
Further, they know that the two uses of the word to in (11a) and (11b) are not the same. As a result,
they know that they cannot contract going to into “gonna” in (11a) but they can make the contraction
in (11b) as follows:
(14) a. *I’m gonna school.
b. I’m gonna eat.
We know children know the difference because they frequently utter sentences like (14b) but never
utter sentences link (14a). Lastly, children also know that (15a) is ambiguous, while (15b) is not.
(15) a. I am going to work. (ambiguous: work is either a noun or a verb)
b. I’m gonna work. (unambiguous: work is only a verb)
In addition to knowing about two and to, children also know about too. They know, for example,
that (16a) means ‘They came in order to clean the room,’ whereas (16b) means ‘They did so clean
the room.’ That is why the interchange in (17) is grammatical, whereas the one in (18) is not.
(16) a. They came to clean the room. (=They came in order to clean the room.)
b. They did too clean the room. (=They did so clean the room)
(17) a. Question: Why did they come?
b. Answer: To clean the room.

(18) a. Question: Why did they come?
b. Answer: *Too clean the room./*So clean the room.
Children also know that too has a variety of uses like to, so that (19) is ambiguous meaning either
‘They are overly clean’ or ‘They are definitely/emphatically clean.’
(19) They are too clean.
Lastly, children know that clean in (20a) is an adjective whereas clean in (20b) is a verb.
(20) a. They were clean.
b. They did clean.
We know they know this because they will utter only the grammatical sentences in examples like
the following:
(21) a. They were very clean.
b. *They did very clean.
c. They were cleaner.
d. *They did cleaner.
e. *They were clean it.
f. They did clean it.
g. *They were so clean better than us.
h. They did so clean better than us.
In short, when children acquire their native language, they acquire unconscious knowledge of the
parts of speech (nouns and verbs) and of word usage (two, to, and too) long before they go to school
and learn to talk consciously about what they already unconsciously know. They unconsciously
know what a noun is and what a verb is; they just don’t consciously know that a noun is called a
“noun” and a verb is called a “verb.” They will correctly use all three words that are pronounced
[tu] long before they learn to spell them. In fact, although they know the difference between the
three words (two, to, and too) and do not misuse them in speech, many children often misspell them
into adulthood writing sentences like He is to tall. We can tell what children unconsciously know
by listening to what they say and observing whether or not they understand what is being said to
them. It is the task of linguistics to discover the exact nature of that unconscious knowledge and to
formulate a theory that explains how children acquire that knowledge. As this discussion indicates,
both tasks are formidable.
27

28
1.2 THE STUDY OF GRAMMAR
In the preceding discussion, the word GRAMMAR was used in several distinct, yet interrelated
senses. This is typical because the term has a variety of applications. Fundamentally and most
commonly, it is used to mean a description of a language, that is, an explicit characterization of the
structure of the sentences in a particular language. A grammatical characterization of English, for
instance, isolates and categorizes the basic linguistic elements found in English sentences, indicates
how these elements are related to each other, and specifies the manner in which they are arranged
in larger units. Terms like noun and verb are examples of labels which grammarians have given to
some of these elements, and a statement like the subject of an English sentence precedes the
predicate is an example of the kind of rule usually found in a grammar of English.
In a second more technical sense, the word grammar is used to mean a theory of language, that is,
a system of hypotheses formulated to account for the various features of human language in general.
For example, one of the outstanding features of all human languages is the fact that each one
consists of a potentially infinite number of sentences. In effect, every human language allows for
CREATIVITY. Although there are severe constraints on individual structures when they are
combined to form complex sentences, every human language is RECURSIVE. This means that
there are structures that can repeat themselves indefinitely. Thus, it is impossible to make a list of
all of the sentences in the English language; additions to any such list can always be made by
combining the sentences one has already thought of with words like and, but, or, etc. There is no
longest English sentence. For example, (22) could, theoretically, be continued without end.
(22) This is the cat that caught the rat that ate the mouse that lived in the house that Jack was
in when he told Harry that Bill said that Mary wanted to know if Sue would be able to tell
her father that Jane hates fairy–tales because they make her think of an unfortunate
experience which occurred one day when she was visiting her aunt in Idaho and...
These two uses of the word grammar, i.e., grammar as characterization of a particular language
(e.g., English grammar) and grammar as a theory of human language generally (UNIVERSAL
GRAMMAR), are basic to all of the other more specialized or expanded uses of the word. In most
schools of modern linguistics, for example, grammar, as a description of a language, includes the
study of sounds and the way sounds are related to meaning. This usage is merely an expansion of
the more common one in which a grammar only describes the classification and arrangement of the
forms of a language, the syntax. Moreover, it is important to realize that, although there is a
distinction between grammar as description and grammar as theory, nevertheless, each of these basic
uses implies the other. On the one hand, a grammatical description is always written within the
framework of some grammatical theory, however loosely formulated, and, on the other hand,
grammatical theories are proposed so that correct grammatical descriptions can be written.
In addition to this, linguists make a distinction between DESCRIPTIVE GRAMMAR and
PRESCRIPTIVE GRAMMAR. In descriptive grammar, no value judgments are attached to the
characterization, that is, the grammar characterizes whatever speakers say. In prescriptive grammar,
on the other hand, value judgments are attached to the characterization, so that prescriptive

grammars prescribe what one should say and why, for example, that one should say It’s I and not
It’s me, if one wishes to sound educated. Throughout the following discussion, we will be
concerned with only descriptive grammar; therefore, our description must provide for utterances like
It’s me as well as utterances like It’s I.
1.3 SPEECH AND WRITING
One of the most common misconceptions about language is the belief that writing is somehow more
important than speech. It is not difficult to imagine how this misconception arose: writing has
always been considered a sign of intelligence and of civilization; it is generally more permanent and
durable than speech; and, through it, the artistic and scientific achievements of humankind have been
recorded and preserved. Although there may be some justification for this cultural superiority of
writing over speech, nevertheless, it is speech and not writing which is basic both to language itself,
as a means of communication, and to linguistics. There are several reasons for this.
First, writing is a relatively recent invention. Even the most ancient examples of writing that have
been preserved are only about seven thousand years old. On the other hand, it seems safe to say that
humans have been speaking since the evolution of human society many hundreds of thousands of
years ago.
Second, writing is much less widespread than speech. Many languages, even today, exist only in
spoken form. Furthermore, there are many extinct languages about which very little is known,
except that they were spoken in a particular area at a particular time, because these languages were
not written down.
Third, it is speech and not writing which human beings are predisposed to learn. The acquisition
of speech in children proceeds automatically and without conscious attention or specific instruction,
unlike the acquisition of writing skills. Acquiring writing ability, on the other hand, is a voluntary,
deliberate choice.
Fourth, and most important, writing is a reflection of speech. All alphabetic writing systems such
as English are based, in one way or another, on the sound systems they represent, that is, a particular
sign or letter always represents some unit of sound or sounds in the spoken language. It is never the
case that the sound system of a language is based on a particular system of writing.
For all of these reasons, writing has only a marginal importance in most language studies. Linguists
are primarily concerned with describing speech and ascertaining how it is acquired and used.
29

30
1.4 LINGUISTIC ARGUMENTATION
Linguistic competence is clearly related to linguistic performance. The facts of grammar are not
arbitrary but, in large part, are due to facts regarding the nature of human beings including their
cognitive abilities and cognitive limitations. Human beings are able to acquire human languages
because they are genetically constituted to do so. It is simply inconceivable that all native speakers
arrive at the same internal understanding of their language on the basis of trial and error, given that
the native acquisition of any language proceeds rapidly and uniformly and without instruction. In
very specific ways, human languages reflect human abilities. For example, the rules of every human
language are based on LINGUISTIC STRUCTURE, that is, on linguistic units. Consider the
following sentences.
(23) a. Eventually the president will resign.
b. The president eventually will resign.
c. The president will eventually resign.
d. The president will resign eventually.
e. *The eventually president will resign.
The possible positions that the sentence adverb eventually can occupy are not unpredictable. This
is because every English sentence consists of structural units. Using traditional terminology, these
units might be labeled and diagramed as follows:
(24)
One can see from this diagram, that the sentence adverb eventually can occur among any of the units
of the node SENTENCE. There are three such units attached directly to the node SENTENCE:
SUBJECT, AUXILIARY, and PREDICATE. Accordingly, there are four possible positions for
the sentence adverb in (23). Further, sentence (23e) is excluded from the infinite list of grammatical
sentences in English because it violates a structural principle of English, and not for some arbitrary
reason. In short, children do not have to be told that (23e) is ungrammatical: its status is predictable
on the basis of other structural facts about English that they unconsciously know.
Of course, the important question here is how to determine the structural units in the first place, that
is, how to justify a diagram like (24). Such a diagram characterizes English in two ways. First, it
states that English sentences consist of three structural units; second, it states that those units should
have three different labels. Each of these statements must be justified before they can be accepted
as an accurate characterization of English.

At this point, let us consider only the question of the number of units, putting aside the matter of
labels until the chapter on syntax. Linguists determine structural units such as the units of a sentence
by attempting to manipulate the words in that sentence in one of three ways: REFERENCE,
OMISSION, and PLACEMENT. Great caution is needed during such investigations because, as
we have seen, the facts of language are largely hidden. It is easy to be misled. Sometimes various
manipulations produce conflicting results. This merely means that the matter is more hidden than
first imagined and that other supporting data, perhaps even from other languages, must be looked
at. For this reason, the more arguments that can be marshaled to support a particular
characterization, the more justified it is.
Introductory linguistics students make a number of incorrect assumptions when they attempt to show
that a characterization is justified. First, they assume that the analysis should be obvious or, at least,
readily clear after a little thought. Second, they assume that there are unique, formulaic, infallible,
and unambiguous procedures available for them to apply to the problem. Third, they assume that
they do not know what the correct analysis is. All of these assumptions are wrong. The facts of
language are not at all clear, and there are no procedures available by which one can quickly
discover the best characterization of the data. If this were correct, the job of linguistics would have
been completed long ago. In truth, grammatical characterization is only in its infancy. Also, all
native speakers, because they are native speakers, know which sentences, out of an infinite list of
sentences, are grammatical and which are ungrammatical. Native speaker competence demands
such knowledge. Therefore, the problem is to learn how to access the unconscious knowledge one
has by asking the appropriate questions. In regard to (24), consider the following data:
(25) Reference: the elements of sentences can be substituted as units.
a. Who will resign? (Who replaces the president in a question.)
b. The president will announce that he will resign. (He replaces the president in
reference.)
(26) Omission: the elements of sentences are omitted as units.
a. The president will resign if the first lady asks him to _______.
b. The president will resign, and the first lady will _______ too.
(27) Placement: other elements cannot break up units.
a. Eventually the president will resign.
b. The president eventually will resign.
c. The president will eventually resign.
d. The president will resign eventually.
e. *The eventually president will resign.
31

32
Notice that the three manipulations in (25) through (27) have varying results: sometimes
grammatical sentences are produced; other times, ungrammatical ones. In each case, NATIVE
SPEAKER INTUITION can verify the decisions on grammaticality. This does not mean that
native speakers always agree on issues of grammaticality, nor is total agreement crucial. What is
crucial is that, for most of the manipulations, most of the results are agreed upon by most of the
speakers. This makes communication between people possible. Their internalized grammars of
English are substantially the same.
Now observe that the best characterization of the units in (25) through (27) is, in fact, (24); that is,
the three unit structure diagramed in (24) predicts (25) through (27). There is no manipulation
which, for example, demands that president will be analyzed as a unit. Therefore, most linguists
today argue that English sentences have three basic units whatever they choose to call them. As a
foundation for argumentation, linguists use the manipulations of reference, omission, and placement.
This helps to distinguish sentences which superficially seem identical, but which are, in fact, very
different, like the following (A=ambiguous; U=unambiguous):
(28) a. He looked up the street. (A) He looked the street up. (U)
b. He looked up the address. (U) He looked the address up. (U)
c. He looked up her dress. (U) *He looked her dress up.
d. He cleaned up her dress. (U) He cleaned her dress up. (U)
Notice that manipulation of these sentences produces different results, indicating that their structure
is not the same.
1.5 UNIVERSAL GRAMMAR (UG)
Turning to the issue of grammatical realization, an immediate question is, Why are the rules of all
human languages based on linguistic units as opposed to something else? For example, rather than
permuting a sentence adverb like eventually among structural units to produce paraphrases like (23a)
– (23d), why don’t human languages permute units on the basis of some other criterion like
counting? As it happens, there is no rule in any human language that is based on counting per se,
e.g., there is no rule like (30) based on principle (29).
(29) The rules of human language are based on counting.
(30) Place sentence adverbs optionally after every two words.
Rule (30) might predict that eventually could occur in every position marked with a # in (31).
(31) The president # said that # he would # resign as # soon as # he found # the stationery # to
write # with.

The obvious reason that human languages do not contain rules based on principle (29) is that human
beings cannot count and speak at the same time. Try, for example, to reread the present sentence
at normal speed and count the number of words as you reread it. What is the ninth word?
Sometimes languages have rules that position elements initially or finally in sentences. For
example, in a direct question in English, the auxiliary unit (helping verb) comes first as in Will he
do it? Languages even contain rules that position elements in a specific order after an introductory
structural unit. For example, German has a rule that says the verb must be the second structural
element when a sentence begins with a unit other than the subject. But such rules as these are highly
constrained. It is never the case that a language contains a rule for positioning a unit after, say, six
other units. The reason is rooted in human memory constraints. The rules of human language are
based on structural units because human short term memory (STM) is severely constrained.
It is a well known fact that humans can only remember about five to seven bits of information
immediately presented to them; for example, no more than the amount in a local telephone number.
In early childhood, the amount is much less than in adulthood. Gradually, during early development,
as children’s cognitive capacities mature, their ability to operate within STM constraints increases.
But there is an upper limit for all humans. If the information is organized into units or chunks, then
more can be remembered. Read each of the following strings of words at normal speed and then try
to recall them from memory.
(32) a. criteria performance high establish may institutions some
b. some institutions may establish high performance criteria
(32b) is much easier to remember than (32a), and the reason is clearly that (32a) has no structure;
it is simply a list of words. (32b), on the other hand, contains units which chunk or organize the
words. We can diagram (32b) as follows:
(33)
As a result of many considerations like these, linguists have concluded that the difference between
grammatical and ungrammatical sentences in human languages is not generally arbitrary. The
ungrammaticality of sentences like (23e) (*The eventually president will resign) is due to the fact
that they violate some principle such as (34).
(34) The rules of human language are structure dependent.
33

34
Principle (34) is directly related to the nature of human short term memory (STM): if languages did
not have structure dependent rules which organized words into units, sentence length would have
to be very short. Therefore, it follows that the nature of language is directly related to the nature
of human beings, in particular, to human mental abilities and limitations. Notice that there is no a
priori reason why humans should have evolved the capacity (34) rather than (29). Computers can,
in fact, deal with rules based on principles like (29) and routinely do so. If extra-terrestrial beings
were to land on earth speaking languages that contained principle (29), it is clear that humans would
not be able to understand them without paper and pencil analysis.
Modern linguistic theory has uncovered many principles like (34) which help to define the nature
of language and, therefore, the nature of human beings. In linguistics, a principle like (34) is called
a LINGUISTIC UNIVERSAL, and the entire set of such linguistic universals is called CORE
GRAMMAR or UNIVERSAL GRAMMAR (UG). UG defines what can and cannot form part
of a human language; therefore, it forms part of a human’s genetic endowment. That means that
children do not have to learn principles like (34); they are born with tacit knowledge of the fact that
the rules in the language to which they have been accidentally exposed are based on (34) and not
on (29). The reason is clear: human beings must operate within their capacities and limitations.
They cannot spend the afternoon underwater, for example, without some artificial device. They
cannot outrun cougars. They cannot understand languages that contain rules based on counting. It
does not take much experience, if any, for children to learn that they have these limitations. They
are part of the biological makeup of all humans.
UG contains two types of universals: FORMAL UNIVERSALS and SUBSTANTIVE
UNIVERSALS. A formal universal is a stipulation on the organization of grammar and on the form
and functioning of grammatical rules. (34) is an example of a formal linguistic universal. It
stipulates how the rules of natural language must work, in particular, they must be based on
structure.
The second type of linguistic universals, substantive universals, are stipulations on the elements or
features which can occur in natural language. For example, the sounds of language are distinguished
by a number of features like the following:
(35) a. [m] [+CONSONANT, +BILABIAL, +VOICED, +NASAL]
b. [b] [+CONSONANT, +BILABIAL, +VOICED, –NASAL]
c. [p] [+CONSONANT, +BILABIAL, –VOICED, –NASAL]
The letter m signifies the sound [m], which is really an abbreviation for a number of phonological
features. Specifically, [m] is a consonant which means that it is formed by temporarily obstructing
the air as it leaves the vocal apparatus. Further, the obstruction is made by the two lips
([+BILABIAL]) while the vocal cords are vibrating ([+VOICED]). Lastly, the air exits through the
nose in articulating [m], that is, it is a nasal sound ([+NASAL]). Changing any of these features
changes [m] to some other sound. For example, if the air exits through the mouth rather than the
nose, then [b] is produced. If the air exits orally without the vocal cords vibrating, then [p] is
produced. The features given in (35), therefore, define the phonological structure of some English

consonant sounds. The number and types of these features is not arbitrary. For example, all sounds
are either [+NASAL] or [–NASAL]; the air exits through the nose or the mouth.
Air could potentially escape the human head from the lungs via three routes: the mouth, the nose,
and the ears. In fact, when the tympanic membrane (eardrum) is not intact, people can blow smoke
through all these orifices. Despite this, no human language contains a phonological feature
opposition [+AURAL]/[–AURAL] meaning involving the ears/not involving the ears. The reason
for this is transparent: humans cannot use their ears for the production of sounds. There is no a
priori reason why the ear should have evolved in the way it did; conceivably, evolution could have
produced an aural canal with an unimpeded connection to the outside. This would have made it
possible to expel air through the ears normally; however, that did not occur. On the other hand, a
separation of the nasal and oral passages did occur in the evolution of mammals. From the point of
view of UG, human languages utilized this separation globally, so that most natural languages
contain a phonological feature opposition [+NASAL]/[–NASAL]. This feature opposition is
possible because humans have a velum which can be lowered to close off some of the oral passage
and allow nasal modification of the air stream. Thus, we have a clear example of how natural
languages are constrained by human anatomy.
Like formal universals, substantive universals are part of the genetic endowment. They form part
of the tacit linguistic competence that all humans innately possess, quite simply because they are the
direct reflection of man’s capacities and limitations. Children don’t have to learn that the sounds
of their native language won’t require modification through their ears, just as they don’t have to
learn that the rules of their native language are not based on counting. They are born with the
intuitive knowledge that [+AURAL] and [–AURAL] are not possible phonological features of
human languages, just as they are born with intuitive knowledge of principle (34). Further, since
all normal children can breathe either through their nose or their mouth, children intuitively know
that their native language might contain the features [+NASAL] and [–NASAL].
Notice however that every language does not have to contain the same set of universals. Not every
language has the same rules and the same features. For example, English does not use the feature
opposition [±NASAL] to distinguish its vowels, but French does. UG simply defines the limits of
what can be part of a human language because of the capacities and limitations of the human
language apparatus (the brain and the organs of speech and hearing). Therefore, a native speaker’s
tacit linguistic competence consists of two kinds of knowledge: (i) innate knowledge or UG, which
is part of a human being’s genetic endowment and consists of the kinds of rules and features which
any language might contain; and (ii) acquired knowledge, which consists of the particular rules and
features that occur in the native language(s) to which the speaker has been exposed as a child.
In the specification of UG, the close association between grammatical characterization and
grammatical realization becomes clear. Formal and substantive universals, which appear as the rules
and features in the grammars of human languages, are grounded in the linguistic capacities of
humans. Therefore, the discovery of a universal in the course of grammatical characterization can
often lead to a hypothesis about the way humans process information. An example of this is the
postulated formal universal (36a) which, it might be argued, leads to the perceptual strategy (36b).
35

36
(36) a. The clause is the major structural unit in human language; that is, the maximal
specification of the syntactic rules of language is the clause.
b. The clause is the major perceptual unit in human language; that is, if a string of
words can be interpreted as a clause, then it must be interpreted as one.
(36a) claims that the major syntactic unit found in the characterization of human languages is the
CLAUSE, which is a string of words that typically contains one subject and one tensed (present,
past, future) verb. (36b) claims that ordinary language use centers around clause structures, so that
a string of words that could possibly be a clause is probably a clause. Properly, (36a) is part of
grammatical characterization, the result of attempting to characterize the structure of English and
various other languages; (36b) is part of grammatical realization. We can see the operation of (36)
in the following data.
(37) a. The coach thought that John succeeded.
b. The coach thought John succeeded.
c. That John succeeded pleased the coach.
d. * John succeeded pleased the coach.
Notice that the word that can be left out of (37a) to produce (37b). We saw the same operation in
(3). However, if that is left out of (37c), the ungrammatical (37d) results (cf. (4)).
Faced with data like (37), a linguist can assume that the ungrammaticality of (37d) is either arbitrary
or principled. We have argued above that languages would be unlearnable under the conditions
which attend ordinary acquisition if the distinction between grammatical and ungrammatical
sentences were always arbitrary. Recall that linguistic knowledge is unconscious; accordingly,
children are not taught language by their parents. They must discover the grammar of their native
language without the benefit of instruction. Furthermore, since children can learn ANY human
language and do so effortlessly and rapidly, the facts of language acquisition demand that linguists
assume that some process other than trial and error guides children. Data like (37) lead linguists to
assume that (37d) is ruled out by some principle of UG, not by some arbitrary language convention.
Data like (37) from many different languages suggest that the principle entails something like (36).
Students new to linguistic analysis typically believe that a sentence like (37d) is ruled out because
it doesn’t make any sense, that is, it doesn’t mean anything. But this is incorrect as the
grammaticality of (37c) clearly shows (cf. John’s success pleased the coach). What is wrong with
(37d) is not that its meaning is somehow bizarre; rather, it is syntactically ill–formed.
Observe that, in (37d), John succeeded is a clause; it has both a subject (John) and a tensed verb
(succeeded). Principle (36b) demands that it be interpreted as a clause. Since the string of words
John succeeded already has a tensed verb, the clausal unit is fulfilled and pleased the coach, which
contains another tensed verb (pleased), cannot be part of the same clause. The function of that in
(37c) is, in fact, to separate the two clauses. Consider the following:

(38) a. [Something] pleased the coach.
b. [That [John] succeeded] pleased the coach.
c. [John] succeeded and pleased the coach.
The bracketed elements in (38) are subjects. In (38a), the sentence consists of one simple clause
with one subject (something) and one tensed verb (pleased). In (38b), on the other hand, the
sentence consists of two clauses, where the subject of the main clause is itself another clause, John
succeeded. The function of that in (38b) is to subordinate one clause to the other, to separate them.
In (38c), the word and performs a similar function of separating two clauses, only in this case it
coordinates the two tensed verbs (succeeded and pleased), rather than subordinating one to the other.
(37d) eliminates the that, so the two clauses are no longer separated; accordingly, the sentence is
ungrammatical. It contains two tensed verbs that are not separated by either a subordinator such as
that (cf. (38b)) or a coordinator like and (cf. (38c)). More importantly, its ungrammaticality is not
arbitrary: it is predictable from (36). Therefore, children do not have to learn not to say (37d), any
more than they have to learn not to try to look through solid objects. They are born with intuitive
knowledge of the principles in (36), which evidently derive from some as yet unknown fact(s) about
human brain function. These principles predict that all sentences like (37d) will be ungrammatical
in any human language. This includes (37d), as well as (39d) and any others which violate (36).
(39) a. The player who(m) the coach picked won.
b. The player the coach picked won.
c. The player who won did a backflip.
d. *The player won did a backflip.
e. The player won and did a backflip.
Despite these remarks, it is always possible that a language might allow some particular construction
under highly specified conditions even when there is a general constraint against it. After all, it is
a well–known fact that languages do have exceptions. In such cases, it is useful to distinguish what
is expected in language on the basis of universal principles from what is unexpected.
Linguists call rules and structures that are expected UNMARKED, and those that are unexpected,
because they violate general principles, MARKED. The more tightly constrained a principle is by
the nature of human beings, the less likely it is to have exceptions and the more highly marked such
exceptions become. For example, if the equivalent of (39d) were found to be grammatical in some
language, it would be highly marked. As one would expect, marked constructions take children
much longer to acquire than unmarked ones.
1.6 LEVELS OF ADEQUACY
There are several levels of adequacy that a linguist can attain in grammatical characterization. The
first level, called OBSERVATIONAL ADEQUACY, is attained when the facts are noted. The
second level, called DESCRIPTIVE ADEQUACY, is attained when the facts are described with
37

38
generalized principles or rules. The third, highest level, called EXPLANATORY ADEQUACY,
is attained when those principles are related to the nature of the language apparatus of humans. As
a linguist moves from observational to explanatory adequacy, an ever higher level of generality is
reached. At the highest level, linguists begin to relate individual languages like English to UG and,
therefore, to offer a real explanation for linguistic phenomena. In attempting to attain explanatory
adequacy, a linguist is also able to separate out what is arbitrary and language specific, from what
is predictable in terms of UG. As a result, attaining explanatory adequacy in grammatical
characterization helps to solve the grammatical realization problem. Our discussion of the data in
(37) in the previous section was an illustration of this. As a further example, this time from
phonology, consider the following data.
At the level of observational adequacy, we observe that, in many languages, there are four major
points of articulation for distinguishing consonant sounds. Looking again at nasal sounds, we find
the following:
(40) a. [m] articulated at the extreme front of the mouth
b. [n] articulated against the alveolar ridge behind the upper teeth
c. [ñ] articulated against the palate or roof of the mouth
d. [õ] articulated at the extreme back of the mouth
The four nasals occur in the following words:
(41) a. [m] whim, met, simmer
b. [n] win, net, sinner
c. [ñ] Spanish words like mañana (‘tomorrow’)
d. [õ] wing, singer
At the level of descriptive adequacy, many phonological descriptions of these sounds classify the
four nasals as follows:
(42) a. [m] is bilabial [+BILABIAL]; produced with the two lips
b. [n] is alveolar [+ALVEOLAR]; produced by raising the tip of the tongue to
the alveolar ridge, i.e., the ridge behind the teeth
c. [ñ] is palatal [+PALATAL]; produced by raising the front part of the tongue
to a point on the hard palate just behind the alveolar ridge
d. [õ] is velar [+VELAR]; produced by raising the back of the tongue toward
the soft palate or velum
Notice that this description requires four phonological features and a considerable amount of
unnecessary redundancy. For example, in addition to saying that [n] is [+ALVEOLAR], one can
say it is [–BILABIAL], [–PALATAL], and [–VELAR]. To correct this, and reach a higher level of
descriptive adequacy, we can reduce the number of oppositions from four to two by reducing the
number of features from four to two as follows:

(43) a. [+ANTERIOR] articulated in the front of the mouth (at or in front of the alveolar
ridge)
b. [–ANTERIOR] not articulated in the front of the mouth (behind the alveolar ridge)
c. [+CORONAL] articulated with the blade of the tongue (the portion immediately
behind the tip) raised from its neutral position
d. [–CORONAL] not articulated with the blade of the tongue (the portion
immediately behind the tip) raised from its neutral position
With these two features, we describe nasals much more efficiently as follows, reaching a higher
level of descriptive adequacy than (42), which requires four features:
(44) a. [m] [+ANTERIOR, –CORONAL]
b. [n] [+ANTERIOR, +CORONAL]
c. [ñ] [–ANTERIOR, +CORONAL]
d. [õ] [–ANTERIOR, –CORONAL]
These sounds do not have equal frequency or distribution in English. The closest approximation to
a palatal nasal in English occurs in a word like onion. The velar nasal, though it can occur medially
and finally, cannot occur initially. The three primary positions in (44), namely, (44a), (44b), and
(44d), are also the positions for the six stop consonants in English. Consider the following (recall
that [+VOICED] means that the vocal cords are vibrating and [–VOICED] means they are not):
(45) a. [p] [+ANTERIOR, –CORONAL, –VOICED]
[b] [+ANTERIOR, –CORONAL, +VOICED]
b. [t] [+ANTERIOR, +CORONAL, –VOICED]
[d] [+ANTERIOR, +CORONAL, +VOICED]
c. [k] [–ANTERIOR, –CORONAL, –VOICED]
[g] [–ANTERIOR, –CORONAL, +VOICED]
Notice the succinctness of this method of characterization. As we noted in previous sections, a letter
like p is a representation for the sound [p], which itself is an abbreviation for a constellation of
phonological features. Further, all those features relate to the structure of the human vocal apparatus
so that they comprise natural classes of sounds. As it turns out, it is precisely those natural classes
that are operative in phonological processes. For example, consider the following words:
(46) a. impossible, imbalance, immeasurable
b. intangible, indiscrete, innumerable
c. incongruent, ingratitude
Observe that the prefix meaning ‘not’ is pronounced in three ways:
39

40
(47) a. [im] before words beginning with sounds that are [+ANTERIOR, –CORONAL],
namely, [p, b, m]
b. [in] before words beginning with sounds that are [+ANTERIOR, +CORONAL],
namely, [t, d, n]
c. [iõ] before words beginning with sounds that are [–ANTERIOR, –CORONAL],
namely, [k, g]
This is, of course, not arbitrary. Sounds assimilate to each other, that is, they tend to become like
each other, when they are adjacent. We can formulate the following rule for English:
(48) If a prefix ends in a nasal, assimilate that nasal to the following consonant sound.
(48) accounts for the occurrence of the bilabial nasal [m] before the bilabial stops [p] and [b], of the
alveolar nasal [n] before the alveolar stops [t] and [d], and of the velar nasal [õ] before the velar
stops [k] and [g]. Notice that this distribution corresponds to the natural classes formed by
combinations of the features [±ANTERIOR] and [±CORONAL] in (44) and (45). We find
verification for this distribution in the occurrences of the Latin prefix con– and the Greek prefix syn–
both meaning ‘with’ or ‘together’:
(49) a. [m] compatible, combustible, community
b. [n] continue, condense, connect
c. [õ] conclude, congruent
(50) a. [m] sympathy, symbiosis, symmetry
b. [n] syntax, syndrome, synnema
c. [õ] synchrony, syngamy
We also find other cases of assimilation in the phonology of English. For example, note the
alternation of [p] and [b] in the root of inscription, scripture as opposed to scribe, scribble; the
alternation of [s] and [z] to indicate the plural of taps, bets, picks as opposed to tags, beds, pigs.
Similar examples are very common in the languages of the world. In view of this, we can formulate
the following principle for UG, thereby reaching the level of explanatory adequacy:

(51) a. Assimilate [+NASAL] consonants to the place of articulation of a following stop,
that is, to the features [ANTERIOR] and [CORONAL].
b. Assimilate adjacent [–NASAL] consonants in voicing.
Of course, (51) is directly related to the nature of the human vocal apparatus. It is more natural to
make adjacent features similar because it is easier to articulate them when they are similar. Children
engage in such simplification from the earliest stages of language acquisition when their vocal
apparatus is still not completely formed. For example, they say things such as gig kig for big pig
assimilating the stops to the same place of articulation because it is too difficult for them to
articulate a syllable like big which begins with an anterior stop and ends with a nonanterior one; it
is easier to make them both the same, so children say gig.
With regard to levels of adequacy, the above characterization proceeds as follows. If it stops at
simply noting the data in (46), then only observational adequacy would be attained. If it stops at the
rule (48), which describes (44) in generalized terms applicable to other data (cf. (49)), then
descriptive adequacy would be attained. If the characterization continues to principle (51) and
relates (51) to the structure of the human vocal apparatus, then explanatory adequacy has been
attained.
The significance of the highest level is that one can separate what is expected to occur in natural
language from what is not expected to occur. Also, at the level of explanatory adequacy, the real
nature of apparent exceptions is often clarified. To see this, notice that the rule (48) is not always
followed, even in English. Consider the following data involving the prefix non– meaning ‘not,’
which is also of Latin origin:
(52) a. [n] nonperson, nonbiodegradable, nonman
b. [n] nontaxable, nondelivery, nonnative
c. [n] noncombat, nongaseous
Notice that non is always pronounced with a final [n]. We must say, therefore, that, if non is a
prefix, it is an exception to (48). However, the matter does not stop there. The prefixes in– and
con– are of Latin origin, and, in Latin, they also function as prefixes, though they are related to
independent words (in and cum). Similarly, syn–, which is of Greek origin, functions as a prefix in
Greek, though it is related to an independent Greek word (syn). The element non, on the other hand,
never functions as a prefix in Latin; it occurs only as a separate word meaning ‘not’ and in a few
compound words. Even in English, the merger of non with the following word is not as complete
as the merger with in– or con– or syn–. Notice the slight pause and shift of articulatory posture after
saying non in the examples of (52). Such hesitation and adjustment does not occur in the examples
in (46), (49) and (50). This indicates that our characterization has missed something. Even in
English, it is probably wrong to consider non a prefix, though it looks like one. If we treat words
like nonman as compounds like snowman or manhole, then non is not a prefix and, accordingly, is
not subject to rule (48). Thus, an apparent exception, after closer examination, is seen not to be an
exception at all.
41

42
The significance of the above discussion is that it illuminates the nature of linguistic processes and
clearly indicates that languages are not governed solely by arbitrary conventions. As a further
example, consider the following.
The past tense in English verbs is variously pronounced [t] as in kicked, [d] as in hugged, and [cd]
as in hunted (the symbol [c] stands for the sound of “a” in sofa). The occurrence of each of these
variants is predictable in terms of the phonetic environment in which the past tense occurs. If it
occurs after a voiceless consonant, it is realized phonetically as [t], e.g., cooked, hoped, and laughed;
if it occurs after a voiced consonant or after a vowel, it is realized phonetically as [d], e.g., begged,
rubbed, and loved; and, if it occurs after [t] or [d] themselves, it is realized phonetically as [cd], e.g.,
prodded, pointed, and suggested. We can state the following generalizations ([æ] = “a”in bat).
(53) a. [t] after [–VOICED] (backed is pronounced [bækt])
b. [d] after [+VOICED] (bagged is pronounced [bægd])
c. [cd] after [+STOP, +ANTERIOR, +CORONAL], i.e, after [t, d] (batted is
pronounced [bætcd])
This distribution is, of course, largely an assimilatory process (and notice, incidentally, how
irrelevant and unrevealing the spelling of the past tense –ed is to the linguistic analysis). The
exception in distribution is apparently the last environment (53c) where [t, d] do not completely
assimilate in words endings in [t, d] (pointed and prodded). However, other facts of English indicate
that this distribution is, in fact, not exceptional. English has no final long sounds; there is no
difference between but and butt from a phonological point of view. Furthermore, final stops in
English are unreleased, that is, they are articulated without full explosion (compare the p in pit with
the p’s in tiptop). Therefore, in order to keep the past tense suffix as a recognizable and
pronounceable entity in a word, English must give it the status of a separate syllable in words like
pointed and prodded where it would otherwise not be possible (*pointt and *prodd).
It is clear from the above examples that the correct formation of the past tense in English is not
arbitrary for regular verbs (that the past tense of go is went is irregular and unpredictable and
therefore marked). Of course, it is completely accidental that English should choose the alveolar
stop consonant series [t]/[d]/[cd] to signal past tense. However, given the choice of alveolar stops,
the distribution of these three sounds is not arbitrary. As we have seen, the nature of the human
vocal apparatus makes certain combinations of sounds difficult or impossible to articulate. For
example, it is not possible to articulate a long unreleased final stop. If the past tense is signaled by
an alveolar stop, and that suffix is added to a verb whose root already ends in an alveolar stop, then
there is no choice but to affix the [cd] option. The past tense of hunt cannot be [huntt] or [huntd],
both of which end in consonant clusters that are impossible for English. It must be [huntcd]. For
this reason, children often say hurted and even goed or goded (rather than went), although they have
never heard such forms. If these verbs were regular, the past tense would be exactly what children
invent. In language acquisition children immediately overapply a rule. Later, they begin the process
of making finer and finer distinctions, including learning the marked exceptions. In short, for a good
many verb forms, the past tense is completely predictable given the choice of alveolar stop and the

constraints on human articulation. Children do not have to learn the past tense of every verb
separately, only of the irregular verbs.
This analysis offers a real explanation for the rapidity with which children acquire knowledge of the
past tense in English. The explanation is given in terms of UG, which is simply a reflection of the
human language apparatus. Not only does this description of past tense explain the rapidity of
acquisition, it also explains why children regularly utter forms they never could have heard like
hurted and goed and goded. Children do not know yet that these are irregular verbs. They have
realized that English arbitrarily signals past tense by the suffixation of an alveolar stop to the verb
root. Their intuitive linguistic competence dictates that the distribution of the various alveolar stops
must obey the constraints on articulation. Therefore, children say hurted and goed. In fact, irregular
forms take children a long time to master, precisely because they violate general principles.
1.7 LANGUAGE VARIATION
All humans languages are based on UG, a specification of the kinds of rules and features that can
exist in human languages. As we have seen, this specification directly reflects the nature of the
human language apparatus, which includes the organs of speech and hearing and the brain. The
existence of UG accounts for the fact that any human being can acquire any human language as a
native language with equal facility.
The principles in UG are available to virtually all humans; in fact, there are only three groups of
children who fail to acquire language normally. The first group includes those children who have
been kept severely isolated: one cannot acquire a language if one is not exposed to a language. The
second group includes the profoundly disabled intellectually, children who fail to attain a mental age
above two. Lastly, psychotic children show severe limitations in native language acquisition,
usually because they do not attend to language when they are exposed to it. Since all of these groups
are extreme cases, absence of the ability to acquire language is very rare in the overall population.
It is important to note also that language can develop quite normally even in cases of severe sensory
and motor deprivation. Children who are deaf, mute, and blind can acquire language, as the
accomplishments of Helen Keller attest.
Given the nature of children who fail to acquire a native language and the presence of language in
deaf, mute, and blind children, it is clear that the major component of UG is neurological. There is
something about the nature and functioning of the human brain that makes human language possible.
For this reason, general principles of grammar are very difficult to discover. It is well to keep this
in mind, as we explore the mysteries and intricacies of grammar in later chapters.
In the search for the grammatical principles which underlie all human languages, it is easy to be
mislead. There are many superficial differences among the world’s languages, and even among
dialects of the same language. Cultural and ethnic diversity among different groups of people add
even more superficial complexities. Students new to linguistics are quick to focus on these
43

44
superficial variations and tend to overestimate their importance. For example, because different
languages sound so different, students often think that some are easier to pronounce than others.
This may be true when one acquires a second language as an adult, but it is not true for native
language acquisition: all other things being equal, no language is harder for a child to pronounce
than another. Also, although different languages organize objects in the world differently, there is
no organization that cannot be acquired by every child. The fact that nomadic tribes living in deserts
have many words for different kinds of wind and sand simply tells us that wind and sand are an
important feature of desert life. That is hardly surprising news.
Despite the enormous variety among the world’s languages and the people who speak them, any
child could potentially acquire any language as a native language. This fact is of crucial significance
in modern linguistic theory. While there is currently much debate over what the principles of UG
are and what grammatical model best expresses them, virtually all linguists agree that the grammars
of human languages reflect innate, biologically determined characteristics of human beings.
1.8 LANGUAGE AND CULTURE
Culture and language are totally different objects of inquiry. Culture is conventional, subject to
whim and fancy, and alterable by consensus. For example, what is considered appropriate attire on
a airplane or in a concert hall today would not have been accepted twenty years ago. The number
of taboo words and taboo subjects has dropped precipitously in the same period. It is possible to
decide whether or not we can burn the flag, address elders by their first names, eat meat on Friday,
or publicly breast-feed an infant. These aspects of culture, like all others, endure or change at the
whim of the society that institutionalizes them.
The structure of language, on the other hand, is not conventional, not subject to whim and fancy, and
not alterable by decree. Consider, for example, the following differences between statements and
direct questions:
(54) Statements Direct Questions
John is going. Is John going?
John has gone. Has John gone?
John will go. Will John go?
Clearly, direct questions in English are formed by inverting the order of the subject and the auxiliary
or helping verb (verbs like is, has, and will). If a sentence lacks an auxiliary, then an appropriate
form of the verb do must be inserted and the tense must be taken off the main verb and put on the
form of do:

(55) Statements Direct Questions Ungrammatical Direct Questions
John went. Did John go? *Do John went? *Went John?
John snores. Does John snore? *Do John snores? *Snores John?
It takes children a relatively long time to work out the details of forming direct questions in English,
and the insertion of do causes people learning English as a second language considerable difficulty.
So, why not get rid of these processes? After all, in indirect questions, these processes don’t apply.
Consider the following indirect questions:
(56) a. Ask Mary if John is going.
b. Ask Mary if John has gone.
c. Ask Mary if John will go.
d. Ask Mary if John went.
e. Ask Mary if John snores.
Why not form direct questions in the same way, i.e., simply prefix statements with a word like if as
many other languages do? For example, let us decree that all the following will be grammatical
direct questions from now on:
(57) a. If John is going? (Replaces Is John going?)
b. If John has gone? (Replaces Has John gone?)
c. If John will go? (Replaces Will John go?)
d. If John went? (Replaces Did John go?)
e. If John snores? (Replaces Does John snore?)
Think of the aggravation that people would be spared. Of course, the idea is absurd. We cannot
change how we form direct questions in English by legislation, mutual consent, or even conspiracy.
Language reflects culture in only the most superficial ways. For example, specific kinship
terminology is influenced by culture. In ancient Rome, there were two different words for uncle,
avunculus (mother’s brother) and patruus (father’s brother) because the distinction was important
in Roman society. In the evolution of Latin to the modern Romance Languages, e.g., French, the
distinction was lost because it was no longer relevant. Such subtleties and changes in usage are
common in languages, as are the societal conventions that dictate what is proper (whether to say I
need to use the bathroom or I gotta take a leak, for example). Thus, the fact that a language has two
words for uncle is entirely conventional and based on societal concerns, whereas the structure of
languages conforms to the capacities and limitations of human cognitive ability, which is entirely
independent of social conventions. It is not an accident, for example, that the sentences of every
human language can be broken up into phrases. The fact is that the human brain cannot process
unstructured material very well so sentences must be organized according to structural principles
that all languages share. The manipulation of phrases is also dictated by such structural principles.
Notice that it is phrases that invert to form direct questions in English, not words:
45

46
(58) Statements Direct Questions
John is going. Is John going?
That man is going. Is that man going?
The man laughing is going. Is the man laughing going?
The man who is laughing is going? Is the man who is laughing going?
The man who is laughing is going? *Is the man who laughing is going?
Every native speaker of English knows that the italicized sequences in (58) are phrases. Of course,
their knowledge is generally unconscious as we have seen. If asked to identify the subjects in those
sentences they might even get the answer wrong. But, at an unconscious level, they know exactly
what all the subjects are because they know it is possible to replace all of them with the pronoun he.
Further, they know they can replace the man laughing with he, but not just the man. The
ungrammaticality of *He laughing is going has nothing to do with English per se; it is
ungrammatical because it violates the structural principles on which all human languages are based.
The structural principles that apply when the phrases in (58) are replaced with pronouns operate in
other uses of pronouns. Consider the following sentences and notice that the first them in (59b)
refers to the children while the second them refers to the matches; further, the interpretation of the
pronouns parallels (59c), which is grammatical, not (59d), which is ungrammatical.
(59) a. The children should not play with the matches.
b. Why did you give them them?
c. Why did you give the children the matches?
d. *Why did you give the matches the children?
Perhaps more interesting, is the fact that the reduction of the word them to [cm] in rapid speech also
follows general principles. Note that (60a) and (60b) are grammatical, but (60c) is not.
(60) a. Why did you give cm them?
b. Why did you give cm cm?
c. *Why did you give them cm?
Grammatical principles dictate how pronouns are interpreted even when the interpretation varies in
sentences that look almost identical like the following:
(61) a. John appeared to Bill to behave himself. (himself = John)
b. John appealed to Bill to behave himself. (himself = Bill)
The referent for himself in these sentences is corroborated in the following:
(62) a. *She appeared to him to behave himself. (himself = she)
b. She appealed to him to behave himself. (himself = him)

Given data like the above, it is clear that cultural matters do not underlie the principles that apply
in the formation of questions, in the replacement of phrases with pronouns, in the reduction of words
and phrases during rapid ordinary speech, and in grammatical processes generally. Rather, such
principles are based on grammatical structure, which, in turn, is based on the nature of the human
language apparatus.
1.9 SUMMARY
As we have seen, there is a high level of abstractness associated with the characterization of
linguistic competence; principle (34) – that the rules of human language are structure dependent –
is hardly a subject of discussion among five–year–olds. Trying to explain it even to an educated
adult is not easy. Also, there is a large measure of creativity associated with linguistic performance;
most of the sentences children experience are novel. Despite this, many psychologists and linguists
believed, until quite recently, that the linguistic abilities of human beings could be attributed to a
complex interaction of elementary learning mechanisms, such as imitation, conditioning,
association, and the like. Biologists have also shared this belief and attempted to document the
separate skills needed to explain human language by citing antecedents found elsewhere in the
animal kingdom. However, it has now become quite clear that the acquisition and use of language
by ordinary human beings cannot be explained as the product of simple learning mechanisms, each
with a precedent elsewhere in the animal kingdom. This is substantiated by a number of
observations which we summarize here.
First, human beings are capable of learning any of the world’s languages. Obviously, a child is not
predestined to acquire a specific language but learns the language of the community into which he
has been placed by chance.
Second, from the inception of speech, a child does not merely imitate his parents or respond to
stimuli but produces entirely novel sound sequences, most of which he has not specifically heard.
One of the outstanding characteristics of both language acquisition and use is their productivity.
Children say things like I goed there, which they clearly have never heard.
Third, all normal children learn their language with remarkable rapidity and uniformity; and, even
non–normal children, such as those who suffer from severe sensory deprivation, acquire linguistic
proficiency. A human can be deaf, mute, and blind and still acquire a human language.
Fourth, the data on the basis of which children learn their language is of a highly fragmentary sort;
many of the sentences they hear deviate considerably from perfect well–formedness. Despite this,
all children learn to distinguish those utterances which are grammatical sentences from those which
are not. And all children, despite their particular home environment, arrive at essentially the same
internalized grammar.
Fifth, while the members of the linguistic community do correct children on occasion, the level of
instruction never approaches the order of complexity and abstractness which must be assumed to
account for the normal use of language.
47

48
Sixth, there is no one–to–one correspondence between sound and meaning in any human language,
whereas all known animal communication systems have very restricted one–to–one correspondences
between signals and meaning. Human languages all have sound sequences that mean more than one
thing, i.e., ambiguities like (63); they also have multiple sound sequences that can mean the same
thing, i.e., paraphrases like (64).
(63) a. It’s too hot to eat.
b. Each of the sisters said that they turned on their brothers yesterday.
(64) a. The boy called up the girl.
b. The boy called the girl up.
On the basis of these observations and many others, it appears that language acquisition is possible
in children because all human languages, despite their superficial diversity, are of a universally
well–defined type and because all children have an innate mental capacity to learn this particular
type of communication system. Consequently, the essential problem for linguistics is to construct
a theory of human language which is sufficiently rich to account for the diversity among the world’s
languages, but which, at the same time, is constrained enough to account for the facts of ordinary
language acquisition. The formal representation of such a theory entails a grammar of linguistic
universals or UG, the investigation of which necessarily links the study of language structure to the
study of human psychology and biology, and ultimately, to the study of the nature of man. The
innateness hypothesis of language acquisition claims only that children are able to learn their native
language because human languages all reflect innate human abilities.
A beginning student of linguistics must realize that it is of crucial importance to study the structure
of sentences with a view toward describing what can and can not be said in a language, rather than
prescribing what should and should not be said, and to approach the problem of grammatical
characterization with a view toward discovering the underlying principles upon which languages are
based. The approach in theoretical linguistics is consequently very different from the approach
taken in most language classes in the schools, including language arts classes and English classes,
even though some of the same terminology is used.
Theoretical linguists do not generally do grammatical analysis to help improve a teacher’s ability
to teach languages or a student’s ability to learn languages. Theoretical linguists deal with native
speakers who already completely know their native language. Theoretical linguists do grammatical
analysis in an effort to characterize native linguistic competence which they believe is directly
related to the nature of human beings. Thus, it is crucially important to try to find out what exactly
causes some sentences to be grammatical and others, ungrammatical.
This approach to the study of language seeks to explain linguistic phenomena, not merely observe
and describe their existence. As we saw above in the characterization of the past tense in English,
it is possible to attain the level of explanatory adequacy by relating the distribution of [t]/[d]/[cd]
in regular English verbs to principles of UG which are based on the structure and function of the
human vocal apparatus. As a result, this characterization of the past tense attains the level of

explanatory adequacy because it relates linguistic phenomena to the nature of man. Further, it offers
an explanation for why children learn the past tense so rapidly despite their lack of instruction and
why they regularly produce forms they never could have heard. In actuality, the characterization
predicts that children will say hurted and goed. Of course, explanation leading to prediction is
precisely what all science is about.
49

EXERCISES FOR CHAPTER ONE
1. What must a descriptive grammar of English say about the word up in the following
sentences? Can you justify your opinions?
a. The price has been upped.
b. They can’t handle the ups and downs of life.
c. He stood up his date.
d. Bob has to be the upest guy on the team.
e. He lives up the street.
f. Let’s up the ante.
g. Sue gets up her children at six.
h. It’s up for grabs.
I. The surf is up.
j. (1) He put up the money.
(2) He put the money up.
(3) He put the money up there.
(4) He put the money up in that case.
2. All of the following sentences are ambiguous. How can you disambiguate them?
a. The thief hit the lady with the hat.
b. The repair man will look up the street.
c. John left the house messy.
d. The prisoner of war spoke foolishly.
e. John slipped on the boots in the kitchen.
f. They hired Spanish teachers.

g. They are all finished.
h. They do so love a good meal.
I. We made them idols.
j. They made idols.
3. The following are grammatical English sentences. Why are they so difficult to understand?
a. Spain was where district and provincial administrators were concerned a country of
corrupt officials.
b. The horse raced past the barn fell.
c. Where Al had had had Sue had had had had had had had had the teacher’s approval.
d. Americans Italians and Germans really like.
e. That that is is that that is not is not that it it is.
4. The following excerpts are from songs, games, magazines, and various other sources. What
makes them special? Turn them into sentences that are unremarkable.
a. They need to adore me so Christian Dior me; it’s vital you sell me, so Machiavell’ me.
b. Freedom from incrustations of grime is contiguous to rectitude.
c. Like my mother is a total space cadet, she like makes me do the dishes and clean the cat
box, I am sure, that’s like gross, barf out!
d. I have hear of yo and your classes on languages for foreigner people. I am very
interesting about take some courses on your school. I have the necessity to learn
English for development very well in my profession in Mexico City. Gratefully
beforehand any information.
e. Nowadays, the only sexuality about which journalese is coy tends to be homosexuality,
and that is adequately covered by he has no close female friends or he is not about to
settle down.
51

52
5. Find the words which express the main idea in each of the following sentences. If you can,
do this by identifying the subject and the main verb.
a. When the dancer faces the audience and with one foot describes a rond de jambe à terre
en dedans so that the foot describes on the ground a semi–circle from back to front, the
head should be inclined to the same side as the foot that makes the movement.
b. To be born, or at any rate bred, in a hand–bag, whether it had handles or not, seems to
me to display a contempt for the ordinary decencies of family life that reminds one of
the worst excesses of the French Revolution.
c. It was from these missions the Jesuit fathers carried the word of God to the high and
undiscovered plateau to those indians still existing in their natural state and received in
return martyrdom.
d. When in the Course of human Events, it becomes necessary for one People to dissolve
the Political Bands which have connected them with another, and to assume among the
Powers of the Earth, the separate and equal Station to which the Laws of Nature and of
Nature’s God entitle them, a decent Respect to the Opinions of Mankind requires that
they should declare the causes which impel them to the Separation.
e. In vertebrates, mechanisms for maintaining body temperature by adjusting heat
production and heat loss, a feat lower animals, who are at the mercy of their
environment, cannot achieve, have evolved.
f. In looking forward to the moment which is intended to terminate the career of my public
life, my feelings do not permit me to suspend the deep acknowledgment of that debt of
gratitude which I owe to my beloved country for the many honors it has conferred upon
me.
g. As we are now entering upon a book in which the course of our history will oblige us
to relate some matters of a more strange and surprising kind than any which have
hitherto occurred, it may not be amiss in the prologomenous or introductory chapter to
say something of that species of writing which is called the marvelous.
6. Why do linguists make a distinction between linguistic competence and linguistic
performance? Why is competence the focus of linguistic description? Should it be?
7. Itemize some of the differences between native language acquisition and second (foreign)
language acquisition. Consider characteristics of the learner and the circumstances of
learning.

8. Here are two typical definitions from traditional grammar:
The SUBJECT of a sentence is the phrase which specifies the topic of the sentence.
The DIRECT OBJECT of a verb specifies the person or thing directly affected by the action
described in the verb.
What problems arise when these definitions are applied to the subjects and direct objects in
the following sentences?
a. John burned the money.
b. John earned the money.
c. Sue suffered a blow to the head.
d. This book reads well.
e. Bill received a promotion.
9. Here are some further definitions from traditional grammar of three parts of speech:
A VERB is a word that expresses an action, occurrence, or state of being.
A NOUN is the name of a person, place, or thing.
An ADJECTIVE is a word that describes a noun.
Using these definitions, find the verbs, nouns and adjectives in the following sentences:
a. They won on their final at–bat.
b. Do–it–yourself–ers get things done.
c. He Fred Astaired her around the ballroom.
d. The River is not your average save–the–farm film.
e. All carry–on luggage must be stowed.
f. These French fries are so ketchuped–up, they’re soggy.
53

54
Can you think of other ways of defining the parts of speech? Try to find criteria like the
following:
A VERB is a word that can carry time distinctions (play, played).
A NOUN is a word that can carry number distinctions (book, books).
An ADJECTIVE is a word that can be compared (tough, tougher, toughest).
10. Each of the following pairs of unpunctuated sentences consists of a statement and a yes/no
question, which is a question to which one can get a yes or no answer. As a native speaker of
English you know which member of each pair is the yes/no question. How do you recognize
a yes/no question? Formulate a rule for forming a yes/no question.
a. (1) will that man go
(2) that man will go
b. (1) the little girl has gone
(2) has the little girl gone
c. (1) the boy with the red hair isn’t going
(2) isn’t the boy with the red hair going
d. (1) all of them went
(2) did all of them go
e. (1) could there be more than one right answer
(2) there could be more than one right answer
Having formulated a rule for forming yes/no questions in English, consider how you might
teach a four–year–old child that rule.

CHAPTER TWO: PHONETICS AND PHONOLOGY
PHONETICS and PHONOLOGY are the two branches of grammar which are concerned with the
study of the sounds of human language. The distinction between the two fields is as follows:
phonetics deals primarily with the speech sound itself, including the way in which it is produced,
transmitted, and perceived; phonology deals more with the organization of speech sounds into sound
systems. The difference is similar to the difference between studying the materials out of which
buildings are constructed (e.g., bricks, concrete, steel, etc.) and studying the way in which a building
is constructed out of these materials (e.g., by alternating a layer of brick with a layer of concrete, by
encasing the steel in concrete, etc.). PHONETICIANS study the raw materials, the sounds;
PHONOLOGISTS study the systems formed from these sounds.
2.1 PHONETICS
PHONETICIANS describe and classify the sounds of human language in the following three ways:
(i) in terms of the way in which they are produced by the vocal apparatus; (ii) in terms of the
physical properties of the sound wave emanating from the speaker; and (iii) in terms of the effect
the sound wave has on the various parts of the ear. These three methods of description and
classification are known respectively as ARTICULATORY phonetics, ACOUSTIC phonetics, and
AUDITORY phonetics. By far, the most common of the three is articulatory phonetics; hence, in
the discussions which follow, the terminology of articulatory phonetics will be employed.
The diagram in Figure One represents a cross–section of the human head, showing the parts of the
vocal apparatus. In the production of speech sounds, the most important of these parts are the vocal
cords. These are two elastic membranes located in the larynx or Adam’s Apple.
In normal breathing, air is forced from the lungs, up the trachea, through the vocal cords, and out
of the mouth or the nose or both. If the air is not obstructed in any way by the parts of the vocal
apparatus, then no sound will be produced. Conversely, when one or more of the parts of the vocal
apparatus form an obstacle in the path of the air, then a sound results.
2.1.1 VOWELS
There are two basic types of sound segments in human speech: VOWELS and CONSONANTS.
Vowels are typically produced when the air is modified by the vibration of the vocal cords and when
the tongue is held in specific positions in the mouth. The vibration of the vocal cords is known as
VOICING. Sounds articulated with the vocal cords vibrating are [+VOICED]; those without, are
[–VOICED]. Voicing is a feature of all English vowels, whereas the position of the tongue is what
distinguishes one vowel from another. For example, the sound of a in father is produced when the
tongue is in a low position in the back of the mouth, and the sound of i in machine is produced when
the tongue is in high position in the front of the mouth. In the production of both vowels, the vocal
cords are vibrating.

56
FIGURE ONE: THE VOCAL APPARATUS
A traditional chart of some vowel sounds specified by the relative positions of the tongue in the
mouth is given in Figure Two, which is the kind of representation one is likely to find in various online
sources. The symbols in brackets are those used by many phoneticians all over the world. Each
PHONETIC SYMBOL is independent of specific languages and spelling conventions. Thus, the
symbol [i] represents the sound of i in the English word machine, of eo in the English word people,
of ie in the German word sieben, of ie in the French word vie, of i in the Spanish word hijo, etc.
The vowels in Figure Two are distinguished by three horizontal tongue positions from the front of
the mouth to the back (FRONT, CENTRAL, and BACK) and three vertical tongue heights (HIGH,
MID, and LOW). In addition, vowels produced with considerable muscular effort (TENSE) are

distinguished from those that don’t involve such effort (LAX). If these differences are considered
features, then Figure Two amounts to a claim that eight difference feature contrasts are necessary
to distinguish the twelve English vowels: [±FRONT], [±CENTRAL], [±BACK], [±HIGH], [±MID],
[±LOW], [±TENSE], and [±LAX]. That claim must be examined to see if it is justified.
FIGURE TWO: ENGLISH VOWELS
FRONT CENTRAL BACK
[i] as in beat [u] as in boot TENSE
[w] as in bit [] as in put LAX
[e] as in bait [o] as in boat TENSE
[e] as in bet [c] as in butted []] as in bought LAX
[a] as in pot TENSE
[æ] as in bat [] as in but LAX
HIGH
MID
LOW
In the Introduction above, we talked about the importance of providing solutions to problems that
are the most cost-effective. Also, in Chapter One, we discussed the importance of attaining
descriptive adequacy, that is, in describing language with the most generalized principles possible
to cover all the data. These two matters, cost-effectiveness and descriptive adequacy, are directly
related. Employers are looking for employees who can provide them with solutions to problems that
satisfy the most factors or constituents in the most effective way (see Page 5). Linguists are looking
for descriptions of language that account for all the data in the most generalized way. Given Figure
Two, the question is this: is it necessary to have eight contrasting features to describe the twelve
English vowels. Is that the most cost-effective description? The answer is, No.
The features [±HIGH], [±FRONT], etc. in Figure Two contain a number of redundancies, that is,
unnecessary duplications. For example, there seems to be no difference between [–BACK] and
[+FRONT], [+BACK] and [–FRONT], [+TENSE] and [–LAX], [–TENSE] and [+LAX], and so on.
Furthermore, it seems excessive to have three horizontal positions, three vertical positions, as well
as the features TENSE and LAX, to describe the twelve English vowels (note the six empty cells
in Figure Two). In an attempt to reach the highest possible level of descriptive adequacy, linguists
seek to remove such unnecessary redundancies and excesses from their descriptions. Therefore, it
is more common today to replace the classification in Figure Two with the one in Figure Three.
Using the terminology from the Introduction (see page 14), Figure Three is more cost-effective than
Figure Two because it uses a fewer number of features to distinguish all the vowel sounds. In Figure
Three, we have eliminated the feature contrasts [±FRONT], [±CENTRAL], and [±MID], reducing
the number of features to describe tongue position from six to three. In addition, we have added the
57

58
feature [±ROUND] to distinguish rounded vowels from unrounded vowels, specifically, the vowel
[]] from the vowel [c].
FIGURE THREE: DISTINCTIVE FEATURES FOR ENGLISH VOWELS
i w e e æ u o ] a c
HIGH + + – – – + + – – – – –
LOW – – – – + – – – – + + –
BACK – – – – – + + + + + + +
ROUND – – – – – + + + + – – –
TENSE + – + – – + – + – + – –
There is more involved in choosing the feature inventory in Figure Three over Figure Two than just
simplicity. The number of features is, of course, important. But more important is whether or not
the feature inventory will be sufficient to describe all the data. In linguistics, that means all the
vowels in all the languages of the world. It is, therefore, an empirical issue. We must take the above
feature inventory and test it against other data in other languages to see if it is satisfactory. Notice
that is exactly what has to be done in the workplace. For example, if a manufacturer decides to
make a change in the number of options available for a product (size, shape, color, weight, etc.), the
effects of that change must be studied against all the relevant issues that could be impacted by the
change (customer satisfaction, cost, distribution, marketing, etc.). Such extensive study minimizes
unforseen consequences which can be very costly to a business. Theoretical changes work in the
same way, whether one is dealing with linguistics, psychology, physics, or any other discipline. For
just this reason, linguistic theory, like all other theories in all other disciplines, is a constantly
evolving theory in which changes are proposed and subsequently modified, extended, abandoned,
and so on. It is important for students to see that the choice of the features in Figure Three above
those in Figure Two is an entirely empirical issue that must be tested and justified; therefore, we will
return to this matter at several points below.
The technical definitions for the distinctive features of English vowels are as follows:
• HIGH – NONHIGH — High sounds are produced by raising the body of the tongue above
the level that it occupies in neutral position; nonhigh sounds are not produced with such a
raising of the body of the tongue.
• LOW – NONLOW — Low sounds are produced by lowering the body of the tongue below
the level that it occupies in neutral position; nonlow sounds are produced without such a
lowering of the body of the tongue.

• BACK – NONBACK — Back sounds are produced by retracting the body of the tongue from
the neutral position; nonback sounds are produced without such a retraction.
• ROUND – NONROUND — Rounded sounds are produced with a narrowing of the lips;
nonrounded sounds are produced without such narrowing.
• TENSE – NONTENSE (LAX) — Tense sounds are produced with a deliberate, accurate, and
maximally distinct articulatory gesture that involves considerable muscular effort; lax sounds
are produced rapidly and somewhat indistinctly.
In addition to the unitary sounds charted above in Figure Three, English contains a number of
DIPHTHONGS, which are vowels followed by a GLIDE ([w] or [y]). The true diphthongs are []y]
as in soy and soil; [ay] as in sigh, nice, guy, lie, lye, sign, aisle, and choir; and [aw] as in cow, house
and doubt. In most dialects, the [–LOW, +TENSE] vowels of English are also articulated with a
glide, particularly in word final position: [siy] see, [sey] say, [suw] sue, [sow] sew.
The glide portion of these examples is not equivalent phonetically to a [w] or [y] as these sounds
would appear in words like wet ([wet]) and yet ([yet]). Therefore, other texts may represent the
above examples as follows: []w] as in soy, [aw] as in sigh, [a] as in cow, [iw] as in see, [ew] as in
say, [u] as in sue, and [o]as in sew. For simplicity, in this book, all glides are represented as [w]
and [y], e.g., [wet] wet, [kaw] cow, [suw] sue, [sow] sew; [yet] yet, [s]y] soy, [siy] see, [sey] say.
2.1.2 CONSONANTS
Consonants are produced in a variety of different ways. They are classified in terms of the place in
the oral cavity (Figure One) where they are articulated and also the manner of their articulation. A
chart of some consonants specified according to this method of classification, as well as the features
[±CORONAL] and [±ANTERIOR] discussed in Chapter One, is given in Figure Four.
2.1.2.1 STOPS
Some involve the complete stoppage of the air flow at some point in the vocal tract. Such
consonants are known appropriately as STOPS. These include [p h ] (the sound of p in pan), [t h ] (the
sound of t in tan), and [k h ] (the sound of c in can and k in keen).
The point in the vocal tract at which the stoppage of air occurs is used to distinguish one stop
consonant from another. [p h ], which is produced by stopping the air with both lips, is called a
BILABIAL stop; [t h ], which is produced by placing the tongue against the alveolar ridge (see Figure
One), is called an ALVEOLAR stop; and [k h ], which is formed with the tongue against the velum,
is called a VELAR stop.
59

60
MANNER OF
ARTICULATION
STOP
[–VOICED]
[–VOICED]
[+VOICED]
FRICATIVE
[–VOICED]
[+VOICED]
AFFRICATE
[–VOICED]
[+VOICED]
NASAL
FIGURE FOUR: ENGLISH CONSONANTS, LIQUIDS, AND GLIDES
PLACE OF ARTICULATION
ANTERIOR NONANTERIOR
NONCORONAL CORONAL NONCORONAL
BILABIAL
p h (pun)
p (spun)
b (bun)
LABIO–
DENTAL
f (fie)
v (vie)
INTER–
DENTAL
› (thigh)
ð (thy)
ALVEOLAR
t h (ton)
t (stun)
d (done)
s (sue)
z (zoo)
(ALVEO)
PALATAL
š (shoe)
ž (beige)
č (chew)
:9 (Jew)
VELAR GLOTTAL
k h (come)
k (scum)
g (gum)
[+VOICED] m (sum) n (sun) õ (sung)
LIQUID
[+VOICED]
[+VOICED]
GLIDE
[–VOICED]
[+VOICED]
l (lay)
D (cater)
(whey)
w (way) y (yea)
r (ray) ł (ail)
§ (cotton)
h (hay)

The superscript “h” in the transcription of stops indicates a puff of air called ASPIRATION that
is emitted when the stops are released; this puff of air is missing when these stops follow [s]. The
distinction is apparent by placing a tissues in front of the mouth and noticing first that the tissue
moves when saying pan [p h æn], tan [t h æn], and can [k h æn] and second that the tissue does not move
when saying span [spæn], Stan [stæn], and scan [scæn]. Compare similar pairs like pit/spit, pin/spin,
tall/stall, cool/school, etc.
In the production of [p h ], [p], [t h ], [t], [k h ], and [k], the vocal cords do not vibrate; hence, these
sounds are called VOICELESS ([–VOICED]) stops. Other stops, for example, [b] as in bet, [d] as
in debt, and [g] as in get, are formed by a complete stoppage of the air flow with an accompanying
vibration of the vocal cords. These stops are called VOICED ([+VOICED]). The symbol [§]
represents a glottal stop that appears in a few words like button [b§cn] and cotton [ka§cn] or, in
some dialects, a word like bottle [ba§cl].
2.1.2.2 FRICATIVES
A second group of consonants, called FRICATIVES, is formed by a closure in the vocal tract,
which, though not complete as in the articulation of stops, is sufficiently constricted to cause
turbulence in the air flow, thereby producing a hissing sound. This group includes the following
sounds: [f] (wafer), [v] (waiver), [›] (ether), [ð] (either), [s] (racer), [z] (razor), [š] (mesher), an
[ž] (measure).
Like stops, fricatives are classified according to the point in the vocal tract at which the obstruction
in the air flow occurs: [f] and [v] are produced by a constriction formed with the lower lip and upper
teeth; they are called LABIODENTAL fricatives; [›] and [ð], produced by a constriction behind
the upper teeth, are INTERDENTAL; [s] and [z], produced by a constriction against the alveolar
ridge, are ALVEOLAR; [š] and [ž], produced by a constriction against the palate, are PALATAL.
The sound [ž] occurs initially only in the word genre which was borrowed from French. Though
relatively uncommon in English, [ž] does appear medially (vision, measure, usual, Asian, etc.) and
finally (beige, massage, camouflage, etc.).
2.1.2.3 AFFRICATES
Closely allied to the articulation of stops and fricatives is a group of consonants called
AFFRICATES. These consonants combine a complete closure at some point in the vocal tract, as
in the articulation of stops, with a turbulent release that produces a hissing sound, as in the
articulation of fricatives. Examples of these consonants are [č], the underlined sounds in chunk,
lecher, and search, and [:9], the underlined sounds in junk, ledger, and surge. These are the only
affricates in English, but there are several more in the other languages of the world.
61

62
2.1.2.4 NASALS
A fourth group of consonants, called NASALS, are produced by a closure in the mouth and a
simultaneous lowering of the velum so that the air escapes through the nasal passage (see Figure
One). English has three nasal consonants. They are distinguished from each other by the point in
the mouth where the closure is made. [m], the sound of m in Kim, is a bilabial nasal; [n], the sound
of n in kin, is an alveolar nasal; and [õ], the sound of ng in king is a velar nasal. Each of the English
nasals is a voiced consonant. The velar nasal [õ] never occurs initially in a word or syllable.
2.1.3 LIQUIDS AND GLIDES
In addition to consonants and vowels, languages contain two other classes of sounds, LIQUIDS and
GLIDES.
2.1.3.1 LIQUIDS
LIQUIDS include [l], the sound of l in low, and [r], the sound of r in row. [l] is produced with a
closure made by the center of the tongue at the alveolar ridge, in such a way that the air is allowed
to pass out of the mouth along the sides of the tongue. Since the air passes along the sides of the
tongue, l–sounds are often referred to as LATERAL consonants. In contrast to the articulation of
[l], [r] usually involves a closure, or near–closure, made with the sides of the tongue, so that the air
escapes from the mouth over the central portion of the tongue. A liquid sound can also be produced
by making the tongue FLAP against the alveolar ridge. This sound, represented as [D] or [], is
heard instead of [t h ] or [d] in words like water and ladder.
2.1.3.2 GLIDES
The last group of sounds includes [y], the sound of y in yell, and [w], the sound of w in well. These
consonants are either called GLIDES or, because of their similarity to the vowels [i] and [u],
SEMIVOWELS. Typically, they are produced like vowels since they do not involve a complete
stoppage of the air flow, but they are distributed like consonants since they occur before and after
vowels, as in the word wow.
The four groups of sounds (vowels, consonants, liquids and glides) are distinguished by the
following distinctive features:
FIGURE FIVE: THE MAJOR PHONOLOGICAL CATEGORIES
FEATURES CONSONANTS VOWELS GLIDES LIQUIDS
CONSONANTAL + – – +
VOCALIC – + – +

The definitions for the remaining distinctive features are as follows:
• SONORANT – NONSONORANT (OBSTRUENT) — Sonorants are sounds produced with
a vocal tract cavity configuration in which spontaneous voicing is possible; obstruents are
produced with a cavity configuration that makes spontaneous voicing impossible. [Note:
sonorants are usually hummable.]
• CONSONANTAL – NONCONSONANTAL — Consonantal sounds are produced with a
sustained vocal tract constriction at least equal to that required to produce fricatives;
nonconsonantal sounds are produced without such a constriction.
• VOCALIC – NONVOCALIC — Vocalic sounds are produced with an oral cavity in which
the most radical constriction does not exceed that found in the high vowels [i] and [u], and
with the vocal cords positioned so as to allow spontaneous voicing; in producing nonvocalic
sounds one or both of these conditions are not satisfied.
• CONTINUANT – NONCONTINUANT (STOP) — In the production of continuant sounds,
the primary constriction in the vocal tract is not narrowed to the point where the air flow past
the constriction is blocked; in stops the air flow through the mouth is effectively blocked.
• NASAL – NONNASAL — Nasal sounds are produced with a lowered velum which allows
the air to escape through the nose; nonnasal sounds are produced with a raised velum so that
the air from the lungs can escape only through the mouth.
• SIBILANT – NONSIBILANT — Sibilant sounds are produced by forcing the air through
a narrow opening produced by a groove in the midline of the tongue. Typically, sibilant
sounds have a hissing quality; non–sibilant sounds do not have this quality.
• CORONAL – NONCORONAL — Coronal sounds are produced with the blade of the
tongue (the portion immediately behind the tip) raised from its neutral position; noncoronal
sounds are produced with some other articulator than the blade of the tongue.
• ANTERIOR – NONANTERIOR — Anterior sounds are produced with an obstruction that
is located in front of the palato–alveolar region of the mouth; nonanterior sounds are produced
without such an obstruction.
• VOICED – NONVOICED (VOICELESS)— Voiced sounds are produced with the vocal
cords vibrating; voiceless sounds are produced without such vibration.
Figure Six categorizes English consonants using these nine distinctive features; it also includes the
four relevant features discussed above with vowels. Again, in terms of cost-effectiveness, Figure
Six is preferred to Figure Four because it removes redundancies and contradictions and, at the same
time, appears to be adequate to describe the data investigated so far.
63

FEATURES
FIGURE SIX: DISTINCTIVE FEATURES FOR ENGLISH CONSONANTS, LIQUIDS & G
STOPS NASALS FRICATIVES AFF* L
p b t d k g § m n õ f v › ð s z š ž č :9 l
SONORANT – – – – – – – + + + – – – – – – – – – – +
CONSONANTAL + + + + + + + + + + + + + + + + + + + + +
VOCALIC – – – – – – – – – – – – – – – – – – – – +
CONTINUANT – – – – – – – – – – + + + + + + + + – – +
NASAL – – – – – – – + + + – – – – – – – – – – –
SIBILANT – – – – – – – – – – – – – – + + + + + + –
ANTERIOR + + + + – – – + + – + + + + + + – – – – +
CORONAL – – + + – – – – + – – – + + + + + + + + +
VOICED – + – + – + – + + + – + – + – + – + – + +
HIGH – – – – + + – – – + – – – – – – + + + + –
LOW – – – – – – + – – – – – – – – – – – – – –
BACK – – – – + + + – – + – – – – – – – – – – –
ROUND – – – – – – – – – – – – – – – – – – – – –
*AFFRICATES

2.1.4 REVIEW OF ISSUES
At this point, it is important to examine the choices we have made in the feature inventories above.
For vowels, we noted that Figure Two involved eight feature oppositions: [±FRONT],
[±CENTRAL], [±BACK], [±HIGH], [±MID], [±LOW], [±TENSE], and [±LAX]. Figure Three
involved five: [±HIGH], [±LOW], [±BACK], [±TENSE], and [±ROUND]. We determined that the
reduced number of features in Figure Three was preferable. The argument was largely about
simplicity, but other factors are also relevant. For example, the feature opposition [±ROUND] must
be part of the inventory to account for phenomena that exist in the world’s languages. There is no
way to know this ahead of time; one must test the hypothesis by examining new data. In Turkish,
for example, an unrounded vowel in a word cannot be followed by a rounded vowel. The
phenomenon, called VOWEL HARMONY, requires that only certain combinations of vowels in
a word are possible. Clearly, we cannot account for the data in Turkish without the feature
[±ROUND]. Thus, quite apart from issues of simplicity, the empirical evidence demands that our
feature inventory includes the opposition [±ROUND]. Even in English the feature is important:
only vowels that are [+BACK, –LOW] are [+ROUND] ([u, , o, ]]).
The chart of consonants in Figure Four specifies two sets of features to indicate place of articulation.
One set includes only the two oppositions [±ANTERIOR] and [±CORONAL]. The other includes
seven oppositions: [±BILABIAL], [±LABIODENTAL], [±INTERDENTAL], [±ALVEOLAR],
[±PALATAL], [±VELAR], and [±GLOTTAL]. On the basis of simplicity alone, we would naturally
choose the set with two as opposed to the one with seven. But again, this is an empirical issue. In
our discussion of levels of adequacy in Chapter One (page 38 ff.), we saw that the data favor the first
set with just [±ANTERIOR] and [±CORONAL]. We observed that the prefix meaning ‘not’ is
pronounced in three ways in English ([wm] as in impossible, [wn] as in intangible, and [wõ] as in
incurable):
(1) a. [wm] before sounds that are [+ANTERIOR, –CORONAL], namely, [p, b, m]
b. [wn] before sounds that are [+ANTERIOR, +CORONAL], namely, [t, d, n]
c. [wõ] before sounds that are [–ANTERIOR, –CORONAL], namely, [k, g]
We also saw that the distribution is not arbitrary. Sounds assimilate to each other, that is, they tend
to become like each other, when they are adjacent. The rules in (1) account for the occurrence of
the bilabial nasal [m] before bilabial sounds, of the alveolar nasal [n] before alveolar sounds, and
of the velar nasal [õ] before velar sounds. The distribution is succinctly accounted for with the
features [±ANTERIOR] and [±CORONAL]. In Latin, the roots of nouns undergo different rules
depending on whether they end in a consonant that is [+CORONAL] ([t] and [d]) or [–CORONAL]
([p], [b], [k], and [g]). In many places below, we will find additional empirical support for the two
feature oppositions [±ANTERIOR] and [±CORONAL].
The chart of consonants in Figure Four involves a set of six feature oppositions for manner of
articulation: [±STOP], [±FRICATIVE], [±AFFRICATE], [±NASAL], [±LIQUID], and [±GLIDE].
The chart of consonants in Figure Six also involves a set of six feature oppositions for manner of
articulation, but they are different oppositions: [±SONORANT], [±CONSONANTAL],
65

66
[±VOCALIC], [±CONTINUANT], [±NASAL], and [±SIBILANT]. In this case, sheer number is
not the issue. So, we need to see which set is more empirically justified. The data indicate that the
second set is. First, notice that the second set includes the oppositions [±CONSONANTAL] and
[±VOCALIC], which allowed us to generalize the differences between the four major types of
sounds in natural languages, namely, vowels, consonants, liquids and glides (see Figure Five).
Second, the inventory in Figure Six also includes the opposition [±SIBILANT]. This opposition
operates in a number of phonological processes in English. For example, the plural of nouns is
indicated by the suffix [cz] only in words that end with a sound that is [+SIBILANT], i. e., [s, z, š,
ž, č, :9] as in bases, mazes, sashes, garages, arches, judges, etc. The data indicate that [±SIBILANT]
must be included in the inventory. Thus, in terms of both simplicity and descriptive adequacy, we
have elected to use the feature oppositions in Figure Three for vowels and Figure Six for consonants,
liquids, and glides. We have eliminated redundancies and excesses and included additional feature
oppositions that are empirically necessary like [±ROUND] and [±SIBILANT]. At the same time,
the features we have chosen seem to be the minimum set required to account for all the data.
As we have seen, the kinds of considerations made in linguistics must also be made in many other
areas. Let us consider an example from the retail business. It is generally thought that a store must
provide customers with several different varieties of a product in several different sizes and shapes
in order to satisfy the needs of as may different customers as possible. Thus, in the average
supermarket, one finds, for example, many different brands of mayonnaise in containers of many
different sizes and shapes. Jim Senegal, the founder of Costco, had a very different marketing
strategy. He reasoned that cutting down on the number of choices would help customers make their
selection more easily, and, in the end, they would purchase the items Costco sells. The result is that
Costco only offers about 4,000 different items compared to the 16,000-150,000 items stocked by
many competitors. Walmart offers over 100,000 different items. Senegal’s strategy has worked.
“In the last 27 years Costco has become the fourth largest retailer in the United States and the eight
largest in the world.” (Costco Case Study, 2011; http://themarketplaceoflife.blogspot.com/2011/01/)
Senegal’s strategy was risky since is went against the conventional wisdom in retail about the
number of choices customers want. Note that there are redundancies and excesses when a
supermarket stocks many different brands of mayonnaise in containers of many different sizes and
shapes. The questions are: If the supermarket gets rid of the redundancies (different brands with
exactly the same ingredients) and the excesses (multiple shapes, sizes, and quantities), will the
resulting selection satisfy as many customers? Will customers shop elsewhere? Or will customers
simply make a choice from the selections they are given? Several studies have shown that
customers report greater satisfaction when their original options are limited. The Costco business
model is a case in point. It includes turnover on a very limited number of selections for each type
of product, e.g., only one size and brand of mayonnaise. That strategy combined with the addition
of other innovations – volume purchasing, efficient distribution practices, and reduced handling of
merchandise – have helped to make Costco so successful. It is not an accident that Costco carries
only a very limited number of items.

For many students, the study of language in a linguistics course appears to involve exotic issues that
have no relationship to events in the real world. As we have seen, this is not the case. The questions
we have considered in this Chapter involve redundancy, consistency, adequacy, generality,
efficiency, and cost, which are exactly the same questions that arise in real world problems such as
the selection of brands and packaging of mayonnaise. When students can see this larger picture,
they can also see that only the data are different. In the Introduction, we noted that it is exactly the
larger picture – the ability to make connections at an abstract level between seemingly unrelated
matters – that is crucial to adaptability, to flexibility, to mobility, and to success.
2.1.5 SUPRASEGMENTALS
There are, of course, many more sounds in the world’s languages than the ones that have been
described above. In addition, human speech is characterized by several SUPRASEGMENTAL
features, that is, vocal modification executed along with the pronunciation of vowels and
consonants. Two of these suprasegmental features are PITCH, i.e., the relative tone at which
individual sound segments or groups of segments are produced, and STRESS, i.e., the relative
emphasis (or accent) given to individual sound segments or groups of segments. Features like pitch
and stress are as important in speech as vowels and consonants. Often, they serve to distinguish
words from each other. For example, permit must be pronounced with heavy stress on the first
syllable [pérmwt] if it is used as a noun, but with heavy stress on the second syllable [permw´t] if it
is used as a verb. Notice that this variation of stress is an example of how the sound system of a
language is directly related to the classes of forms in a language. It is impossible to describe this
variation without reference to whether permit is used as a noun or as a verb. There are many similar
cases in English, e.g., contract, pervert, present, and so on. Consider also pairs of examples like the
following:
(2) a. Look at the black bird. (white house, long horn, light weight, light year, etc.)
b. Look at the blackbird. (White House, longhorn, lightweight, light-year, etc.)
One cannot describe the stress given to sequences like black+bird, unless one considers whether the
sequence is used as a compound noun or as an adjective–noun construction. Examples such as these
attest to the interrelationship of sound and form in the grammar of a language.
2.1.6 SYLLABLES
There are as many syllables in a word as there are vowel sounds. Note the following paying
attention to sound, not spelling:
(3) a. One syllable: I my mine strengths
b. Two syllables: enough pygmy mining strengthens
c. Three syllables: Michigan Missouri Tennessee strengthening
67

68
As these examples show, a syllable must contain a vowel. It also can begin and end with no
consonants or one or more consonants. Using the symbol “C” for consonant and “V” for vowel, the
following are examples of one syllable words that begin and end with one consonant, a CVC pattern:
(4) CVC: pin seam gain suit zoom beaut
We could simply repeat Cs to indicate multiple consonants, but since the number of successive
consonants can fairly large (cf. strengths [streõg›s]), linguists often use a superscript to indicate the
most number of consonants and subscript to indicate the least number of consonants. In the
following notations, the superscript on the symbol “C” indicates the maximum number of
consonants (n=any number greater than 1); the subscript indicates the minimum number (0=none).
(5) a. All one syllable words: # C n
0 V C n
0
b. All two syllable words: # C n
0 V C n
0 V C n
0
c. All three syllable words: # C n
0 V C n
0 V C n
0 V C n
0
Here are some examples of one, two, and three syllable words using this notation:
(6) a. All one syllable words: # Cn 0 V Cn I: a
0
y
I’m: a ym
my: m a y
mine: m a yn
streams: str i mz
strengths: str e õg›s
Flint: fl w nt
b. All two syllable words: # C n
0 V C n
0 V C n
0#
enough i n f
pygmy: p w gm i
mining: m a yn w õ
strengthens: str e õg› c nz
Utah: y u ta
Milford: m w lf c rd
c. All three syllable words: # Cn 0 V Cn 0 V Cn 0 V Cn Michigan: m w š c g c
0
n
Missouri: m c z u r i
Tennessee: t e n c s i
instructive: w nstr kt c v
strengthening: str e õg› c n w õ
fiasco: f i y æ sk o
#
#
#
#
#

The last syllable is called the ULTIMA; the next to last syllable, the PENULT; the third to last
syllable ANTEPENULT. Words with accents on each of these syllables are given in the following
examples where the X indicates the position of the accent; the exact phonetic value for unaccented
vowels may vary from speaker to speaker ([c], [w], [], etc.):
(7) a. Accent on the ANTEPENULT: X
Michigan: m w´ š c g c n
buffalo: b ´ f c l o
asterisks: æ´ st c r c sks
America: c m é r c k c
asparagus: c sp æ´ r c g c s
b. Accent on the PENULT: X
pygmy: p w´ gm i
Utah: y ú t a
Missouri: m c z ú r i
aroma: c r ó m c
verenda: v c r æ´ nd c
chianti: k i y á nt i
c. Accent on the ULTIMA: X
streams: str í mz
enough: i n ´ f
Tennessee: t e n c s í
Illinois: w l c n ]´y
For practice, there are additional examples of stress placement in English words on pages 95 and
96. The answers to those practice examples are on pages 106 and 107.
2.2 PHONOLOGY
With a complete phonetic system for the identification and classification of speech sounds available,
phonologists can begin their work. Essentially, their task is to discover the ways in which the
sounds of language are systematized. This includes finding out (i) which sounds, out of all the
sounds a speaker of some language utters, are the linguistically significant sounds for that language,
and (ii) which rules govern the organization and distribution of these sounds with respect to each
other.
In analyzing English, for example, phonologists observe that all English speakers pronounce p in
several different ways. As we have seen, the sound [p h ] is followed by a perceptible puff of air,
called ASPIRATION; this occurs in the articulation of a word like pin. In a word like spin, on the
other hand, the p is not aspirated. As before, we use a superscript “h” for aspiration: the aspirated,
voiceless, bilabial stop in English is [p h ], and the unaspirated, voiceless, bilabial stop is [p].
69

70
In examining English, phonologists also observe that there are no pairs of words which are
distinguished by aspiration, in the way that there are many pairs which are distinguished by other
phonetic features, such as [–VOICED] ([p h ]) and [+VOICED] ([b]), e.g., plank and blank, tap and
tab, rapid and rabid, and so on. Lastly, they observe that [p] occurs only after the sound [s] and that
[p h ] occurs everywhere else. Thus, in spot, spin, spoke, and Spain, the p is unaspirated, but in pot,
pin, poke, and pain, the p is aspirated.
These observations are important because they reveal that the distinction between [p] and [p h ], i.e.,
unaspirated versus aspirated, is not a significant distinction in the phonology of English, unlike the
distinction between [p h ] and [b], i.e., voiceless versus voiced. Thus, while it is possible to predict
when [p], as opposed to [p h ], will occur in English words, it is not possible to predict when [p h ], as
opposed to [b], will occur. This distinction is crucial in phonological analysis.
Phonologists often call the distinctive classes of sounds in a language the PHONEMES of that
language, and refer to the positional variants of phonetically similar sounds as the ALLOPHONES
of a particular phoneme. To distinguish phonemes from allophones, the former are placed in slanted
lines, e.g., /p/ and /b/, and the latter retain their phonetic symbolization, e.g., [p] and [p h ]. The
relationship between a phoneme and its allophonic variants is expressed in the form of a rule
included in the grammar of a language. For example, the relationship between /p/, [p], and [p h ] in
English is expressed in two rules as follows:
(8) a. The phoneme /p/ is realized phonetically as its allophone [p] when it occurs in the
environment after [s], e.g., in the word spot.
b. The phoneme /p/ is realized phonetically as its allophone [p h ] in most other
environments, e.g., in the word pot.
It is important to realize that sounds which are phonemic, i.e., distinctive, in one language may be
allophonic, i.e., non–distinctive, in another language. For example, while [p h ] and [p] are allophones
of the same phoneme /p/ in English, they are two different phonemes in Chinese. To put it
differently, while [p h ] and [p] do not distinguish pairs of words in English, they do distinguish pairs
of words in Chinese. The Chinese word pai, if pronounced with a [p h ], means ‘branch,’ and, if
pronounced with a [p], means ‘weeds.’ It is very difficult for English speakers to master this
distinction when they attempt to learn Chinese; very often, they confuse the two sounds.
Conversely, speakers of many Asian languages have great difficulty in distinguishing the phonemes
/l/ and /r/ in English, because these two sounds are not phonemic in their native language. Without
considerable practice, a native speaker of Japanese, for example, might say rorripop when she
means lollipop. When English speakers hear rorripop, they find it hard to believe that the Japanese
speaker can’t easily hear the difference between what she’s said and lollipop. Similarly, a Chinese
speaker will find it hard to believe that an American cannot easily distinguish between the distinct
sounds [p h ] and [p].
Continuing with their analysis of English, phonologists observe that the other voiceless stop
phonemes in English, namely, /t/ and /k/, show the same allophonic variations as /p/. Thus, in stop,

stub, skin, school, the t and k sounds are unaspirated, but in top, tub, kin, and cool, they are aspirated.
This is not unexpected, since /p/, /t/, and /k/ form a NATURAL CLASS of sounds, which is a class
that shares the same set of distinctive features. In this case, the three sounds form the natural class
of [+CONSONANT, –CONTINUANT, –NASAL, –VOICED]. Furthermore, the same natural class
has another allophonic variation. In final position or at the end of a syllable, they become
unreleased. Notice that the final sounds in cap, pat, and tack are not fully exploded in the way the
initial sounds of the same words are; similarly, the p sound at the end of the first syllable in captain
and the k sound at the end of the second syllable in refracted are also not fully exploded. We will
use the symbol "°" to indicate lack of release. For example, we have [kæp°] (cap) and [kæp°t h wv]
(captive). The environment in which "°" occurs is predictable by rule: voiceless stops are
unreleased in final position of a syllable or word. Since the feature RELEASED is predictable in
English, it is not distinctive.
In Chapter One, we discussed the classification of several distinctive features for English, such as
[±NASAL], [±VOICED], [±ANTERIOR], and [±CORONAL]. We now have two nondistinctive
features in English: [± ASPIRATION] and [±RELEASE]. Summarizing, we have contrasts
between distinctive and nondistinctive features like the following:
(9) DISTINCTIVE FEATURES.
a. [± VOICED]
1. [–VOICED] [p], [t], [k]
2. [+VOICED] [b], [d], [g]
b. [± ANTERIOR]
1. [–ANTERIOR] [k], [g]
2. [+ANTERIOR] [p], [b], [t], [d]
c. [± CORONAL]
1. [–CORONAL] [p], [b], [k], [g]
2. [+CORONAL] [t], [d]
(10) NONDISTINCTIVE FEATURES.
a. [± ASPIRATED]
1. [–ASPIRATED] little breathiness
(i) [p] spade [sped]
(ii) [t] steam [stim]
(iii) [k] school [skul]
71

72
2. [+ASPIRATED] considerable breathiness
b. [± RELEASED]
(i) [p h ] paid [p h ed]
(ii) [t h ] teem [t h im]
(iii) [k h ] cool [k h ul]
1. [–RELEASED] no explosion
(i) [p°] mop [map°]
(ii) [t°] rate [ret°]
(iii) [k°] nook [nk°]
2. [+RELEASED] explosion
(i) [p h ] paid [p h ed]
(ii) [t h ] teem [t h im]
(iii) [k h ] cool [k h ul]
Phonologists discover the distinctive features of a language by looking for a pair of words
distinguished by one and only one sound. Such a pair is called a MINIMAL PAIR. If two sounds
occur in a minimal pair, then they occur in identical environments. This means that the two sounds
are distinctive and distinguish words. (11) contains examples of minimal pairs; (12) does not.
(11) MINIMAL PAIRS Good Examples (SPELLING IRRELEVANT):
a. mace/base [mes]/[bes]
b. ether/either [i›cr]/[iðcr]
c. known/sewn/shone [non]/[son]/[šon]
d. phlegm/Clem [flem]/[k h lem]
e. who’d/hood [hud]/hd]
f. bird/heard [bcrd]/[hcrd]
(12) Not Examples (SPELLING IRRELEVANT):
a. buff/bluff [bf]/[blf]
b. choose/loose [čuz]/[lus]
c. though/tough [ðo]/[tf]
d. chef/chief [šef]/[čif]
e. great/heart [gret]/[hart]
f. beard/heard [bird]/[hcrd]

As we have had occasion to notice many times in this book, an important objective of linguistics is
the construction of formal grammars. A formal grammar is simply one that is perfectly explicit and
testable. Precision is essential in linguistics, as it is in all other sciences. Without precision,
hypothetical principles and rules cannot be evaluated and tested with confidence. If hypotheses
cannot be tested, then substantive conclusions cannot be drawn. Rules such as those in (8) are too
informally written to satisfy scientific criteria. Therefore, linguists usually replace them with a
SLASH–DASH notation as follows:
FORMAL PHONOLOGICAL RULES:
(13) a. [+STOP, –VOICED] ÷ [–ASPIRATED] / [s] ___
a voiceless stop is unaspirated after [s]
b. [+STOP, –VOICED] ÷ [–RELEASED] / {+, #}
a voiceless stop is unreleased before a syllable boundary or a
word boundary
Each symbol in this notation has a precise meaning. As a result the rules are explicit and testable.
The slash–dash notation, which will be used throughout this text, observes the following
conventions:
THE SLASH–DASH NOTATION:
(14) a. ÷ means ‘is’
b. (...) items in parentheses are optional
c. {...} items in curly brackets or braces are alternatives
d. / means ‘in the environment of’
e. ___ means ‘before’ or ‘after’ or ‘between’
f. / ___x means ‘before x’
g. / x___ means ‘after x’
h. / x___y means ‘between x and y’
i. # means ‘word boundary’
j. #___ means ‘after a word boundary,’ that is, ‘at beginning of word’
k. ___# means ‘before a word boundary,’ that is, ‘at end of word’
l. + means ‘syllable boundary’ or ‘morpheme boundary’
m. +___ means ‘after a syllable or morpheme boundary’
n. ___+ means ‘before a syllable or morpheme boundary’
Summarizing the theoretical apparatus developed so far, we say that phonological representations
occur in pairs: a phonemic or abstract representation coupled with a phonetic or real representation.
The two representations are related by phonological rules.
The necessity for both levels of representation can be illustrated further in the following way (see
Figure Seven below). Suppose one English speaker utters the sentence If you hit me, I’ll hit you.
73

74
What the speaker actually says in the second part of this sentence is [ayl hwču]. The verb, therefore,
appears to be hitch [hwč]. But, although the listener has heard [hwč], what the listener understands
is /hwt/ (hit). The real or superficial or phonetic representation, the one actually spoken and heard,
is [hwč]; the abstract or underlying or phonemic representation, the one intended and understood, is
/hwt/. The relationship between the two is accounted for by phonological rules like those in (13),
specifically, (15a) and (15b).
(15) a. [t] ÷ [č] / ___ [y]
b. [y] ÷ Ø / [č]___
(15a) changes a [t] to a [č] before [y]; (15b) deletes a [y] after [č]. The symbol "Ø" means ‘zero’;
so (15b) says that a [y] becomes zero (is deleted) after [č]. These rules represent the change of the
sound [t] to [č]. The derivation from underlying form to surface form proceeds as follows:
(16) UNDERLYING FORM: /hwt yu/
APPLICATION OF RULE (15a): hwč yu
APPLICATION OF RULE (15b): hwč u
SUPERFICIAL FORM: [hwču]
This derivation is intended to represent the native speaker’s knowledge (competence) of the intended
meaning of what is actually uttered, namely, [hwču].
To an introductory linguistics student, a derivation like (16) often appears to be the result of
hocus–pocus. This is not the case. The derivation proceeds from facts: in the example cited, there
is a discrepancy between what is uttered and heard ([hwču]), on the one hand, and what is intended
and understood (/hwt/), on the other. Further, all native speakers of English know this unconsciously;
it is part of their linguistic competence. If this were not the case, the person hearing [hwču] in the
context given, would not understand the intended meaning. These facts are givens; there is nothing
that a linguist can do but attempt to describe and explain them. It is not the job of linguistics to alter
the way people communicate (a hopeless task, incidentally). The postulation of an abstract
underlying form, related by rule to a real superficial form, is merely an effort to describe observed
phenomena. Again, linguistic theory proceeds in the appropriately scientific manner. Notice that
the example here is not isolated; in ordinary speech, the spoken form rarely matches the intended
meaning word for word:
(17) Spoken Form Intended Meaning Written Form
a. [:9uwancgo] /du yu want tu go/ Do you want to go?
b. [ičycdwnc] /it y]r dwncr/ Eat your dinner.
c. []lcvcmcvleft] /]l f ðem hæv left/ All of them have left.
d. [kwigcwawt] /kd wi go awt/ Could we go out?

FIGURE SEVEN: PHONOLOGICAL AND PHONETIC REPRESENTATIONS
PHONEME ALLOPHONE
DISTINCTIVE FEATURES
[±VOICED]
[±ANTERIOR]
NONDISTINCTIVE FEATURES
[±ASPIRATED]
[±RELEASED]
NOT PREDICTABLE BY RULE PREDICTABLE BY RULE
NONREDUNDANT REDUNDANT
FOUND IN MINIMAL PAIRS NOT FOUND IN MINIMAL PAIRS
FOUND IN OVERLAPPING
DISTRIBUTION (IN THE SAME
ENVIRONMENTS)
SLANTED LINE NOTATION
/spat/ spot
FOUND IN COMPLEMENTARY
DISTRIBUTION (NOT IN THE SAME
ENVIRONMENTS)
SQUARE BRACKET NOTATION
[spat°] spot
UNDERLYING (DEEP) SUPERFICIAL (SURFACE)
ABSTRACT REAL
Examples: Examples:
Will he hit me?
/wwl hi hwt mi/
Will he hit you?
/wwl hi hwt yu/
Willie hit you?
/wwli hwt yu/
Will he hitch you up?
/wwl hi hwč yu p/
Willie hitched you up?
/wwli hwč+d yu p/
Will he hit me?
[wwli hwt° mi]
Will he hit you?
[wili hwču]
Willie hit you?
[wwli hwču]
Will he hitch you up?
[wwli hwču p°]
Willie hitched you up?
[wwli hwčt°ču p°]
75

76
2.3 PHONOTACTICS
Further investigation of English phonology reveals that there are general restrictions on the
sequences of sounds that are permissible in the language. There are ACCIDENTAL GAPS,
sequences of sounds that just don’t happen to occur, and, NONACCIDENTAL GAPS, sequences
of sounds that are not possible. The permissible sequences are specified in the PHONOTACTIC
RULES of the language, that is the rules concerned with the sequential arrangements of sounds in
larger units. Consider the following:
(18) ACCIDENTAL GAP (possible words): [blwd], [stæno], [sprim]
(19) NONACCIDENTAL GAP (impossible words): [ftwd], [zdæno], [spnim]
In analyzing the PHONOTACTICS of English, phonologists would observe that an English word
cannot end with two nasal consonants even if it is spelled with two. We see this in a root such as
/kalmn/ which loses the second nasal in the word column [kalm], but retains it before the suffix
–ist in the word columnist [kalmnwst]. Compare also autumn/autumnal and hymn/hymnal. The loss
of the second nasal at the end of a word is accounted for by the following phonotactic rule:
(20) [+NASAL] ÷ Ø / [+NASAL]___#
In addition to (20), the phonotactic rules of English do not permit other consonants to occur before
nasals at the beginning of a word. Notice that both consonants are pronounced only when they occur
in separate syllables:
(21) a. mnemonic [nemánwk] ‘helping the memory’
amnesia [æmnéža] ‘loss of memory’
b. pneumonia [ncmónya] ‘respiratory disease’
apnea [æ´ pniya] ‘temporary loss of breathing’
c. know [nó] ‘grasp in the mind’
acknowledge [æknálc:9] ‘recognize’
The data in (21) indicate that English also has the following rule:
(22) [+CONSONANT] ÷ Ø / #____[+NASAL]
The presence of rules like (22) is what makes some words retaining their foreign spelling look so
strange as compared to their pronunciation, e.g., mnemonic, pneumonia, and pterodactyl, which
come to English from Ancient Greek. Still, the deleted consonant must be considered part of the
root so that the words can be related. Thus, the root for know must be /kno/ in order to relate know
and acknowledge. As before, we must appeal to an abstract level of representation to do so.

In addition to specific phonotactic rules such as those above involving nasals, there are more
generalized phonotactic rules. For example, if an English word begins with three consonants (CCC),
the first one must be [s], the second one must be a voiceless stop ([p], [t], or [k]), and the third one
must be a liquid ([l] or [r]).
(23) a. spl- splash [splæš] spr- spread [spred]
b. stl- * str- stream [strim]
c. skl- sclaff [sklæf] skr- scream [skrim]
Notice that, so far, there are no English words that begin with [stl]. Given the other possibilities,
that gap should probably be considered an accidental gap.
2.4 MORPHOLOGY
As one might expect, the study of the sound system of a language cannot be carried out in complete
isolation from the other components of grammar. The phonological rules, in particular, very often
involve reference to MORPHEMES, which are the minimal units of meaning in a language.
Consider the following (ignoring details involving aspiration and release):
(24) MORPHEME: a minimum unit of distinctive meaning.
a. [spikcr] ‘speaker’; 2 syllables; 2 morphemes:
/spik/ ‘to utter words’
/cr/ ‘agent’
b. [kriyetcr] ‘creator’; 3 syllables; 2 morphemes.
/kriyet/ ‘to bring into existence’
/cr/ ‘agent’
c. [dalcr] ‘dollar’; 2 syllables; 1 morpheme.
d. [æntlcr] ‘antler’; 2 syllables; 1 morpheme
e. [bwgcr] ‘bigger’; 2 syllables; 2 morphemes
/bwg/ ‘large in size’
/cr/ ‘more’
As the examples in (24) show, morphemes are not necessarily equivalent to either words or
syllables. Further, a particular sequence of sounds can be a morpheme in one word and not another:
notice [cr] in speaker is a morpheme, whereas the [cr] in antler is not. Similarly [pik] is a
morpheme in peeks [piks], but not in speaks [spiks]. Lastly, observe that the morpheme /cr/ either
indicates an agent or expresses the comparative (more).
77

78
Morphemes have different phonetic variants in different contexts just as phonemes do. Paralleling
the PHONEME/ALLOPHONE relation, we have the MORPHEME/ALLOMORPH relation.
An ALLOMORPH is a conditioned phonetic variation of a morpheme. Consider the following:
(25) ALLOMORPH: a conditioned PHONETIC variation of a morpheme:
a. Negative Prefix:
[wm]: impossible, imbalance
[wn]: intangible, indiscrete
[wõ]: incorrigible, ingratitude
b. The indefinite article:
[]: a book, a use, a hotel
[æn]: an owl, an honor
For the grammarian, one point of particular interest in morphological analysis concerns the nature
of the variations which specific morphemes show in their phonetic shape. Occasionally, this
variation is peculiar and reflects the idiosyncratic speech habits of individual speakers. For example,
some English speakers pronounce abdomen with heavy stress on the first syllable, while others
pronounce it with heavy stress on the second syllable. In such cases as this, the rival forms are said
to be in FREE VARIATION.
More commonly, a variation in the pronunciation of a morpheme is predictable and reflects specific
rules in the grammar of a particular language, sometimes called MORPHOPHONEMIC or
MORPHOPHONOLOGICAL RULES. For example, the plural morpheme in English nouns is
variously pronounced [cz] as in busses, [s] as in bucks, and [z] as in bugs.
The occurrence of each of these allomorphic variants is predictable in terms of the phonetic
environment in which the plural morpheme occurs. If the morpheme occurs after [s], [z], [š], [ž],
[č], or [:9], it is realized phonetically as [cz], e.g., dresses, sizes, ashes. Otherwise, if it occurs after
a voiceless consonant, it is realized phonetically as [s], e.g., tanks, pipes, slots; if it occurs after a
voiced consonant or after a vowel, it is realized phonetically as [z], e.g., bulbs, trees, eggs. Utilizing
the theory of distinctive features, we can formally represent these rules as follows (where PL means
‘plural’):
(26) a. PL ÷ [cz] / [+SIBILANT] ___
b. PL ÷ [s] / [–VOICED] ___
c. PL ÷ [z] / [+VOICED] ___
These same variations are also found in the morpheme which indicates the third person singular
present tense of verbs, e.g., [cz] as in blesses and pleases, [s] as in attempts and pollutes, and [z] as
in confides and specifies. They also appear in contractions (he’s for he is) and in possessives, e.g.,
[cz] as in Bess’s and Josh’s, [s] as in Pat’s and Skip’s, and [z] as in Budd’s and Don’s.

2.5 THE BIOLOGICAL BASIS OF PHONOLOGY
In Chapter One, we discussed two fundamental problems for linguistic theory, the grammatical
characterization problem and the grammatical realization problem. We pointed out that the ultimate
objective of linguistic research is to uncover the specifications of UG, the collection of features and
principles which are potential characteristics of any human language. This goal is achieved by
relating the facts of grammatical characterization to the facts of grammatical realization.
The rules we have considered so far are descriptively adequate (see Chapter One) as far as the
examples cited are concerned; however, they are not general enough. We can attain a higher level
of descriptive adequacy. In this regard, linguistics is not different from any other discipline,
business, or enterprise. Everyone is searching for the most cost-effective solution to problems.
Consider, first, rule (15a) and notice that both members of the pair [t]–[č] are voiceless. If our
theory is correct, we expect the other members of this natural class to behave similarly, that is, we
expect the grammar to contain (27).
(27) [d] ÷ [:9] / ___ [y]
In fact, English does have (27), as (28) shows.
(28) a. If you hide me, I’ll hide you.
b. If you hide me, /ay wwl hayd yu/
c. If you hide me, [ayl hay :9u].
Since the sounds involved form a natural class, we can collapse (15a) and (27) to (29), thereby
attaining a higher level of descriptive adequacy.
(29) [–SIBILANT, +ANTERIOR] ÷ [+SIBILANT, –ANTERIOR] / ___[y]
This rule appears to be correct; however, we have gone too far. The feature constellation
[–SIBILANT, +ANTERIOR] includes many consonants other than [t] and [d]; specifically, it
includes [p], [b], [m], [n], [›], [ð], [l], [r], and [y]. None of these sounds becomes an affricate before
[y]; therefore, we amend (29) to (30):
(30)
79

80
Notice that (30) involves a change of two distinctive features; the other features remain unchanged.
The result is that an alveolar stop ([t] or [d]) followed by the anterior glide ([y]) becomes an affricate
([č] or [:9], respectively). It is important to realize that one letter or sound does not turn into another.
The letter t and the sound [t] are merely abbreviations for constellations of distinctive features
derived from characteristics of the human vocal apparatus. It is these features that change in the
operation of phonological rules. In (30), two of the features in the feature cluster change value; the
others remain unchanged.
Notice also that our concern for formalism, that is, precision, now pays off, because only now can
we ask the really important questions: Why does the grammar of a human language contain a rule
like (30)? Do children born in an English speaking environment have to learn this rule specifically,
as a peculiar fact about English, in the way that they must learn that the past tense of go is went?
The answer to these questions involves a dynamic that exists between the structure of the human
vocal apparatus and the general phonology of English. This dynamic is influenced by a number of
factors.
First, notice that rapid articulation of the sound sequence [tyu] sounds very much like [ču], and the
articulation of [dyu] sounds very much like [:9u]. The reason for this is due to the position the
tongue must assume during the articulation of the sequences [tyu] and [dyu]. Thus, it would be quite
natural for speakers to merge [t]/[d] and [yu] across a word boundary, substituting [ču] for [tyu] and
[:9u] for [dyu] in rapid articulation.
Second, this substitution would be particularly likely under two conditions: (i) if the phonology of
the language includes the sounds [č] and [:9] as phonemes, and (ii) if the language does not have a
phonological distinction between [ty] and [č], or [dy] and [:9]. These two conditions both exist in
English. The sounds [č] and [:9] are common in all dialects (chew, church, chat, chair, cheer; Jew,
judge, jet, jam, jeer, etc.). Furthermore, most dialects of English do not have phonetic sequences
of [t] or [d] followed by [y] unless there is a pause in pronunciation. The usual pronunciation of
tune, for example, is [tun], not [tyun].
For these reasons, in rapid speech, it is natural for an English speaker, knowing the overall
phonology of English, to apply (30). What this means is that, in the overall phonology of any given
language, some phonological rules may follow naturally given other factors. The suggestion,
therefore, is that an English speaking child does not have to learn (30) specifically as a peculiar,
isolated fact about English.
This analysis clearly indicates that some phonological processes are at least partially predictable.
In other words, children learning a language are not faced with the task of acquiring a set of rules
that are isolated, idiosyncratic, and arbitrary. Apparently, language acquisition is possible precisely
because a significant portion of the grammar of all natural languages is a direct reflection of the
nature of the apparatus used to produce them, that is, of the articulatory and auditory organs and the
brain.

This issue is important enough to require careful scrutiny. Therefore, let us consider a more
extended example and return to the rules for forming the plural given in (26). As we discussed in
the Introduction (Page 12 ff.), there is an alternation in the allomorph which signals past tense in
English that is similar to the rules in (26). Consider the following (where PST means ‘past tense’):
(31) a. PST ÷ [cd] / {[t], [d]} ___
b. PST ÷ [t] / [–VOICED] ___
c. PST ÷ [d] / [+VOICED] ___
Rule (31a) correctly specifies past tense forms like relented ([rilentcd]) and applauded
([cplawdcd]); (31b), disliked ([dwslaykt]) and laughed ([læft]); (31c), nagged ([nægd]), riled
([rayld]), and relayed ([riled]). Notice that all past tenses of regular verbs in English are spelled
either –ed or –d; but, this is not relevant to their pronunciation. Compare riled ([rayld]) and relayed
([riled]). As we have seen, the rules of language are based on speech, not writing.
The important question now is, Can the characterizations in (26) for the plural and (31) for the past
tense be related to universal grammar? The answer is, Yes. The following sections specify that
relationship.
2.5.1 FIRST SET OF OBSERVATIONS
Notice that (26) and (31) are very similar. Both sets of rules begin with a rule that deals with the
allomorphic variant that occurs when the morpheme and the stem to which it is suffixed are
HOMORGANIC, that is, share the same or very similar set of distinctive features.
(26a) deals with the suffixation of a [+SIBILANT, +CORONAL] morpheme to a word which ends
in a sound that is [+SIBILANT, +CORONAL]. (31a) deals with the suffixation of an alveolar stop
([–CONTINUANT, –NASAL, +ANTERIOR, +CORONAL]) to a word which ends in an alveolar
stop. In both sets of rules, a separate syllable is created by inserting the neutral vowel [c] in order
to distinguish the morpheme from the stem to which it is being attached. Furthermore, if this were
not done, some sequences violating English (if not, universal) phonotactics would be the result, e.g.,
*[æss] and *[cplawdd].
In the remaining rules in both sets, a very common phonological process occurs. This process,
called ASSIMILATION, involves making adjacent sound segments similar in distinctive feature
composition. In (26) and (31) the assimilation involves the feature [±VOICED]. Assimilation in
natural languages is so common that it is considered the UNMARKED or expected case. Lack of
assimilation in suffixation is the MARKED or atypical case.
Given these remarks, we specify the morphemes in question as (32), and we replace (26) and (31)
with the more general principles in (33).
81

82
(32) a. The PST morpheme in English is the suffix [d].
b. The PL morpheme in English is the suffix [z].
(33) a. If a suffix and the final segment to which it is attached are homorganic, insert the
neutral vowel [c] between the stem and the suffix.
b. In all other cases, assimilate the suffix and the final segment in voicing.
In order to express principles like (33) formally, we must have a way to represent a variable feature
value. This is done using Greek letters in the spot in front of a feature where a “+” or “–” usually
occurs. For example, we express voicing assimilation (33b) as follows:
(34) [+CONSONANT] ÷ [αVOICED] / [αVOICED] + ___
This rule says that a consonant beginning a suffix gets a value of voicing (either “+” or “–”) that is
the same as the final segment before the morpheme boundary. Notice that “+” has two meanings:
before a feature as in [+VOICED], it indicates a positive value for the feature; by itself in a rule like
(34), it indicates a morpheme boundary.
Sometime, several features assimilate at the same time. In this case, more than one Greek letter is
necessary to separate the assimilating features from each other. For example, consider again the
negative prefix given in (25a), repeated here:
(35) a. [wm]: impossible, imbalance
b. [wn]: intangible, indiscrete
c. [wõ]: incorrigible, ingratitude
In this set of data, there is an assimilation of the features ANTERIOR and CORONAL to those of
segment beginning the word. This means that the value (“+” or “–”) of the features ANTERIOR and
CORONAL of the consonant beginning the word must be the same as the prefix. We can express
this as follows:
(36) [+NASAL] ÷ [αANTERIOR, βCORONAL] / ___+ [αANTERIOR, βCORONAL]
In this rule, we must keep the assimilating values for ANTERIOR separate from those of
CORONAL; hence, two Greek letters are needed. The value of “α” and “β” must be the same in
both places where each occurs in the rule. In short, (36) abbreviates the following four rules:
(37) [+NASAL] ÷ [+ANTERIOR, –CORONAL] / ___+ [+ANTERIOR, –CORONAL]
(38) [+NASAL] ÷ [–ANTERIOR, +CORONAL] / ___+ [–ANTERIOR, +CORONAL]
(39) [+NASAL] ÷ [+ANTERIOR, +CORONAL] / ___+ [+ANTERIOR, +CORONAL]
(40) [+NASAL] ÷ [–ANTERIOR, –CORONAL] / ___+ [–ANTERIOR, –CORONAL]

Clearly, a rule like (36) will be needed to express the homorganicity in (33a). For example, if we
wish to express the insertion of [c] before the past tense morpheme [d], then (41) is needed.
(41)
Similarly, (42) is needed for the insertion of [c] before the plural morpheme [z]:
(42)
Using the α–notation, we can collapse these to (43), the formal expression of (33a) .
(43)
2.5.2 SECOND SET OF OBSERVATIONS
We have now correctly identified the morphemes for the plural and for the past tense (cf. (32)),
repeated here as (44).
(44) a. The PST morpheme in English is the suffix [d].
b. The PL morpheme in English is the suffix [z].
We have also formally expressed the distribution of the allomorph in (33a)/(43) and (33b)/(34),
repeated here together as (45).
83

84
(45) a. If a suffix and the final segment to which it is attached are homorganic, insert the
neutral vowel [c] between the stem and the suffix.
b. In all other cases, assimilate the suffix and the final segment in voicing.
[+CONSONANT] ÷ [αVOICED] / [αVOICED] + ___
How does this characterization in (44) and (45) relate to what speakers know? It is clear that native
speakers of English know what the two morphemes are and how they are distributed. English
speakers do not have to hear the plural of every noun and the past tense of every verb in order to
produce the correct forms. In fact, if speakers are given nonsense words and asked to make them
plural or past tense, they produce forms which obey (44) and (45). Moreover, speakers make up
words in the correct number and tense all the time. For example, on a recent TV commentary of the
US Figure Skating Championships, Peggy Fleming said, She two–footed that landing. Members of
the television audience did not wince at the new verb two–foot; everyone routinely understood it.
Further, the correct allomorph was uttered and understood, even though the expression must have
been a neologism to most who heard it. Examples like this occur very frequently.
Young children also produce forms that they never could have heard: past tenses like [god] ‘goed’
and [›rod] ‘throwed’; plurals like [mænz] ‘mans’ and [fts] ‘foots.’ These overgeneralizations give
clear evidence that the substance of (44) and (45) is unconsciously known to native speakers of
English from a very early age.
Do native speakers (unconsciously) know (44) and (45) in the form given? The answer is not clear.
While the substance of (44) and (45) can be expressed in many different ways, the characterization
given is the most generalized. Science usually tries to account for natural phenomena with the most
generalized descriptions possible. But there is no way to determine whether or not people do the
same when they learn their native language. In fact, it may be the case that the rules of language are
redundantly specified. Future psycholinguistic and neurolinguistic investigations may help to settle
this issue.
At this point, linguistics is in the same position as other disciplines, businesses, and enterprises. We
have produced the best possible product our ingenuity and science have allowed, namely, (44) and
(45). It remains to be seen how well our product fares upon further scrutiny.

2.5.3 THIRD SET OF OBSERVATIONS
Summarizing, we can say that (44) and (45) reach a very high level of descriptive adequacy. They
adequately account for all the known cases, and do so in a highly generalized way. Further, the
substance of (44) and (45) is unconsciously known to every native speaker.
Consider next the question whether children have to learn both (44) and (45). It seems clear that
children are not born with unconscious knowledge of (44), because (44) contains facts peculiar to
English. There is no reason in principle why, say, the meanings of the morphemes couldn’t be
reversed in some other language. The rule (45a) is also peculiar to English; however, its effect is
to differentiate morphemes, so we would expect other languages to have rules of similar design.
The status of (45b) is another matter. The rule (45b) is a good candidate for UG status, that is, it is
possible that (45b) is part of every human being’s genetic makeup. This would mean that it follows
directly and naturally from the nature of the human language apparatus, and that a particular form
in a particular language violating (45b) would be MARKED in the sense described above. Human
languages generally have a voicing assimilation rule, apparently because it is easier for a human
vocal apparatus to articulate adjacent sounds with the same voicing than with opposite voicing. This
is particularly true at the end of a word. As a result, we say that children do not have to learn (45b);
it is part of their genetic endowment; it is the UNMARKED case.
While this discussion is far from conclusive, it does show that all the facts in the characterization
of a language are not equal. Some facts may be highly marked, and, therefore, take children a long
time to master. Other facts may reflect inner characteristics of man, and, therefore, be part of the
innate knowledge that children bring to native language acquisition. Because children have a human
language apparatus, some features and principles are more likely to occur in a human language than
others.
2.6 THE PHONOLOGICAL AND LEXICAL COMPONENTS OF A GRAMMAR
The PHONOLOGICAL COMPONENT of a grammar consists of the list of phonological rules.
These are rules like (34), (36), and (43), which relate underlying forms to superficial forms. Also
included in the phonological component are phonotactic rules like (20) and phonological
REDUNDANCY RULES like (46).
(46) a. [+HIGH] ÷ [+HIGH, –LOW]
b. [+LOW] ÷ [+LOW, –HIGH]
(46a) accounts for the fact that a segment that is [+HIGH] cannot also be [–LOW], and (46b)
account for the reverse. These are redundantly specified features, that is, features that are
predictable given other features.
85

86
Vowels have many redundant features. Since only vowels are [+TENSE], we can specify the
redundancies as follows:
(47) [+TENSE] ÷ [+TENSE, +SONORANT,–CONSONANTAL, +VOCALIC,
+CONTINUANT, –NASAL, –SIBILANT, –ANTERIOR,
–CORONAL, +VOICED]
Since all the features except TENSE in (47) are redundant, they are generally left out of charts of
distinctive features for vowels (cf. Figure Three on Page 58).
The LEXICAL COMPONENT (LEXICON) of a grammar is the dictionary of morphemes and
words, as well as lexical redundancy rules and rules of word formation. (38) is an example of
lexical redundancy.
(48) [+HUMAN] ÷ [+HUMAN, +ANIMATE, +CONCRETE]
The rule (48) accounts for the fact that all HUMAN nouns are also ANIMATE and CONCRETE.
The lexicon also contains morphemes like (49).
(49) Morphemes:
a. /trst/ trust, trusts, trusted, mistrust, mistrustful, entrust, untrustworthy, etc.
b. /red/ raid, raids, raided, raiding, raider, raiders
From these components, we can produce underlying representations like (50).
(50) Underlying phonological representations:
a. /trst+s/
b. /trst+d/
c. /mws+trst/
d. /mws+trst+fl/
Application of various phonological rules to (50) produces the following phonetic representations,
respectively:
(51) Phonetic representations (ignoring details involving aspiration and release):
a. [trsts]
b. [trstcd]
c. [mwstrst]
d. [mwstrstfl]

To these forms, various stress rules must apply assigning the appropriate intonation contours to
words. We will return to a discussion of the lexicon in subsequent chapters.
2.7 ENGLISH SPELLING
It is no secret that English words are not spelled phonetically. In fact, lists of spelling nightmares
like the following can be found in every student’s notebook:
(52) a. rafter / laughter / slaughter
b. slay / sleigh / sleight / slight
c. ear / earn / urn / earth / hearth
d. though / thought / through / tough
e. for / fore / four / fourteen / forty
Often, the same sound has many different spellings:
(53) a. heard / word / bird / curd / nerd
b. rays / raise / raze / weighs / phase / obeys
c. weird / feared / veered / adhered
d. birth / berth / worth / earth
Likewise, similar spellings can signify widely different sounds:
(54) a. beard [bird] / heard [hcrd]
b. comb [kom] / tomb [tum] / bomb [bam]
c. meat [mit] / great [gret] / threat [›ret] / heart [hart]
d. cough [k]f] / bough [baw] / dough [do] / rough [rf] / through [›ru]
One frequently hears desperate cries for reform of the English spelling system. While there are
many peculiarities, it is clear that a system of phonetic spelling would undermine the considerable
regularity between English spelling and English morphology. For example, consider the various
spellings of the reduced vowel schwa (c) in the second syllable of fallacy, remedy, family, history,
luxury. In these, as in other cases, the spelling derives from the spelling of the root word as
comparisons reveal:
(55) a. Spelled a: fallacy/fallacious; comparable/compare; legacy/allegation
b. Spelled e: remedy/remedial; manager/managerial; competent/compete
c. Spelled i: family/familial; president/preside; immigrate/migrate
d. Spelled o: history/historian; consolation/console; janitor/janitorial
e. Spelled u: luxury/luxurious; injury/injurious; mercury/mercurial
87

88
Similarly, silent letters frequently show up in related words:
(56) a. Silent n: hymn/hymnal; column/columnar, columnist; damn/damnation
b. Silent g: resign/resignation; sign/signal, designate; phlegm/phlegmatic
c. Others: muscle/muscular; bomb/bombard; soften/soft; know/acknowledge
There are also many regularities. For example, consider the vowel alterations in (47) where the
silent e indicates that the preceding vowel must be tense.
(57) Lax/Tense Vowel Alternations
a. [ay]/[w]: divine/divinity; ignite/ignition; wide/width; type/typical
b. [ey]/[æ]: profane/profanity; sane/sanity; bathe/bath; shade/shadow
c. [iy]/[e]: serene/serenity; redeem/redemption; sleep/slept; meter/ metric
d. [aw]/[]: profound/profundity; abound/abundant; south/southern
e. [ow]/[a]: verbose/verbosity; know/knowledge; nose/nostril; phone/phonic
These vowel alternations are rule governed as are the consonant alternations in (58) and (59).
(58) Stop/Fricative Alternations
a. [t/[s]: democrat/democracy; president/presidency; pirate/piracy
b. [k/[s]: electric/electricity; public/publicity; medical/medicine
c. [s]/[š]: express/expression; digress/digression; race/racial; space/spatial
d. [z]/[ž]: revise/revision; infuse/infusion; excise/excision; confuse/confusion
e. [t/[š]: relate/relation; ignite/ignition; native/nation; president/presidential
f. [k/š]: logic/logician; magic/magician; practical/practitioner
g. [d]/[ž]: persuade/persuasion; corrode/corrosion; decide/decision
h. [d]/[s] persuade/persuasive/; corrode/corrosive; divide/divisive
I. [z]/[s] abuse (verb)/abusive; diffuse/diffusive; advise/advice; choose/choice
(59) Stop/Affricate Alternations
a. [d]/[:9]: grade/gradual; residue/residual; fraud/fraudulent; proceed/procedure
b. [t/[č]: right/righteous; Christ/Christian; quest/question; moist/moisture
c. [g]/[:9]: allegation/allege; regal/regicide; rigor/rigid; gregarious/egregious
Basically, changing English spelling to a system that is phonetic would obscure many of the regular
allomorphic variations in the language. A much better approach would be to teach students
something about English phonology and morphology pointing out that English spelling reveals
underlying regularities. Discussion of the rules governing these alternations requires considerable
technical mastery and is beyond the capabilities of elementary school children trying to cope with
English spelling. However, it would certainly be both possible and helpful to point out the
regularities as the words entered children’s vocabulary studies.

2.8 SUMMARY OF RULE FORMALISM
SYMBOL MEANING
1. [x] x is a PHONETIC symbol or feature
2. /x/ x is a PHONEMIC symbol or feature
3. Ø null
4. ÷ is
5. / in the environment of
6. ___ before, after, or between
7. / ___ x before x
8. / x ___ after x
/ x ___ y between x and y
9. # word boundary
/ # ___ at beginning of word
/ ___ # at end of word
10. + morpheme boundary
/ + ___ after a morpheme boundary
/ ___ + before a morpheme boundary
11. C CONSONANT
12. V VOWEL
13. (x) x is an OPTIONAL element
14. {x, y, z} x, y, and z are ALTERNATIVE elements
15. [x, y, z] x, y, and z are a cluster of features
2.9 NOTES ON SYLLABLES
A word has as many syllables as it has vowels. Each vowel can be preceded and followed by any
number of consonants. The superscript on the symbol “C” indicates the maximum number of
consonants (n=any number greater than 1); the subscript indicates the minimum number (0=none).
1. All one syllable words: # C n
0 V C n
0
2. All two syllable words: # C n
0 V C n
0 V C n
0 #
3. All three syllable words: # Cn 0 V Cn 0 V Cn 0 V Cn 0
4. Vowel in the last syllable (ULTIMA): / __ C n
0
5. Vowel in the next to last syllable (PENULT): / __ C n
0
6. Vowel in the third to last syllable (ANTEPENULT): / __ C n
0
VC n
0
VC n
0
VC n
0
#
#
#
#
#
89

90
VOWELS:
SUMMARY OF ENGLISH SOUNDS AND SPELLING
[i] [iy] [iw] beat, beet, Caesar, people, we, machine, receive, believe
[w] bit, into, been, abyss, fountain, button, language
[u] [uw] [u] boot, who, duty, through, too, to, two, new, nuisance, blue
[] put, wood, would, butcher, woman
[e] [ey] [ew] bait, fate, great, eight, rain, reign, rein, they, hay, gauge
[e] bet, dead, said, guest, many, friend, says, jeopardy
[c] butted, roses, alone, sofa, the
[o] [ow] [o] boat, so, sew, own, though, plateau, toe, soul, brooch
[]] bought, caught, chalk, saw, awful, author, sore, soar, four
[æ] bat, laugh, plaid
[] but, does, tough, flood, ton, country
[a] [Y] pot, palm, far, sergeant, ox
DIPHTHONGS:
[]y] []w] boy, soil
[ay] [aw] sigh, nice, guy, lie
[aw] [a] cow, house
GLIDES:
[w] wish, queen, weather, wear (for some, also when, whether, where)
[ ] when, whether, where
[y] [j] yes, beyond, use, feud
[h] high, whole
LIQUIDS:
[l] live, along
[ł] drill, willing
[r] [‹] rich, car, bury, furry
[D] [] water, later, latter, ladder, writer, rider, party

NASALS:
SUMMARY OF ENGLISH SOUNDS AND SPELLING
[m] mess, hymn, hum, swimming
[n] neat, kin, winning, pneumonia, knife
[ñ] [ny] [nj] onion, union
[õ] sing, think, tongue
STOPS:
[p h ] pain, prune, plaque, appear, shrimps
[p] spit, spun, spring, expose
[b] broom, crabs, rubber
[t h ] tip, tough, pterodactyl, intend, attack
[t] stick, stun, string, extend
[d] dig, bad, killed, heard, mending
[k h ] keep, cat, chord, quite, pickle, accommodate
[k] skit, scum, school, screw, excuse
[g] give, bag, egg, rogue
FRICATIVES:
[f] first, stuff, telephone, rough
[v] value, sliver, of
[›] thin, breath, teeth, ether, both, thing
[ð] then, breathe, teethe, either
[s] sin, this, cross, science, ceiling, race, psychology
[z] zoo, drizzle, was, xylophone, scissors
[š] [•] show, chute, flash, issue, conscious, sure, mansion, racial
[ž] [¥] measure, azure, prestige, illusion
AFFRICATES:
[č] [t•] chin, kitchen, question, righteous, much
[:9] [d¥] jig, judgment, ridge, gem, graduation, region
91

EXERCISES FOR CHAPTER TWO
1. Transcribe the following English words (Answers in Appendix A):
rich ridge ring wring wrist
bush butch budge box buzz
gem Jim sham chum gum
ice eyes east eased oozed
race raise rays rise rose
lose loose louse chose choose
cease seize seas says size
chef chief chic sheik shack
worth earth teethe bathe bath
them this thumb thing then
debt vex Scotch broth knot
maize quiche squish queue shrew
though cough rough bough through
heard word bird curd nerd

weird feared veered cared cord
meat great threat heart cart
chair jeer share shear cheer
wheel whale weigh who how
witch which watch wretch wedge
coat code root rude read
laud loud laid led lewd
knifed sighed prized healed shipped
view dew do poor pour
latch loaf roof tune pure
car oar her hear whore
George vague vast don’t they’re
oath both moth mouth myth
tit Tut tat toot tot
who’d wood would wooed hood
93

94
2. Transcribe the following English words (Answers in Appendix A):
music physics psychic sonic helix
passion cashier cushion caption cashmere
thistle thimble tumble dangle dinghy
champagne cartoon Khartoum careen gangrene
immune impugn impinge syringe obscene
racial racer razor sugar quota
pleasure pledger leisure azure seizure
scissors vision fusion ocean mission
beauty booty purely poorly pearly
quitter water rider writer raider
joining whining singing finger languor
cavil navel ravel gable squabble
Mary marry merry quarry query
numbskull zigzag homegrown deathblow shipshape
outboard marshlands whiplash pastime pitchfork
cutthroat whetstone hangman onslaught crabgrass

3. Transcribe the following words including placement of primary stress (´). You can ignore
aspiration and release. (v) = verb; (n) = noun.
agree debris ennui Pawnee marquee
angry money sunny gypsy litchi
survey (v) Bombay bouquet inveigh parfait
survey (n) bomb bay Tuesday foray subway
review canoe shampoo construe eschew
preview igloo curfew yahoo Zulu
below outgrow plateau Bordeaux although
bellow meadow hobo cocoa depot
outcry (v) apply goodbye sky high untie
outcry (n) ally cacti fish fry pigsty
soda tchotchke bwana chutzpah geisha
4. Transcribe the following words including placement of primary stress (´). You can ignore
aspiration and release. (v) = verb; (n) = noun.
broccoli apogee fricassee symphony pedigree
shillelagh Chianti bologna confetti Hatari
guarantee bourgeoisie chimpanzee Tennessee jamboree
95

96
runaway Saturday ricochet protégée workaday
matinée San José Chevrolet Santa Fé overplay
rendevous residue peekaboo honeydew avenue
Timbuktu kangaroo misconstrue overthrew hitherto
mistletoe Mexico embryo cameo Scorpio
fiasco concerto soprano falsetto crescendo
overflow (v) apropos status quo cheerio to-and-fro
amplify rockaby occupy prophesy underlie
subpoena agenda charisma pariah pagoda
5. Transcribe the following words including placement of primary stress (´). You can ignore
aspiration and release. (v) = verb; (n) = noun.
apostrophe Menominee Penelope fraternity telegraphy
Ypsilanti kamikaze Cincinnati insincerely poison ivy
pistachio Arapaho imbroglio adagio portfolio
armadillo Filipino virtuoso Sacramento lucky fellow
Arabia azalea Bulgaria et cetera America
hullabaloo Kalamazoo Tippecanoe

6. In each of the following, supply English words that contain the sound given in as many
different spellings as you can. Underline the letters representing the sound in the answer.
[f] Answer: physics, fish, enough
a. [i] Answer:
b. [r] Answer:
c. [k] Answer:
d. [š] Answer:
e. [y] Answer:
f. [e] Answer:
g. [o] Answer:
h. [:9] Answer:
I. [e] Answer:
j. []] Answer:
7. Supply the symbol corresponding to each of the following phonetic descriptions.
voiced velar stop consonant Answer: [g]
a. lax mid back rounded vowel Answer:
b. voiced alveolar liquid Answer:
c. voiceless palatal affricate consonant Answer:
d. tense low back unrounded vowel Answer:
e. voiceless alveolar fricative consonant Answer:
f. voiced palatal liquid Answer:
g. voiceless bilabial stop consonant Answer:
97

98
h. tense high front unrounded vowel Answer:
i. voiced velar nasal consonant Answer:
j. voiced labiodental fricative consonant Answer:
8. Write a phonetic description for each of the following symbols.
[g] Answer: voiced velar stop consonant
[o] Answer: tense mid back rounded vowel
a. [h] Answer:
b. [v] Answer:
c. [e] Answer:
d. [o] Answer:
e. [ž] Answer:
f. [›] Answer:
g. [æ] Answer:
h. [ñ] Answer:
I. [č] Answer:
j. [c] Answer:
9. In each of the following, all but one of the sounds listed exhaustively comprise a natural class
of sounds for English. Identify the sound that does not belong in the natural class AND name
the feature or features that define the class.
[k], [d], [g], [b] Choice: [k]
Class: [+VOICED, +STOP]
a. [a], [e], [], [æ] Choice:
Class:

. [k], [õ], [h], [g] Choice:
Class:
c. [z], [č], [ð], [ž], [v] Choice:
Class:
d. [t], [›], [z], [ð], [s] Choice:
Class:
e. [u], [w], [o], [e], [i] Choice:
Class:
f. [w], [h], [y], [§] Choice:
Class:
g. [u], [o], [a], [], []] Choice:
Class:
h. [m], [w], [v], [b], [p] Choice:
Class:
I. [u], [i], [e], [w], [] Choice:
Class:
j. [õ], [m], [n], [h], [ñ] Choice:
Class:
10. State the following rules using formal devices (Answers in Appendix A).
Example: A nasal consonant is deleted before another nasal consonant at the beginning of
a word.
Answer: [+CONSONANT, +NASAL] ÷ Ø / #___ [+CONSONANT, +NASAL]
a. All intervocalic voiceless stops become voiced.
99

100
b. Nontense vowels in the final syllable of a word are stressed when they are followed by
two or more consonants.
c. Delete all final consonants when the following word begins with a consonant.
d. Reduce all unstressed vowels to schwa.
e. Palatalize all alveolar stops before [y].
f. Devoice all final fricatives.
g. Aspirate all initial voiceless stops.
h. Nasalize all vowels before a nasal consonant.
i. A lax vowel in the antepenultimate syllable of a word is stressed.
j. Palatalize all alveolar stops before [y] when they are preceded by a vowel or when they
are initial.

11. In Classical Latin, nouns have different endings (CASES) depending on their use in a
sentence. For example, a noun functioning as the SUBJECT of a sentence has a
NOMINATIVE CASE ending. One functioning as a POSSESSIVE phrase has a GENITIVE
CASE ending. Consider the following, where a macron over a vowel means the vowel is long:
NOMINATIVE GENITIVE MEANING
arks arkis enclosure
duks dukis leader
daps dapis feast
rēks rēgis king
sēps sēpis snake
falanks falangis phalanx
urps urbis city
pleps plebis people
a. In terms of these data, what are the morphemes for the NOMINATIVE and the
GENITIVE?
b. What is the Latin morpheme for each of the following?
(1) enclosure
(2) leader
(3) feast
(4) king
(5) snake
(6) phalanx
(7) city
(8) people
c. State any phonological rule(s) applicable in the above Latin data.
101

102
12. Consider the following additional Classical Latin data:
NOMINATIVE GENITIVE MEANING
li)s li)tis strife
fraws frawdis deceit
frōns frōntis brow
frōns frōndis leaf
dens dentis tooth
sors sortis lot
a. What are the morphemes for the NOMINATIVE and the GENITIVE?
b. Give the Latin morphemes for each of the following:
(1) strife
(2) deceit
(3) brow
(4) leaf
(5) tooth
(6) lot
c. State any phonological rule(s) in addition to those from the previous problem.

APPENDIX A: ANSWERS TO TRANSCRIPTION EXERCISES
1. rich ridge ring wring wrist
[rwč] [rw:9] [rwõ] [rwõ] [rwst]
bush butch budge box buzz
[bš] [bč] [b:9] [baks] [bz]
gem Jim sham chum gum
[:9em] [:9wm] [šæm] [čm] [gm]
ice eyes east eased oozed
[ays] [ayz] [ist] [izd] [uzd]
race raise rays rise rose
[res] [rez] [rez] [rayz] [roz]
lose loose louse chose choose
[luz] [lus] [laws] [čoz] [čuz]
cease seize seas says size
[sis] [siz] [siz] [sez] [sayz]
chef chief chic sheik shack
[šef] [čif] [šik] [šik] [šæk]
worth earth teethe bathe bath
[wcr›] [cr›] [t h ið] [beð] [bæ›]
them this thumb thing then
[ðem] [ðws] [›m] [›iõ] [ðen]
debt vex Scotch broth knot
[det] [veks] [skač] [br]›] [nat]
maize quiche squish queue shrew
[mez] [k h iš] [skwwš] [k h yu] [šru]
though cough rough bough through
[ðo] [k h ]f] [rf] [baw] [›ru]
heard word bird curd nerd
[hcrd] [wcrd] [bcrd] [k h crd] [ncrd]
103

104
weird feared veered cared cord
[wird] [fird] [vird] [k h ærd] [k h ]rd]
meat great threat heart cart
[mit] [gret] [›ret] [hart] [k h art]
chair jeer share shear cheer
[čær] [:9ir] [šær] [šir] [čir]
wheel whale weigh who how
[wil] [wæl] [we] [hu] [haw]
witch which watch wretch wedge
[wwč] [wwč] [wač] [reč] [we:9]
coat code root rude read
[k h ot] [k h od] [rut] [rud] [rid]
laud loud laid led lewd
[l]d] [lawd] [led] [led] [lud]
knifed sighed prized healed shipped
[nayft] [sayd] [p h rayzd] [hild] [šwpt]
view dew do poor pour
[vyu] [du] [du] [p h ur] [p h ]r]
latch loaf roof tune pure
[læč] [lof] [ruf] [t h un] [p h yur]
car oar her hear whore
[k h ar] [or] [hcr] [hir] [hor]
George vague vast don't they're
[:9or:9] [veg] [væst] [dont] [ðer]
oath both moth mouth myth
[o›] [bo›] [m]›] [maw›] [mw›]
tit Tut tat toot tot
[t h wt] [t h t] [t h æt] [t h ut] [t h at]
who'd wood would wooed hood
[hud] [wd] [wd] [wud] [hd]

2. music physics psychic sonic helix
[myuzwk] [fwzwks] [sayk h wk] [sanwk] [hilwks]
passion cashier cushion caption cashmere
[p h æšcn] [k h æšir] [khšcn] [k h æpšcn] [k h æžmir]
thistle thimble tumble dangle dinghy
[›wscl] [›wmbcl] [t h mbcl] [dæõgcl] [diõgi]
champagne cartoon Khartoum careen gangrene
[šæmp h en] [k h art h un] [k h art h um] [k h crin] [gæõgrin]
immune impugn impinge syringe obscene
[wmyun] [wmp h yun] [wmp h wn:9] [scrwn:9] [absin]
racial racer razor sugar quota
[rešcl] [rescr] [rezcr] [šgcr] [k h woDc]
pleasure pledger leisure azure seizure
[p h ležcr] [p h le:9cr] [ležcr]/[ližcr] [æžcr] [sižcr]
scissors vision fusion ocean mission
[swzcrz] [vwžcn] [fyužcn] [ošcn] [mwšcn]
beauty booty purely poorly pearly
[byuDi] [buDi] [p h yurli] [p h ]rli] [p h erli]
quitter water rider writer raider
[k h wwDcr] [w]Dcr] [rayDcr] [rayDcr] [reDcr]
joining whining singing finger languor
[:9]yniõ] [wayniõ] [siõiõ] [fiõgcr] [læõgcr]
cavil navel ravel gable squabble
[k h ævcl] [nevcl] [rævcl] [gebcl] [skwabcl]
Mary marry merry quarry query
[mæri] [mæri] [meri] [k h wari] [k h wiri]
numbskull zigzag homegrown deathblow shipshape
[nmskl] [zwgzæg] [homgron] [de›blow] [šwpšep]
outboard marshlands whiplash pastime pitchfork
[awtb]rd] [maršlændz] [wwplæš] [p h æst h aym] [p h wčf]rk]
cutthroat whetstone hangman onslaught crabgrass
[kt›rot] [wetston] [hæõmæn] [ansl]t] [kræbgræs]
105

106
3. agree debris ennui Pawnee marquee
[cgrí] [dcbrí] [anwí] [p]ní] [markí]
angry money sunny gypsy litchi
[æ´ õgri] [m´ni] [s´ni] [:9ípsi] [líči]
survey (v) Bombay bouquet inveigh parfait
[scrvé] [bambé] [boké] [wnvé] [parfé]
survey (n) bomb bay Tuesday foray subway
[s´rve] [bámbe] [túzde] [f]´re] [s´bwe]
review canoe shampoo construe eschew
[rcvyú] [kcnú] [šæmpú] [kcnstú] [cšú]
preview igloo curfew yahoo Zulu
[prívyu] [w´glu] [kérfyu] [yáhu] [zúlu]
below outgrow plateau Bordeaux although
[bcló] [awtgró] [plætó] [b]rdó] [ólðó]
bellow meadow hobo cocoa depot
[bélo] [médo] [hóbo] [kóko] [dípo]
outcry (v) apply goodbye sky high untie
[awtkráy] [cpláy] [gdbáy] [skayháy] [cntáy]
outcry (n) ally cacti fish fry pigsty
[áwtkray] [æ´lay] [kæ´ktay] [fw´šfray] [pw´gstay]
soda tchotchke bwana chutzpah geisha
[sódc] [čáčkc] [bwánc] [h´tspc] [géšc]/[gíšc]
4. broccoli apogee fricassee symphony pedigree
[brákcli] [æ´pc:9i] [frw´kcsi] [sw´mfcni] [pédcgri]
shillelagh chianti bologna confetti Hatari
[šcléli] [kiyánti] [bclóni] [kcnféDi] [hctári]
guarantee bourgeoisie chimpanzee Tennessee jamboree
[gærcntí] [buržwazí] [čwmpænzí] [tencsí] [:9æmbcrí]
runaway Saturday ricochet protégée workaday
[r´ncwe] [sæ´Dcrde] [ríkcše] [prótc:9e] [w]´rkcde]
matinée San José Chevrolet Santa Fé overplay
[mætcné] [sænhozé] [ševrclé] [sæn(t)cfé] [ovcrplé]

endevous residue peekaboo honeydew avenue
[rándevu] [rézcdu] [píkcbu] [h´nidu] [æ´vcnu]
Timbuktu kangaroo misconstrue overthrew hitherto
[twmbktú] [kæõgcrú] [mwskcnstrú] [ovcr›rú] [hwðcrtú]
mistletoe Mexico embryo cameo Scorpio
[mw´sclto] [mékscko] [émbriyo] [kæ´ miyo] [sk]´rpiyo]
fiæsko concerto soprano falsetto crescendo
[fiyæ´ sko] [kcnčéDo] [scpráno] [f]lséDo] [krcšéndo]
overflow (v) apropos status quo cheerio to-and-fro
[ovcrfló] [æprcpó] [stæDcskwó] [čiriyó] [tucnfró]
amplify rockaby occupy prophesy underlie
[æ´mplcfay] [rákcbay] [ákycpay] [práfcsay] [´ndcrlay]
subpoena agenda charisma pariah pagoda
[scpínc] [c:9éndc] [kcrw´zmc] [pcráyc] [pcgóDc]
5. apostrophe Menomini Penelope fraternity telegraphy
[cpástrcfi] [mcnáncmi] [pcnélcpi] [frctérncDi] [tclégrcfi]
Ypsilanti kamikaze Cincinnati insincerely poison ivy
[wpsclæ´ nti] [kamckázi] [swnscnæ´Di] [wnscnsírli] [p]yzcnáyvi]
pistachio Arapahoe embroglio adagio portfolio
[pcstæ´ šiyo] [cræ´pcho] [embróliyo] [cdážiyo] [p]rtfóliyo]
armadillo Filipino virtuoso Sacramento lucky fellow
[armcdw´lo] [fwlcpíno] [vcrčuwózo] [sækrcménto] [lkifélo]
Arabia azalea Bulgaria et cetera America
[crébiyc] [czéliyc] [bclgæ´riyc] [etséDcrc] [cmérckc]
hullabaloo Kalamazoo Tippecanoe
[hclcbclú] [kæl̀ m̀ zú] [twpckcnú]
107

108
APPENDIX B: PHONOLOGICAL RULES
1. a. All intervocalic voiceless stops become voiced.
b. Voiceless stops become voiced when intervocalic.
c.
d.
e.
2. a. Nontense vowels in the final syllable of a word are stressed when they are followed by
two or more consonants.
b. Nontense vowels become stressed when followed by two or more consonants in the
final syllable of a word.
c. [–TENSE] ÷ [+STRESS] / _____________ # C n
2
3. a. Delete all final consonants when the following word begins with a consonant.
b. Consonants become zero when final and when the following word begins with a
consonant.
c. C ÷ Ø / _____________ C #
4. a. Reduce all unstressed vowels to schwa.
b. Unstressed vowels become schwa.
c. [–STRESS] ÷ [c]

5. a. Palatalize all alveolar stops before [y].
b. Alveolar stops become palatal before [y].
c.
d.
e.
6. a. Devoice all final fricatives.
b. Fricatives become voiceless when final (before #).
c.
d.
e.
7. a. Aspirate all initial voiceless stops.
b. Voiceless stops become aspirate when initial (after #).
c.
109

110
d.
e.
8. a. Nasalize all vowels before a nasal consonant.
b. Vowels become nasal before nasal consonants.
c. V ÷ [+NASAL] / _______ [C, +NASAL]
9. a. A lax vowel in the antepenultimate syllable of a word is stressed.
b. Nontense vowels become stressed in antepenultimate syllables.
c. [–TENSE] ÷ [+STRESS] / _______ C n
0 V C n
0 V C n
0 #
10. a. Palatalize all alveolar stops before [y] when they are preceded by a vowel or when they
are initial.
b. Alveolar stops become palatal before [y] when they are preceded by a vowel or by #.
c.
d.
e.

1. Data.
APPENDIX C: SLASH–DASH NOTATION
ROOT MEANING EXAMPLES
/skrwb/ write scribble /skrwb+cl/ [skrwbcl]
script /skrwb+t/ [skrwpt]
/fræg/ break fragment /fræg+ment/ [frægment]
refract /ri+fræg+t/ [rifrækt]
2. First Set of Rules.
a. [bt] ÷ [pt]
b. [gt] ÷ [kt]
3. Revision One.
4. Revision Two.
5. Revision Three.
[b] ÷ [p] / _________ [t]
[g] ÷ [k] / _________ [t]
111

112
APPENDIX D: PHONOLOGY PROBLEM
Data in phonetic transcription from the language Hip:
SINGULAR PLURAL MEANING
tip tiben girl
pit piden boy
kip kiben child
pik pigen police
tab taben teacher
bid biden dogs
gip giben car
dag dagen cat
tap _____ house
___ giden lions
1. What are the stop phonemes in Hip? Justify your answer with minimal pairs.
2. What is the phonemic form for children?
3. What is the phonetic form for houses?
4. What is the phonemic form for lions?
5. State the rules which accounts for the phonetic differences between the singular and the plural
forms where they occur?
6. What is the morpheme for PLURAL?
7. What is the morpheme for SINGULAR?

APPENDIX E: MORPHOLOGY PROBLEM
Consider the following phonetic transcriptions and translations of Blatin:
pelam I wash nores you work katim I jump
pelat he washes noret he works katis you jump
pelabas you washed norebam I worked katibam I jumped
pelabat he washed norebas you worked katibat he jumped
pelabim I will wash norebis you will work katibim I will jump
pelabis you will wash norebit he will work katibit he will jump
Identify the following morphemes:
first person (I) ______________________
second person (you) ______________________
third person (he) ______________________
present tense ______________________
past tense ______________________
future tense ______________________
What do the following mean:
pelas ______________________
norebim ______________________
How does one say the following in Blatin:
He will wash. ______________________
I work. ______________________
He jumps. _____________________
113

114
APPENDIX F: ENGLISH VOWEL SHIFT
1. e [–HIGH, –LOW, –BACK, +TENSE] [beð] [sen] [eksplen]
(bathe) (sane) (explain)
æ [–HIGH, +LOW, –BACK, –TENSE] [bæ›] [sænwDi] [eksplænctori]
(bath) (sanity) (explanatory)
2. o [–HIGH, –LOW, +BACK, +TENSE] [noz] [fon] [telcskop]
(nose) (phone) (telescope)
a [–HIGH, +LOW, +BACK, +TENSE] [nastrcl] [fanwk] [telcskapwk]
(nostril) (phonic) (telescopic)
3. i [+HIGH, –LOW, –BACK, +TENSE] [slip] [mitcr] [ridim]
(sleep) (meter) (redeem)
e [–HIGH, –LOW, –BACK, –TENSE] [slept] [metrwk] [ridempšcn]
(slept) (metric) (redemption)
4. u [+HIGH, –LOW, +BACK, +TENSE] [luz] [gus] [fud]
(lose) (goose) (food)
] [–HIGH, –LOW, +BACK, –TENSE] [l]st] [g]sliõ] [f]dcr]
(lost) (gosling) (fodder)
**************
5. ay [–HIGH, +LOW, +BACK, +TENSE] [wayd] [layn] [wgnayt]
(wide) (line) (ignite)
w [+HIGH, –LOW, –BACK, –TENSE] [ww›] [lwnicr] [wgwšcn]
(width) (linear) (ignition)
6. aw [–HIGH, +LOW, +BACK, +TENSE] [saw›] [cbawnd] [profawnd]
(south) (abound) (profound)
[–HIGH, +LOW, +BACK, –TENSE] [sðcrn] [cbndcnt] [profndcDi]
(southern) (abundant) (profundity)

(1) Root /æg/ ‘do’:
APPENDIX G: ENGLISH MORPHOLOGY
/PHONOLOGICAL FORM/ [PHONETIC FORM] SPELLING
a. /æg + ent/ > [e:9cnt] agent
1. Vowel Shift: æ > e in certain contexts
sanity/sane; profanity/profane; explanatory/explain; gratitude/grateful
2. Velar Softening: g > j in certain contexts
rigor/rigid; regal/regicide; allegation/allege; analogous/analogize
b. /æg + ent + I/ > [e:9cnsi] agency
president/presidency; democrat/democracy; pirate/piracy; regent/regency
c. /æg + t/ > [ækt] act
regal/direct; fragment/refract; segment/bisect; pigment/depict
d. /æg + t + wv/ > [æktwv] active
e. /æg + t + wv + wti/ > [æktwvcDi] activity
f. /æg + t + wv + wti + z/ > [æktwvcDiz] activities
g. /wn + æg + t + wv/ > [wnæktwv] inactive
h. /æg + t + wcn/ > [ækšcn] action
relate/relation; devote/devotion; omit/omission; complete/completion
I. /trænz + æg + t + wcn + æl/ > [trænzækšcncl] transactional
j /ri + æg + t + ]r + z/ > [riæktcrz] reactors
(2) Root /reg/ ‘rule’:
a. /reg + ent/ > [ri:9cnt] regent
1. Vowel Shift: e > I in certain contexts
serenity/serene; severity/severe; delicacy/delicious; metric/meter
2. Velar Softening: g > j in certain contexts
rigor/rigid; regal/regicide; allegation/allege; analogous/analogize
b. /reg + ent + I/ > [ri:9cnsi] regency
c. /di + reg + t/ > [dcrekt] direct
d. /di + reg + t + wcn + z/ > [dcrekšcnz] directions
e. /wn + reg + ul + ær + wti + z/ > [wregyulærwDiz] irregularities
regular/regulate; tabular/tabulate; circular/circulate; particular/particulate
115

116
(3) Root /fræg/ ‘break’:
a. /fræg + ment/ > [frægmcnt] fragment
b. /fræg + wl/ > [fræ:9cl] fragile
c. /ri + fræg + t/ > [rifrækt] refract
(4) Root /leg/ ‘law’:
a. /leg + æl/ > [ligcl] legal
b. /leg + ws + let + wcn/ > [le:9cslešcn] legislation
c. /wn + leg + wtwm + et/ > [wlc:9wDcmct] illegitimate
(5) Root /leg/ ‘choose’:
a. /I + leg + ebwl/ > [elc:9cbcl] eligible
b. /I + leg + t + wcn/ > [ilekšcn] election
c. /leg + end + æri/ > [le:9cndæri] legendary
(6) Root /seg/ ‘cut’:
a. /seg + ment/ > [segmcnt] segment
b. /seg + t + wcn/ > [sekšcn] section
c. /dis + seg + t/ > [daysekt] dissect
(7) Root /pwg/ ‘paint’:
a. /pwg + ment/ > [pwgmcnt] pigment
b. /di + pwg + t/ > [dipwkt] depict
c. /pwg + t + or + I + æl/ > [pwktoricl] pictorial
(8) Root /mæg/ ‘great, more’:
a. /mæg + swm + æl/ > [mækscmcl] maximal
b. /mæg + ws + ter + I + æl/ > [mæ:9cstericl] magisterial
c. /mæg + or/ > [me:9cr] major
(9) Root /]g/ ‘increase’:
a. /]g + ment/ > []gment] augment
b. /]g + t + wcn/ > []kšcn] auction
c. /wn + ]g + ur + et/ > [wn]gyuret] inaugurate
(10) Root /pæg/ ‘fasten, agree’:
a. /pæg + ent/ > [pæ:9cnt] pageant
b. /wn + pæg + t/ > [wmpækt] impact
c. /pro + pæg + et/ > [prapcget] propagate

CHAPTER THREE: SYNTAX
3.1 LINGUISTIC ARGUMENTATION REVISITED
Linguistics, like all sciences, is concerned with explaining natural phenomena, that is, things which
occur naturally in the world and which people can perceive with their senses. The particular
phenomenon investigated in linguistics is language. Linguists look at all human languages that
naturally occur in the world and attempt to understand and describe their properties. One property
that linguists have discovered is that all natural languages consist of an infinite number of sentences
which are constructed in specific ways following specific rules. For example, all the sentences in this
paragraph are English sentences constructed correctly in terms of the rules of English grammar. The
area of linguistics that is specifically concerned with sentence structure is called SYNTAX.
It is possible to look at the structure of sentences and provide a variety of descriptions to account for
the way those sentences are formed. These various descriptions are actually hypotheses about how
sentences are constructed, and like all scientists, linguists must justify their hypotheses with
independent, objective evidence. Let us consider an example of what such evidence looks like.
3.1.1 LINGUISTIC ARGUMENTATION: SENTENCE TYPES
In most English classes, teachers say that there are three types of complete sentences in English.
There are statements like She is careful, questions like Is she careful, and commands like Be careful,
as well as other examples like the following:
(1) a. Statements: She is careful. That man will try to win. They don’t drive too fast.
b. Questions: Is she careful? Will that man try to win? Don’t they drive too fast?
c. Commands: Be careful. Try to win. Don’t drive too fast.
English teachers also point out that all English sentences must have a subject, which is usually the
person or thing that performs the action in the verb. For example, the subjects for the statements and
questions in (1) are the italicized phrases in (2).
(2) a. Statements: She is careful. That man will try to win. They don’t drive too fast.
b. Questions: Is she careful? Will that man try to win? Don’t they drive too fast?
When we consider the commands in (1c), it appears that they have no subjects. Generally, English
teachers say that the subject of all commands is an understood you. In other words, (1c) is
understood as follows:
(3) Commands: (You) be careful. (You) try to win. (You) don’t drive too fast.
Actually, the statement, “the subject of all commands is an understood you,” is a hypothesis, and like
all hypotheses, it must be justified with evidence. When asked to justify this particular hypothesis
about commands, people often assert that commands are directed at the person (or persons) one is

118
talking to and that such a person is referred to with the pronoun you. The pronoun for the speaker,
called the FIRST PERSON, is I; the pronoun for the person spoken to, the addressee, called the
SECOND PERSON, is you. Since commands are addressed to the second person, the subject must
be the second person pronoun, you.
While this seems logical, it is not proof because it relies on the judgement and opinion of the person
making the assertion about the data. It is subjective. Since different people can view the same
phenomenon in different ways, what is needed as proof of the above description is independent,
objective evidence – evidence that is not based on the way in which a particular person views the
data.
To see how we can support a hypothesis with objective evidence, consider first the following
sentences paying particular attention to the italicized words which are called REFLEXIVE
PRONOUNS:
(4) a. I see myself.
b. The little girls dressed themselves.
c. Mary voted for herself.
Notice that the reflexive pronouns refer back to the subject, that is, in (4a) myself refers back to I, in
(4b) themselves refers back to the little girls, and in (4c) herself refers back to Mary. In fact, if the
reflexive pronoun does not agree with the subject, then the sentence is ungrammatical as follows:
(5) a. *I see yourself. (lack of agreement in person, first person versus second person)
b. *The little girls dressed herself. (lack of agreement in number, singular versus plural)
c. *Mary voted for himself. (lack of agreement in gender, female versus male)
Given the above, we can make the following statements about reflexive pronouns in English.
(6) a. Reflexive pronouns occur after the main verb and refer back to the subject.
b. Reflexive pronouns must agree with the subject in person, number, and gender.
Returning to the issue of commands, notice that the only reflexives that are possible in a command
are the second person reflexive pronouns, yourself and yourselves:
(7) a. Dress yourself/yourselves.
b. *Dress myself/himself/herself/ourselves/themselves.
Since reflexive pronouns must agree with the subject in person (cf. (6b)) and since the only reflexive
pronouns possible in commands are second person, then the subject of a command must be second
person, that is, the addressee(s) or you. This argument constitutes independent, objective evidence
for the assertion, “the subject of all commands is an understood you.” Notice that all native speakers
of English unconsciously know the facts listed in (6) even if they have never heard the term reflexive

pronoun. We know that is correct because native speakers utter sentences like (4) and never utter
ones like (5).
Like other scientists, linguists look for more than one argument to support their hypotheses; therefore,
let us explore the matter further.
Consider next tag questions, in which the subject of the main sentence is repeated in a question
tagged onto a statement, such as the following:
(8) a. He didn’t do it, did he?
b. They will go, won’t they?
c. *She wasn’t there, was he?
d. *I’ll be invited, won’t they?
Notice that tag questions after a command must contain the second person pronoun you; hence, the
subject of a command must be an understood you.
(9) a. Dress yourself, won’t you (please)?
b. Don’t bother me, will you?
c. *Dress yourself, won’t he?
d. *Don’t bother me, will she?
As a third proof, consider idioms like the following and notice that the POSSESSIVE PRONOUNS
(in italics) must agree with the subject of the sentence; if they don’t the sentence is ungrammatical:
(10) a. She gave the children her full attention.
b. The man lost his temper.
c. *She gave the children my full attention.
d. *The man lost your temper.
Notice that idioms like the above can only contain second person possessive pronouns when they
occur in commands; hence, the subject of commands must be second person or you:
(11) a. Give the children your full attention.
b. Don’t lose your temper.
c. *Give the children my full attention.
d. *Don’t lose her temper.
The three arguments above provide justification for the hypothesis, “the subject of all commands is
an understood you.” Such justification is not derived from the opinion of the person making the
119

120
assertion, that is, it is not subjective. Instead, it is derived entirely from the objective observation of
facts about English sentences as they occur naturally in everyday speech. The arguments can be
verified or falsified through additional examination of other data; therefore, they constitute
independent evidence to support the hypothesis.
3.1.2 LINGUISTIC ARGUMENTATION: PARTS OF SPEECH
When trying to describe some aspect of syntactic structure, linguists always rely on independent
criteria to support their analyses. Let us consider here another example with the following sentences:
(12) a. John looks Italian.
b. John speaks Italian.
These two sentences appear to have the same structure, that is, they appear to be made up of the same
types of phrases with the same parts of speech (noun, verb, etc.). In fact, native speakers know that
the word Italian is used differently in the two sentences. We can prove that by examining other data
like the following.
(13) a. John looks very Italian.
b. *John speaks very Italian.
(14) a. *John looks fluent Italian.
b. John speaks fluent Italian.
Native speakers would agree that (13a) is grammatical, but (13b) is not, and that (14a) is
ungrammatical while (14b) is perfectly well-formed. These facts indicate the word Italian is used
differently in the two sentences. Specifically, in (12a) it functions as an adjective, while in (12b) it
functions as a noun. Confirmation of this hypothesis comes from examining other words whose
function is unquestioned (adjective or noun) and conjoining those words with Italian in the two
contexts under investigation (after looks and speaks):
(15) a. John looks tall, handsome, Jewish, and Italian (so he’s right for the part in the movie).
b. John speaks Hebrew, Sanskrit, Italian, and several other languages.
(16) a. *John looks Hebrew, Sanskrit, Italian, and several other languages.
b. *John speaks tall, handsome, Jewish, and Italian.
A linguist must take great care when examining data. It is easy to be mislead, which is why linguists
search for as many arguments as they can. For example, consider the following:
(17) a. John looks more Italian (than Bill).
b. John speaks more Italian (than Bill).

Given these examples, it appears that Italian functions in the same way in the two sentences since
both sentences allow the comparative more. Further examination of other comparative-type
expressions reveals that, in fact, they are different as we have hypothesized:
(18) a. *John looks too much Italian.
b. John speaks too much Italian.
(19) a. John looks too Italian (to play that part in the movie).
b. *John speaks too Italian.
The additional data indicate that the grammaticality of both examples in (17) is only an apparent
falsification of our hypothesis that Italian functions differently in the two sentences in (12). There
must be some other fact which makes both of the examples in (17) good. In those cases, it is because
the comparative more can freely occur before both adjectives and nouns:
(20) a. John is more handsome, educated, courteous, etc. (than Bill).
b. John has more money, grandchildren, hair, etc. (than Bill).
3.1.3 LINGUISTIC ARGUMENTATION: PHRASES
Consider now one final example of linguistic argumentation using the following sentences:
(21) a. My niece hit the bus stop.
b. My niece heard the bus stop.
c. My niece saw the bus stop.
These sentences appear to involve the same succession of elements with only the verbs being
different (hit, heard, and saw):
(22) a. My niece - hit - the bus stop
b. My niece - heard - the bus stop
c. My niece - saw - the bus stop
Other data, however, indicate that the structure of these sentences is quite different. The difference
hinges on the way the word stop is analyzed. Notice that the word stop can be used either as a noun
(This bus makes too many stops) or a verb (The bus stopped at the corner). When stop functions as
a noun, other words can combine with it to form a COMPOUND, which is a phrase composed of two
or more elements that act like a single unit (bus stop, mail box, hair dryer, time machine, etc.). When
stop functions as a verb, it can be modified by an adverb (suddenly stop, quickly go, happily sing,
easily succeed, etc.). Thus, in (21a), the word stop is a noun and part of the compound bus stop; in
(21b), the word stop is a verb. Further, in (21c), either analysis is appropriate (noun or verb) so the
example is ambiguous.
121

122
We can support this analysis of the sentences in (21) with data like the following:
(23) a. *My niece hit the bus suddenly stop.
b. My niece heard the bus suddenly stop.
c. My niece saw the bus suddenly stop.
(24) a. My niece hit both the mail box and the bus stop.
b. *My niece heard both the mail box and the bus stop.
c. My niece saw both the mail box and the bus stop.
What the above sentences indicate is that, after the verb hit, the words bus and stop can form a
compound. Sentence (24a) is grammatical because the word and connects two compound nouns
(mail box and bus stop). On the other hand, sentence (23a) is ungrammatical because the word
suddenly separates the two elements of what is supposed to be a compound (cf. *mail probably box,
*hair thoroughly dryer, etc.).
After the verb heard in (23b), the words bus and stop do not form a compound and the word stop is
itself a verb. As a result, it can be modified by the adverb suddenly, and (23b) is grammatical. On
other hand, (24b) is ungrammatical under the ordinary interpretation of the verb hear and the objects
mail box and bus stop. Judgements of grammaticality are based on ordinary language in ordinary
circumstances, i.e., ones in which mail boxes and bus stops do not talk or make noise by themselves
though that is certainly imaginable. Significantly, (24b) only makes sense under such special
circumstances.
The example in (23c) is unambiguous allowing only the reading in which stop is a verb. As expected,
(24c) is also unambiguous having only the opposite reading that involves conjoining two compound
nouns (mail box and bus stop). Other data that support our analysis include the following:
(25) a. The bus stop was hit by my niece.
c. *The bus stop was heard by my niece. (Bus stops do not ordinarily make noise.)
c. The bus stop was seen by my niece. (unambiguous; stop must be a noun)
(26) a. *It was the bus that my niece hit stop. (Elements of a compound cannot be separated.)
b. It was the bus that my niece heard stop.
c. It was the bus that my niece saw stop. (unambiguous; stop must be a verb)

(27) a. *My niece hit the busses stop. (The first element in a compound cannot be plural.)
b. My niece heard the busses stop.
c. My niece saw the busses stop. (unambiguous; stop must be a verb)
All these examples add independent and objective support for our analysis of the phrase structure of
the sentences in (21).
3.1.4 LINGUISTIC ARGUMENTATION: STRATEGIES
The three examples of linguistic argumentation given here should be viewed as a model for how to
proceed in syntactic analysis. With these examples, we can formulate four strategies that can be used
not only for linguistic analysis, but also for the analysis of any problem. Here is a summary of these
four strategies.
(28) Strategy One: Proceed in a step-wise fashion.
It is often useful to scan data for patterns and regularities. If nothing appears to fall into place,
try to find some commonality in a few examples in the data and proceed from that starting
point to gradually incorporate other examples. Separate examples into mutually exclusive
classes defined by the same types of criteria, so that each of your examples can only fall into
one class. If some examples or some data don’t fit into the criteria, put them aside. It is better
to leave some examples out of your analysis than to fit them in by relaxing your criteria.
(29) Strategy Two: Search for generalizations.
An analysis says very little if it does not lead to generalizations; therefore, your description of
the data should result in a statement of some general principles. As above, it is better to leave
some elements out of an analysis if their inclusion requires giving up a generalization or
violating a well-established principle. Perhaps they are misunderstood cases or exceptions.
An analysis that describes most of the data with interesting, contrastive generalizations is better
than one that includes all the examples but says little of interest. A proposal that covers most
concerns with well-supported data is better than one which tries to include everything by
adding extra conditions and special cases.
(24) Strategy Three: Make your hypotheses and your proposals data-driven.
Your hypotheses and your proposals should arise from an examination of the data. Always
begin with the examples and show how they drive your descriptive apparatus, terminology, and
representations. Do not propose a hypothesis unless there are data to support it. Once you
make a proposal make sure your representations (symbols, discussion, etc.) always follow it.
The more you show how that data drive your proposals, the more robust your proposals will
be, so much so that your proposals become inevitable.
123

124
(25) When presenting your hypothesis or making your proposal, consider your audience.
You should choose your examples carefully, be consistent, and remove extraneous material
from your description. You should only mention what is essential to make your point, and
don’t introduce unnecessary and potentially distracting complexities. Your proposal should
be presented in such a way that it can be easily followed and understood. Presentation matters.
3.2 ORDERING CONSTRAINTS
With the above discussion about the nature of linguistic argumentation in mind, let us now turn to
the focus of this chapter, describing the structure of English sentences and developing a theory of
syntax. We begin by noting that sentences are not random strings of words. Rather, the words of
every sentence are arranged into units called PHRASES. Further, there is both a left to right
constraint on the order of phrases, called LINEAR ORDER, and a top to bottom constraint on the
order of phrases, called HIERARCHICAL ORDER. Learning English consists in part of mastering
these ordering constraints. Failure to observe the constraints on linear order can lead to
ungrammaticality at any time during language acquisition. In adult language, (30b) is
ungrammatical; in child language, (31b) in ungrammatical for the meaning given.
(30) a. The boy ran away.
b. *Boy the away ran.
(31) a. Daddy chair. (= That is daddy’s chair.)
b. *Chair daddy. (= That is daddy’s chair.)
In addition, failure to observe linear order can lead to meaning changes in both adult language (32)
and child language (33).
(32) a. John saw Bill.
b. Bill saw John.
(33) a. Mommy find. ( = Mommy come and find me; I’m hiding.)
b. Find mommy. ( = I will find mommy.)
Changes in hierarchical order often result in ambiguity, that is, multiple meanings. For example, the
phrase the Spanish history books can be interpreted like the Spanish math books or like the art history
books, as follows:

(34) a. John has the Spanish history books. (= the history books in Spanish)
John has the Spanish math books.
b. John has the Spanish history books. (= the books on Spanish history)
John has the art history books.
As we noted above, SYNTAX is the area of linguistics that is specifically concerned with sentence
structure, including the linear and hierarchical ordering of phrases. Formally, a PHRASE is a word
or group of words that functions as a unit, which means that the language treats it as a whole. Phrases
are identified by one or more of three syntactic processes mentioned in the Chapter One and reviewed
here.
3.2.1 TESTING HYPOTHESES: REFERENCE, OMISSION, PLACEMENT
REFERENCE is a binding relationship between phrases. Some phrases get their meaning only from
identification with other phrases. For example, in John hurt himself, the reflexive pronoun himself
gets its meaning from its identification with John. We say himself refers to or is bound to John, and
the relationship between the two is an instance of reference. Reference occurs in simple sentences
like (35), in complex sentences like (36), and between sentences like (37) and (38).
(35) a. The painter cut himself.
(The phrase himself refers to the phrase the painter.)
b. The painters painted each other.
(The phrase each other refers to the phrase the painters.)
(36) a. The portrait of himself that Rembrandt liked the most is the one he painted last.
b. After the children finished dinner, they brushed their teeth.
125

126
(37) a. Impressionist painters seem to use more blue than expressionist painters.
b. At least, I think they do.
(The phrase they refers to the phrase impressionist painters; the phrase do refers to
the phrase use more blue)
(38) a. The children should not be playing with the paints.
b. Why did you give them them?
(the first them refers to the children; the second them refers to the paints)
OMISSION is the deletion of a phrase. Repeated occurrences of phrases may be left out. In such
cases, the omitted phrase is understood to be identical to the given phrase. For example, one can say
John likes apples more than he likes bananas or John likes apples more than bananas.
PLACEMENT is the occurrence of a phrase in a particular position. For example, the adverb phrase
readily can be placed either before or after the verb it modifies: John readily accepted or John
accepted readily; however, it cannot be placed at the beginning of a sentence: *Readily, John
accepted.
The three syntactic processes of reference, omission, and placement are possible because sentences
are organized into phrases. The sentences of every human language contain phrases. Furthermore,
phrase structure is a feature of the earliest sentences used by children regardless of the language they
are learning. The exact character of phrases varies from language to language, but there is no known
language which lacks phrase structure. Therefore, phrase structure is not an arbitrary convention;
it must be rooted in human biology.
3.2.2 LANGUAGE AND MEMORY
The biological origins of phrase structure become clear when one examines the nature of human
memory. It is a well–known fact that human beings have very limited SHORT TERM MEMORY
(STM), which is the immediate memory for new information, such as a telephone number just
received from the operator. This contrasts with our virtually limitless LONG TERM MEMORY
(LTM), which is the storage and recall of information that is not new, such as one’s mother’s maiden
name.
Generally, STM is thought to be limited to about seven (±2) bits of information. STM can be
facilitated by organizing information into units called CHUNKS. For example, the string of numbers
in (39a) is much more readily recalled when given as (39b).
(39) a. 1–4–9–2–2–0–0–1–1–7–7–6–1–9–9–7
b. 1492 – 2001 – 1776 – 1997

Given the reality of the chunk in psychological studies and the phrase in linguistic studies, it seems
plausible to derive (40).
(40) The sentences of a natural language consist of phrases because human STM requires
organizing a string of words into chunks to facilitate processing.
In short, a phrase is a chunk of a sentence. Syntax is the study of phrases. A linguist specifies phrase
structure by examining the units that participate in reference, omission, and placement. This
methodology is theory–independent; in fact, this is the method by which linguists develop theories
of grammar. We can see this if we examine the development of grammatical theory in this century,
specifically, the development from traditional grammar to structural grammar to generative grammar.
3.3 TRADITIONAL AND STRUCTURAL GRAMMAR
The oldest grammatical model in western civilization is TRADITIONAL GRAMMAR, a
grammatical model that has evolved from the language studies of the Greeks and Romans. Such
grammars contain definitions like (41).
(41) A SENTENCE is a structured expression consisting minimally of a SUBJECT and a
PREDICATE. The subject specifies the topic of the sentence, and the predicate specifies what
is asserted of the topic. Predicates always contain a VERB, which is a word that serves to
express an action, occurrence or state of being; sometimes predicates also contain an OBJECT,
which specifies the person or thing directly affected by the action defined in the verb. Since
they must at least contain a verb, predicates are also referred to as VERB PHRASES. Subjects
and objects are phrases that contain a NOUN, which is the name of a person, place, or thing,
and a DETERMINER (DET), which is a word like the, this, or that.
According to these definitions, a sentence like (42a) might be graphically represented by the TREE
DIAGRAM (42b).
(42) a. The passengers obey the stewardess.
b.
127

128
While (42b) appears to be a fairly adequate representation of the phrases in (42a), there are problems.
The first concerns an auxiliary or helping verb like must when it appears in a similar sentence like
The passengers must obey the stewardess. (41) does not tell us what to do with it, in particular, it
does not tell us whether an auxiliary is part of the SUBJECT node or the SENTENCE node or the
VERB PHRASE node. This is important because we would like to be able to devise an explicit set
of rules that describe the phrase structure of all English sentences. Since the left to right order of the
elements in a sentence must be considered, we have three possibilities for connecting an auxiliary
(AUX) with the rest of the sentences. We can represent the choices as follows, where the dotted lines
indicate the possible attachments of AUX:
(43) a. The passengers must obey the stewardess.
b.
If AUX is part of the SUBJECT, then the tree diagram for an English sentence is (44a), and the
PHRASE–STRUCTURE RULES (PS–RULES) that describe English sentences are (44b), where
the arrow can be interpreted as ‘consists of.’
(44) a.
b. (1) SENTENCE ÷ SUBJECT + VERB PHRASE
(2) SUBJECT ÷ DET + NOUN + AUX
(3) VERB PHRASE ÷ VERB + OBJECT
(4) OBJECT ÷ DET + NOUN

If AUX is part of the VERB PHRASE, then the tree diagram for an English sentence is (45a), and
the PS–RULES that describe English sentences are (45b).
(45) a.
b. (1) SENTENCE ÷ SUBJECT + VERB PHRASE
(2) SUBJECT ÷ DET + NOUN
(3) VERB PHRASE ÷ AUX + VERB + OBJECT
(4) OBJECT ÷ DET + NOUN
Lastly, if AUX is part of the SENTENCE, then the tree diagram for an English sentence is (46a), and
the PS–RULES that describe English sentences are (46b).
(46) a.
b. (1) SENTENCE ÷ SUBJECT + AUX + VERB PHRASE
(2) SUBJECT ÷ DET + NOUN
(3) VERB PHRASE ÷ VERB + OBJECT
(4) OBJECT ÷ DET + NOUN
Traditional grammarians were not concerned much with the objective justification of the statements
they made about grammar. The choice between analyses like the above was often, quite simply, a
matter of opinion. All modern theories of grammar take a very different point of view. All modern
theories of linguistics (those of the twentieth century) attempt to ground their analyses in terms of
129

130
structural principles which can be justified, that is, proven. The further into the twentieth century one
gets in the development of grammatical theory, the more stringent the requirement for proof becomes.
To determine which of the above sets of rules is correct, we can use the processes of reference,
omission, and placement. Consider first (44), the hypothesis that AUX is part of SUBJECT. A
well–known fact about English is that subjects and objects can be replaced with pronouns like they,
them, she, her, etc. The use of pronouns is very common in language: by replacing a phrase with a
pronoun we do not have to keep repeating the whole phrase. Instead of saying The stewardess told
the passengers that the passengers had to obey the stewardess, we say The stewardess told the
passengers that they had to obey her. Rules of reference tell us what they and her refer to.
Furthermore, specific pronouns must be used in specific positions: subjects must be replaced with
pronouns like I, he, she, we, they, etc., and objects must be replaced with pronouns like me, him, her,
us, them, etc. Thus, alongside of The passengers must obey the stewardess, we have sentences like
(47), but none like (48).
(47) a. They must obey her.
b. She must obey them.
(48) a. *Them must obey she.
b. *Her must obey they.
Following (44), we produce a fully grammatical sentence when we attempt to substitute the
SUBJECT node, which contains the AUX, and the OBJECT node with the appropriate pronouns:
(49) a. [The passengers must] obey [the stewardess].
b. [They] obey [her].
However, if the nouns in SUBJECT and OBJECT position of (44) are reversed as in (50a), then
substitution of SUBJECT and OBJECT with pronouns produces the ungrammatical (50b).
(50) a. [The stewardess must] obey [the passengers].
b. *[She] obey [them]. (cf. She obeys them).
The ungrammaticality of (50b) tells us that the analysis in (44) is wrong. If the rules were correct,
then we would not get an ungrammatical sentence when we applied them in various circumstances.
We therefore must reject (44). Intuitively, (44) seems incorrect for another reason. Subjects and
objects appear to be the same kinds of phrase; yet, (44) claims that AUX is part of SUBJECT.
Clearly, AUX can never be part of OBJECT: *They obey the stewardess must. If subjects and objects
are the same kinds of phrase, we ought to be able to get AUX in either. As a result, we must reject
(44) which means that our alternatives are (45) or (46). To help us choose between (45) and (46),
let us consider the following data:

(51) a. Generally the passengers must obey the stewardess.
b. *The generally passengers must obey the stewardess.
c. The passengers generally must obey the stewardess.
d. The passengers must generally obey the stewardess.
e. *The passengers must obey generally the stewardess.
f. *The passengers must obey the generally stewardess.
g. The passengers must obey the stewardess generally.
(52) a. *Strictly the passengers must obey the stewardess.
b. *The strictly passengers must obey the stewardess.
c. *The passengers strictly must obey the stewardess.
d. The passengers must strictly obey the stewardess.
e. *The passengers must obey strictly the stewardess.
f. *The passengers must obey the strictly stewardess.
g. The passengers must obey the stewardess strictly.
These data indicate that there are two broad classes of adverbs in English: SENTENCE ADVERBS
like generally that modify the whole sentence, and MANNER ADVERBS like strictly that modify
the verb phrase. Only the diagram in (46) and the rules in (46b) capture this classification, that is,
we arrive at a very efficient characterization of the two classes of adverbs with the following logic:
(53) a. If the rules in (46b) specify sentence structure, then we can say that sentence adverbs
occur around the three major sentential elements, namely, the four positions indicated
by the pound sign as follows:
SENTENCE ÷ # SUBJECT # AUXILIARY # VERB PHRASE #
131
b. If the rules in (46b) specify verb phrase structure, then we can say that manner
adverbs occur either at the beginning or the end of the verb phrase, namely, the two
positions indicated by the pound sign as follows:
VERB PHRASE ÷ # VERB + OBJECT #
The diagram and rules in (45) do not lead to such an efficient characterization. In short, the
PLACEMENT of adverbs proves what the units in sentences are. Furthermore, these rules predict
that (54a) is grammatical, and that (54b) is not.
(54) a. The passengers must generally strictly obey the stewardess.
b. *The passengers must strictly generally obey the stewardess.
There are sentences that corroborate the choice of (46). Notice that the VERB PHRASE node in
(46a) is substantiated by its possible OMISSION in a sentence like The passengers must obey the
stewardess if the pilot orders them to (obey the stewardess). If the VERB PHRASE included AUX,
as in (45), then repetition of the VERB PHRASE would produce *The passengers must obey the

132
stewardess if the pilot orders them to must obey the stewardess. Again, following the rules logically
we see that they are confirmed or rejected by the sentences that result, specifically whether the rules
lead to other sentences that are grammatical or ungrammatical.
As final confirmation of (46), consider the use of do it in a sentence like The pilot told the passengers
that they must obey the stewardess and actually had no trouble convincing them to do it (do it = obey
the stewardess). Again, if the AUX were part of the VERB PHRASE, then do it would have to mean
must obey the stewardess not obey the stewardess. Indeed, there is no pronominal form like do it in
English that stands for a VERB PHRASE that contains an AUXILIARY. If one existed, we would
have evidence supporting (45).
As a result of these considerations, we prove that English sentences contain three basic units;
therefore, the analysis in (46) is justified. Stated this way, a grammatical description is not a matter
of opinion. It is a matter of fact. Furthermore, we attain a very high level of adequacy by basing our
description on the concept of phrase, since phrase structure is the result of chunking, and chunking
is the result of biological constraints on human memory. We have a direct connection between the
nature of man and the nature of grammar.
As we saw in the Chapter One, there are three levels of adequacy that a linguist can attain in
grammatical analysis. The first level, called OBSERVATIONAL ADEQUACY, is attained when
the facts are noted. The second level, called DESCRIPTIVE ADEQUACY, is attained when the
facts are described with generalized principles or rules, e.g., (46). The third and highest level, called
EXPLANATORY ADEQUACY, is attained when those principles are related to the nature of the
language apparatus of man. As a linguist moves from observational to explanatory adequacy, an ever
higher level of generality is reached. At the highest level, linguists begin to relate individual
languages like English to human language in general and, therefore, to offer a real explanation for
linguistic phenomena. In attempting to attain explanatory adequacy, a linguist is also able to separate
out what is arbitrary and language specific, from what is predictable in terms of the nature of the
human language apparatus (the brain and the vocal and auditory organs).
Over a period of many years, traditional grammar gave way to STRUCTURAL GRAMMAR
because the latter could provide more descriptively adequate characterizations of natural language.
Traditional definitions such as those in (41) have been replaced by PS–RULES like those in (46b).
Thus far, we have justified a model of English which contains tree structures like (46a). However,
there are improvements that can be made to (46) that will allow us to attain a higher level of
descriptive adequacy. Notice that any word or group of words that can be a subject of an English
sentence can also be an object (with a meaning change, of course):
(55) a. The handsome prince rescued the beautiful princess.
b. The beautiful princess rescued the handsome prince.
(56) a. The new tenor from Italy met the members of the orchestra.
b. The members of the orchestra met the new tenor from Italy.

A categories like VP (Verb Phrase) or NP (Noun Phrase) is a STRUCTURAL CATEGORY. It
abbreviates a class of syntactic structures that can occur in certain positions in a syntactic
HIERARCHY. On the other hand, a category like subject or object is a FUNCTIONAL
CATEGORY, which names what function or role a particular structure plays in the sentence. Thus,
in (56a) and (56b), the new tenor from Italy is structurally a noun phrase (NP), consisting of the noun
tenor preceded by the determiner the and the adjective new, and followed by the prepositional phrase
from Italy; functionally, this NP is the subject of (56a) and the object of (56b).
This additional information about English syntax, and more specifically about the sentence in (42a),
can be expressed if the tree diagram in (42b) is replaced with the one in (57), an entirely structural
representation (S = Sentence; NP = Noun Phrase; VP = Verb Phrase; N = Noun; V = Verb):
(57)
3.4 TENSE AND THE INTERNAL STRUCTURE OF SENTENCES
The representation in (57) describes the structure of the sentence The passengers must obey the
stewardess. Naturally, our rules must do more than describe the structure of just one sentence since
there is an infinite number of sentences in every language. Also, some VP do not contain direct
objects and some NP do not contain determiners:
(58) a. The passengers can deplane.
b. Passengers must obey stewardesses.
These two facts can easily be accounted for by stipulating that direct object NPs and determiners are
optional. Borrowing the parenthesis notation we used in phonology to indicate optional elements,
we revise the PS-Rules as follows (cf. (46b)):
133

134
(59) a. S ÷ NP + AUX + VP
b. VP ÷ V + (NP)
c. NP ÷ (DET) + N
Other facts about English cannot be so easily accommodated. Consider the following sentences
which appear to contain no AUX:
(60) a. The passenger obeys the stewardess.
b. The passenger obeyed the stewardess.
We might say that such sentences indicate that the AUX should be optional in (59a); however, this
fails to account for the fact that the sentences in (60) have a tense. Example (60a) is in the present
tense (PRS); (60b) is in the past tense (PST). Furthermore, there can be only one tense in each
sentence:
(61) a. The passenger does obey the stewardess.
b. The passenger did obey the stewardess.
c. *The passenger does obeys the stewardess.
d. *The passenger did obeyed the stewardess.
e. *The passenger does obeyed the stewardess.
f. *The passenger did obeys the stewardess.
Summarizing, it is clear that all native speakers of English know the following facts about tense and
the internal structure of sentences:
(62) a. Every sentence (S) must contain a tense.
b. There is only one tense in each S.
c. If there is no auxiliary verb, then tense is marked on the main verb (cf. (60)).
d. If there is an auxiliary verb, then tense is marked on the auxiliary verb (cf. does in
(61a) and did in (61b)).
Of course, the knowledge that native speakers have of these facts is unconscious. That much is clear
because they observe the facts when they produce new sentences, that is, they never say things like
(61c) - (61f). The facts are part of the grammar of English that native speakers have internalized.
The facts exist whether or not there are people around who want to write the facts down as we are
trying to do. When we write down the grammatical facts, all we are doing is bringing those facts to
consciousness. We are not making them up.

Additional facts about tense that native speakers know are the following:
(63) a. Two tense markers are signaled by a change in the shape of the verb: present tense
(PRS) and past tense (PST).
[1] The passenger obeys the stewardess.
[2] The passenger obeyed the stewardess.
[3] The passenger does obey the stewardess.
[4] The passenger did obey the stewardess.
b. One tense marker is signaled by the bare verb stem alone: the imperative tense (IMP),
which is used to express commands.
[1] Obey the stewardess.
[2] Do (not) obey the stewardess.
c. Two other tense markers are signaled by preceding the main verb with an auxiliary
verb called a MODAL: future tense (FUT) and conditional tense (CND).
[1] The passenger will/shall obey the stewardess.
[2] The passenger could/should obey the stewardess.
Since a tense is obligatory and a modal is optional, we should replace (59) with the following where
TP stands for Tense Phrase:
(64) a. S ÷ NP + TP + VP
b. VP ÷ V + (NP)
c. NP ÷ (DET) + N
d. TP ÷ (MOD) + T
e. T ÷ {PRS, PST, IMP, FUT, CND}
In PS–Rules like the above, the PARENTHESES indicate OPTIONAL ELEMENTS and the
BRACES (CURLY BRACKETS) indicate ALTERNATIVE ELEMENTS. These notations, of
course, are the same ones we saw above in Chapter Two in P-Rules (Phonological Rules).
The symbol T stands for TENSE, which is used here as a special abbreviation for the various TIME
and MODE/MOOD distinctions in English. The five tenses in English include the three TIME
distinctions, PRESENT (PRS), PAST (PST), and FUTURE (FUT), and the two MODE/MOOD
135

136
distinctions, CONDITIONAL (CND), and IMPERATIVE (IMP). We have examples like the
following:
(65) a. She lives in a dorm. (lives is present tense, PRS)
b. She lived in a dorm. (lived is past tense, PST)
c. She will live in a dorm . (will is future tense, FUT)
d. She should live in a dorm. (should is conditional mode, CND)
e. Live in a dorm. (live is imperative mode, IMP)
(66) a. She goes to college. (goes is present tense, PRS)
b. She went to college. (went is past tense, PST)
c. She will go to college. (will is future tense, FUT)
d. She should go to college. (should is conditional mode, CND)
e. Go to college. (go is imperative mode, IMP)
Notice that some of these distinctions are not realized as individual words, but as inflectional endings,
that is, changes in the shape of the verb (lives, lived, etc.). Further, in English, inflections aren’t
always an identifiable element at the end of a word: they can appear as modifications of the root or
as nothing at all. For example, went is the past tense of go; will is the modal will in future time;
should is the modal shall in the conditional mode.
In English, a MODAL (MOD) is a word that colors the meaning of the main verb and expresses
notions like possibility (can, may), necessity (must, should), potentiality (might, could), likelihood
(may), obligation (must, should), and so on. The most common modals are can, could, shall, should,
will, would, may, might, must, do, does, did, etc. These words are actually composite forms, e.g.,
does is PRS of do, did is PST of do, etc.
Here is a summary and some other examples of regular and irregular verbs in various tenses:
(67) PRS Singular PRS Plural PST Singular PST Plural FUT
a. obeys obey obeyed obeyed will obey
b. lives live lived lived will live
c. goes go went went will go
d. becomes become became became will become
e. writes write wrote wrote will write
f. dives dive dove/dived dove/dived will dive
g. is are was were will be
h. does do did did will do
i. has have had had will have

3.5 DETERMINERS AND THE INTERNAL STRUCTURE OF NOUN PHRASES
Continuing our investigation of the phrase structure of English, let us now consider the internal
structure of noun phrases like the following (the NPs are in italics):
(68) a. Passengers obey stewardesses.
b. Those passengers obey these stewardesses.
c. All the passengers obey all the stewardesses.
d. All the nervous passengers obey the stewardesses.
(69) a. *Those all passengers obey the stewardesses.
b. *Nervous the passengers obey the stewardesses.
Again, it is clear that native speakers unconsciously know a variety of facts about the internal
structure of NP because they will utter sentences like (68), but not ones like (69). Among other facts,
native speakers know that the left to right order of the elements in an NP must be as follows:
(70) NP ÷ (QNT) + (DET) + (ADJ) + N
In this rule, QNT stands for quantifier (a word like all, both, or each ), DET stands for determiner
(a word like the, this/these, or a/an), and ADJ stands for adjective (a word like nervous, tall, or
happy). Native speakers know which category (QNT, DET or ADJ) each word belongs to because
they always put them in the proper order which, again, is why they utter sentences like (68) but not
ones like (69). They also know exactly how the categories differ from each other, for example, that
only adjectives can be compared:
(71) a. tall, taller tallest
b. nervous, more nervous, most nervous
c. all, *aller, *allest
d. these, *theser, *thesest
In addition to the above, native speakers also know that there are two broad categories of nouns in
English: PROPER NOUNS, which refer to a specific entity (Heather, Daisy, Mary, Michigan,
Oakland University, etc.), and COMMON NOUNS, which are general in their reference (heather,
daisy, girl, state, education, university, etc.).
The major distinction between the two categories of nouns is that proper nouns cannot be specified
by a determiner (*this Michigan, *that Oakland University, etc.), whereas common nouns can (this
state, that university, etc.). Further, an overt determiner is obligatory for singular COUNT NOUNS,
which are nouns that can be counted (one daisy, two daisies, many daisies, etc.), but optional for
plural count nouns and for non-count nouns or MASS NOUNS, which can not be counted (*one
heather, *two heathers, *many heathers, etc.). Here is a summary of the relevant facts (notice the
capital letters on the proper nouns):
137

138
(72) Noun Class Overt Determiner No Overt Determiner
a. Proper: *I like the Heather and the Daisy. I like Heather and Daisy.
b. Singular Count: I like the daisy. *I like daisy.
c. Plural Count: I like the daisies. I like daisies.
d. Non-count (Mass): I like the heather. I like heather.
As we can see above, proper nouns are usually capitalized in writing; however, that “clue” is missing
in speech. Still, the entire classification of nouns given here is unconsciously known by all native
speakers, which is why they only use determiners when they should in speech. Remember that young
children do not have access to things like the rules for capitalization; they speak the language long
before they learn to read and write. The unconscious knowledge they acquire is our focus. Our
grammar must account for that knowledge to be descriptively adequate.
3.5.1 ABSTRACT ELEMENTS: Ø
As the above description indicates, the use of determiners in noun phrases is complex and subtle.
People learning English as a second language (ESL) often make mistakes in using determiners,
putting them in where they shouldn’t occur and leaving them out where they should occur. This is
especially true of people whose native language has no obligatory determiners, such as Japanese.
Adult native speakers of English never make such mistakes. They don’t say things like *I saw child
or *I live in the Michigan. Importantly, the reason that native speakers do not say such things is not
because they have never heard anyone say such things. The number of sentence is infinite so a
speaker could not have heard everything. The reason that native speakers use determiners correctly
is because they have unconsciously internalized the facts about determiners described above. They
unconsciously know the rules. They have not memorized all the good sentences.
Students new to linguistics are often overwhelmed by the multitude of facts that linguists discuss
about grammar. But students need to remember that they unconsciously already know those facts.
For example, they know not to say things like *I saw child or *I live in the Michigan. Students do
have to learn the terminology that linguists use, such as “proper noun,” but they do not have to learn
the facts; they only need to learn how to recall the facts. Thus, a major focus of this book is on
learning to ask the right questions to become consciously aware of the facts one already knows
unconsciously. This is not as easy as it sounds. If it were simple, linguists would have uncovered
all the facts of grammar long ago.
As we have observed, the basic distinction in the above classification of nouns is that proper nouns
cannot be specified by a determiner and common nouns can:
(73) a. *I like the Heather and the Daisy.
b. I like the heather and the daisy.

The grammar we are developing must specify the distinction between proper and common nouns in
some explicit way. Most grammars of English account for that distinction by postulating a
phonologically null determiner, represented by the symbol Ø. Grammarians argue that this
determiner is present when it appears that a common noun does not have a determiner. The full
analysis is as follows:
(74) a. I like the heather. [ay layk ðc heðcr]
b. I like Ø heather. [ay layk heðcr] - phonetically equivalent to (74f)
c. *I like heather. (ungrammatical because there is no determiner before heather)
d. *I like the Heather.
e. *I like Ø Heather. (ungrammatical because of the presence of Ø before Heather)
f. I like Heather. [ay layk heðcr] - phonetically equivalent to (74b)
The example in (74c) is ungrammatical because heather, a common noun, requires Ø when there is
no overt determiner (cf. (74b)). The example in (74e) is ungrammatical because Heather, a proper
noun, cannot be specified by a determiner, including the phonologically empty determiner Ø (cf.
(74f)). The bottom line here is that our grammar must account for the fact that native speakers know
the difference between thousands of proper nouns like Heather, Daisy, Bill, John, Grace, Jack,
Dawn, etc., and thousands of common nouns like heather, daisy, bill, john, grace, jack, dawn, etc.
Note that (74b) and (74f) are spoken and heard in exactly the same way. Postulating Ø enables us
to distinguish the two types of nouns formally.
Summarizing, the above analysis says this: proper nouns never have a determiner; common nouns
always have a determiner; if a common noun lacks an overt determiner, then Ø is present.
The above analysis extends to possessives (Heather’s, women’s, her, etc.) and to quantifiers (all,
both, etc.). In English, an NP can contain either a possessive or a determiner, but not both – another
fact that native speakers unconsciously know. Further, an NP can only contain a quantifier if it also
contains a determiner. Consider the following, where POS represents the possessive suffix
(apostrophe) and notice that POS is in the same position as determiners, including Ø:
(75) a. all the children
b. all Ø children
c. *all POS children (*all’s children)
d. *Heather the children
e. *Heather Ø children
f. Heather POS children (Heather’s children)
g. *her the children
h. *her Ø children
i. her POS children (her children)
139

140
A possessive pronoun like her is actually a composite of she and the possessive morpheme (POS),
just as went is a composite of go and the past tense morpheme (PST) and did is a composite of do and
the past tense morpheme (PST) as we saw above.
Again, all the above facts about noun phrases must be known by native speakers, at least
unconsciously, because they know which nouns can be specified by determiners and which can’t.
It is worth repeating here that the facts are facts in the grammar of English. Students often think that
grammatical facts and rules are things that teachers make up. That is incorrect. Languages have
grammar whether or not there are teachers to talk about the grammar. There is no human language
that does not have a grammar because languages require grammar. Grammar is, as we noted in the
beginning of this chapter, a biological necessity.
What we are trying to do here is to account for the grammatical facts native speakers know in the
most generalized way possible, an objective of all science. All of the revisions we have made in our
formal description of English are DATA-DRIVEN: we observe the data, discover the facts in
English grammar, and then adjust our rules in such a way that our written, descriptive grammar
accounts for the data and the facts (see Chapter One, Section 1.2). Linguists postulate revisions in
their grammatical descriptions and rules only when there are data to support those revisions.
Now notice that possessives themselves are noun phrases; in the following examples people is
modified by the adjective wealthy and donations is modified by the adjective generous:
(76) a. the people’s generous donations
b. those wealthy people’s generous donations
c. those very wealthy people’s very generous donations
d. their very generous donations
Furthermore, the possessive marker (the apostrophe) is a separate element affixed to the end of the
entire possessive noun phrase:
(77) a. the people’s donations [the people]’s donations
b. the people in that room’s donations [the people in that room]’s donations
c. *the people’s in that room donations
Lastly, as we have seen, determiners and possessives cannot specify the same noun:
(78) a. those donations
b. the people’s donations
c. their donations
d. *those the people’s donations *the people’s those donations
e. *those their donations *their those donations
In other words, determiners and the possessive occupy the same position in noun phrases. A
quantifier like all can only be followed by a determiner (all the books, *all’s books), and an ordinary

noun like John or child can only be followed by the possessive marker (John’s book, that child’s
books, *John the book, *that child the books (cf. (75)). The facts require us to modify (70), repeated
here as (80), to (81).
(80) NP ÷ (QNT) + (DET) + (ADJ) + N
(81) NP ÷ ( {QNT, NP} ) + (DET) + (ADJ) + N
A cluster like ( {QNT, NP} ) in above rule, means that either a QNT or an NP can optionally occur
in the first position of an NP.
Continuing to explore the internal structure of noun phrases, we note that noun phrases can contain
whole sentences:
(82) a. John heard a rumor.
b. John heard (that) Mary kissed Bill.
(83) a. A rumor upset John.
b. That Mary kissed Bill upset John.
The direct object of (82a) is a rumor, and the direct object of (82b) in the sentence Mary kissed Bill,
which must be a sentence because it contains a tensed verb (kissed).
To accommodate all the varieties of NP that occur in the above examples, we require the following
PS-Rule for NP:
(84) NP ÷
The rule (84) says that an NP has one of two forms: either a noun, which can be preceded by some
optional elements, or a sentence. Thus, (84) is an abbreviation for all the rules in (85).
(85) a. NP ÷ QNT + DET + ADJ + N
b. NP ÷ QNT + DET + N
c. NP ÷ QNT + ADJ + N
d. NP ÷ QNT + N
e. NP ÷ DET + ADJ + N
f. NP ÷ DET + N
g. NP ÷ ADJ + N
h. NP ÷ N
i. NP ÷ S
Again, observe that the left to right order in PS–RULES is crucial: it expresses the fact that the NP
in (86) are grammatical while those in (87) are not:
141

142
(86) a. the new tenor from Italy
b. the tenor
c. tenors from Italy
(87) a. *the from Italy new tenor
b. *tenor the
c. *from Italy tenors
3.5.2 PHRASE STRUCTURE RULES: FIRST PROPOSAL
Expanding our corpus, we can allow for more complex phrases such as those in (88) if we make the
additions to our rules specified in the right-hand column (note that (88d) is ambiguous).
(88) a. He will not go. Add a negative (NEG) to the TP rule.
b. Mary quickly finished the assignment. Add an adverb (ADV) to the VP rule.
c. They elected Bill president. Add a second NP to the VP rule.
d. The thief hit the lady with the hat. Add a PP (Prepositional Phrase) to both
VP rule and NP rule.
These new examples gives us the set of PS–Rules (89).
(89) a. S ÷ NP + TP + VP
b. TP ÷ (MOD) + T + (NEG)
c. VP ÷ (ADV) + V + (NP) + ( {NP, PP} )
d. PP ÷ P + (NP)
e. NP ÷
f. T ÷ {PRS, PST, FUT, CND, IMP}
These rules will generate the appropriate structure for the sentence in (88). In particular, they account
for the ambiguity of (88d): if the PP with the hat is what the thief used to hit the lady, then the PP
is part of the VP (89b); if the PP with the hat describes what the lady is wearing, then the PP is part
of the direct object NP (89d). As before, the arrow in these rules is an instruction to rewrite the
symbol to the left as the symbol or symbols to the right. Parentheses indicate optional constituents;
braces indicate alternative constituents. The only function of the plus symbol in PS–RULES is to
mark a clear separation of constituents; it is, therefore, optional. Note again that the left to right order
of constituents is an entirely empirical matter; once a particular order is determined from an
examination of possible phrase structures, it cannot be changed unless there is other evidence

indicating that the change is justified. The rules in (89) are not a complete list of all the PS–Rules
for English; for example, notice that conjunctions like and are not included. These rules will be
added to and modified as we proceed.
The name given to grammars formulated in terms of phrase structure rules and tree diagrams is
PHRASE STRUCTURE GRAMMARS; the theory behind them, as we have indicated, is called
STRUCTURAL GRAMMAR. In developing a set of phrase structure rules like (89), a linguist will
attempt to reach the highest level of generality possible, which means proposing rules that describe
the greatest number of cases (out of the infinite number possible) in the simplest way. For example,
postulating tense as occurring after MOD and before V accounts most simply for did play (do PST
play) and played (PST play). Considering only a sentence like they played, one might conclude that
an analysis which places tense to the left of V is unnecessarily abstract. But remember, a grammar
attempts to account for all possible sentences. In English, the tense marker (PRS, PST, etc.) is
affixed to the closet verb, what ever that is (modal or main verb).
In addition to identifying all of the major constituents of a sentence and indicating how these
constituents are related to each other, phrase structure grammars offer a formal way of formulating
such functional notions as subject and object. Notice that in a tree diagram like (57) some of the
nodes in the tree immediately dominate others, that is, are directly above them. For example, the S
node immediately dominates an NP node, a TP node, and a VP node; the VP node immediately
dominates a V node and an NP node; and so on. Using this relationship of immediate dominance,
one can define the subject of a sentence as the NP node which is immediately dominated by S, and
the object as the NP node which is immediately dominated by VP. Similarly, the predicate can be
defined as the VP immediately dominated by S, and the main verb can be defined as the V
immediately dominated by VP. Linguists also use “family relationships” to describe positions in
phrase structure, for example, NP, TP and VP are sisters as well as daughters of S.
Such formal definitions are preferable to definitions like those in (41) because they are explicit, i.e.,
they refer to specific entities standing in unique relationships. A definition like the direct object
specifies the person or thing directly affected by the action defined in the verb, on the other hand, is
not explicit and, therefore, raises many problems. For example, consider the direct objects in the
following sentences:
(90) a. The man will burn the money.
b. The man will earn the money.
Under phrase structure analysis, the sentences in (90) would be assigned to the constituent structure
given in (91).
143

144
(91)
This tree characterizes the object NP of both (90a) and (90b) in the same way. A comparison of the
following sentences indicates that this characterization is correct.
(92) a. How much money did the man burn?
b. How much money did the man earn?
(93) a. The money was burned quickly.
b. The money was earned quickly.
If there were something different about the occurrence of the NP the money in (90a) and (90b), one
would not expect them to behave in the same way. But, the sentences in (92) and (93), and many
others, show that they are, in fact, the same.
Now, according to the definitions in (41), the NP the money in (90a) is the direct object because it
specifies the thing directly affected by the action defined in the verb. This seems defensible. With
regard to the NP the money in (90b), however, a problem arises, because it is difficult to conceive of
the money as being directly affected by the action of earning. This problem becomes even greater
for sentences like the following:
(94) Sue earned a bad reputation.
(95) John suffered a blow on the head.
Generally speaking, this problem always arises when one attempts to describe language on the basis
of definitions like those in (41). Another example occurs with the definition usually given to nouns,
i.e., a noun is the name of a person, place, or thing. This definition combined with (41) leads to
statements like A noun can be the name of a thing which is the topic of an expression. Such
statements reveal little about English sentence structure. On the other hand, a description of a
category like noun, formulated in terms of phrase structure grammar, will explicitly separate those
items which are nouns from those which are not. For this reason, most linguists have abandoned
definitions like those in (41) in favor of more rigorous and formal ways of characterizing sentence
structure, such as those afforded by phrase structure grammar.

3.6 THE ENDOCENTRICITY CONSTRAINT
PS-Rules like (89) and the structural diagrams they generate like (91) also allow us to begin to make
some important generalizations about phrase structure. For example, notice that, in (91), the major
phrases (NP, TP and VP) each dominate a category of the same type (N, T, and V, respectively). The
technical term for this is ENDOCENTRICITY, which means that every constituent X must be
dominated by a phrase of the same type XP, whatever X is. Further, every phrase XP must dominate
an X. What that means, for example, is that noun phrases must contain nouns and verb phrases must
contain verbs. We can formulate this Endocentricity Constraint as follows:
(96) ENDOCENTRICITY CONSTRAINT.
145
a. Each XP must dominate one and only one X; each X must be dominated by one and
only one XP. The constituent X is called the HEAD of the XP. The phrase XP is
called the MAXIMAL PROJECTION of X.
b. Each head X must dominate one and only one word or abstract category (PRS, PST,
FUT, POS, Ø, etc.).
The Endocentricity Constraint stipulates the following:
(97) a. XP ÷ X
b. *XP ÷ Y
There are three constituents in (91) which do not obey the Endocentricity Constraint, DET, MOD and
S. Implementing the Endocentricity Constraint for DET and MOD, we modify (91) as follows (we
will consider S later in this chapter):
(98)
In the above representation, with the exception of S, every X is dominated by an XP, and every XP
dominates X. Further, each head X dominates only one word or abstract category.

146
It is important to note, at this point, that good scientific inquiry always seeks to uncover
generalizations like the Endocentricity Constraint even when there are apparent exceptions such as
S. Also, given the examples we have considered so far, it appears as though DP and MP are
HEAD–ONLY constructions, that is, phrases that only can contain a head.
Returning to our original example, we now have the following representation for The passengers do
obey the stewardess:
(99)
Implementing the Endocentricity Constraint, we can now generalize the internal structure of phrases
in our PS rules as follows:
(100) a. S ÷ NP + TP + VP
b. TP ÷ (MP) + T + (NEGP)
c. VP ÷ (ADVP) + V + (NP) + ( {NP, PP} )
d. PP ÷ P + (NP)
e. NP ÷
f. T ÷ {PRS, PST, FUT, CND, IMP}
g. XP ÷ X
Notice that the structure and rules above are more generalized than those we have had in the past,
because the internal structure of phrases is now more generalized. In our investigation, there has
been a gradual progression toward greater generality from (42b) to (57) to the representation and
rules above. In fact, as we proceed in this chapter, we will uncover evidence supporting the principle
that all phrases have the same internal structure.

3.7 THE LEXICON
In order to generate a tree structure like (99) from rules like (100), our grammar must also contain
a lexicon or dictionary of all the words in the language. We must have a way to ensure that words
are inserted into the appropriate slots in structure, that is, that verbs are inserted under V, nouns under
N, determiners under D, and so on. Recalling our use of features from phonology, let us propose a
lexicon with SYNTACTIC FEATURE distinctions such as the following:
(101) a. truth [+N] N=Noun
b. explain [+V] V=Verb
c. the [+D] D=Determiner
d. from [+P] P=Preposition
Some words can clearly function as more that one category:
(102) a. love [+N; +V]
b. patient [+N; +ADJ]
c. swell [+N; +V; +ADJ]
d. down [+N; +V; +ADJ; +P]
Given feature specifications like (101) and (102), we ensure that words are inserted into tree
structures with the following principle of lexical insertion:
(103) For each X, freely insert a lexical item (word) with the feature specification [+X, ...].
Given (103), only words with the feature [+X, ...] will be inserted into tree structures under the
category X.
In addition to ensuring that words are inserted under the appropriate category, we must also ensure
that they are inserted into the appropriate environments (before and/or after the appropriate
structures). For example, we must ensure that verbs which require a direct object (Transitive Verbs)
are inserted into structures where an NP follows the verb, such as (99). Similarly, verbs which cannot
have direct objects (Intransitive Verbs) should not be inserted into structures before an NP. Linguists
call such restrictions CO-OCCURRENCE RESTRICTIONS. They stipulate the structural
environments in which lexical items can occur. Adopting the formalism we used in phonology for
environments, we have examples like the following (see the Rule Formalism on page 89):
(104) a. obey [+V, +/___NP] Transitive Verb
b. smile [+V, –/___NP] Intransitive Verb
c. break [+V, ±/___NP] Transitive or Intransitive Verb
These entries stipulate that a verb like obey can only be inserted into trees having a transitive
structure, that is, a VP that dominates a V followed by an NP such as (99).
147

148
Using syntactic features like the above, we can make improvements in the rules in (100) to attain an
even higher level of generality and descriptive adequacy. Consider the following examples of adverb
phrases (ADVP) and adjective phrases (ADJP):
(105) ADVP
a. very beautifully
b. as helpfully as Mary
c. so much more skillfully than Mary
(106) ADJP
a. very beautiful
b. as helpful as Mary
c. so much more skillful than Mary
Clearly, adverbs and adjectives are essentially the same type of syntactic category such that one can
easily be turned into the other. The major difference, of course, is that ADVP modify verbs and
ADJP modify nouns, facts which are directly expressed in the PS-Rules (100). Notice that ADVP
is to the left of V and dominated by VP, while ADJP is to the left of N and dominated by NP. Given
these relationships, it is clear that adverbs and adjectives are members of some larger category, which
linguists call ADJUNCT or A. Implementing this observation, we can replace both ADVP and
ADJP with AP, for ADJUNCT PHRASE. An adjunct is a modifier. The position of the AP in
structure will determine whether it is functioning adverbially or adjectivally: in a VP, it will be
adverbial; in an NP, it will be adjectival. Using features, let us say that adverbs are adjuncts that are
non-adjectival and adjectives are adjuncts that are adjectival:
(107) a. [+A, –ADJ] beautifully, helpfully, skillfully, happily, quickly, etc.
b. [+A, +ADJ] beautiful, helpful, skillful, happy, quick, etc.
c. [+A, ±ADJ] hard, fast, daily, etc.
In a similar way, we do not need to specify both QP and NP in (100e) because quantifiers are a type
of noun. In fact, sometimes the same word can function as either a quantifier noun or a nonquantifier
noun. Consider the following (in the “a” examples, the italicized words function as
quantifier nouns; in the”b” examples, as non-quantifier nouns):
(108) a. Half the men were eliminated. (cf. All/both the men were eliminated)
b. We cut the team in half. (cf. We cut the team into two groups.)
(109) a. We made three times the profit. (cf. We made all the profit.)
b. Times are hard. (cf. Decisions are hard.)
(110) a. We made double the profit. (cf. We made twice the profit.)
b. She is my double. (cf. She is my stand-in.)
Quantifier nouns (all, both, etc.) and determiners are actually in complementary distribution with
non-quantifier nouns (John, child, etc.) and the possessive suffix. Determiners can only be preceded

y quantifier nouns and POS can only be preceded by non-quantifier nouns. Consider again
examples like the following:
(111) NP DP HEAD
a. all the books
b. all Ø books
c. *all POS books (*all’s books)
d. *John the books
e. *John Ø books
f. John POS books (John’s books)
Since a quantifier is a type of noun, we can replace ({QP, NP}) with just (NP) in the rule (100e); and,
given the above discussion, we can also replace ADJP with AP in that rule. The new rule is as
follows:
(112) NP ÷
Using features, we can distinguish individual common nouns and quantifiers in the following way:
(113) a. [+N, –Q] book, girl, truth, group, decision, profit, etc.
b. [+N, +Q] all, both, each, twice, etc.
c. [+N, ±Q] half, time, double, etc.
All the data taken together show that determiners, quantifiers, and possessives have the same
function, namely, they specify nouns (this child, all those books, Mary’s decision, etc.). That
indicates that they are members of a larger category, which linguists call SPECIFIERS.
We can also use syntactic features to distinguish the major subcategories of nouns and pronouns as
follows:
(114) a. [+N, +COMMON] heather, jack, state, etc.
b. [+N, –COMMON, –PRO] Heather, Jack, Michigan, etc.
c. [+N, –COMMON, +PRO, +REFLEXIVE] herself, himself, itself, etc.
d. [+N, –COMMON, +PRO, –REFLEXIVE] she, he, it, etc.
Proper nouns ([+N, –COMMON, –PRO]) share many features with pronouns ([+N, –COMMON,
+PRO]). Both subcategories of nouns cannot occur after quantifiers, determiners and adjectives or
before PP:
(115) a. *That happy Richard laughed.
b. *A tall she married a short him.
149

150
We can account for these co-occurrence restrictions borrowing the formalism for redundancy rules
in Chapter Two (cf. (46) on page 85):
(116) a. [–COMMON] ÷ [–\NP__ ]
b. [–COMMON] ÷ [–\DP__ ]
c. [–COMMON] ÷ [–\AP__ ]
d. [–COMMON] ÷ [–\__ PP]
These redundancy rules state that all non-common nouns (proper nouns and pronouns) cannot occur
after quantifiers, determiners, and adjectives or before PP.
Given the Endocentricity Constraint, we can collapse the four redundancy rules above into one rule
as follows:
(117) [–COMMON] ÷ [–\ XP __, –\ __XP]
This redundancy rule states that all non-common nouns cannot occur after or before any other type
of phrase within an NP.
3.8 A NOTE ON SCIENTIFIC INQUIRY
In the preceding sections, we have discussed some attempts to develop an adequate theory of
grammar beginning with traditional grammar and culminating in a version of structural grammar that
reaches a high level of generality with the Endocentricity Constraint. Reviewing this, we can see
how linguistics has been able to deepen and broaden its characterizations of human language in an
effort to attain ever higher levels of adequacy. The search for descriptively and explanatorily
adequate models of grammar continues today.
It is often dismaying to come to a discipline for the first time and find that it does not provide a
succinct, unified, and fully justified theory. Linguistics is such a discipline; however, although there
is no grammatical theory which is championed by every living linguist, the issues today are quite
clearly drawn, and so are the objectives of research. The discovery of the grammatical facts
themselves is much more significant than the details of particular representations of those facts.
No matter what grammatical model is proposed, it must be able to provide a characterization of all
sentences in such a way that it accounts for the native speaker’s knowledge of them. This involves,
among other things, accounting for ungrammatical sentences in a principled way, that is, in a way
which predicts that they will be ungrammatical. The facts to be accounted for are the same regardless
of the theory invented to characterize them. The objective of the discipline is to provide a theory
which will be consistent with the ever–expanding corpus of data. In this respect, linguistics is no
different from any other science. Consider, for example, the study of celestial objects.
Astronomy is one of the oldest sciences. Even before 2000 B.C., crops were sowed according to the
movements of the sun and moon. Modern theories of the universe which displace the earth from the

center are due to Copernicus (1473–1543). There have been many modifications over the centuries
even to the point that separate areas of inquiry or methodology have been identified like radio
astronomy, astrophysics, and spectroscopy. The discovery of pulsars, quasars, black holes, and other
stellar phenomena has engendered modifications in existing theories of the universe. As our
knowledge of the universe has expanded, the field of astronomy has expanded, and a number of
distinct theories for a number of distinct issues have emerged, e.g., theories on the origin of the
universe, the big bang theory, string theory, brane theory, and so on. In the process, the explanatory
power of the combined theories has increased. The discovery of some new stellar phenomena never
nullifies everything that precedes, that is, astronomy does not begin again from scratch. Rather,
existing theories are appropriately modified or, if the phenomena are not understood, left entirely
unmodified despite apparent conflicts or inconsistencies. It is always possible that putative
exceptions are not exceptions at all, only misunderstood cases.
People have been studying language almost as long as they have been speculating on the heavens.
Terms like verb which form part of the theories described in this book were coined by ancient
grammarians. However, it is only in the last fifty years or so that linguists have been concerned with
many of the issues discussed in this book, in particular, those issues formulated here as the
grammatical characterization problem and the grammatical realization problem. In short, students
of linguistics must understand that new models of characterization are inevitable as new data are
discovered. During this evolution, the issues have become more clearly defined, and the
characterizations have become more explanatorily adequate. More than anything else, this process
and progress of scientific inquiry must be understood.
In addition, students must understand that grammars are theories about language. In writing
grammars, linguists attempt to describe observed phenomena, such as the different ways in which
they observe native speakers using proper nouns and common nouns. Sometimes, the description will
make use of abstract elements such as Ø postulated above. In this regard, linguists are no different
from other scientists. For example, physicists postulate the existence of dark matter, that is,
“invisible” matter, to account for abnormalities in the observed rotation of galaxies. No one has ever
seen dark matter anywhere in the universe, just as no one has ever heard Ø. Nevertheless, scientists
assume that these theoretical constructs exist so that they can account for the phenomena they do
observe.
In our discussion, various approaches to grammatical characterization are adopted, discussed, and
either discarded or maintained as our inquiry proceeds. To some, this may seem to be a waste of
time. Why not ignore the historical perspective and just discuss the right analysis? The answer is
quite simple: there is no right analysis. There is only the analysis which covers the known facts at
the present time. Furthermore, science sometimes takes one or even a few steps backwards, returning
to an older theory once discarded and resurrecting it with modifications.
Since our understanding of language is increasing all the time and analyses, consequently, can be
expected to be falsified by newly discovered data, the most important thing to be learned is how to
do grammatical analysis to begin with. This prepares students for the changes that are inevitable in
so young an area of inquiry as formal linguistics. Therefore, our focus must be on the methods and
objectives of linguistic inquiry. At every stage in this inquiry, we will formulate hypotheses and
151

152
examine their consequences for the theory of language. This process of hypothesis development is
common not just to linguistics, but to scientific method in general.
Everything said here about scientific inquiry applies to the analysis of problems that confront society,
businesses, industry, government, and all other institutions and enterprises. As we noted in the
Introduction, the problems facing today’s workers are complex and subtle, as difficult to solve as the
nature of the human mind, language, and the universe. What we have done throughout this book is
devise steps for breaking down problems into manageable parts, for making proposals that are robust
and workable, for testing hypotheses, and for finding solutions that satisfy the most factors and
constituents in the best way possible. Linguistic analysis has been our means to develop expertise
in problem solving. At an abstract level, formulating a theory of language, which is our specific
problem, is like all problems.
3.9 X–BAR SYNTAX
Returning to our main discussion, let us continue to explore the nature of an adequate theory of
syntax. Thus far, our discussion has pivoted around the concept of SYNTACTIC UNIT. To find
these units, we considered the three syntactic processes of REFERENCE, OMISSION, and
PLACEMENT. We have the following definitions:
(118) a. PHRASE: a word or group of words that functions as syntactic unit in
REFERENCE, OMISSION, and/or PLACEMENT.
b. HEAD: the minimal element of a phrase.
A phrase whose head is a noun is a noun phrase; one whose head is a verb, a verb phrase; one whose
head is a preposition, a prepositional phrase; and so on. The phrases we have examined in the
sections above have been relatively simple. Consider now the following increasingly complex
examples of noun phrases:
(119) a. those paintings
b. all those paintings
c. all those paintings of landscapes
d. all those stolen paintings of landscapes
e. all those stolen paintings of landscapes of the Louvre’s
f. all those recently stolen paintings of landscapes of the Louvre’s
g. all those ten recently stolen paintings of landscapes of the Louvre’s
h. all those ten recently stolen Impressionist paintings of landscapes of the Louvre’s on
sale
If we attempt to diagram the NP’s in (119) in accordance with the mechanisms we have so far
developed for English phrase structure, we encounter a number of immediate problems that require
a solution. The most difficult problem concerns the fact that both traditional and early structural
grammar provide only the most meager structural framework for basic syntactic units. Quite clearly,
we cannot suspend all the structures in (119h) , for example, from the same NP node. Such an

analysis ignores the fact that NP’s have internal hierarchical structure as well as internal linear
structure. To see this, consider the simple NP in (120) provided by the rules developed so far.
(120)
A diagram like (120) leaves a number of important facts unaccounted for. First, note that we have
examples like (121).
(121) a. John saw those Spanish paintings, and Bill saw these Italian ones.
b. John saw those Spanish paintings, and Bill saw these ____.
The word ones in (121a) substitutes for the noun paintings indicating that paintings is an internal unit
within its own phrase. That fact is accounted for by (120): paintings is under one node by itself, the
N node. However, the same diagram cannot account for the gap in (121b). In this sentence, the gap
stands for Spanish paintings which is not identified as a unit in (120). For Spanish paintings to be
a unit the two elements, the AP (Spanish) and the N (paintings), must be under one node, separate
from the DP (those) node. This means that a better diagrammatic representation of (120) is (122),
where the "?" identifies the missing, as yet unlabeled, node.
(122)
This deficit in the internal structure of phrases has not gone unnoticed in modern grammatical
analysis, and, in the 1970’s, a phrase structure framework was developed that ultimately came to be
known as X–BAR SYNTAX. This theory replaces diagrams like (122) with ones like (123), in
which progressively larger syntactic units of the same phrasal type are indicated by progressively
higher primes; for example, N´ designates a larger unit than N. In the original version of this theory,
primes were not used; rather a bar was placed over a symbol, e.g., N¯. Since bars are difficult to
produce on most typewriters and many word processors, they are generally replaced with primes as
follows:
153

154
(123)
Notice that, in (123), Spanish paintings is contained in a separate node labeled N´. This node
represents a separate, intermediate noun phrase level between N and NP. The representation claims
that phrases, in particular, noun phrases, must have more internal hierarchical structure than the
amount usually accorded them. One of the many examples that demands this is (121b) which
contains a gap that stands for the unit Spanish paintings, represented by the N´ in (123). The
representation in (123) replaces the "?" of (122) with a new intermediate noun phrase level and,
accordingly, directly accounts for the gap in (121b). Basically, X–Bar syntax recognizes more
syntactic levels than the two traditionally recognized, that is, more than just XP over X.
Before continuing, a somewhat trivial matter should be mentioned – one which seems to help students
read complicated phrase structure diagrams. In X–Prime (or X-Bar) notation, it is useful to line up
all the head elements (XP, X´, X) in a straight vertical sequence making the location of the head of
each separate phrase straightforward. Thus, the representation of (123) is (124).
(124)
Note that the X´ phrase is explicitly mentioned in a tree only if it branches; thus, N´ is mentioned but
neither D´ nor A´ is. Also, for simplicity, we can represent tree diagrams like the above linearly as
LABELED BRACKETS:
(125) [ NP [ DP [ D those ] ] [ N´ [ AP [ A Spanish ] ] [ N paintings ] ] ]

Note that each bracket on the left is paired with one on the right beginning with the smallest units and
proceeding to increasingly larger units. The following diagram illustrates which brackets are paired
(first the X phrase, then the X´ phrase if there is one, and lastly the XP phrase):
(126) ¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
¯¯¯¯¯¯¯¯¯¯ ¯¯¯¯¯¯¯¯¯¯¯¯¯
¯¯¯¯¯¯ ¯¯¯¯¯¯¯¯ ¯¯¯¯¯¯¯¯¯
[ NP [ DP [ D those ] ] [ N´ [ AP [ A Spanish ] ] [ N paintings ] ] ]
Returning to the main discussion, consider now a phrase like (127), which can have either of the two
meanings indicated.
(127) those Spanish teachers
a. those teachers who are Spanish
b. those teachers of Spanish
In addition, we have the following examples:
(128) a. those Spanish Braille teachers
b. *those Braille Spanish teachers
(129) a. John saw those Spanish math teachers, and Bill saw these ________.
b. John saw those Spanish math teachers, and Bill saw these Italian ones.
These phrases require that compound nouns like Braille teachers in (128a) and math teachers in
(129) be contained under a separate N node, giving structures like the following:
(130) (131)
155

156
The above representations contain the compound nouns Braille teachers and math teachers. In such
compounds, the first element is actually the COMPLEMENT of the head, it completes it’s meaning:
a Braille teacher is a person who teaches Braille; a math teacher is a person who teaches math. In
these representations, both Braille teachers in (130) and math teachers in (131) are contained under
one N´ node. Such representations account for the use of ones in examples like (129b). Observe also
that these representations can account for the ambiguity of a phrase like (127). If it means (127a),
then Spanish is a descriptive adjective and branches off the higher N´ if it means (127b), then
Spanish is part of a compound and branches off the lower N´. Further, this analysis predicts that both
should be able to occur together, and that is correct: those Spanish Spanish teachers is possible;
where the first Spanish (an adjective phrase or AP) refers to origin, and the second Spanish (a noun
phrase or NP) refers to language.
These data indicate that the order of elements within an NP is fixed: SPECIFIERS must dominate
ADJUNCTS and ADJUNCTS must dominate COMPLEMENTS. We have no examples like (132).
(132) a. *(John met) those math Spanish teachers.
b. *(John met) Spanish those teachers.
c. *(John met) math Spanish those teachers.
In the following two diagrams, only (133) is grammatical because the adjunct (AP) Spanish
(indicating national origin) dominates the complement (NP) Spanish (indicating the language). In
(134), the two constituents are reversed so the phrase is not grammatical (cf. *those math Spanish
teachers).
(133) (134) *

The data that we have observed and the facts that we have discovered indicate that the Endocentricity
Constraint (96) must be expanded as follows:
(135) ENDOCENTRICITY CONSTRAINT.
a. Each XP must dominate one and only one X; each X must be dominated by one and
only one XP. The constituent X is called the HEAD of the XP.
b. Each head X must dominate one and only one word or abstract category (PRS, PST,
FUT, POS, Ø, etc.).
c. Between XP and X, multiple recursions of X´ are possible.
d. Specifiers must dominate adjuncts; adjuncts must dominate complements.
3.10 GENERALIZING PHRASE STRUCTURE: S AND NP
While the revisions that we have implemented reach a much higher level of descriptive adequacy than
the rules of traditional grammar, they still leave many generalizations unexpressed. For example, in
every language, there are abundant instances of cross-categorial principles, that is, principles which
spread over many different categories like sentences (S) and noun phrases (NP). Consider the
following:
(136) S: The people generously donate to that charity.
NP: The people’s generous donations to that charity...
(137) S: The barbarians suddenly destroyed the city.
NP: The barbarians sudden destruction of the city...
(138) S: John steadfastly refuses to do anything illegal.
NP: John’s steadfast refusal to do anything illegal...
(139) S The Mafia illegally shipped drugs into the country during June.
NP: The Mafia’s illegal shipments of drugs into the country during June...
Notice that in examples like (136) and for any given verb/noun pair like donate and donation, it is
possible to substitute the same kinds of subjects (the people, Americans, etc.), the same kinds of
complements (to that charity, to Detroit, etc.) and the same kinds of adjuncts (generous, generously,
etc.). Some of these generalizations can be expressed by the principles we have formulated. For
example, consider the following representations for the sentence and noun phrase in (136):
157

158
(140) S: The people generously donate to that charity.
(141) NP: The people’s generous donations to that charity...

Given these structures, we can make the necessary generalizations about complements and adjuncts
because they occupy exactly parallel positions. Both adjuncts generous and generously are
dominated by an X´ over another X´; the complements to that charity follow the head X, which is
dominated by an X´. Generalizations regarding the subjects, however, cannot be expressed, because
the above diagrams do not completely match each other; in particular, there seems to be an extra VP
node in (140). Furthermore, the symbol S does not obey the Endocentricity Constraint. We can solve
these problems if we replace (140) with the following structure:
(142) S: The people generously donate to that charity.
The structure in (142) exactly parallels (141) so all relationships can be generalized between
sentences and noun phrases. Just as DP (POSSESSIVE) specifies the head noun, TP (TENSE)
specifies the head verb; the subjects of both structures is the phrase the people. Again, both adjuncts
generous and generously are dominated by an X´ over another X´; the complements to that charity
follow the head X, which is dominated by an X´.
Essentially, (142) represents a sentence as a large verb phrase (VP), which is appropriate since the
main constituent of a sentence is the verb.
3.11 ADJUNCTS (MODIFIERS) AND COMPLEMENTS
There is one remaining problem with the above structures and that concerns the distinction between
adjuncts (modifiers) and complements. Notice first that the distinction between adjuncts and
complements is real. Adjuncts are always higher in the tree structure than complements, and the
same phrase often can have either function. We have examples like the following:
159

160
(143) Prehead NP position.
a. the cotton sack
[1] ‘the sack made of cotton’ or ‘the sack for cotton’
[2] the cotton cotton sack
[3] the cotton potato sack
[4] *the potato cotton sack
b. a grand piano
[1] ‘a really nice piano’ or ‘a type of piano’
[2] a grand grand piano
[3] a terrific upright piano
[4] *an upright terrific piano
(144) Posthead NP position.
a. the choice of the men
[1] ‘the choosing of the men’ or ‘the choice by the men’
[2] the choice of the men of the men
[3] the choice of the men by the men
[4] *the choice by the men of the men
b. the rumor that they made up
[1] ‘the rumor about their making up’ or ‘the rumor which they made up’
[2] the rumor that they made up that they made up
[3] the rumor that they made up which they made up
[4] *the rumor which they made up that they made up
(145) Posthead VP position.
a. He decided on the boat.
[1] ‘He chose the boat’ or ‘He was on the boat when he decided.’
[2] He decided on the boat on the boat.
[3] He decided on his vote on the boat.
[4] *He decided on the boat on his vote.
b. He reads the most.
[1] ‘He reads the most things’ or ‘He reads the most often.’
[2] He reads the most the most.
[3] He reads the most books the most often.
[4] *He reads the most often the most books.

The problem with the structures (140) and (142) concerns the ambiguities in the above examples;
specifically, there is no way to tell whether a phrase is functioning as an adjunct or as a complement
in representations like the following:
(146) a grand piano
(147) He decided on the boat.
The constituents grand and on the boat in the above trees may either be adjuncts or complements
given our statement of the Endocentricity Constraint (135). In both cases, the ambiguities are
structural, not lexical, that is, the ambiguities pertain to syntactic relationships, not the meaning of
words. In a sentence like I found the bug, the ambiguity derives from the multiple meanings of the
word bug (insect, computer problem, etc.). Other examples of LEXICAL AMBIGUITIES are as
follows:
(148) a. The bandit was shot by the bank. (bank = financial institution, edge of river, etc.)
b. The students are revolting. (revolting = disgusting, uprising, etc.)
161

162
Other examples of STRUCTURAL AMBIGUITIES occur in the following examples, where the
constituents can be grouped in more than one way:
(149) a. The farmer looked over the fence.
[1] The farmer - looked over - the fence.
[2] The farmer - looked - over the fence.
b. The man who left suddenly returned.
[1] The man - who left suddenly - returned.
[2] The man - who left - suddenly returned.
The structural ambiguity in (146) and (147) cannot be resolved by different hierarchical groupings
because the problem does not involve the left-to-right hierarchical order of the constituents. To solve
this problem, we might propose to make a distinction between an X´ over another X´, on the one
hand, and an X´ over an X, on the other, as follows:
(150) a. XP ÷ ...X´... Specifier Rule
b. X´ ÷ ...X´... Adjunct Rule
c. X´ ÷ ...X... Complement Rule
This proposal means that He decided on the boat would have the following two representations,
where on the boat is an adjunct in the first and a complement in the second:
(151) a. He decided on the boat. (on the boat is a adjunct; it is where he decided)

. He decided on the boat. (on the boat is a complement; he chose the boat)
But this proposal will not work because complements can be recursive just as adjuncts can. Note first
that we can have multiple adjuncts, that is, recursive X´ levels:
(152) a. That pretty, little, yellow house is for sale.
b. He drives a sleek, green, speedy Maserati.
Similarly, complements and compound nouns can be stacked recursively:
(153) a. That kitchen appliance factory guarantee is very generous
(cf. The factory guarantees the kitchen appliance.)
b. The terrorist cell police raid stunned everyone.
(cf. The police raided the terrorist cell.)
c. The sexual harassment policy court decision was of national interest.
(cf. The court decided on the policy on sexual harassment.)
d. She filled out the customer satisfaction store survey.
(cf. The store surveys customer satisfaction.)
e. They dedicated a cancer patient radiation treatment facility.
(cf. A facility treats cancer patients with radiation.)
According to (150b) and the proposals we have made, a phrase like kitchen appliance factory
guarantee would have the following representation, which would make kitchen appliance an adjunct,
rather than a complement (direct object) of guarantee:
163

164
(154)
To solve this problem, we will make one final revision to phrase structure levels. Let us assume a
three level version of the Endocentricity Constraint such that each level is distinctly specified with
numbers and specifiers are always daughters of X3, adjuncts are always daughters of X2, compounds
and complements are always daughters of X1, and the head is always X0. Just as the original
Endocentricity Constraint stipulated that each phrase minimally consists of an XP over an X, the new
revision stipulates that each phrases minimally consists of an X3 over an X0. This new phrasal
architecture yields diagrams like the following (note again that intermediate levels are only
mentioned in a diagram if they branch):
(155) a. b.

With this revision, we can diagram the noun phrase the kitchen appliance factory guarantee and its
related sentence (The factory guaranteed the kitchen appliance) as follows:
(156) a.
b.
Note that in both of the above trees, kitchen appliance is a complement (X1 level constituent). Since
each level is uniquely specified with a different number, recursion is possible at all levels without
ambiguity.
165

166
Adding a quantifier and adjuncts (both adjectives and an adverb) to the above phrase, we have
examples like all those unexpected, very generous kitchen appliance factory guarantees:
(157)

With all the above revisions, it is now possible to accurately represent ambiguous phrases in which
constituents can be interpreted either as an adjunct or as a complement:
(158) a. [ N3 [ N2 [radio station] [ N1 night manager]]] = a night manager at a radio station
b. [ N3 [ N1 [radio station] [ N1 night manager]]] = a night manager of a radio station
167
A radio station manager can be either a night manager at a radio station (different
managers at the same station), in which case radio station is an adjunct (N2), or a
night manager of a radio station (someone who manages the whole radio station), in
which case radio station is a complement (N1).
(159) a. peanut oil preservatives = oil preservatives made of peanuts
[ N3 [ N2 peanut [ N1 oil preservatives]]]
b. peanut oil preservatives = oil preservatives for peanuts
[ N3 [ N1 peanut [ N1 oil preservatives]]]
c. peanut oil preservatives = preservatives made of peanut oil
[ N3 [ N2 [peanut oil] [ N0 preservatives]]]
d. peanut oil preservatives = preservatives for peanut oil
[ N3 [ N1 [peanut oil] [ N0 preservatives]]]
(160) a. American embassy sniper attack = attack of/on the American embassy by snipers
[ N3 [ N1 American embassy [ N1 sniper [ attack]]]]
b. American embassy sniper attack = attack at/from the American embassy by snipers
[ N3 [ N2 American embassy [ N1 sniper [attack]]]]
c. American embassy sniper attack = an attack by American embassy snipers
[ N3 [ N1 American embassy sniper [attack]]]
d. American embassy sniper attack = an American attack by embassy snipers
e. etc.
[ N3 [ N2 American [ N1 embassy sniper [attack]]]]

168
Returning to our original examples, we have the following representations in X3 notation (cf. (141)
and (142)):
(161) N3: the people’s generous donations to that charity...
(162) V3: The people generously donate to that charity.

Summarizing, we have the following generalizations:
(163) a. Elements that branch from X3 are SPECIFIERS
in N3 Determiners, possessive marker (POS), quantifiers
in V3 Subjects, tenses
b. Elements that branch from X2 are ADJUNCTS
in N2 Numerals (three, third, etc.), adjectives (silky dress)
Compound elements expressing source or material (silk dress)
in V2 Expressions of time and place, adverbs, predicate nominatives
c. Elements that branch from X1 are COMPLEMENTS and COMPOUNDS
in N1 Complements, direct objects
Compound elements expressing complements (evening dress)
in V1 Complements, direct objects
d. Elements that branch from X0 are parts of words
in N0 Affixes that form nouns (-er in singer, teacher, etc.)
in V0 Prepositions that form two part verbs (put on, throw out, etc.)
Lastly, we reformulate the Endocentricity Constraint as follows:
(164) a. Every phrase has three syntactic levels above the X0 level, which is the head and
contains a word or abstract category (PRS, PST, POS, Ø, and so on).
169
b. Every word or abstract category must be dominated by an X0 which, in turn, must be
dominated by an X3. Intermediate levels (X2 and/or X1) are explicitly mentioned in
diagrams and labeled brackets only if the phrase contains X2 and/or X1 level elements
other than the head. Implicitly, intermediate levels are always there.
c. There may be multiple recursions of every level; however, each higher level of a
phrase must be greater than or equal to the level below, that is, all the specifiers must
be higher than all the adjuncts which, in turn, must be higher than all the
complements.
X3 above X3 above X2 above X2 above X2 above X1 above X1 above X0
[ N3 all [ N3 those [ N2 pretty [ N2 little [ N2 yellow [ N1 ranch [ N1 vacation [ N0 homes ]]]]]]]]
*X3 above X2 above X3 above X2 above X1 above X2 above X1 above X0
*[ N3 those [ N2 pretty [ N3 all [ N2 little [ N1 ranch [ N2 yellow [ N1 vacation [ N0 homes ]]]]]]]]

170
3.12 RESIDENTIAL GRAMMAR (RG)
The major unresolved issue in the previous discussion is that our theory still does not provide a way
to express the relationships between syntactic categories like verbs (V3) and nouns (N3). Let us,
therefore, re–examine the matter. Traditionally, syntactic categories have been viewed as atomic,
that is, not composed of other elements. The data from the last few sections suggest that that view
is incorrect. As we have seen, syntactic categories share characteristics which we have accounted
for with features in much the way that sounds share characteristics in phonology.
Feature theory is a standard part of phonological descriptions. For example, consider again the
following features:
(165) a. [+VOICED] produced with vocal cord vibration
b. [–VOICED] not produced with vocal cord vibration
(166) a. [+NASAL] involving exhalation through the nose
b. [–NASAL] not involving exhalation through the nose
(167) a. [+ANTERIOR] articulated at the front of the mouth
b. [–ANTERIOR] not articulated at the front of the mouth
Given these features, we can say that the sounds for m as is mass, b as in bass, p as in pass, and g as
in gas, are as follows:
(168) a. [m] [+VOICED, +NASAL, +ANTERIOR]
b. [b] [+VOICED, –NASAL, +ANTERIOR]
c. [p] [–VOICED, –NASAL, +ANTERIOR]
d. [g] [+VOICED, –NASAL, –ANTERIOR]
Thus, a sound like [m] is nothing more than an abbreviation for a specific set of phonological
features. Let us propose a similar feature analysis for syntax.
We begin by noting the difference between verbs and all other categories, in particular, that verbs are
the only category that carries tense (broke), the only category that shows mode (may break) and
aspect (has broken) distinctions, and the only category that can occur in present and past participial
forms (breaking and broken). For these and other reasons to be discussed below, we make the
following division:
(169)

The feature [+VBL] specifies those categories which are inherently marked for distinctions in tense,
aspect, voice, and/or mode; [–VBL] specifies categories that are not so marked. [–VBL] categories
also fall neatly into two major classes, those that behave like nouns, and those that do not.
A distinctive attribute of nouns is that they are inherently marked for distinctions in number, person,
gender, and/or case. Let us specify the former as [+NML]; the latter, [–NML]. The same distinction
occurs in some verb forms. Verb forms that are inherently marked for nominal distinctions [+NML]
include verbal nouns like the gerund basing in Basing the argument on so many assumptions is
unwise; verb forms that are not inherently marked for nominal distinctions include tensed verbs like
bases in He bases his argument on many assumptions, in which the singular number on bases is a
result of agreement and is not an inherent feature of the tensed verb (cf. They base their argument
on many assumptions). To capture these distinctions, we modify (169) to (170) and refer to
nonnouns as CHARACTERIZERS since they characterize a head.
(170)
[+NML] categories differ from characterizers in a number of ways. First, [+NML] categories have
a number of morphological specifications that characterizers lack. They occur before specific
suffixes like –ment, –ness, –ing, etc. (basement, baseness, basing). They show number distinctions:
base versus bases. They occur in the possessive: the base’s shape and the shape of the base.
Second, [+NML] categories fulfill a number of specific syntactic functions, which means they occupy
specific positions in syntactic hierarchies. For example, they can be SUBJECT or OBJECT. Third,
[+NML] categories are always distributed before [–NML] categories on every X level in both
prehead and posthead positions. Notice that in every diagram in this monograph, when two
categories are suspended from the same level, [+NML] categories (NOUNS) always precede [–NML]
categories (CHARACTERIZERS and VERBS).
We now have four major syntactic categories in language, which RG specifies as follows:
(171) a. Verbs (V): [+VBL, –NML] go, goes, went, gone, going
b. Nouns (N): [–VBL, +NML] woman, women, she, all
c. Characterizers (C): [–VBL, –NML] happy, happily, probably, on, after
d. Verbal Nouns (VN): [+VBL, +NML] swimming, eating, reading
Given these major categorial distinctions, we note that many parts of speech are distinguished from
each other mainly by what kind of phrase they can occur in and where they can occur within that
phrase, that is, they have a specific residence. Formally, we define a RESIDENCE as a specific
171

172
position in syntactic structure which has a specific semantic or syntactic function. The theory based
on the concept of residence is called RESIDENTIAL GRAMMAR (RG).
The morphosyntactic feature oppositions [±VBL, ±NML] are not sufficient to distinguish the major
syntactic categories that occur in the world’s languages. For example, within the category verb, there
are subclasses which include finite verbs, nonfinite verbs (infinitives, participles, etc.), auxiliaries,
and modals. Within the category characterizer, there are prepositions, subordinators (subordinating
conjunctions), adjectives, adverbs, determiners, and so on. To specify these subclasses, we will use
two features [±OPEN PHRASE] ([±OPH]) and [±OPEN CLASS] ([±OCL]) defined as follows:
(172) a. [+OPH]: occurring as the head of a phrase that freely contains other elements
b. [–OPH]: occurring as the head of a phrase that does not freely contain other
elements
(173) a. [+OCL]: unlimited in number
b. [–OCL]: limited in number
The syntactic feature specification [±OPH] separates categories whose internal phrase structure is
open in the sense that it can freely contain specifiers (words like the, this, and that), quantifiers
(words like all, some, and much), or adjuncts (adjectives and adverbs), from other categories whose
internal phrase structure is highly restricted. The [±OPH] distinction is most clearly seen in the
variety of structures possible for common nouns on the one hand, which can be specified, quantified,
and modified, and pronouns and proper nouns on the other hand, which cannot:
(174) a. That very old woman left early./*That very old she left early.
b. All the tall women walked in./*All the tall they walked in.
c. A woman that is tall will get the job./*Mary that is tall will get the job.
The feature [+OCL] specifies those categories which form an open class to which an unlimited
number of items may be added. Since classical antiquity, most grammarians have observed the sharp
distinction between the very large number of nouns, verbs, adjectives, and adverbs that occur in
languages as opposed to the rather small fixed number of prepositions, conjunctions, determiners,
pronouns, etc. The opposition [±OCL] is intended to capture this dichotomy. For the most part, in
English, open class words are marked by distinct inflectional suffixes: nouns show variations in
number (boy, boys); verbs show variations in tense (plays, played); and adjectives and adverbs show
variations in comparison (fast, faster, fastest; long, longer, longest). Examples of [+OCL] and
[–OCL] categories are the following:
(175) a. [+OCL]: nouns, verbs, adjectives, adverbs
b. [–OCL]: auxiliaries, modals, pronouns, determiners, prepositions, etc.
Within a particular kind of phrase, certain categories are restricted to X3 level, others to X2 level,
and still others to X1 level. For example, all categories denoting specification (DETERMINERS)
or quantification (QUANTIFIERS) are X3 Level residents. All descriptive modifiers of the head are
X2 Level residents. All elements of compounds (history in history teacher) and complements

(history in teacher of history and teach history) are X1 Level residents. Therefore, with reference
to the X hierarchy, we recognize the following level features:
(176) a. [+X3L] can freely occur immediately dominated by X3 (N3, V3, or C3)
b. [–X3L] cannot freely occur immediately dominated by X3 (N3, V3, or C3)
(177) a. [+X2L] can freely occur immediately dominated by X2 (N2, V2, or C2)
b. [–X2L] cannot freely occur immediately dominated by X2 (N2, V2, or C2)
(178) a. [+X1L] can freely occur immediately dominated by X1 (N1, V1, or C1)
b. [–X1L] cannot freely occur immediately dominated by X1 (N1, V1, or C1)
We note further, that some categories are restricted to phrases with specific heads, that is, can be
suspended from some X level only when X is a specific phrase type. DETERMINERS can only
occur in noun phrase, suspended from N3; ADJECTIVES can occur in either noun phrases or verb
phrases, suspended from N2 or V2; MANNER ADVERBS can occur only verb phrases suspended
from V2. Therefore, we recognize the following environment features:
(179) a. [+ENH] can freely occur immediately dominated by N3, N2, or N1
b. [–ENH] cannot freely occur immediately dominated by N3, N2, or N1
(180) a. [+EVH] can freely occur immediately dominated by V3, V2, or V1
b. [–EVH] cannot freely occur immediately dominated by V3, V2, or V1
(181) a. [+ECH] can freely occur immediately dominated by C3, C2, or C1
b. [–ECH] cannot freely occur immediately dominated by C3, C2, or C1
Lastly, we recognize that some categories are restricted to prehead position (DETERMINERS) and
others to posthead position (NEGATIVES). We, therefore, add these feature oppositions:
(182) a. [+PRH] can freely occur before the head of a phrase
b. [–PRH] cannot freely occur before the head of a phrase
(183) a. [+PSH] can freely occur after the head of a phrase
b. [–PSH] cannot freely occur after the head of a phrase
Ultimately, the concept of residence is linked to the syntactic features of RG. These are proposed
as substantive syntactic universals, meaning that the languages of the world must construct their
syntactic inventories out of these features, just as they must construct their phonological inventories
out of substantive phonological features. Despite this constraint, it is important to note that the theory
does not demand that all languages choose the same features in the same combinations. Quite the
contrary occurs. Different languages select different sets of features out of the pool of possible
features, and even the same language, at different points in its history, utilizes features in different
ways. For example, at an earlier stage of English, word order constraints were less stringent than
they are today; therefore, the features [±PREHEAD] and [±POSTHEAD] did not have the
173

174
significance they have in modern English. The task of children acquiring a language is to discover
which system they have been exposed to by chance.
Note that, with features, we can directly account for the relationship between examples in which
constituents occupy different positions, such as the available books and the books available as in
(184).
(184) a. b.
Recall that such diagrams omit unbranching intermediate X levels, by convention. Thus, N2 is
present because it branches; but N1, not relevant here, is omitted. If the phrases were the available
history books and the history books available, then both N2 (dominating available) and N1
(dominating history) would be present in the diagrams.
In addition, individual levels must be recursive to distinguish units: [ N3 the [ N2 intelligent [ N2 young
[ N men]]]] contrasts with [ N3 the [ N2 young [ N2 intelligent [ N men]]]]. In both of these examples, there
are two successive, separate N2 levels distinguishing the meanings. In the first case, [ N3 the [ N2
intelligent [ N2 young [ N men]]]], one is talking about the young men who are intelligent. But in the
second, [ N3 the [ N2 young [ N2 intelligent [ N men]]]], one is taking about the intelligent men who are
young.
This method of recursion is also necessary to account for the use of ones in sentences like John met
the young men from Cleveland, and Bill met the old ones, where ones refers to the subunit men from
Cleveland. Such examples contrast with others like John met the young men from Cleveland, and
Bill met the ones from New York, where ones refers to the subunit young men.
Since all categories on the X2 Level are modifiers of the head, it no longer matters whether the
adjective precedes the noun or follows it. We simply state idiosyncratic facts about individual words
in the lexicon next to each lexical entry. This means that the word available has a dictionary
specification like [–VBL, –NML, +X2L, +PRH, +PSH] meaning that it is an X2 Level characterizer
(Adjective) that can occur in either prehead or posthead position.
We can extend this analysis to show the relationship between a courageous man and a man of
courage as in (185) and a Braille teacher and a teacher of Braille as in (186).

(185) a. b.
(186) a. b.
The structure of courage in (185b) is called a genitive of quality; notice that, as an X2 Level element,
it serves as a modifier of the head, exactly like the adjective courageous in (185a). The structures
in (186) show the relationship between compounds and complements. In English, this is most clearly
seen in nouns, although it exists for other categories like verbs (bartend versus tend bar) and
adjectives (life supporting versus supporting life).
The of in (185b) and (186b) is not a full preposition introducing a prepositional phrase. Actually, it
is little more than a phonological filler necessary in English to separate the head from posthead
constituents. In English, noun phrases can contain very long sequences of compound elements as we
have seen. Further, the last noun in a sequence of nouns is the head of the phrase and only the head
can carry number distinctions. Consider the following examples which show that only the head in
a sequence of nouns can show number distinctions and it is the head of the subject NP that the verb
agrees with:
(187) a. The kitchen appliance factory guarantee is/*are generous.
b. The kitchen appliance factory guarantees *is/are generous.
175

176
c. *The kitchen appliance factories guarantee is/are generous.
d. *The kitchen appliances factory guarantee is/are generous.
e. *The kitchens appliance factory guarantee is/are generous.
f. A ten thousand word paper is/*are required.
g. *A ten thousand words paper is/are required.
Also, as the following data show, the head (in italics) of an NP must be separated from posthead
constituents by a characterizer (double underlined):
(188) a. The votes against gun control *is/are greater than expected.
b. John’s reliance on his parents is/*are pathetic.
c. The birds in that tree *is/are nesting.
d. The language teacher is/*are in the auditorium.
If there is no overt characterizer separating the head noun from posthead elements, then of must be
inserted directly after the head. We have contrasts like the following:
(189) a. His love stories *is/are charming.
b. His love of stories is/*are charming.
c. The language teachers *is/are monitored by the principal.
d. The language of teachers is/*are monitored by the principal.
e. The number systems *is/are complicated.
f. The number of systems is/*are complicated.
g. The student letters *is/are very complimentary.
h. The student of letters is/*are very complimentary.
All of the above intricacies about English noun phrases seem to involve the rule of verb-subject
agreement. Without agreement, many English sentences are uninterpretable. There is a very large
number of words in English that can be either a noun or a verb, e.g., school, show, play, book, table,
chair, man, etc. Consider the following and observe the number on the verb:
(190) a. The school shows play here. (Subject is shows; the verb play is plural.)
b. The school show plays here. (Subject is show; the verb plays is singular).
c. The school shows plays here. (Subject is school; the verb shows is singular.)
d. The schools show plays here. (Subject is schools; the verb show is plural.)
Without agreement, the sentences are uninterpretable:

(191) a. *The school show play here.
b. *The schools shows plays here.
Since verbs agree with the head of the subject phrase, it is important that speakers be able to identify
the head. The three constraints mentioned above facilitate that identification: (i) only the head can
show number distinctions; (ii) the head must be separated from posthead constituents by a
characterizer; and (iii), if there is no characterizer after the head, then of must be inserted effectively
marking the head. It is interesting that the word that is inserted to mark the head is of, which is
perhaps the most generic preposition in English, often pronounced [c].
Given this analysis, we might have an explanation for why the past tense has become regular in
English for all persons and numbers, but the present tense has not. Compare the following.
(192) a. Past Tense:
played for all persons and numbers: She/They played.
b. Present Tense:
plays for third person singular; play elsewhere: She plays versus They play.
The basic problem is that singular and plural noun inflection is identical to verb inflection:
(193) a. play: either a singular noun or a plural present
b. plays: either a plural noun or a singular present
There are many nouns in English that also function as verbs. On the other hand, there are only a few
past tense verbs that are also nouns, and they are all irregular, e.g., cut and hit. Further, locating the
head of an English noun phrase is not a trivial matter. There are many opportunities for ambiguity.
All these potential ambiguities are eliminated by the rule of subject verb agreement. Quite simply,
without the inflectional differences between singular and plural count nouns and between singular
and plural present verbs, speakers would be unable to locate the head of a noun phrase. Thus, the
present tense of verbs cannot become regular
We can generalize the above description of the English N3 to V3 and account for the very large
number of nominals that are related to verbs, e.g., destruction and destroy, evolution and evolve,
advancement and advance, shipment and ship, etc. As we have seen, this is essential if we are to
generalize notions like subject–of and object–of across categories: we want to be able to relate the
N3 his insistence on that to the V3 he insisted on that, the N3 her unexpected refusal to comply to
the V3 she unexpectedly refused to comply, and so on. Consider the following representations which
illustrate the close relationships between V3 and N3:
177

178
(194) The Mafia illegally shipped drugs into the country during June.
(195) the Mafia’s illegal shipments of drugs into the country during June...

The representations in (194) and (195) make significant improvements in the search for descriptive
adequacy. For example, notice that such representations as the above make possible a generalized
statement of subject and object. The subject N3 (the Mafia) has the same relationship to the V3 head
shipped in (194) as it has to the N3 head shipment in (195). Similarly, the object N3 (drugs) has the
same relationship to both V1 in (194) and N1 in (195); the adverb illegally is in the same relationship
to shipped as the adjective illegal is to shipment. In short, the RG framework goes a long way in
solving the problems in previous discussions.
In (194), PST is the morpheme for the past tense discussed earlier (see Page 136). Thus, TENSE
markers in V3 occupy the same residence as DETERMINERS and POSSESSIVES in N3. In (195),
POS stands for the possessive suffix, which is usually either [s] as in book’s (the book’s cover), [z]
as in John’s (John’s son), or [cz] as in Bess’s (Bess’s boss). Notice that, in forming a possessive like
Mafia’s, the noun Mafia is under the N3 and POS is under the C3. This is necessary because English
contains expressions like the boy with the stick’s hand is sore, where the possessive phrase consists
of the N3 the boy with the stick followed in a separate node by the possessive suffix. Notice that we
do not say *The boy’s with the stick hand is sore.
With these improvements, we are able to represent a complex example like (119h) as follows:
(196) all those ten recently stolen Impressionist paintings of landscapes of the Louvre’s on sale
179

180
Another example of the relationship between V3 and N3 is the following:
(197) The friendly flight attendant patiently described the safety precautions to the nervous
passengers in their native language.
(198) the friendly flight attendant’s patient description of the safety precautions to the nervous
passengers in their native language...

Observe again that, by convention, nodes do not appear in trees explicitly unless they branch:
(199) this very young student (200) this Renaissance history student
(201) this very curious very young Renaissance history student
181

182
The complete RG morphosyntactic feature matrix is summarized in Figure Eight. Elsewhere in this
text, syntactic categories are often referred to using the following abbreviations:
V: VERB ADV: MANNER ADVERB
AUX : AUXILIARY PPN: PREPOSITION
MOD : MODAL SCJ: SUBORDINATING CONJUNCTION; SUBORDINATOR
N: NOUN CCJ: COORDINATING CONJUNCTION; COORDINATOR
PRO: PRONOUN CPL: COMPLEMENTIZER
QNT: QUANTIFIER DET: DETERMINER
ADJ: ADJECTIVE DGR: DEGREE WORD
One of the most significant advantages of the above framework is that it enables us to state the formal
universal (202):
(202) [+NML] categories must precede [–NML] categories in both prehead and posthead position
on every X level.
Given the matrix in Figure Eight, the potential difficulty of constructing a revealing set of phrase
structure rules is solved. All possible phrase structure sequences in English are reducible to one
PHRASE STRUCTURE FILTER (PSF), namely, (203a) which is equivalent to the tree in (203b).
(203) a. [X n ([+NML]) ([–NML]) X m ([+NML]) ([–NML]) ]
b.
where m # n
(203) asserts that each X level must dominate a lower X level with an equal or lesser numerical value,
e. g., X2 over X2, X1, or X0 but not X2 over X3. Further, (203) asserts that each X level may have
at most one [+NML] and one [–NML] category, in that order, to the right and/or to the left of the
head X.
Given (203), we have solved the categorial problem and can eliminate from our grammar complex
and unrevealing PS–Rules like those in (89).

FIGURE EIGHT: FEATURES FOR ENGLISH MORPHOSYNTACTIC CATEGORIES
CATEGORIES & EXAMPLES VBL NML OPH OCL PRH PSH ENH EVH ECH X1L X2L X3L
VERB: go, goes, went + – + + – + + + + + + +
AUX: have, has, had + – – – + + – + + – – –
MODAL: can, could, must + – – – + – – – + – – –
NOUN: boy, milk, Mary – + + + + + + + + + + +
PRONOUN: she, he, herself – + – – + + + + + + + +
QUANTIFIER: all, both, half – + – – + – + – – – – +
DETERMINER: a, the, this – – – – + – + – – – – +
COMPLEMENTIZER: that, if – – – – + – – + – – – +
DEGREE WORD: so, too, this – – – – + – + – – – – +
PREPOSITION: in, out, after – – + – – + + + + + + +
SUBORDINATOR: since, after – – + – – + + + + – + +
COORDINATOR: and – – – – – + + + + + + +
ADJECTIVE: happy, big – – + + – + + + – + + –
MANNER ADVERB: happily – – + + + + – + – – + –
SENTENCE ADVERB: probably – – – + + + – + + – – –
VBL: VERBAL inherently marked for distinctions in tense, aspect, voice, and/or mode
NML: NOMINAL inherently marked for distinctions in number, person, gender, and/or case
OPH: OPEN PHRASE occurring as the head of a phrase that freely contains other elements
OCL: OPEN CLASS unlimited in number
PRH: PREHEAD can freely occur before the head of a phrase
PSH: POSTHEAD can freely occur after the head of a phrase
ENH: ENVIRONMENT OF N can freely occur immediately dominated by N3, N2, or N1
EVH: ENVIRONMENT OF V can freely occur immediately dominated by V3, V2, or V1
ECH: ENVIRONMENT OF C can freely occur immediately dominated by C3, C2 or C1
X1L: X1 LEVEL can freely occur immediately dominated by X1 (N1, V1, or C1)
X2L: X2 LEVEL can freely occur immediately dominated by X2 (N2, V2, or C2)
X3L: X3 LEVEL can freely occur immediately dominated by X3 (N3, V3, or C3)
183

184
(203) also makes clear the relationship between a tree diagram such as (203b) and an equivalent,
linear representation such as (203a), referred to as LABELED BRACKETS. For example, the
following representations are equivalent; again, note that labeled brackets always occur in left–right
pairs as the arrows indicate:
(204) ¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
¯¯¯¯ ¯¯¯¯¯¯ ¯¯¯¯¯¯¯ ¯¯¯¯¯
a. [ V3 [ N3 he] [ C3 PST] [ V1 [ V0 leave] [ N3 her] ] ]
b.
Note that her is the object of leave and is part of V1. As before, PST + leave = left.
Labeled brackets are difficult to read and, therefore, often simplified. For example,
[ N3 he] above is technically [ N3 [ N0 he]]; however, since the internal structure of the
noun phrase is clear, it is given in simplified form. If only the V1 is important to a
discussion, the above might even be simplified to [ V3 he PST [ V1 leave her]].
(205) ¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
¯¯¯¯ ¯¯¯¯¯¯ ¯¯¯¯ ¯¯¯¯¯¯¯¯¯¯¯¯¯¯
a. [ V3 [ N3 he] [ C3 PST] [ V2 [ V0 leave] [ C3 happy/happily] ] ]
b.
Note that happy/happily is a modifier of the predicate leave and is part of V2 (the
distinction between X1 and X2 is discussed further on the next page).

As the above diagrams show, the most significant difference between RG and other versions of
X–Bar syntax is the level distinctions, in particular, the status of the X2 level. In all phrases, the X2
level contains descriptive modifiers of the head. These may be either characterizers or nouns;
compare (206a) and (206b).
(206) a. a courageous act b. an act of courage.
In V3, the V2 level dominates manner adverbs and, in copulative constructions, predicate modifiers
of the subject:
(207) a. He acted courageously. b. He was courageous.
The parallelism between V2 and N2 can be seen in the following representations:
(208) a. It smells bad. [ V3 [ N3 it] [ C3 PRS] [ V2 [ V0 smell] [ C3 bad] ]]
b. It has a bad smell. [ V3 [ N3 it] [ C3 PRS] [ V1 [ V0 have] [ N3 [ C3 a] [ N2 [ C3 bad] [ N0 smell] ]] ]]
c. Its smell is bad. [ V3 [ N3 [ N3 it] [ C3 POS] [ N0 smell] ] [ C3 PRS] [ V2 [ C3 bad] ]]
185

186
The need for three X levels above the head X proceeds from the following assumptions we have
made all along:
(209) a. The rules of language are structure dependent, that is, they refer to structural units
or phrases.
b. The rules refer to whole phrases, not pieces of phrases. In terms of a tree diagram,
this means that rules make reference to nodes.
Given (209), consider the RG structure for this young history student:
(210)
Verification of each level proceeds as follows:
(211) a. The entire phrase (the whole N3) is replaced in the sentence This young history
student said that he would pass. (he = this young history student)
b. The anaphor one can occur with N2 elements: John met this young history student,
and Mary met that one. (one = young history student)
c. Conversely, one cannot occur with N1 elements: *John met a history student, and
Bill met a biology one. (one = student)
The need for the distinction between N2 and N1 is confirmed by a phrase like the Spanish Spanish
teacher. Without two levels below the, there would be no way to disambiguate the senses of
Spanish. The first one means from Spain; the second, of Spanish. Thus, the phrase equals the
teacher of Spanish from Spain (compare (212a) and (212b)).

(212) a. the Spanish Spanish teacher b. the teacher of Spanish from Spain
Residency requirements for this phrasal architecture predict the grammaticality of (213a) and (213b),
and the ungrammaticality of (213c) and (213d).
(213) a. the French math teacher
b. the teacher of math from France
c. *the math French teacher
d. *the teacher from France of math
In addition, these structures account for the ambiguity of phrases like the practical nurse, an
intellectual historian, and the issue of student grants. All of these facts result from the following
residency requirements of English: Descriptive modifiers of the head must hang from X2; elements
of compounds and complements must hang from X1.
Lastly, observe that this system allows the first member of a compound noun to be any one of the
three major syntactic categories: in math teacher the noun head (teacher) is preceded by another
noun (math); in physical therapist the head noun (therapist) is preceded by a characterizer
(physical); and, in go cart, the head noun (cart) is preceded by a verb (go).
187

188
The three level hypothesis for V3 can be confirmed with reference to the following structure:
(214) Sue will get the mail on Monday.
Given (209), verification of the three V nodes proceeds as follows:
(215) a. The existence of the V3 level is indicated by its pronominalization in the following
sentence: Sue will get the mail on Monday; at least, that’s what Bill says. (that =
Sue will get the mail on Monday)
b. The existence of the V2 level is indicated by its omission in the following sentence:
Sue will get the mail on Monday, and Mary will ___ too. (get the mail on Monday
is left out.)
Further, V2 level elements can occur after the pro–form do so as in the following:
Sue will get the mail on Monday, and Mary will do so on Tuesday. (do so = get the
mail)
c. The existence of the V1 level is indicated by its omission in the following sentence:
Sue will get the mail on Monday, but Mary won’t ____ until Tuesday. (get the mail
is left out.)
Further, V1 elements can NOT occur after the pro–form do so as the following
indicates: *Sue will get the mail on Monday, and Mary will do so the pay checks on
Tuesday.

The existence of V1 (direct object and object complement position) as opposed to V2 (predicate
nominative position, referring back to subject) also resolves the ambiguity in (216).
(216) a. John left the house messy.
(i) messy an object modifier:
[ V3 [ N3 John] [ C3 PST] [ V2 [ V1 [ V0 leave] [ N3 the house] [ C3 messy ] ] ] ]
(ii) messy a predicate adjective, subject to agreement rules:
[ V3 [ N3 John] [ C3 PST] [ V2 [ V1 [ V0 leave] [ N3 the house] ] [ C3 messy] ] ]
b. John defied Bill to get even.
(i) to get even a complementary infinitive with subject Bill:
[ V3 [ N3 John] [ C3 PST] [ V2 [ V1 [ V0 defy] [ N3 Bill] [ C3 to get even] ] ] ]
(ii) to get even a purposive infinitive with subject John:
[ V3 [ N3 John] [ C3 PST] [ V2 [ V1 [ V0 defy] [ N3 Bill] ] [ C3 to get even] ] ]
189

190
c. The boys made good models.
(i) make a transitive verb like build:
[ V3 [ N3 the boys] [ C3 PST] [ V2 [ V1 [ V0 make] [ N3 good models] ] ] ]
(ii) make a copulative verb like become:
[ V3 [ N3 the boys] [ C3 PST] [ V2 [ V1 [ V0 make] ] [ N3 good models] ] ]
In addition, the V1/V2 distinction explains the grammaticality of (217), and predicts the
ungrammaticality of (218).
(217) a. John ate the food raw, nude.
b. John left the house unlocked, drunk as a skunk.
(218) a. *John ate the food nude, raw.
b. *John left the house drunk, unlocked.
The V1/V2 distinction also accounts for the ambiguity of sentences like those in (219):
(219) a. They left good friends.
b. They left the country great men.
c. They considered the candidates very concerned about the future of the country.
The most striking feature of these examples is the parallelism between the N1/N2 distinction and
the V1/V2 distinction, clearly seen in the following:
(220) a. a young math teacher
[ N3 [ C3 a] [ N2 [ C3 young] [ N1 [ N3 math] [ N0 teacher ] ] ] ]
b. a teacher of math that young (is rare)
[ N3 [ C3 a] [ N2 [ N1 [ N0 teacher] of [ N3 math] ] [ C3 that young] ] ]
c. He taught math young.
[ V3 [ N3 he] [ C3 PST] [ V2 [ V1 [ V0 teach] [ N3 math] ] [ C3 young] ] ]
(221) a. a superb math teacher
[ N3 [ C3 a] [ N2 [ C3 superb] [ N1 [ N3 math] [ N0 teacher ] ] ] ]
b. a teacher of math that superb (is rare)
[ N3 [ C3 a] [ N2 [ N1 [ N0 teacher] of [ N3 math] ] [ C3 that superb] ] ]
c. He taught math superbly.
[ V3 [ N3 he] [ C3 PST] [ V2 [ V1 [ V0 teach] [ N3 math] ] [ C3 superbly] ] ] ]

3.13 TRANSFORMATIONAL GENERATIVE GRAMMAR
By themselves, phrase structure rules can describe a great number of the individual sentences that
occur in languages; however, if our corpus is expanded to include more types of constructions, it
appears that some relationships between sentences will not be able to receive adequate
characterizations with phrase structure rules alone. For example, consider sets of English sentences
like the following and note that each pair means basically the same thing:
(222) a. The city must burn this book.
b. This book, the city must burn.
(223) a. The frightened cat ran under the bed.
b. Under the bed, the frightened cat ran.
(224) a. He will never be very strong.
b. Very strong, he will never be.
The construction illustrated in (222b), (223b), and (224b) is TOPICALIZATION, the placement
of a phrase in sentence initial position for emphasis. Such examples rarely occur in isolation, which
is why they sound somewhat strange. However, in context they are not unusual. For example,
during a discussion of the books to be burned, (222b) would sound quite appropriate.
Students often claim that they would never utter topicalized phrases, but careful attention to
language reveals that, in fact, they do. Further, whether one utters topicalized phrases or not is
completely irrelevant. The fact is that such sentences are immediately understood by native speakers
in context. Given this, the sentences become part of the corpus of data that a linguist must analyze.
As our discussion of prescriptive versus descriptive grammar in Chapter One indicated, modern
linguistics is concerned with describing what is possible, not what is considered correct.
These three pairs of sentences begin with phrases that occur before the NP subject phrase. In (222b),
the subject phrase is preceded by another N3; in (223b), by a prepositional phrase; and, in (224b),
an adjective phrase. We could accommodate these data by adding a PS-Rule like the following
where X3 is a phrase with a [–VBL] head (noun, preposition, or adjective):
(225) V3 ÷ (X3) + N3 + T3 + V2
But this will not do for several reasons. First, the optional phrases at the beginning of (225) are
understood as the complements of the verbs, that is, phrases that complete the meaning of the verbs.
For instance, this book in (222b) is the object of burn. Our structural definition of object says that
the object is the N3 dominated by the V1. While this definition holds for (222a), it fails for (222b)
because the initial NP of (225) is not dominated by V1. Furthermore, our structural definition of
subject is now compromised. Given (225), there are now two N3 dominated by V3. Worse yet, if
we adopt (225), the whole classification of English verbs comes to grief. Broadly speaking, there
are two types of verbs in English: TRANSITIVE VERBS like catch, which must take a direct
object; and, INTRANSITIVE VERBS like die, which never do:
191

192
(226) a. John caught Bill.
b. *John caught.
(227) a. John died.
b. *John died Bill.
If, to account for (228), we allow transitive verbs to occur in verb phrases without a following N3,
we lose the distinction between transitives and intransitives.
(228) Foul balls like that, no one can catch.
Somehow, we must indicate that the initial phrases in (222b), (223b), (224b), and (228) are
complements of the main verb. Since a phrase structure solution via (225) failed, another method
must be sought.
PS–RULES generate tree structures with phrases in specific positions. Suppose that we now allow
grammars to contain other rules that move elements from such positions to new positions, that is,
transform one structure into another. For example, suppose that, as before, the PS–RULES in (89)
generate a structure like (229).
(229)
Now suppose a TRANSFORMATIONAL RULE moves the object NP to the front of the sentence
leaving behind a TRACE (a kind of footprint) of that object in the form of a co–indexed NP. The
result is (230), where "[e]" indicates the EMPTY NP TRACE designating the original home of the
object.

(230)
This produces the sentence (231a) or, more accurately, (231b).
(231) a. This book, the city must burn.
b. [ NP this book] i the city must burn [ NP e] i
A crucial point in the representation (231b) is that there really is a gap in the sentence at the point
of the [e]. When speakers hear a sentence that begins with a “displaced” object, they must wait for
the gap in the sentence before they can assign an interpretation to that “displaced” phrase. Consider
(52).
(232) a. This book, you absolutely must try to get your students to convince each other to tell
their friends to read.
b. [ NP this book] i you absolutely must try to get your students to convince each other to
tell their friends to read [ NP e] i
Given representations like the above, we can say that [ NP e] i is bound to [ NP this book] i via
REFERENCE, not unlike the way pronouns are bound to their referents in sentences like John i saw
himself i in the mirror.
A grammar that contains transformational rules is called a TRANSFORMATIONAL
(GENERATIVE) GRAMMAR (TG). In such grammars, a sentence like (231a) is associated with
two structures: the structure generated by the phrase structure rules, that is, (229); and the structure
generated by the transformational rules, that is, (230). Technically, we say that phrase structure
rules generate the DEEP STRUCTURE of sentences, and transformational rules generate the
SURFACE STRUCTURE. Crucially, functional relations like subject and object are not altered
by transformational rules. However, the rules do switch the order of constituents in a sentence
meaning that the order in the deep structure can be different from the order in the surface structure.
More importantly, our description of the grammars of human languages has now changed.
Originally, we claimed that human languages could be described by PS-Rules alone. Now, we must
say that they require an additional rule, the Transformational Rule or T-Rule.
193

194
3.14 NONTRANSFORMATIONAL GENERATIVE GRAMMAR
With transformational rules, we say that human languages cannot be adequately described without
the following two assumptions:
(233) a. Every sentence has two structural levels, an abstract underlying structure roughly
equivalent to what is understood, and a real superficial structure roughly equivalent
to what is heard.
b. The order of phrases in the abstract underlying structure can be different from the
order of phrases in the real superficial structure.
It seems quite clear that (233a) must be assumed. The syntactic processes of REFERENCE,
OMISSION, and PLACEMENT demand it. For example, there is no way to understand the
superficial structure (234a) without the underlying representation (234b).
(234) a. John will sing if you ask him to.
b. John will [VP sing ] i if you ask him to [VP e ] i
On the other hand, (233b) is not a necessary assumption. That assumption is linked to
transformational grammar. Rather than postulating a new rule type that moves constituents to
different positiions leaving behind a co-indexed empty category, supposed we allow [e] to be
directly generated as follows:
(235) XP ÷ [e]
If we allow empty phrases to be generated directly via PS–Rules, then we can give up (233b). All
that remains is to propose principles for finding the referent for [e] something that we need to do
anyway to account for sentences like the following which contain co-indexed constituents:
(236) a. After John i got up, he i had a big breakfast.
b. After he i got up, John i had a big breakfast.
c. First John i entered the room, and then the SOB i insulted the hostess.
Therefore, we do not need transformational rules at all. What we need is a mechanism for relating
[NP e] to full noun phrases, something that is needed anyway. Again, it is useful to stop at this point
and take stock of what we are trying to accomplish. Native speakers of every language completely
understand every possible sentence in their language. The sentence that you are now reading is a
sentence which you have probably never encountered before in your life, and yet you understand
it immediately as you read it. It must be that, at some unconscious level, you understand precisely
the relationships we are trying to uncover. The number of sentences in every language is infinite;
learning a language does not consist of memorizing a list of the possible sentences. Rather, it must
consist of learning the principles upon which every possible sentence is constructed. These
principles form native speakers’ unconscious knowledge of their language. Throughout this

discussion, we have been trying to ascertain just what this knowledge is, that is, what must native
speakers know to understand all the sentences we have been considering.
3.15 COMMAND RELATIONS IN PHRASE STRUCTURE
We are now in a position to formalize relationships like SUBJECT and OBJECT and to account for
so–called displaced phrases. Consider first (229) recast with the RG innovations as (237).
(237)
In a diagram like (237), it is possible to follow a path from one X category to another. Some paths
prove to be crucial in determining syntactic relationships. One particularly useful relationship in
syntax is the C–COMMAND (for CATEGORY COMMAND) relationship, which we define as
follows:
(238) A node α C–commands a node β if the node immediately above α dominates β.
Applying (238) to the diagram in (237), we see that the N3 the city C–commands the verb burn
because the node immediately above that N3 is the top V3 node and that top V3 node dominates the
verb burn. Similarly, the N3 this book C–commands burn because the node immediately above that
N3 is the V1 in the diagram and that V1 dominates the verb burn.
As we will see, this relationship is at the basis of all the rules for finding the referents of pronouns
in English. For now, we will use it to formally define the functional notions subject–of and
object–of:
(239) a. The subject of some node β is the N3 which most immediately C–commands and
precedes β.
b. The object of some node β is the N3 which most immediately C–commands and
follows β.
The phrase most immediately means ‘is closest to’ or ‘is separated by the fewest number of
intervening nodes in the tree.’
195

196
Given (238) and (239), we determine the subject and object of a verb such as the one diagramed in
(237) as follows:
(240) a. The N3 above city C–commands and precedes the verb burn. The N3 is α; the V0
node burn is β; the node immediately above α, which is the topmost V3 in the
diagram, dominates β. Since α precedes β, α (the city in this example) is the subject
of β (burn).
b. The N3 above book C–commands and follows the verb burn. The N3 is α; the V0
node burn is β; the node immediately above α, which is V1, dominates β. Since α
is to the right of β, α (this book in this example) is the object of β (burn).
The same definitions will work for all the other relevant diagrams in this text. For example, The
Mafia is the subject and drugs the object of ship in (197) and shipment in (198); and Sue is the
subject and mail the object of get in (214).
As we have seen, RG permits the terminal nodes of (203) to dominate two kinds of empty
categories, that is, categories that are phonologically null: [u], which is an understood element in
a sentence like Let’s eat [u]; and [e], which is a bound ANAPHOR, that is, an item that requires a
referent. The availability of [e] together with (238) will account for the interpretation of so–called
displaced constituents. Consider again (231) recast with the RG innovations as (241).
(241)
As in (237), city is the subject of burn. Importantly, book is not the subject. Although the N3 above
book C–commands burn, it does not immediately C–command it. The N3 above city is closer.

In (241), the empty N3 most immediately C–commands and follows burn, so it is the object;
however, it is also an anaphor, so it must have a referent. To find the referent, we use the following
concepts.
(242) IMMEDIATE NEIGHBORHOOD: An immediate neighborhood embraces all phrases
that are dominated by the first X3 above an X3 level prehead characterizer (DET or POS
in N3; CPL or TNS in V3).
(243) EXTENDED NEIGHBORHOOD: An extended neighborhood embraces all X3
recursions above the immediate neighborhood.
(244) THE EMPTY CATEGORY CONDITION (ECC): An empty syntactic category can
only occur in a position that can contain a fully specified phrase.
Given these specifications, consider (241). The tense characterizer PRS is an X3 Level prehead
characterizer; therefore, all items within the first V3 above it are in its immediate neighborhood.
This includes the anaphor [N3 e]. The phrase this book is in a V3 recursion above the immediate
neighborhood; hence, it is in an extended neighborhood. Both this book and the [e] are phrases of
the same type; accordingly, this book is the referent for [e] and the object of burn even though it
precedes the subject.
While the definitions and conditions presented in this section have been simplified for expository
purposes, they nonetheless satisfy all the major difficulties that have been presented in previous
sections. There are, quite naturally, some more difficult examples to be considered, but the essential
direction of the solution presented here is correct and extends, with some elaboration, to other very
complicated examples.
The crucial result of our investigation is that it is possible to formalize even complex relationships
in terms of linear and hierarchical structure. We have succeeded in reducing all phrases to variations
on one frame (203). Furthermore, we have provided formal, explicit definitions of relations like
SUBJECT, OBJECT, REFERENCE, and so on. We have done this without transformational rules.
In short, our entire solution is base on the concept phrase that is rooted in human cognitive capacity.
3.16 EMPTY CATEGORIES
As we have seen, sentences sometimes contain missing elements or gaps which speakers must fill
either from information elsewhere in the sentence or from some general principles. Examples of
sentences with gaps are the following:
(245) a. You can come if you want to ____.
b. He shrugged ____.
197

198
Native speakers know intuitively that the gaps in the above sentences should be filled as follows:
(246 ) a. You can come if you want to come.
b. He shrugged his shoulders.
There are two major types of empty categories in language: [e], a bound element that has a referent
somewhere inside the same sentence, and [u], an understood element. Consider the following
examples, where [1] is the sentence as it is spoken, where [2] indicates the way it must be
understood, and where the subscripts indicate which two elements are bound (refer to each other):
(247) a. [1] You can come if you want to.
[2] You can come i if you want to [e] i. (The [e] is bound to come.)
b. [1] What did he eat today?
[2] What i did he eat [e] i today. (The [e] is bound to what, the logical object of
eat.)
c. [1] He shrugged.
[2] He shrugged [u]. (The [u] indicates some understood phrase.)
Note that [e] is bound to some referent elsewhere in the sentence, but [u] is unbound. In other
words, the meaning of [e] is recovered from information given elsewhere in the sentence, whereas
the meaning of [u] must be supplied by some other general principles. In the case of (247c), shrug
belongs to class of English verbs, body-part verbs, which allow the omission of a specific body-part.
Membership in this class must be learned. For example, the class includes blink, clap, nod, and
wave, but not braid, crane, pluck, or sprain. While it’s possible to say She blinked, it is not possible
to say *She sprained. Speakers must simply learn which verbs allow an understood body-part.
As the above examples indicate, an empty category can only occur in a position that can contain a
fully specified phrase. Each [e] and [u] above can be replaced by a fully specified phrase. [u]
occurs in a position that can contain an NP; [e] occurs in a position that can contain its referent.
Conversely, [e] and [u] cannot occur where fully specified phrases cannot occur:
(248) a. John persuaded Bill that Mary should go.
b. *John persuaded Bill Mary to go.
c. *Who i did John persuade Bill [e] i to go?
(249) a. He donates money to charities.
b. He donates [u] to charities.
c. *He donates charities money.
d. *He donates charities [u].
Given the above, we can add the following condition to our grammar:

(250) THE EMPTY CATEGORY CONDITION (ECC)
An empty syntactic category can only occur in a position that can contain a fully specified
phrase.
The data above indicate clearly that grammars must contain empty categories as well as rules for
where empty categories can or must occur. Understandably, the matter can be complicated.
Consider, for example, the following sentence:
(251) Who did John promise that Bill would visit?
This sentence has two possible meanings, which are disambiguated by the following representations:
(252) a. Who i did John promise [e] i that Bill would visit [u]?
b. Who i did John promise [u] that Bill would visit [e] i?
In (252a), the question being asked is who John promised, whereas in (252b) the question being
asked is who Bill would visit. Note that only [e] is bound to who, and that empty categories can
occur in the middle of sentences.
For a sentence like Who did John promise that bill, the grammar must generate the following:
(253) Who i did John promise [e] i that bill?
Note that (254) should not generated because then there would be no slot for an [e] bound to who:
(254) *Who did John promise [u] that bill?
Declarative variants of (254) do tolerate a [u] in object position for some speakers:
(255) a. John promised me that bill.
b. John promised [u] that bill. (cf. I never promised you/[u] a rose garden.)
Crucially, [e] and [u] must occur in every WH–question where a verb allows a [u] object; otherwise,
grammars would not produce the correct reading for classic examples like the following:
(256) a. What i did the teacher read [u] to the children from [e] i?
b. *What i did the teacher read [e] i to the children from [u]?
An important feature of empty categories is that they block contraction so that grammars of natural
languages must contain the following condition:
199

200
(257) THE CONTRACTION BLOCK CONDITION (CBC)
Contraction is not possible over an empty category.
Examples of this condition involving [e] include the following:
(258) a. Who i did you want [e] i to win the race?
b. *Who did you wanna win the race?
(259) a. Could i we [e] i have gone there?
b. *Could we’ve gone there? (cf. We’ve gone there.)
The empty [e] found in passive constructions also blocks contraction:
(260) a. He i was dared [e] i to challenge her.
b. *He was [dærDc] challenge her. (cf. Did he [dærDc] challenge her?)
(261) a. John i is said [e] i to be a good teacher.
b. *John is [seDc] be a good teacher.
The empty category [u] also blocks contraction:
(262) a. Sue will now dress [u] to get ready for the prom. (reflexive object understood)
b. *Sue will now [dresc] get ready for the prom.
(263) a. They met [u] to discuss the wedding. (reciprocal object understood)
b. *They [meDc] discuss the wedding.
(264) a. Sue will nod [u] to indicate her approval. (understood body part)
b. *Sue will [naDc] indicate her approval.
(265) a. They eat [u] to live. ([u] = some type of food)
b. *They [iDc] live.
The CBC also extends to idioms such as the following where a very specific N3 is understood:
(266) a. Did they get a chance to see that play when they were in town?
b. Did they get [u] to see that play when they were in town.
c. *Did they [geDc] see that play when they were in town.
Examples like (266) contrast with the following where no empty category occurs:
(267) a. They have got to see that play.
b. They’ve [gaDc] see that play.

Consider now a sentence like (268), which must be related to the sentences in (269).
(268) When was Sally interviewed?
(269) a. Someone interviewed Sally at a particular time.
b. Sally was interviewed.
c. Someone interviewed Sally.
The question (268) is clearly related to the statements in (269). Notice that (268) presupposes, in
fact, that there was someone who actually interviewed Sally at a particular time. As a result, the
grammar should give (268) the representation in (270).
(270) When i was j Sally k [e] j interviewed [e] k [e] i?
201
a. The empty category with subscript “i” marks the ordinary position of a phrase
expressing time (compare (269a)).
b. The empty category with subscript “j” marks the ordinary position of the auxiliary
in a sentence (compare (269b)).
c. The empty category with subscript “k” marks the ordinary position of a direct object
(compare (269c)).
It is clear from examples such as the ones above that empty categories are necessary to give
sentences the full interpretation of their meaning. All native speakers know that there is a gap in a
sentence like He can come if he wants to, and that the gap can only be filled by come. As a result,
the complexity that empty categories introduce into a grammar is unavoidable if one wishes to
impart meaning to sentences.
Note that the matter is not simply one of providing a meaningful interpretation to a sentence. Empty
categories are justified in representations on purely linguistic grounds. Observe that the following
sentences are ungrammatical:
(271) a. *He wants to.
b. *He admires.
(271a) is ungrammatical because the infinitive construction (the one introduced by to) requires a
verb phrase after the to; similarly, (271b) is ungrammatical because admire is a transitive verb,
meaning it requires a direct object. The following representations satisfy these constraints:
(272) a. He can come i if he wants to [e] i. ([e] i = come)
b. Who i does he admire [e] i?
As these examples indicate, inserting empty categories into syntactic representations is a necessity.
Actually, it is a biological necessity since human beings are not clairvoyant. Elements cannot be
left out of sentences unless there are principles that can recover them, because human beings can’t

202
read minds. In fact, we can propose the following universal condition operative in all human
languages:
(273) THE RECOVERABILITY OF DELETION CONDITION (RDC)
All gaps in sentences must be recoverable.
The ECC, the CBC and the RDC are natural conditions for a grammar to contain. Human beings
are not clairvoyant (generally). Since the ECC stipulates that an empty category can only occur in
a position where a fully specified category can occur, empty categories are associated with positions
where speakers expect something to occur. The RDC stipulates that an empty category in those
positions must be recoverable. The position occupied by [u] is one that is usually filled by a fully
specified phrase; the one occupied by [e] is the position occupied by its referent in a given sentence.
Contraction over empty categories would have the effect of eliminating those positions; in essence,
speakers would be unable to fill the empty positions with appropriate phrases because the empty
positions would have disappeared in contraction.

3.16 THE ENGLISH AUXILIARY
3.16.1 THE FACTS AND GENERALIZATIONS
Chomsky’s original analysis of the English auxiliary, along with subsequent work in the framework
of transformational grammar (TG), revealed a number of important facts about the English auxiliary,
which are summarized in (274).
(274) a. Every English independent clause (S/V3) must contain a tense (TNS) marker:
1. PRS (PRESENT): hunts, has hunted, is hunting
2. PST (PAST): hunted, had hunted, was hunting
3. FUT (FUTURE): will hunt, will have hunted, will be hunting
4. CND (CONDITIONAL): could/would/should hunt
5. IMP (IMPERATIVE): hunt/do hunt
203
b. The verb that carries the TNS marker is always the first verb in the verbal sequence.
This verb also carries agreement: She is hunting/*They is being hunted/*She have
been hunting.
c. Every English clause must also contain a main verb, which is the last verb in the
verbal sequence. This main verb may be empty: The women have hunted, and the
men have [e] too.
d. Every English clause may also optionally contain a MODAL, PERFECTIVE,
PROGRESSIVE, and/or PASSIVE element in that order. Thus:
TNS (MOD) (PERF) (PROG) (PASS)
e. The PERF, PROG, and PASS each consist of an auxiliary verb and a participial affix
which always appears on the next verb in the sequence, whatever that verb is (the
PROG affix is always –ing; the PERF and PASS affixes are –ed in regular verbs):
PERF = have + PFP (PERF PARTICIPLE AFFIX): has hunt–ed
PROG = be + PGP (PROGRESSIVE PARTICIPLE AFFIX): is hunt–ing
PASS = be + PSP (PASSIVE PARTICIPLE AFFIX): is hunt–ed

204
f. Further examples of participles are:
PROGRESSIVE PARTICIPLES PASSIVE/PERFECTIVE PARTICIPLES
hunting hunted
going gone
weeping wept
being been
having had
doing done
These facts are realized in English in sixteen possible auxiliary sequences:
(275) a. Women hunted. TNS
b. Women could hunt. TNS + MOD
c. Women had hunted. TNS + PERF
d. Women were hunting. TNS + PROG
e. Men were hunted. TNS + PASS
f. Women could have hunted. TNS + MOD + PERF
g. Women could be hunting. TNS + MOD + PROG
h. Men could be hunted. TNS + MOD + PASS
i. Women have been hunting. TNS + PERF + PROG
j. Men had been hunted. TNS + PERF + PASS
k. Men were being hunted. TNS + PROG + PASS
l. Women could have been hunting. TNS + MOD + PERF + PROG
m. Men could have been hunted. TNS + MOD + PERF + PASS
n. Men could be being hunted. TNS + MOD + PROG + PASS
o. Men have been being hunted. TNS + PERF + PROG + PASS
p. Men could have been being hunted. TNS + MOD + PERF + PROG + PASS
Listing the possible combinations is hardly an adequate description of English: it attains only
observational adequacy. The TG analysis has gone much further. It notes that all the above
combinations are variations of the sequence in (3).
(276) TNS (MOD) (have + PFP) (be + PGP) (be + PSP) V
The major coup was realizing that the affixes (TNS, PFP, PGP, and PSP) are always associated with
the following verb, whichever happen to come next in the sequence. Given this, a transformational
analysis is necessary; it must take a phrase structure sequence like (3) and distribute the affixes to
the following verb, whatever that might be. Consider the following examples:

(277) TNS + V
PST + hunt
.)))0)))hunted
(278) TNS + MODAL + V
CND + can + hunt
.)))0))))could
hunt
(279) TNS + MODAL + PERF + PROG + V
CND + can + have + PFP + be + PGP + hunt
.)))0)))- .))0))- .)))0)))-
could have been hunting
(280) TNS + PERF + PROG + PASS V
PRS + have + PFP + be + PGP + be + PSP + hunt
.)))0))))- .))0))- .))0)))- .))0))has
been being hunted
205

206
3.16.2 PROBLEMS WITH THE CLASSIC TG ANALYSIS
Despite the elegance of this treatment of the English auxiliary, it was soon realized that it could not
be maintained in its original form as described above. Three major problems were encountered.
First, it was realized that the auxiliary elements (TNS, MOD, PERF, PROG, PASS) do not have the
same range of occurrence possibilities. In particular, the following split exists:
(281) TNS and MOD cannot occur in infinitives (to go) and participles (going):
a. I wanted to be seen.
*I wanted to was seen.
I wanted to be able to drive.
*I wanted to can drive.
b. Being seen with her is a sign of success.
*Wasing seen with her is a sign of success.
Being able to drive is a sign of success.
*Canning (to) drive is a sign of success.
(282) PERF, PROG, and PASS can occur in infinitives and participles:
a. PERF:
It was fun [to have done that].
[Having done that] he left.
b. PROG:
I want [to be working] when he arrives.
?[Being hunting] when you should be working is bad.
c. PASS:
(283) BASIC SPLIT:
He tried [to be seen].
[Being invited to such parties] is not important.
(MOD) TNS – (PERF) (PROG) (PASS) V
This split is confirmed in two ways:

(284) An negative (not) can modify either constituent:
a. He can not attend. (ambiguous)
b. He can’t just not attend.
(285) When the modal occurs, it inverts in questions:
a. Can he go? (cf. He can go.)
b. Won’t they go? (cf. They won’t go.)
Adopting either a TG or RG framework, we can account for the above split with the following
diagrams:
(286) You can not attend.
[ S [ NP you] [ AUX [ TNS [ MOD can]] PRS [ NEG not] ] [ VP [ V attend] ]]
[ V3 [ N3 you] [ C3 [ C0 [ V3 can]] PRS [ C3 not] ] [ V0 attend] ]
[ S [ NP you] [ AUX [ TNS [ MOD can]] PRS ] [ VP [ NEG not] [ V attend] ]]
[ V3 [ N3 you] [ C3 [ C0 [ V3 can]] PRS ] [ V2 [ C3 not] [ V0 attend] ]]
The second problem with the classic TG analysis of the auxiliary concerns the use of do. If a clause
contains only TNS, then the verb do must occur in questions:
(287) a. He will go. Will he go?
b. He has gone. Has he gone?
c. He is going. Is he going?
(288) a. He went. *Went he?
b. He went. Did he go?
This do acts like a modal for two reasons:
(289) a. It carries TNS, agreement, and NEG:
Doesn’t – he – want that?
Don’t – they – want that?
Did – they – want that?
b. It does not occur in infinitives and participles:
*I want to do go there.
*Doing go there is fun.
207

208
This do must be distinguished from the main verb do as follows:
(290) MODAL MAIN VERB
a. He does that.
b. He does see that.
c. He does do that.
d. He doesn’t do that.
e. He does so do that.
But, do (unlike the other modals, e.g., can) cannot occur with PERF, PROG, and/or PASS:
(291) a. He can be working.
b. *He does be working.
c. He could have gone.
d. *He did have gone.
If we treat do as a modal, we can add a constraint to rule out occurrences of do with PERF, PROG,
and/or PASS. Recalling that PERF, PROG, and PASS all consist of a verb, we can add filter (292)
to our grammar (where W, X, Y and Z are variables):
(292) a. TG: *[ S W do X TNS Y V Z]
b. RG: *[ V3 W do X TNS Y V Z]
This filter, like most syntactic filters in English, is under a clause constraint, meaning that the listed
morphological elements do and TNS must be in the same clause (S or V3). This proviso is
necessary to disallow *he does be going while allowing does he think he is going.
Given this analysis, do occupies the same residence as the modals generally do in addition to main
verb position:
(293) He does not do that.
[ S [ NP he] [ AUX [ TNS [ MOD does] PRS] [ NEG not] ] [ VP do that] ]
[ V3 [ N3 he] [ C3 [ C0 [ V3 does] PRS] [ C3 not] ] [ V0 do that] ]
The third problem with the classic TG analysis of the auxiliary is the most serious. It concerns the
constituency of the various elements. It is here that RG provides an analysis that is superior to the
classic TG analysis. In particular, the classic TG analysis leaves the following facts unexpressed:

(294) The PROG and PASS can occur, as participles, without their associated verbs:
a. The man lurking in the shadows is the culprit.
The man who is lurking in the shadows is the culprit.
b. The man discovered in the bushes is a flasher.
The man who was discovered in the bushes is a flasher.
c. We saw the man hunting.
d. We saw the man hunted.
(295) PGP and PSP forms are ambiguous:
a. an imposing Avon lady
an Avon lady imposing on her neighbors (PARTICIPLE)
an Avon lady imposing in size (ADJECTIVE)
b. a broken vase
a vase accidentally broken by a child (PARTICIPLE)
a vase so broken that it can't be fixed (ADJECTIVE)
These facts can be accommodated within the RG framework by observing that participial markers
behave like characterizers, in particular, like prepositions. Thus, (296a), (296b), (296c), and (296d)
are analogous structures; significantly, the post-verbal characterizers (the C3 off V2 to the right of
V0) are independent structures, thus allowing for participial examples like those in (294):
(296) a. The jars started to break.
[ V3 [ N3 the jars] [ C3 PST] [ V2 [ V0 started] [ C3 [ C1 [ C0 to ] [ V3 [ V0 break] ]]]]
b. The jars started breaking.
[ V3 [ N3 the jars] [ C3 PST] [ V2 [ V0 started] [ C3 [ C0 [ C0 PGP ] [ V3 [ V0 breaking] ]]]]
c. The jars were breaking.
[ V3 [ N3 the jars] [ C3 PST] [ V2 [ V0 were] [ C3 [ C0 [ C0 PGP ] [ V3 [ V0 breaking] ]]]]
d. The jars were broken.
[ V3 [ N3 the jars] [ C3 PST] [ V2 [ V0 were] [ C3 [ C0 [ C0 PSP ] [ V3 [ V0 broken] ]]]]
209

210
If we treat be as a main verb and treat PGP and PSP as characterizers, we solve the above problems
in (294) and (295). Specifically, we take advantage of the fact that RG makes transparent the many
left-of head/right-of-head alternations that occur in English.
English is a head–center language, that is, one that tolerates elements both before and after the head;
many phrases can occur alternatively in prehead or posthead position. For example, we have noun
phrase pairs like the following:
(297) a. [ N3 the [ N2 Spanish [ N1 math [ N0 teacher]]]]
b. [ N3 the [ N2 [ N1 [ N0 teacher] of math] from Spain]
We have other examples of prehead/posthead alternations like the following:
(298) a. V3: Ultimately, peace will prevail/Peace will prevail, ultimately.
That guy there, he’s my friend./He’s my friend, that guy there.
b. N3: enough money/money enough
as many too many marbles/as many marbles too many
c. C3: quite far down the road/down the road quite far
two miles down the road/down the road two miles
(299) a. V2: enthusiastically sang the aria/sang the aria enthusiastically
often sang the aria/sang the aria often
b. N2: a sleeping baby/a baby sleeping
a 5000 foot high mountain/a mountain 5000 feet high
a courageous man/a man of courage
c. C2: directly down the road for a mile/down the road directly for a mile
especially for Sue/for Sue especially
In English, complements are generally posthead, although there are productive alternatives in noun
phrases (history student versus student of history; *history studied versus studied history) and
participial phrases (French speaking students versus students speaking French). Depending on the
language (head center, head initial, head final), complements generally either precede or follow the
head.
When a verb is embedded in prehead position of PGP or PSP it forms a kind of compound adjective
with the affix; in posthead position, the embedded verb is the complement of the affix:

(300) a. The lady was imposing (PROGRESSIVE).
[ V3 [ N3 the lady] [ C3 PST] [ V2 [ V0 was] [ C3 [ C0 [ C0 PGP ] [ V3 [ V0 imposing] ]]]]
b. The lady was imposing ( DERIVED ADJECTIVE 1).
[ V3 [ N3 the lady] [ C3 PST] [ V2 [ V0 was] [ C3 [ C0 [ V3 [ V0 imposing] ] [ C0 DA1 ] ]]]
(301) a. The jars were broken (PASSIVE).
[ V3 [ N3 the jars] [ C3 PST] [ V2 [ V0 were] [ C3 [ C0 [ C0 PSP ] [ V3 [ V0 broken] ]]]]
b. The jars were broken (DERIVED ADJECTIVE 2).
[ V3 [ N3 the jars] [ C3 PST] [ V2 [ V0 were] [ C3 [ C0 [ V3 [ V0 broken] ] [ C0 DA2 ] ]]]
If we allow that a main verb can be empty [e] and bound to be in the TNS characterizer, we account
for the occurrence of be in questions:
(302) The jars all were broken.
[ V3 [ N3 the jars] [ C3 [ C0 PST] ] [ V2 [ N3 all] [ V0 were] [ C3 broken]] ]
(303) The jars were all broken (PASSIVE). [= All the jars were broken (PASSIVE).]
[ V3 [ N3 the jars] [ C3 [ C0 were i [ C0 PST]] ] [ V2 [ N3 all] [ V0 [e] i] [ C3 broken]] ]
In the process, we also account for the ambiguity of all.
(304) The jars were all broken (ADJ). [= All the jars were broken (ADJ).]
[ V3 [ N3 the jars] [ C3 [ C0 were i [ C0 PST]]] [ V2 [ N3 all] [ V0 [e] i] [ C3 broken]] ]
(305) The jars were all broken (ADJ). [= The jars were totally broken.]
[ V3 [ N3 the jars] [ C3 [ C0 were i [ C0 PST]]] [ V2 [ V0 [e] i] [ C3 [ N3 all] broken]] ]
(306) The jars were all all broken (ADJ). [= All of the jars were totally broken (ADJ).]
[ V3 [ N3 the jars] [ C3 [ C0 were i [ C0 PST]]] [ V2 [ N3 all] [ V0 [e] i] [ C3 [ N3 all] broken]] ]
The above analysis solves a multitude of problems for English syntax. Notice that the participial
characterizer phrases are all suspended from the V2 level. This is the residence in RG of predicate
nominatives: items suspended from this position refer back to the subject. Consider the following
sentences and their associated RG diagrams:
211

212
(307) The boys made good cooks. (make is copulative)
[ V3 [ N3 the boys] [ C3 [ C0 PST]] [ V2 [ V0 made] [ N3 good cooks] ] ]
(308) The boys made good cakes. (make is transitive)
[ V3 [ N3 the boys] [ C3 [ C0 PST]] [ V1 [ V0 made] [ N3 good cakes] ] ]
Observe that the only structural difference between (307) and (308, 309) is that the postverbal N3
in (307) comes off V2, while the one in (308, 309) comes off the V1. This is in accordance with the
RG residential requirements: predicate nominatives are V2 level elements; direct objects are V1
level elements. Since predicate nominatives refer back to the subject, they must agree with the
subject. In short, the syntactic hierarchy of RG accounts automatically for all of the following:
(309) Faulty agreement between subject and V2 level predicate nominative phrase:
a. *The boys became a good cook.
b. *The boy made good cooks.
c. *The boy is pregnant.
d. *The boys made (became) coffee.
e. *The coffee is broken.
(310) No need for agreement between subject and V1 level direct object phrase:
a. The boys made a good cake.
b. The boy made good cakes.
c. The boy made the girl pregnant.
d. The boys made coffee.
(311) There is often ambiguity between V1 and V2 level elements.
a. John left the house messy. (= John was messy when he left the house.)
[ V3 John PST [ V2 [ V1 [ V0 left] [ N3 the house]] [ C3 messy]]]
b. John left the house messy. (= The house was left messy by John.)
[ V3 John PST [ V1 [ V0 left] [ N3 the house] [ C3 messy]]

(312) The order must be V1 level elements before V2 level elements in post verbal position.
a. John [ V2 [ V1 left the house unlocked] ]
b. John [ V2 [ V1 left the house] dazed ]
c. John [ V2 [ V1 left the house unlocked] dazed ]
d. *John [ V2 [ V1 left the house dazed] unlocked ]
Since the structure of (312) parallels (307) as regards predicate nominative position, agreement
between participles and subjects is automatically accounted for:
(313) a. The terrorist was undaunted by threats.
1. *Threats undaunted the terrorist.
2. *The lamp was undaunted by threats.
b. The men were undivided.
1. *That man was undivided.
2. *We undivided those men.
c. The men were divided (ambiguously adjective or participle).
1. I am divided. (adjective only)
2. *These conflicts are dividing me.
Of course, an added plus for this analysis is that case/number/gender agreement between subjects
and predicate nominatives (including participles) in languages like Latin, French, etc., is directly
accounted for.
Given the above, it follows that we can analyze PERFECTIVE have in either of two ways (recall
that the emphatic so is part of the TNS characterizer):
(314) a. They [could have so] [gone].
b. They [could so] [have gone].
There is, in fact, support for assuming that have can occur in each one of the two above positions.
It must be in V3 when it occurs in infinitives and participles:
(315) a. For him [to have done that] is awful.
b. [Having done that] he left.
In cases like (315), we know that have cannot be in the tense characterizer, because TNS cannot
occur in infinitives and participles. However, have must be in the TNS characterizer in questions
and in certain elliptical clauses:
213

214
(316) a. Have they done that?
b. He has done that, and so has she _______.
The two positions for have in (314) are, in fact, fairly fixed positions in English syntax: the tense
characterizer and the main verb. Perhaps, for this reason, have seems to float back and forth
between the two possible structures, required in the one to account for (315), and in the other to
account for (316). We document this fluctuation with the following two representations for
PERFECTIVE have:
(317) Perfective have is part of the TNS characterizer:
[ V3 [ N3 he] [ C3 [ C0 [ V3 [ V0 has]] PRS ]] [ V0 gone] ]
(318) Perfective have is part of the verb phrase:
[ V3 [ N3 he] [ C3 PRS ] [ V2 [ V3 [ V0 has]] [ V0 gone]] ]
3.16.3 RESOLUTION
The complete English auxiliary is, therefore, the following:
(319) The jars could so have been being broken.
[ V3 [ N3 the jars] [ C3 [ C0 could [ C0 CND]] so ]
[ V3 have [ V2 [ V0 been]
[ C3 [ C0 [ C0 PGP] [ V3 [ V2 [ V0 being]
[ C3 [ C0 [ C0 PSP] [ V3 [ V0 broken]]]] ]]]]]]]
Additional support for the above analysis comes from the behavior of so-called "floating" quantifiers
like all:
(320) a. All of the men could have gone fishing.
b. The men all could have gone fishing.
c. The men could all have gone fishing.
d. The men could have all gone fishing.
e. *The men could have gone all fishing.
Observe that floating quantifiers can "float" up to the main verb (gone in (320)).We see the same
paradigm in (320) in all of the following:

(321) a. All of the men could have been six-feet tall.
b. The men all could have been six-feet tall.
c. The men could all have been six-feet tall.
d. The men could have all been six-feet tall.
e. *The men could have been all six-feet tall.
(322) a. All of the men could have been fishing.
b. The men all could have been fishing.
c. The men could all have been fishing.
d. The men could have all been fishing.
e. *The men could have been all fishing.
(323) a. All of the men could have been seen.
b. The men all could have been seen.
c. The men could all have been seen.
d. The men could have all been seen.
e. *The men could have been all seen.
(324) a. All of a men could have been aware of a problem.
b. The men all could have been aware of a problem.
c. The men could all have been aware of a problem.
d. The men could have all been aware of a problem.
e. *The men could have been all aware of a problem.
(325) a. All of the jars could have been all broken.
b. The jars all could have been all broken.
c. The jars could all have been all broken.
d. The jars could have all been all broken.
e. *The jars could have been all all broken.
Of course, be is the main verb of all of these sentences in the RG analysis. Accordingly, all of the
data are accounted for. There is no version of TG that can make the same claim.
Notice, in addition, that the above analysis treats all uses of be (copular, progressive, and passive)
in the same way, which is necessary given the form of tag questions like the following:
(326) a. They were aware of the problem, weren’t they?
They weren’t aware of the problem, were they?
b. They were fishing, weren’t they?
They weren’t fishing, were they?
c. They were seen, weren’t they?
They weren’t seen, were they?
Given (326), we see that all uses of be repeat be in the tag with opposite polarity. In this regard, all
uses of be work the same way as modals and perfective have:
215

216
(327) a. They could have been aware of the problem, couldn’t they?
They couldn’t have been aware of the problem, could they?
b. They have been aware of the problem, haven’t they?
They haven’t been aware of the problem, have they?
Thus, all auxiliary verbs and all uses of be repeat those verbs in the tag, as opposed to non-auxiliary
verbs, which require a form of do:
(328) a. They went fishing, didn’t they?
They went fishing, did they not?
*They went fishing, weren’t they?
*They went fishing, went they not?
b. They have new cars, don’t they?
They have new cars, do they not?
*They have new cars, haven’t they?
In short, verbs that do not require do-support form their tags without do:
(329) a. *They do be aware of the problem.
b. *They do be fishing.
c. *They do be seen.
d. *They do have gone.
e. *They do can go.
(330) a. They do go fishing.
b. They do have new cars.
c. They do do a good job.
d. They do can tomatoes.
The data above suggest that modals, perfective have, and all uses of be form one class of verbs, say
[+AUX], while all other verbs form another class, say [–AUX].

3.16.4 SUMMARY OF NOMINALS AND VERBALS IN RG
(331) Progressive Participles (PGP) and Derived Adjectives (DA1).
a. Participle (the lady charming the man): [ C3 [ C0 [ C0 PGP] [ V3 [ V0 charming]]]]
b. Adjective (a very charming lady): [ C3 [ C0 [ V3 [ V0 charming]] [ C0 DA1]]]
(332) Passive Participles (PSP) and Derived Adjectives (DA2).
a. Participle (charmed by her flattery): [ C3 [ C0 [ C0 PSP] [ V3 [ V0 charmed]]]]
b. Derived Adjective (a very charmed life): [ C3 [ C0 [ V3 [ V0 charmed]] [ C0 DA2]]]
(333) Gerundial Nominals (GN) and Derived Nominals (DN).
a. GN (charming candy from a baby): [ N3 [ N0 [ N0 GN] [ V3 [ V0 charming]]]]
b. DN: (such charming of snakes): [ N3 [ N0 [ V3 [ V0 charming]] [ N0 DN]]]
(334) Summary of Verbal Structures.
a. Verb resides in posthead position.
[ C3 [ C0 [ C0 PGP] [ V3 [ V0 charming]] ]]
[ C3 [ C0 [ C0 PSP] [ V3 [ V0 charmed]] ]]
[ N3 [ N0 [ N0 GN] [ V3 [ V0 charming]] ]]
b. Verb resides in prehead position.
[ C3 [ C0 [ V3 [ V0 charming]] [ C0 DA1] ]]
[ C3 [ C0 [ V3 [ V0 charmed]] [ C0 DA2] ]]
[ N3 [ N0 [ V3 [ V0 charming]] [ N0 DN] ]]
217

218
3.16.5 PARSES ILLUSTRATING THE ENGLISH AUXILIARY
(335) Imperative Tense: interview them
(336) Present Tense: he interviews them
(337) Past Tense: he interviewed them

(338) Future Tense: he will interview them
(339) Conditional Tense: he would interview them
219

220
(340) Present Tense + Perfective Aspect: he has interviewed them
(341) Present Tense + Progressive Aspect: he is interviewing them

(342) Past Tense + Passive Voice: they were interviewed by him
221

222
(343) Conditional Tense + Perfective Progressive Aspect:
he should have been interviewing them
Structure below need for sentences like He should’ve been interviewing them.
Structure below needed for sentences like Should he have been interviewing them?

(344) Conditional Tense + Perfective Aspect + Passive Voice:
they might have been interviewed by him
223

224
(345) Conditional Tense + Perfective Progressive Aspect + Passive Voice:
they really couldn't all have been being interviewed by him

(346) Conditional Tense + Perfective Progressive Aspect + Passive Voice:
they each could not actually have been being interviewed by him
Another parse has the word not specifying (and, therefore, negating) the modal, which would underlie
the sentence They each couldn’t actually have been being interviewed by him.
225

226

EXERCISES FOR CHAPTER THREE
1. Draw TG diagrams for the DEEP STRUCTURE of the following sentences:
a. The book may be in the library.
b. John will not read those magazines.
c. Mary will send a letter to the company.
d. John reads those magazines.
e. The roses border the fence.
f. The accident occurred at the corner.
g. The plant fell out the window.
h. The boy could take the garbage out.
I. The boy could take out the garbage.
j. The book was not in the library.
2. Collapse each of the following sets of rules into one rule using formal abbreviatory devices
(parentheses and curly brackets):
a. The symbol QNT = QUANTIFIER (more, all, half, etc.)
NP ÷ DET + N
NP ÷ QNT + DET + N
NP ÷ QNT + N
NP ÷ N
b. The symbol AP = ADJECTIVE PHRASE
VP ÷ V
VP ÷ V + PP
VP ÷ V + AP
VP ÷ V + NP
3. As we have noted in the text, the PS–Rules given in (35) are incomplete. They will have to be
revised to accommodate each new structure we consider. Such revisions must be justified by
argument, and the details of proposals must be carefully worked out. Which of the following
sentences can be generated by (89)? Which cannot and, therefore, will require that (89) be
revised? What sort of revisions are necessary?
a. The gardener should water those shrubs with a fine spray.
b. The nearsighted soprano fell into the orchestra pit.
c. They will probably read the assignment reluctantly.
d. A new cook from India prepared all the meals.
e. John thinks Bill will come.
227

228
4. All of the following sentences have the same basic phrase structure, that is, they contain the
same basic units. Keep this in mind as you provide TG diagrams for them.
a. The soprano sang the aria with skill.
b. The young soprano from a small town in the south of France sang the aria with skill.
c. The soprano sang the aria from the third act of an opera by Verdi with skill.
d. The soprano sang the aria with the skill of a seasoned veteran.
e. The young soprano from a small town in the south of France sang the aria from the third
act of an opera by Verdi with the skill of a seasoned veteran.
5. Structurally ambiguous sentences must be given a different diagram for each structural
ambiguity. For example, the sentence Bill may slip on the boots has the following two
diagrams in TG format:
a. For the meaning on the boots, Bill may slip:
b. For the meaning Bill may slip the boots on:

Provide disambiguating diagrams for each of the following ambiguous sentences:
a. The thief hit the lady with the hat.
b. The repair man will look up the street.
c. John left the house messy.
d. The prisoner of war spoke foolishly.
6. What problems, if any, do the following ambiguous sentences pose for the TG model of
grammar?
a. They hired Spanish teachers.
b. They are all finished.
c. They do so love a good meal.
d. We made them idols.
e. They made idols.
7. Draw RG diagrams for the following sentences:
a. The book may be in the library.
b. John will not read those magazines.
c. Mary will send a letter to the company.
d. John reads those magazines.
e. The roses border the fence.
f. The accident occurred at the corner.
g. The plant fell out the window.
h. The boy could take the garbage out.
I. The boy could take out the garbage.
j. The book was not in the library.
8. Draw diagrams for the following sentences in RG framework. What are the advantages of the
X–Bar notation?
a. The gardener should water those shrubs with a fine spray.
b. The nearsighted soprano fell into the orchestra pit.
c. They will probably read the assignment reluctantly.
d. A new cook from India prepared all the meals.
e. John thinks Bill will come.
229

230
9. Draw diagrams for each of the meanings of the following sentences in the RG framework.
Does the X–Bar notation allow you to reach a higher level of descriptive adequacy?
a. The thief hit the lady with the hat.
b. John left the house messy.
c. The repair man will look up the street.
d. The prisoner of war spoke foolishly.
e. They hired Spanish teachers.
f. They made idols.
10. Show that the following sentences must be assigned different structural descriptions no matter
which theory of grammar is adopted.
a. (1) The boys made good cooks.
(2) The boys made good cakes.
Do the same with the following pair:
b. (1) The cars passed in the tunnel.
(2) The boys passed in the tests.

APPENDIX A: ANSWERS TO EXERCISES
1. Draw TG diagrams for the following sentences:
a. The book may be in the library.
b. John will not read those magazines.
c. Mary will send a letter to the company.
231

232
d. John reads those magazines.
e. The roses border the fence.
f. The accident occurred at the corner.
g. The plant fell out the window.

h. The boy could take the garbage out.
I. The boy could take out the garbage.
j. The book was not in the library.
2. Collapse each of the following sets of rules into one rule using formal abbreviatory devices
(parentheses and curly brackets):
a. NP ÷ (QNT) + (DET) + N
b. VP ÷ V + ({ PP, AP, NP})
233

234
3. a. (89) can generate the structure for The gardener should water those shrubs with a fine
spray:
b. (89) cannot generate the structure for The nearsighted soprano fell into the orchestra pit.
The phrase orchestra pit is a compound noun made up of two nouns (orchestra and pit).
The rule (89d) for rewriting NP does not allow an NP to contain two nouns. Note that
orchestra is a noun, not an adjective: it can be made plural (orchestras) and it cannot be
made comparative (*more orchestra).
c. (89) cannot generate the structure for They will probably read the assignment reluctantly.
(89) contains no rule for adverbs. Probably is a sentence adverb (a constituent of S), so
(89a) must be revised to allow an adverb. Reluctantly is a manner adverb (a constituent
of VP), so (89b) must also be revised to allow an adverb.
d. (89) cannot generate the structure for A new cook from India prepared all the meals. All
is a quantifier, and (89d) does not allow for the generation of a quantifier in an NP. The
top line of (89d) would have to be revised as follows:
(QNT) + (DET) + (ADJ) + N + (PP)
e. (89) can generate the structure for John thinks Bill will come.

4. All of the following sentences have the same basic phrase structure, that is, they contain the
same basic units. Keep this in mind as you provide TG diagrams for them.
a. The soprano sang the aria with skill.
b. The young soprano from a small town in the south of France sang the aria with skill.
c. The soprano sang the aria from the third act of an opera by Verdi with skill.
d. The soprano sang the aria with the skill of a seasoned veteran.
e. The young soprano from a small town in the south of France sang the aria from the third
act of an opera by Verdi with the skill of a seasoned veteran.
235

236
5. Provide disambiguating diagrams for each of the following ambiguous sentences:
a. The thief hit the lady with the hat.
(1) =It was with the hat that the thief hit the lady.
(2) =It was the lady with the hat that the thief hit.
b. The (repair) man will look up the street.
Note: Repair man cannot be generated by the rules in (89); it is a compound noun like
orchestra pit in exercise (3b) above. Still, the sentence remains ambiguous without the
word repair in it.
(1) = It is up the street that the man will look.

(2) = The man will look the street up.
c. Ron left the house messy.
(1) = The house was messy when Ron left it.
(2) = Ron was messy when he left the house.
237
Note: While this diagram would appear to be satisfactory to account for the meaning in
which messy modifies Ron, it will prove to be descriptively inadequate when other
sentences are considered. For example, we know that a manner adverb like reluctantly
must be part of the VP (see exercise (5d) below and the explanation for (3c) above).
There are sentences like Ron left the house messy reluctantly, which are still ambiguous
as to whether it is Ron or the house that was left messy. Thus, the above diagram is
descriptively inadequate: since reluctantly must part of the VP in both readings of the
sentence Ron left the house messy reluctantly (it is a manner adverb in both readings),
then messy cannot be outside of the VP as in the above diagram. The standard theory
fails in this case.

238
d. The prisoner of war spoke foolishly.
(1) = The manner in which the prisoner of war spoke was foolish.
(2) = It was foolish of the prisoner of war to speak. (foolishly is a sentence adverb)
6. What problems, if any, do the following ambiguous sentences pose for TG?
Within the TG framework, there is no way to disambiguate any of the following sentences, that
is, PS-Rules, like those in (89), cannot handle the subtle distinctions made.
a. They hired Spanish teachers means either They hired teachers who are Spanish or They
hired teachers of Spanish. Notice that we have They hired Spanish Spanish teachers.
b. They are all finished means either All of them are finished or They are completely
finished. Notice that we have They are all all finished.
c. In They do so love a good meal, so emphasizes do as in They do too love a good meal or
it emphasizes love as in He really loves a good meal.
d. We made them idols means We made them into idols (like We elected them officers) or
We made idols for them (like We bought them presents).
e. They made idols means either They became idols or They created idols. The verb make,
therefore, can be used either as a transitive verb (one that takes a direct object) or as a
copulative verb (one that has a complement which refers back to the subject.

7. Draw RG diagrams for the following sentences:
a. The book may be in the library. (be is a copulative verb; note the V2)
b. John will not read those magazines. (read is a transitive verb; note the V1)
c. Mary will send a letter to the company. (send is a transitive verb; note the V1)
239

240
d. John reads those magazines. (read is a transitive verb; note the V1)
e. The roses border the fence. (border is a transitive verb; note the V1)
f. The accident occurred at the corner. (occur is an intransitive verb; note the V2)
g. The plant fell out the window. (fall is an intransitive verb; note the V2)

h. The boy could take the garbage out. (take is a transitive verb; note the V1)
i. The boy could take out the garbage. (take is a transitive verb; note the V1)
j. The book was not in the library. (be is a copulative verb; note the V2)
241

242
8. Draw diagrams for the following sentences in RG framework. What are the advantages of the
X–Bar notation?
a. The gardener should water those shrubs with a fine spray.
b. The nearsighted soprano fell into the orchestra pit.

c. They will probably read the assignment reluctantly.
d. A new cook from India prepared all the meals.
e. John thinks Bill will come.
243

244
9. a. The thief hit the lady with the hat.
(1) = It was with the hat that the thief hit the lady.
(2) = It was the lady with the hat that the thief hit.

. Ron left the house messy.
(1) = The house was messy when he left it.
(2) = Ron was messy when he left the house.
245

246
c. The repair man will look up the street.
(1) = The repair man will look the street up.
(2) = It is up the street that the repair man will look.

d. The prisoner of war spoke foolishly.
(1) = It was foolish of the prisoner of war to speak.
(2) = The way the prisoner of war spoke was foolish.
247

248
e. They hired Spanish teachers.
(1) = teachers who were Spanish (2) = teachers of Spanish
f. They made idols
(1) = They became idols. (2) = They built idols.
10. a. The following data indicate that The boys made good cooks and The boys made good
cakes must have different structural descriptions:
(1) a. *The boy made good cooks.
b. The boy made good cakes.
(2) a. *Good cooks were made by the boy.
b. Good cakes were made by the boy.
b. The following data indicate that The cars passed in the tunnel and The boys passed in the
tests must have different structural descriptions:
(1) a. *The cars passed the tunnel in.
b. The boys passed the tests in.
(2) a. It was in the tunnel that the cars passed.
b. *It was in the tests that the boys passed.

APPENDIX B: LATIN SYNTAX
Latin verbs must agree with their subjects in number:
1. a. Agricola lupam necat. The farmer kills the wolf.
b. Agricola lupas necat. The farmer kills the wolves.
2. a. Agricolae lupam necant. The farmers kill the wolf.
b. Agricolae lupas necant. The farmers kill the wolves.
Agreement patterns between subjects and verbs:
3. a. Singular: Subject–a–Ø verb–a–t
regin–a–Ø voc–a–t
b. Plural: Subject–a–e verb–a–nt
regin–a–e voc–a–nt
4. a. *Agricola lupas necant.
b. *Agrocilae lupam necat.
Latin adjectives and numerals must agree with the nouns they modify in both number and case:
5. a. Laeta agricola malas lupas duas necat.
‘The happy farmer kills the two bad wolves.’
b. Amicae feminae duae aegras aegricolas duas curant.
‘The two friendly women take care of the two sick farmers.’
Phrase Structure Rules for Latin:
6. a. S ÷ NP + VP
b. VP ÷ NP + V
c. NP ÷ ADJ n
0 + N + (NUM)
249

250
7.
Rules for Case Marking and Rules for Agreement:
8. NP dominated by S is subject; all elements of that NP are in the nominative case (–Ø if
singular, –e if plural).
9. NP dominated by VP is object; all elements of that NP are in the accusative case (–m if
singular, –s if plural).
10. If the subject is singular, verb ends in –t; if the subject is plural, verb ends in –nt.
After these rules apply to the above tree, the grammar generates the sentence Aminae feminae duae
aegras agricolas duas curant.
Vocabulary:
11. Nouns (quoted in the nominative singular feminine form):
lupa ‘wolf’ luna ‘moon’ regina ‘queen’ agricola ‘farmer’
stella ‘star’ casa ‘cottage’ fabula ‘story’ domina ‘mistress’
12. Adjectives (quoted in the nominative singular feminine form):
mala ‘bad’ pulchra ‘pretty’ multa ‘many’ clara ‘famous’
bona ‘good’ amica ‘friendly’ inimica ‘unfriendly’ laeta ‘happy’
13. Verbs (quoted in the infinitive form):
liberare ‘free’ vocare ‘call’ laudare ‘praise’ aedificare ‘build’
necare ‘kill’ curare ‘care for’ adorare ‘worship’ superare ‘conquer’
14. Others:
non ‘not’ et ‘and’

APPENDIX C: SUMMARY OF TREE STRUCTURES
1. Traditional grammar (confuses functional and structural categories):
2. Structural grammar and TG (not enough internal levels for phrases):
3. X–bar grammar and RG:
4. Sentence with a transitive verb (John studies Latin):
251

252
5. Noun phrase structure (John’s study of Latin...):
6. Noun phrase structure (all the/Ø study of Latin...):

7. Sentence with an intransitive (John died happy/happily) or copulative verb (John was happy):
253

254

4.1 PRELIMINARY REMARKS
CHAPTER FOUR: SEMANTICS
Of all of the areas of grammar which linguists have studied, by far the most puzzling and problematic
is semantics, the area which concerns the meanings of words and sentences. Despite the efforts of
numerous linguists, philosophers, psychologists, and anthropologists, a number of very fundamental
questions remain unanswered both in regard to the meaning of words and to the meaning of
sentences. Consider, for example the question of what a particular word, say, table, means. It is clear
that when native speakers of English use the word table, they are referring to a specific object and
recognize in that object certain identifiable characteristics which make the use of table appropriate.
Still, it is exceedingly difficult to determine what these identifiable characteristics are, considering
all of the various types of objects which are called a table. Even words such as dog and cat present
problems. Although a biologist could give a highly specific definition of both of these animals by
referring to particular anatomical and physiological features, it is obvious that the average speaker
does not distinguish a dog from a cat by referring to such features, e.g., that a dog has two upper
molar teeth while a cat has only one. When words such as beauty, love, and joy are brought into the
picture, the problem of providing definitions reaches nearly unmanageable proportions, for words
such as these, in addition to referring to intangible objects, also vary considerably in meaning from
speaker to speaker and language to language. Semantic analysis typically raises wicked problems
in the sense mentioned in the Introduction (Page 5).
In translating from one language to another, these problems are compounded because the sphere of
objects named by a particular word is rarely the same from language to language. For example, the
English word uncle can refer either to one’s mother’s brother or one’s father’s brother (among
others). In Latin, these two relatives are distinguished by separate words: avunculus is one’s
maternal uncle, and patruus is one’s paternal uncle. This distinction was important in ancient Rome
to designate who a minor’s guardian would be upon the death of parents. In general, kinship terms
vary greatly from language to language, reflecting social custom and law. The same can be said
about words to designate color, shape, time, place, and many other categories. One successful
approach to describing such cultural differences has been to break the categories down to features
and describe the different classes as different arrangements of features. We will take this approach
in analyzing English prepositions. The important point to keep in mind now is that any child can
learn any of these systems as part of his native language. Such cultural differences, therefore, must
reflect innate abilities of human perception and categorization.
4.2 DENOTATION AND CONNOTATION
In an effort to cope with the problem of definition, semanticists have recognized a number of useful
distinctions in the meaning of a word. One of these is the distinction between DENOTATION and
CONNOTATION. The denotative meaning of a word is its specific, exact meaning, excluding all
of its extended uses and emotional colorations. On the other hand, a connotative meaning is any
meaning which carries some special implication or association. Thus, in (1) the word fox is used in

256
its denotative meaning to refer to a specific animal, while in (2) it is used in a connotative meaning
to imply a crafty person.
(1) A hungry fox was seen prowling around the chicken coop.
(2) Mary doesn’t trust John, because he’s such a fox.
4.3 SYNONYMY, HOMONYMY, AND ANTONYMY
Other terms often used by semanticists in defining words and indicating relationships among them
include SYNONYM, HOMONYM, and ANTONYM. Two words are synonyms if they are
pronounced differently, but mean the same thing, e.g., huge and enormous (adjectives), oculist and
eye doctor (nouns), imprison and incarcerate (verbs) ; they are homonyms if they are pronounced
the same, but have different meanings, e.g., bank (river’s edge) and bank (financial institution); they
are antonyms if they mean opposite things, e.g., hot and cold (adjectives), love and hate (nouns or
verbs).
Antonyms fall into two groups: complementary and gradable. Complementary anonyms are absolute
opposites like male and female, dead and alive, present and absent, etc., in which the negative of one
is synonymous with the other. Thus, if one is not male, one is female; or, if one is not female, one
is male. Gradable anonyms are not absolute opposites; the negative of one is not synonymous with
the other, e.g., fast and slow, big and little, etc.
4.4 HYPONYMY/HYPERNYMY AND MERONYMY/HOLONYMY
Semanticists also make distinctions between HYPONYMY/HYPERNYMY which is the
member/class distinction; for example, robin is a hyponym of bird, and bird is a hyponym of animal.
Such relationships are TRANSITIVE, which means that the reverse relationship also holds: animal
is a hypernym of bird, and bird is a hypernym of robin.
Closely related is the distinction between MERONYMY and HOLONYMY, that is, the part/whole
relationship; for example, ceiling is a meronym or room, and room is a meronym of house. This
distinction is also transitive: house is a holonym of room, and room is a holonym of ceiling.
4.5 IDIOMS
Semanticists also make a distinction between idiomatic and non–idiomatic meaning. An IDIOM is
a highly specialized expression whose meaning typically cannot be deduced from the meaning of the
words out of which it is composed. Expressions like shoot the breeze, be up tight, and so long are
examples of idioms. In most instances, idioms are fixed in usage and cannot be expanded or
modified. For example, one cannot say *We shot a southwesterly breeze yesterday.

(3) a. We were shooting the breeze.
1. *We were shooting a southwesterly breeze.
2. We were cheerfully shooting the breeze.
3. We shot the tigers./The tigers were shot by us.
4. We shot the breeze./*The breeze was shot by us.
b. He was pulling your leg.
257
1. Were you pulling my (her, John’s) leg?
2. Were you pulling (*yanking, tugging) my leg (*arm, hair).
3. Sue’s leg’s been pulled so many times that she doesn’t trust anyone anymore.
On the other hand, idioms must be assigned a syntactic structure because they are subject to rules.
For example, consider the rule of Particle Movement (the movement of an intransitive preposition)
as it applies in the following:
(4) Particle Movement.
a. The teacher handed out the assignment.
The teacher handed the assignment out.
b. The teacher spoke about the assignment.
*The teacher spoke the assignment about.
c. John gave away the show.
John gave the show away.
d. John beats around the bush.
*John beats the bush around.
e. Further examples: bark up the wrong tree, bring down the house, go off the deep end,
jump down one’s throat, lay down the law, let down one’s hair, make up one’s mind,
pay through the nose, sit on the fence
4.6 AMBIGUITY AND PARAPHRASE
In the study of meaning, linguists have also been concerned with AMBIGUITY and
PARAPHRASE. There are a great many pairs of words, phrases or sentences in all languages which
are ambiguous, that is, which are phonologically identical but semantically distinct; others are
paraphrases, that is, phonologically distinct but semantically similar.

258
(5) AMBIGUITY (one sound, more than one meaning).
a. Lexical level, the level of the word.
1. After making the right turn at the corner,...
2. After making the left turn at the corner,...
b. Phrasal level.
1. He looked over the fence.
2. He looked under the fence.
c. Sentential level.
1. Visiting relatives can be boring.
2. The chickens are ready to eat.
3. What worried John was being ignored by everybody.
(6) PARAPHRASE (one basic meaning, more then one sound).
a. Lexical level.
1. They murdered/killed the president.
2. Mary picked up the pail/bucket.
b. Phrasal level.
1. John resembles Bill./John is similar to Bill.
2. John is a father./John has children.
c. Sentence level.
1. The bees are swarming in the garden./The garden is swarming with bees.
2. John is easy to please./It is easy to please John.
(7) Disambiguating sentences with paraphrase.
a. Moving vehicles can be hazardous.
1. Vehicles which are moving can be hazardous.
2. It can be hazardous to move vehicles.

. The baby is too young to stand up.
1. The baby is too young for someone to stand up.
2. The baby is too young to stand up by itself/himself/herself.
Most native speakers are unaware of ambiguity when it occurs because context favors one meaning
over others. The task of the linguist is to ascertain what differences, if any, there are between each
member of the ambiguous pair. Very often, the members can be shown to be syntactically distinct.
For example, when sentence (7a) receives the first interpretation, the word vehicles is the subject of
the word move, and when the sentence receives the second interpretation, vehicles is the object of
move.
4.7 PRAGMATICS
Beyond these considerations, most of the problems in sentence meaning which linguists have been
especially interested in concern either the relationship between semantics and the other components
of grammar, particularly syntax and phonology, or the relationship between semantics and such
extra–linguistic or PRAGMATIC facts as the beliefs and knowledge a person has about the world.
The interaction of phonology and semantics is clearly illustrated in the following set of examples
(upper–case type indicates heavy stress):
(8) a. After John complimented Mary, SHE complimented HIM.
b. *After John complimented Mary, she COMPLIMENTED him.
c. *After John complimented Mary, SHE insulted HIM.
d. After John complimented Mary, she INSULTED him.
Notice that the pronouns in the second half of these sentences receives stress only when the meaning
of the verbs in both halves is identical. When the meaning of the verbs in both halves is not identical,
then the verbs themselves receive the heavy stress. This is an example of how the phonological rules
of a language can depend on semantic structure.
Lastly, notice that the interpretation of the following sentences depends on specific beliefs that people
have.
(9) a. After John called Mary a republican, SHE complimented HIM.
b. After John called Mary a republican, she COMPLIMENTED him.
c. After John called Mary a republican, SHE insulted HIM.
d. After John called Mary a republican, she INSULTED him.
(10) a. Both of John’s parents are schizophrenics, and it looks as though John is crazy too.
b. Both of John’s parents are skydivers, and it looks as though John is crazy too.
c. Both of John’s parents are engineers, and it looks as though John is crazy too.
259

260
For most speakers of English, the sentiment expressed in (10a) would be considered reasonable, since
they know that schizophrenia is a form of mental illness, and the one in (10c) would be considered
unreasonable, since most people do not equate engineering with insanity. The status of (10b), on the
other hand, would probably vary. If people are skydivers, they are likely to find it unreasonable;
however, other people, who believe that anyone willing to jump out of a plane without immediately
opening a parachute is insane, would find (10b) reasonable.
4.8 THEMATIC RELATIONS: CASE GRAMMAR
One of the best understood areas of semantics concerns the roles that various noun phrases fulfill in
a sentence, roles such as AGENT and INSTRUMENT. The research was originally carried out by
Charles Fillmore in the 1960s and became known as CASE GRAMMAR because of its similarity
with the study of grammatical case in traditional grammar. Over the past thirty years, case relations
have been investigated by many other linguists in a variety of theoretical frameworks. Many scholars
have pointed out that the SEMANTIC FUNCTION (AGENT or INSTRUMENT) that a noun phrase
has in a sentence often appears to be of greater significance than its SYNTACTIC FUNCTION
(SUBJECT or OBJECT). For example, although the syntactic function of the noun phrase the door
in (11) changes, its semantic function does not.
(11) a. The door will open. door is subject
b. The janitor will open the door. door is object
c. The door will open with this key. door is subject
d. This key will open the door. door is object
There are many similar cases in English and other languages such as those in (12).
(12) a. John broke the dish. dish is object
b. The dish broke. dish is subject
c. John started the car. car is object
d. The car started. car is subject
e. John woke up the baby. baby is object
f. The baby woke up. baby is subject
Consider also the various roles that the subject children plays in the following:
(13) a. Children walk easily.
b. Children fall easily.
c. Children frighten easily.
d. Children cry easily.

The role of a noun phrase in relation to a particular verb determines not only how it will be
understood in a sentence, but also how it will behave. For example, in (14) Mary can be understood
as either an ACCOMPANIER or as an INSTRUMENT.
(14) John broke the window with Mary.
As (15) shows, different roles cannot usually be combined.
(15) a. John broke the window. (John is agent)
b. A hammer broke the window. (hammer is instrument)
c. *John and a hammer broke the window.
Relations like AGENT, ACCOMPANIER and INSTRUMENT, which specify the role that a noun
phrase plays in a sentence, are called CASE RELATIONS or THEMATIC RELATIONS. In his
earliest formulations, Fillmore proposed the following case relations:
(16) AGENTIVE: the case of the typically animate perceived instigator of the action identified
in the verb (often simply called AGENT).
a. JOHN opened the door.
b. The door was opened BY JOHN.
(17) INSTRUMENTAL: the case of the inanimate force or object causally involved in the
action or state identified in the verb.
a. THE HAMMER broke the window.
b. The window was broken WITH THE HAMMER.
c. THE HURRICANE destroyed the village.
(18) OBJECTIVE: the semantically most neutral case; typically the case of the thing directly
affected by the action or state identified in the verb. (Not to be confused with the notion
direct object; often called PATIENT or THEME)
a. Someone opened THE DOOR. (door is both direct object and objective case.)
b. THE DOOR opened. (door is subject, but still objective case.)
(19) DATIVE: the case of the usually animate entity affected by the state or action identified
in the verb (often called EXPERIENCER or RECIPIENT).
a. John gave the book TO BILL.
b. BILL received the book.
c. The movie appeals to HIM.
d. THE SNOW melted.
261

262
(20) POSSESSIVE: the case of the possessor.
a. JOHN’S book is on the table.
b. JOHN owns that book.
(21) LOCATIVE: the case which identifies the location or spatial orientation of the state or
action identified in the verb.
a. It happened IN ITALY.
b. The book is ON THE TABLE.
(22) GOAL: the case which identifies the direction of the state or action identified in the verb.
a. John flew TO ITALY.
b. John ran INTO THE HOUSE.
(23) SOURCE: the case which identifies the origin of the state or action identified in the verb.
a. John flew FROM ITALY.
b. John ran OUT OF THE HOUSE.
Given an inventory of thematic relations like the above, a linguist can specify the roles that noun
phrases play with particular verbs. For example, we can characterize the verb open as follows:
(24) a. It requires an OBJECTIVE and tolerates an INSTRUMENTAL and/or AGENTIVE.
b. If only the OBJECTIVE NP occurs, it must be the subject:
[1] The door can open. ACTIVE
[2] The door will be opened. PASSIVE
c. If the INSTRUMENTAL also occurs, either the OBJECTIVE or the
INSTRUMENTAL can be subject.
[1] The door can open with a key. ACTIVE
[2] The door can be opened with a key. PASSIVE
[3] A key can open the door. ACTIVE
d. If the OBJECTIVE and the AGENTIVE occur, either can be subject.
[1] A janitor can open the door. ACTIVE
[2] The door can be opened by a janitor. PASSIVE

e. If all three cases occur, either the AGENTIVE or the OBJECTIVE can be subject, but
the INSTRUMENTAL cannot.
[1] A janitor can open the door with a key. ACTIVE
[2] The door can be opened with a key. PASSIVE
[3] *A key can open the door by a janitor.
As these examples indicate, the thematic relation of an NP in English is signaled overtly by a
preposition unless that NP is a subject or object. For example, the instrumental relation is signaled
by the preposition with except when the instrument is in subject position (cf. (24c3)). Fillmore
argued that this indicates that the case relations should be directly represented in DEEP
STRUCTURE, and that subject and object should be derived from full prepositional phrases in
English. For example, he proposed that the deep structure of (25a) would be something like (25b).
(25) a. The janitor will open the door with the key.
b. [ S [ AUX will] [ VP [ V open] [ PP of the door] [ PP by the janitor] [ PP with the key] ]]
Rules of SUBJECT FORMATION and OBJECT FORMATION then turn these structures into
structures of the more familiar type like (26).
(26) [ S [ NP the janitor] [ AUX will] [ VP [ V open] [ NP the door] [ PP with the key] ]]
4.9 THE MEANING AND USE OF THE WORD CASE
In a description such as the above, an important distinction must be made between THEMATIC or
CASE RELATIONS and the traditional use of case as a label for a particular morphological category.
For example, in Latin, we find inflectional endings like the following:
(27) a. Nominative (–s suffix): the case of the subject.
b. Accusative (–m suffix): the case of the direct object.
[1] Marcus (subject) Publium (direct object) adjuvat. = Marcus helps Publius.
[2] Marcum (direct object) Publius (subject) adjuvat. = Publius helps Marcus.
A term like subject or direct object is a label for a SYNTACTIC RELATION (SYNTACTIC
FUNCTION); these indicate specific positions that a noun phrase occupies in a SYNTACTIC
HIERARCHY like (?). Notice that the subject is the NP immediately dominated by the S node, and
the object is the NP immediately dominated by the VP node. A term like agent or agentive is a label
for a THEMATIC RELATION (SEMANTIC FUNCTION); these specify the various roles that a
noun phrase can play in a sentence. In English, these are usually indicated by prepositions; in
languages like Japanese, they are indicated by postpositions; and in languages like Latin, they are
frequently indicated by inflectional endings called case endings.
263

264
Quite frequently in languages, these functions overlap, so there is some terminological confusion.
Thus, the word case is used for both a morphological category (a case or inflectional ending) or a
semantic category (a case or thematic relation). Compare the following examples.
(28) a. English:
John gave the book to Mary.
b. Japanese:
Taroo wa Hanako ni hon o ageta
Taro SUBJECT Hanako TO book OBJECT gave
‘Taro gave a book to Hanako.’
wa: postposition indicating subject/topic, in this example, the agentive relation
ni: postposition indicating indirect object, in this example, the dative relation
o: postposition indicating direct object, in this example the objective relation
c. Latin:
Marcu–s Publi–o libru–m dabat
Mark–SUBJECT Publius–TO book–OBJECT gave
‘Mark gave a book to Publius.’
–s: nominative case ending indicating subject; here, the agentive relation
–o: dative case ending indicating indirect object; here, the dative relation
–m: accusative case ending indicating direct object; here, the objective relation
As (28c) shows, a term like nominative or accusative is a label for a GRAMMATICAL or
MORPHOLOGICAL FORM. But these terms are also used for both syntactic and semantic
functions. For example, in (28c), Marcus carries the morphological marker for the nominative case
indicating the subject, a syntactic function. But, at another level of analysis, Marcus (still marked
nominative) is also Fillmore’s agentive case, which is a semantic function. Consider the following
example from English:
(29) I saw him.
a. him is direct object:
SYNTACTIC RELATION or SYNTACTIC FUNCTION.
b. him is accusative case:
GRAMMATICAL or MORPHOLOGICAL FORM.
c. him is objective case:
THEMATIC RELATION or SEMANTIC FUNCTION.

Unfortunately, there is even further terminological confusion. Some grammarians use the term
objective case to refer to the grammatical form of the direct object. The term possessive has been
used for all three of the above.
4.10 PROBLEMS WITH CASE GRAMMAR
Regardless of terminology, CASE GRAMMAR has much to recommend it, though it does leave
some facts unaccounted for. For example, this approach leaves unexplained the fact that languages
use the same prepositions (grammatical case/postpositions) to express apparently different thematic
relations.
(30) a. to as Fillmore’s Dative and Goal:
[1] The movie appeals to him. (Dative)
[2] John flew to Italy. (Goal)
b. of as Fillmore’s Agentive, Objective and Possessive:
[1] the crying of babies
[2] the taking of pictures
[3] the shooting of hunters
[4] the pictures of those painters
[a] the pictures painted by those painters
[b] the pictures with those painters in them
[c] the pictures those painters own
Furthermore, CASE GRAMMAR cannot explain why languages use the same markers (prepositions,
postpositions, inflectional endings) for both positional and nonpositional thematic relations as in (31).
(31) [+POSITIONAL] [–POSITIONAL]
He put it into the vase. He turned it into a vase.
He took it out of a vase. He made it out of a vase.
He headed for freedom. He died for freedom.
He withdrew from the heat. He died from the heat.
Although the same marker is used, there is a clear difference between the phrases contrasted in (31).
For example, only a [+POSITIONAL] phrase can be substituted with here, there, where, etc.:
(32) a. Over there is where he put the vase.
b. *Over there is where he turned the vase.
265

266
4.11 RG SEMANTICS
The central hypothesis articulated in the preceding chapters is that the categories of language are
natural classes of features reflective of the nature of the human language apparatus. Further, we have
postulated that languages differ because they select and organize these universal and biologically
determined features differently. If semantic analysis is to parallel the analysis of syntax and
phonology, which one would expect, then one should look first at the categories linguists have
postulated to account for meaning. The largest group of such categories comprises case or thematic
relations.
Over the past twenty–five years, a number of linguists have attempted to incorporate the case
relationships of traditional grammar into formal grammatical theory. All of these efforts have one
thing in common: they attempt to explicate the nature, distribution, and manifestation of thematic
relations like AGENT, INSTRUMENT, SOURCE, GOAL, etc.
Within the entire group of thematic relations that have been proposed in the literature, a basic
division occurs between positional [+PST] and nonpositional [–PST] relations. In the former group,
we have such relations as PLACE and DIRECTION; in the latter, AGENT and PATIENT. For some
reason that is not clear, the former group has been given short shrift by modern theorists, and most
of the attention has been focused on the nonpositional relations. This is a fundamental error, because
the positional relations are more concrete and, therefore, undeniably easier to specify, and because
the nonpositional relations are clearly related to and derived from the positional relations.
Languages, in general, do not have two separate sets of thematic markers (prepositions, postpositions,
grammatical cases, etc.), one for relations that are [+PST], and another for those that are [–PST].
Thus, for example, in the ancient Indo–European languages, all separative relations are expressed by
the ablative case, if there is one. This includes actual perceived movements in the direction away,
e.g., SOURCE, as well as expressions of freedom, deprivation, agency, abandonment, etc. We have
such examples as the following in Latin (33) and Sanskrit (34):
(33) a. Eius milites ab opere (ABLATIVE) vacabant.
‘His soldiers are free from work.’
(Caesar, De Bello Civili, 3, 21)
b. Oculis (ABLATIVE) se privavit.
‘He deprived himself of his eyes.’
(Cicero, De Finibus Bonorum et Malorum, v. 42)
(34) a. nāsmād ganāç (ABLATIVE) chidyate.
‘The people are not cut off from him.’
(Catapatha–Brahmana, 14, 5, 1, 10)
b. sa evainam varunapāçān (ABLATIVE) muñcati.
‘He releases him from Varuna’s snare.’
(Taittiriya–Samhita, 2, 1, 2, 2)

In ancient Greek, which has no ablative case, the separative functions of the ablative are expressed
by the genitive:
(35) a. apostereî me tōn chrēmátōn (GENITIVE).
‘He deprives me of my property.’
(Isocrates, 12, 35)
b. tōn epitēdeíōn (GENITIVE) ouk aporēsomen.
‘We shall not be without provisions.’
(Xenophon, Anabasis, 2, 2, 11)
Whatever the defects of traditional terminology, in the Indo–European languages which have an
ablative, it remains that the ablative is the "from–case"; therefore, all governors (verbs, adjectives,
etc.) having to do with separative notions are construed with the ablative, independent of their
(syntactic) categorial membership. When the ablative is lost as a morphologically distinct case as
in Greek, there is order in the way the functions of the ablative are distributed among the remaining
cases. That order is specified by semantic classes like separative notions. The same situation obtains
in other languages with grammatical case, including those as diverse as Innuit and Finnish: semantic
relationships are grouped into classes, and these classes are realized by the same grammatical case.
When languages do not have elaborate case systems, but express thematic relations with prepositions
(English, Welsh, etc.) or postpositions (Japanese, Hindi, etc.), we find the same basic criterion of
semantic commonality underlying the use of these particles. Consider, for example, the following
data from Italian, which use the same preposition for AGENT, SOURCE, and GOAL:
(36) a. Il meccanico ha riparato la macchina.
the mechanic has repaired the car
‘The mechanic has repaired the car.’
b. La macchina è stata riparata dal meccanico (AGENT).
the car has been repaired by the mechanic
‘The car has been repaired by the mechanic.’
(37) a. Giovanni èsce dal negozio (SOURCE).
Giovanni comes from the store
‘Giovanni is coming from the store’
b. Giovanni trema dal freddo (CAUSE).
Giovanni trembles from the cold
‘Giovanni is trembling from the cold’
Therefore, it is not an accident, for example, that from occurs in separative phrases like the following,
whether or not real movement occurs:
267

268
(38) a. He ran from his office (his father, is responsibilities, etc.)
b. He is back from Europe (the market, unconsciousness, etc.).
c. Keep this away from the children.
d. She can’t tell puce from fuchsia.
e. He can’t find any relief from pain.
f. They’ll be here an hour from now.
g. We got a note from the dean.
h. He died from overexposure.
I. We made it from these instructions.
j. We made it from the materials we had.
In examples like the above, there is a direct connection between positional phrases (from his office)
and nonpositional phrases (from fuchsia), however one wishes to term this connection (denotative
) connotative; literal ) figurative; literal ) metaphorical; central ) extended; etc.). Furthermore,
these examples are typical: as we will see, similar sets exists for the other prepositions in English
(to, in, out, over, etc.); and, one can easily construct other sets for non–Indo–European languages,
e.g., the postpositions ni, e, kara, de, etc. in Japanese.
Theories of case grammar cannot account for data such as those above in a way that relates the
positional and nonpositional uses of prepositions. In Fillmore’s system, for example, it is a complete
accident that a preposition like English to shows up as both GOAL (He went to New York.) and
DATIVE/EXPERIENCER (New York appeals to him.). This is a serious loss of descriptive
adequacy, because the same two relations (GOAL and DATIVE/EXPERIENCER) are realized by
the same postposition in Japanese (ni), the same inflectional case affix in ancient Indo–European
languages like Latin (dative case), the same case in non–Indo–European languages (Finnish allative
case), and so on. When such a wide range of languages realizes ostensibly different relations in the
same way, it is clear that the relations have something in common, and that a descriptively adequate
theory of language must discover and formalize what that commonality is if it is to make revealing
generalizations about natural language. That such generalizations must be made is indicated by sets
of data like the following (an example of literal motion toward a place occurs in the a sentences; of
non–literal motion in the b sentences; of no perceptible motion at all, but rather of the person affected
in the c sentences):
(39) English (same preposition, to).
a. She drove the car to New York. (A question like Where did she drive the car (to)?
is completely acceptable and asks about literal motion toward some unknown place.)
b. She gave the book to him./She gave him the book. (To him cannot be the answer to
a question like Where did she give the book?; in fact, such questions are marginal, at
best.)
c. The book appeals to him. (*Where does the book appeal?)

(40) Latin (same grammatical case, dative).
a. Hostes finibus (dative) appropinquant.
enemy border approach
‘The enemy approaches the border.’
b. Litteras mihi (dative) nuntius reddidit.
letter to me messenger delivered
‘The messenger delivered a letter to me.’
c. Mihi (dative) placet.
to me it pleases
‘It pleases me.’
(41) Italian (same preposition, a).
a. Mio fratello va a scuola ogni giorno.
my brother goes to school every day
‘My bother goes to school every day.’
b. Paulo scrive una lettera a Maria.
Paul wrties a letter to Mary
‘Paul is writing a letter to Mary.
c. A mio fratello piace il romanzo.
to my brother pleases the novel
‘The novel pleases (is pleasing to) my brother.’
(42) Sanskrit (same grammatical case, dative).
a. rājā vanāya (dative) pratis;t;hati.
king forest sets out
‘The king sets out for the forest.’
b. mahyam; (dative) pustakam; dehi.
to me book give
‘Give me the book.’
c. rocate viprebhyah (dative).
it is pleasing to the Brahmans
‘It is pleasing to the Brahmans.’
269

270
(43) Hindi (same postposition ko).
a. vah apne deś ko laut; gayā. (McGregor 1995: 54)
he own country to return went
‘He went back to his own country.’
b. us ādmī ko tīn pustkem< dījie. (McGregor 1995: 54)
that man to three books give
‘Give three books to that man.’
c. landan Ram ko pasand hai.
London Ram to pleasing is
‘Ram likes London.’
(44) Welsh (same preposition, i).
a. Ddaru nhw fynd i Iwerddon. (King, G. 1993: 187)
did they go to Ireland
‘Did they go to Ireland?’
b. Rhoddias fwyd i gath. (Thorne 1993: 405)
I gave food to cat
‘I gave food to a cat.’
c. Dylem ni maddau i’n gelynion. (Thorne 1993: 405)
necessary we forgiving to–our enemies
‘We ought to forgive our enemies.’
(45) Japanese (same postposition, ni).
a. Mary–wa Tokyo–ni itta.
Mary Tokyo–to went
‘Mary went to/toward Tokyo.’
b. Mary–wa sensei–ni tegami–o kaku.
Mary teacher–to letter writes
‘Mary writes a letter to the teacher.’
c. John–ni eiga–ga omoshiroi
John to movie is interesting
‘The movie is interesting to John.’

(46) Finnish (same grammatical case, allative).
a. Menimme asemalle (allative). (Olli 1958: 147)
we went to the station
‘We went to the station.’
b. Annoin miehelle (allative) rahaa. (Olli 1958: 147)
I gave man money
‘I gave the man some money.’
c. Jumala on laupias syntisille (allative). (Eliot 1890: 155)
God is merciful to sinners
‘God is merciful to sinners.’
(47) Turkish (same grammatical case, dative).
a. Türkiye’ye (dative) döndüler. (Lewis 1967: 36)
Turkey they returned
‘They returned to Turkey.’
b. Mektubu Ali’ye (dative) gösterdim. (Lewis 1967: 36)
letter Ali I showed
‘I showed the letter to Ali.’
c. Tütün size (dative) dokunur–mu. (Deny 1971: 187)
tobacco you annoy–question
‘Does tobacco annoy you?’
In RG, case relations like GOAL, EXPERIENCER, SOURCE, AGENT etc. are actually categorial
labels for constellations of semantic features. In the examples (39) to (47), the commonality in
GOAL and EXPERIENCER is the feature [+CONJUNCTURAL] which, roughly speaking, denotes
association or union. In examples (36) and (37) and all the examples in (39), the commonality in
SOURCE, AGENT, and CAUSE is [+DISJUNCTURAL], which, again roughly speaking, denotes
dissociation or separation. Therefore, the use of one thematic marker (preposition, postposition,
grammatical case, etc.) for a variety of thematic relations can be attributed to the presence of identical
features in those relations. This approach overcomes the loss of descriptive adequacy that all the
theories of case relations mentioned above have shared. In those systems, the common features
associated with thematic relations are not expressible, and it becomes a complete accident that the
same marker is used across relations.
The loss of descriptive adequacy in past theories of case grammar leads to a loss of explanatory
adequacy as well. It is very important that work in semantics, like all work in grammatical
description, be generalized across analyses. The study of thematic relations involves one area of
271

272
semantics. The descriptions used in this area must be relatable to descriptions used in other areas of
semantics and to the whole grammar in general. But a theory such as Fillmore’s case grammar is not
easily relatable to other work. For example, in an attempt to represent semantic relations among
different expressions, researchers have constructed semantics networks, which are systems of
connections that link together expressions from the same semantic field, such as color terms (red,
puce, dye, etc.), or animal categories (domestic animals, mammals, animals that live in water, etc.)
or expressions of separation (from, expel, source, etc.). Such semantic networks have been widely
studied in the computer representation of language. But thematic relations like SOURCE and GOAL
cannot be related in any direct way to the concepts which form such semantic networks, as we will
see when we consider semantic networks below.
Further, concepts like SOURCE and GOAL cannot be related to concepts used in other semantic
descriptions. For example, Roger Schank has proposed a set of semantic primitives out of which
meanings can be described. One of his primitives is EXPEL, which refers to the expulsion of an
object from the body of an animal; EXPEL is part of the meaning of verbs like sweat, spit, and cry.
Another of Schank’s primitives is INGEST, which refers to the taking in of an object; INGEST is part
of the meaning of verbs like eat, smoke, and breathe. Clearly, EXPEL is related to SOURCE and
INGEST is related to GOAL. But Fillmore’s theory and Schank’s theory are stated in such as way
that this relationship cannot be precisely specified.
In short, there is little transportability between these systems, so that the valuable insights of each
cannot be gathered into one framework. Yet, the grammatical facts of natural language, in particular,
the distribution of thematic markers, clearly indicate that there must be a connection between
thematic relations and semantic fields in general. That is, the same feature which shows up in
relations like SOURCE and AGENT ([+DISJUNCTURAL], e.g., from) should form part of the
definition of words like aversion, deprive, empty, and so on; and, that feature should also show up
in the definition of a primitive like EXPEL if a theory contains such a primitive. Similarly, the same
feature that shows up in relations like GOAL and EXPERIENCER ([+CONJUNCTURAL], e.g., to)
should form part of the definition of words like inclination, supply, fill, and the like and show up in
a putative primitive like INGEST.
This feature based approach to thematic relations provides an explanation for why the same
groupings of markers occur repeatedly in natural languages. Additionally, since features like
[+DISJUNCTURAL] and [+CONJUNCTURAL] can be directly defined in terms of human and
machine perceptual systems, this approach also specifies the connection between perception and
language. Any system which does not make this connection fails to attain explanatory adequacy.
The central thesis of RG semantics, therefore, is that thematic relations that are [–PST] are metaphors
for those that are [+PST], so that the same features that distinguish and relate the latter group are
employed in distinguishing and relating the former. Furthermore, we will see that thematic relations
like SOURCE and AGENT are categorial abbreviations for constellations of semantic features, just
as syntactic categories like NOUN and VERB are abbreviations for constellations of syntactic
features. Lastly, we will see that the semantic features from which complex semantic categories are
built are features that can be grounded in perception (human or machine).

4.12 POSITIONAL THEMATIC RELATIONS IN RG
We began the discussion of thematic relations by making a distinction between expressions which
indicate positional relations like He ran from his office and those which do not like He died from
overexposure. To this, we must add a fundamental breakdown between those positional relations
which refer to position in time [+TEMPORAL] and those which refer to position in space
[–TEMPORAL]. We can loosely define these features as follows:
(48) a. [+POSITIONAL] ([+PST, ±TMP]): having the primary focus on location,
orientation, or movement in space or time, e.g., He went FROM CALIFORNIA TO
NEW YORK, He remained FROM SUNRISE TO SUNSET.
273
b. [–POSITIONAL] ([–PST]): not having the primary focus on location, orientation, or
movement in space or time, e.g., FROM MY DESCRIPTION, he believes the movie
will appeal TO HIM.
(49) a. [+TEMPORAL] ([+PST, +TMP]): focusing on time, e.g, He remained FROM
SUNRISE TO SUNSET.
b. [–TEMPORAL] ([+PST, –TMP]): focusing on place, e.g., He went FROM
CALIFORNIA TO NEW YORK
Given a perceptual apparatus (human or machine), these definitions can be made very precise. For
example, a computer equipped with a vision system and an internal clock can assign specific values
to each of these features: the definition of [+PST] relations involving movement can be equated with
changes in positional vectors over time; the definition of [+PST] relations not involving movement
can be equated with a lack of change in positional vectors over time; the definition of [–PST]
relations can be equated with the lack of a positional vector altogether; and so on. Thus, the features
can be used to bridge the gap between the concrete (measurable, literal) and the abstract
(metaphorical), a very important objective in semantic analysis.
The vector components for space include values of length, width and depth although spatial relations
do not always refer to all three components. For example, a region may be viewed as a surface in
examples like They are on the lawn, where the depth of the lawn is not relevant. On the other hand,
a region can also be viewed as an area in examples like They are in the lawn, where the depth is
relevant. Compare also They are on the bed and They are in the bed; Look at the smudge on the
mirror and Look at your reflection in the mirror; The ants are on the floor and The termites are in
the floor.

274
4.13 SOME BASIC FEATURE DISTINCTIONS
Within the compass of positional relations, languages make a fundamental distinction between
relations that involve three dimensions and those that do not, e.g., in English, the difference between
in and on. We will specify this distinction as first and second order positional arguments as follows:
(50) [±FIRST ORDER] ([±FST]).
a. [+FST]: Thematic relations which are [+PST, +FST] express relationships relative
to a point, line or surface. Depth is not involved so that “surface” refers to a region
with only length and width.
Examples: on the table, off the table, at the door; on Tuesday, at six o’clock.
b. [–FST] (SECOND ORDER): Thematic relations which are [+PST, –FST] express
relationships relative to area or volume. Depth is involved so that “area” refers to a
region with length, width, and height; compare He is on the field (surface) versus He
is in the field (area).
Examples: in the room, out of the office; in March, out of office hours.
Within each of these categories, natural languages have positional markers to indicate whether there
is contact or lack of contact between the object and the location. If the location is a container, e.g.,
a building, then part of the container includes its interior space; hence, anything within that interior
space is viewed as being in contact with a part of the container. This detail is crucial to
understanding the possible meaning of sentences like The balloon floated around in the smoke–filled
room and She put the (porcelain) elephant into her purse. We will express the feature of contact
with the following opposition:
(51) [±PROXIMAL] ([±PRX]).
a. [+PRX]: involving contact between the object and the location.
Examples: on the floor, in the drawer; on Tuesday, in March.
b. [–PRX]: not involving contact between the object and the location.
Examples: at the door, near his office; around noon, near July 4.
Thematic relations that are [+PST, ±FST, ±PRX] do not necessitate motion, although motion is
possible, e.g., He walked out of his office versus He is out of his office. Such an option is not
available with all positional markers. For example, positional uses of to typically involve movement
of some kind (cf. He has never been to Europe, where the use of to necessitates interpreting been as
gone). Positional uses of to without movement as in They stood back to back are less common.

To account for the differences between motion and rest, as well as the direction of motion (to versus
from) and the location of rest (on/in versus off of/out of), we posit the feature oppositions
CONJUNCTURAL ([±CNJ]) and DISJUNCTURAL ([±DSJ]). Informally, we say that disjunctural
relations ([+DSJ]) are defined as those that involve parting, dissociation, withdrawal, detachment,
or separation. Conjunctural relations ([+CNJ]) are defined as those that involve joining, association,
advance, attachment, or union. Significantly, in any movement, there is simultaneously both a
withdrawal from one location (the SOURCE) and an advance to some other location (the GOAL).
In actual usage, one or the other is generally emphasized: depart from, advance to, go from, go to,
etc. In cases of rest like the cup is on the table, neither parting nor joining is involved.
These two interacting feature oppositions, [±CNJ] and [±DSJ], can be used to account for the
differences between movement and rest in space or time as follows:
(52) INVOLVING TWO LOCATIONS AND INVOLVING MOVEMENT:
a. [–DSJ, +CNJ]: involving movement toward one location in space or time, the
SOURCE location being understood or not mentioned:
Examples: He went as far as/up to the river; He’ll be here until/up to Friday.
b. [+DSJ, –CNJ]: involving movement away from one location in space or time, the
GOAL being understood or not mentioned:
Examples: He left from Detroit; He’s been here since Easter.
(53) INVOLVING ONE LOCATION AND NOT INVOLVING MOVEMENT:
a. [–DSJ, –CNJ]: not involving movement and relating to one location only, specifically
the place where some object is located:
Examples: He is in Europe; The meeting is in June.
275
b. [+DSJ, +CNJ]: not involving movement and relating to one location only,
specifically the place where some object is not located. Temporal examples like out
of season are uncommon.
Examples: He is away from his desk, out of the office.
Notice that, for movement to occur, at least two locations are required. In [αDSJ, –αCNJ] relations,
only one location need be mentioned; the other is implied and could be mentioned (He departed from
the city (for the country)). However, in [+DSJ, +CNJ] cases, any second implied location cannot be
mentioned unless a coordinating conjunction or comma intonation is used between the two,
otherwise, the object appears to be in two places at once:

276
(54) a. *It is now off the wall on the table.
b. It is now off the wall and on the table.
The last feature opposition to be presented is one needed to capture the distinction between
pinpointed and widespread positional expressions, e.g., on versus over, and in versus throughout.
We therefore add (55).
(55) [±EXTENSIONAL] ([±EXT])
a. [+EXT]: emphasizing the extent of space or time.
Examples: over the floor, all around the house; over the weekend, all during
February.
b. [–EXT]: emphasizing a small and highly circumscribed space or time.
Examples: on the horse, in the box; on Monday, in July.
A summary of examples of the positional thematic relations is given in Figure Nine.

–DSJ
–CNJ
+DSJ
+CNJ
–DSJ
+CNJ
+DSJ
–CNJ
FIGURE NINE I: PROXIMAL POSITIONAL THEMATIC RELATIONS ([+PST, –TMP, +PRX])
–EXT +EXT
+FST –FST +FST –FST
LOCATIVE1
on, on top of
He was on the horse.
ABSENTIVE1
off, off of
He was off the horse.
ILLATIVE1
on, onto
He got on the horse.
ELATIVE1
off, off of, from
He got off the horse.
LOCATIVE2
in, within
He was in the pool.
ABSENTIVE2
out of
He was out of the pool.
ILLATIVE2
into, in
He jumped into the
pool.
ELATIVE2
out of, from
He jumped out of the
pool.
LOCATIVE3
over, on
He had the patch over
his eye.
ABSENTIVE3
off, off of
He had the patch off
his eye.
ILLATIVE3
over, on
He put the patch over
his eye.
ELATIVE3
off, off of, from
He took the patch off
his eye.
LOCATIVE4
in, through(out)
There was smoke all
through the house.
ABSENTIVE4
out of
All the smoke was out
of the house.
ILLATIVE4
into, through(out)
He got smoke all
through the house.
ELATIVE4
out of, from
He got all the smoke
out of the house.
FIGURE NINE II: NONPROXIMAL POSITIONAL THEMATIC RELATIONS ([+PST, –TMP, –PRX])
–DSJ
–CNJ
+DSJ
+CNJ
–DSJ
+CNJ
+DSJ
–CNJ
–EXT +EXT
+FST –FST +FST –FST
ADESSIVE1
near, at
He was near the tree.
ABESSIVE1
away from, from
He was away from the
tree.
ALLATIVE1
to, toward
He walked toward the
tree.
ABLATIVE1
away from, from
He walked away from
the tree.
ADESSIVE2
near, at
The balloon was near
the ceiling.
ABESSIVE2
away from, from
The balloon was away
from the ceiling.
ALLATIVE2
to, toward
The balloon floated
toward the ceiling.
ABLATIVE2
away from, from
The balloon floated
away from the ceiling.
ADESSIVE3
along, near, at
The signs were along
the road.
ABESSIVE3
away from, from
The signs were away
from the road.
ALLATIVE3
along, near
He put the signs along
the road.
ABLATIVE3
away from, from
He put the signs away
from the road.
ADESSIVE4
among, amid
The planes were amid
the clouds.
ABESSIVE4
away from, from
The planes were away
from the clouds.
ALLATIVE4
among, amid
The planes flew amid
the clouds.
ABLATIVE4
away from, from
The planes flew away
from the clouds.
277

278
Figure Nine reveals an important advantage to having two interacting features (DSJ and CNJ) with
two values (plus and minus) to account for positional relations, namely, it affords us a simple way
of accounting for the many markers that spread over both STATIVE (rest) and NONSTATIVE
(motion) relations. For example, from is either [+DSJ, +CNJ], that is, stative, as in he is from Detroit,
or [+DSJ, –CNJ], that is, nonstative, as in he moved from Detroit. Using the same α–notation from
phonology (Page 82), we can generalize this as [+DSJ, αCNJ].
A system that begins with [±STATIVE] (or [± MOTION]) as primitives and then attempts to
categorize markers in terms of these primitives misrepresents the facts. A marker like from is
essentially separative, i.e., [+DISJUNCTURAL], however one wishes to label this; it is not
essentially either a marker of motion or a marker of rest. Interestingly, English to is not essentially
[+CONJUNCTURAL]; rather, its basic feature is [–DISJUNCTURAL]. There are LOCATIVE uses
(it is stuck to the wall, they are cheek to cheek) and ADESSIVE uses (his back is to the wall, they
stood to one side, it lies to the north). We find similar data in other languages. In French, for
example, en is used for [–DISJUNCTURAL] expressions: aller en France (‘go to France’), entrer
en ville (‘go into town’), vivre en Angleterre (‘live in England’).
In general, thematic markers in the world’s languages have both stative and nonstative uses, and are
not primarily associated with either motion or rest though this certainly does occur. Crucially, even
when this does occur, the relations are still best expressed along the lines of Figure Nine. For
example, many German prepositions (an, auf, hinter, etc.) are used with the dative case for rest and
the accusative case for motion. This simply means that the dative is associated with [αDSJ, αCNJ]
and the accusative with [αDSJ, –αCNJ]. What we do not find in natural language are common
examples where markers are confined to motion or rest and within each category have both
disjunctural and conjunctural meanings, i.e., a marker that means either motion to or motion from,
or one that means either rest in or rest out of. For these reasons, concepts like stative are derivative,
not primitive, in the system described here.
In Figure Nine notice that English frequently neutralizes features so that the same preposition is used
throughout a row. This is, in fact, one of the motivations behind the RG system. One needs to be
able to both pinpoint and generalize prepositional usage. In some instances, there are fine
distinctions available: in, on, over, throughout. In others, the same preposition covers a wide range
of cases, e.g., from and to. In this regard, English is typical of the world’s languages, though, again,
as the examples from French and German above indicate, different languages will distribute
prepositions over different feature spaces.
The most obvious positional features that have nonpositional correlates are the four DSJ/CNJ pairs
of Figure Nine in both their PROXIMAL and NONPROXIMAL variations. Certain nonpositional
thematic relations are viewed by speakers as involving a conjunction (to and into), others a
disjunction (from and out of), and still others a combination of disjunction and conjunction (off and
away) or lack of both (on and in). Accordingly, the same prepositions that occur in the positional
relations show up in the nonpositional ones. We have very clear cut examples like (56) though (59).

(56) a. He turned into a street. [+PST]
b. He turned into a frog. [–PST]
(57) a. He leaned against them. [+PST]
b. He fought against them. [–PST]
(58) a. They roamed about the city. [+PST]
b. They talked about the city. [–PST]
(59) a. They live by water. [+PST] or [–PST]
b. They travel by water. [+PST] of [–PST]
4.14 NONPOSITIONAL THEMATIC RELATIONS
We have argued above that the nonpositional thematic relations are metaphors for their positional
counterparts. To repeat, a [+PST, +DSJ] theme expresses literal (measurable) separation, whereas
a [–PST, +DSJ] theme expresses figurative separation. A theme like SOURCE is [+PST, +DSJ]; one
like CAUSE is [–PST, +DSJ]. Similarly, a [+PST, –DSJ] theme expresses literal (measurable) union,
whereas a [–PST, –DSJ] theme expresses figurative union. Thus, GOAL is [+PST, –DSJ];
EXPERIENCER is [–PST, –DSJ].
A summary of the nonpositional thematic relations is given in Figure Ten. As Figures Nine and Ten
indicate, case relations like ILLATIVE, AFFECTIVE (EXPERIENCER), EFFECTIVE (AGENT),
etc. are actually categorial labels for constellations of semantic features in RG. Thus, semantics
exactly parallels syntax and phonology. Each one of the semantic features can be precisely defined
in terms of human (or machine) vision.
The commonality in ILLATIVE, ALLATIVE and AFFECTIVE is the feature [+CNJ] which denotes
association or union; the commonality in ELATIVE, ABLATIVE, and EFFECTIVE (and all the
examples in (39)) is [+DSJ], which denotes dissociation or separation. Therefore, the use of one
thematic marker (preposition, postposition, grammatical case, etc.) for a variety of thematic relations
can be attributed to the presence of identical features in those relations. This approach overcomes
the loss of descriptive adequacy that all the theories of case relations mentioned above have shared.
In those systems, the common features associated with thematic relations are not expressible, and it
becomes a complete accident that the same marker is used across relations.
279

280
–DSJ
–CNJ
+DSJ
+CNJ
–DSJ
+CNJ
+DSJ
–CNJ
–DSJ
–CNJ
+DSJ
+CNJ
–DSJ
+CNJ
+DSJ
–CNJ
FIGURE TEN I: PROXIMAL NONPOSITIONAL THEMATIC RELATIONS ([–PST, +PRX])
POSSESSOR
Ø
He knows French.
NONPOSSESSOR
Ø
He doesn’t know
French.
–EXT +EXT
+FST –FST +FST –FST
AFFECTIVE
to
Al taught French to Ed.
EFFECTIVE
from, by
Ed learned French from
Al.
ATTRIBUTIVE
in, with, of
He is a man with/of many
hopes.
NONATTRIBUTIVE
out of, without, from
He is a man without any
hopes.
RESULTATIVE
into
Al turned Ed into an
optimist.
COMPOSITIONAL
out of, from
Ed made a optimist out of
Al.
COMPARATIVE
like, with, to
He is similar to his father.
He is like his father.
NONCOMPARATIVE
unlike, from
He differs from his father.
He is unlike his father.
REFERENTIAL
for
For her, the price is right.
EXPEDIENTIAL
by
He got there by plane.
CIRCUMSTANTIAL
with
He does everything with
haste.
NONCIRCUMSTANTIAL
without
He does everything without
haste.
CONSECUTIVE
for
He’s too ill for work.
CAUSAL
from, of
He died from the extreme
cold.
FIGURE TEN II: NONPROXIMAL NONPOSITIONAL THEMATIC RELATIONS ([–PST, –PRX])
POSSESSOR
Ø
He has a Volvo.
NONPOSSESSOR
Ø
He doesn’t have a
Volvo.
–EXT +EXT
+FST –FST +FST –FST
AFFERENTIAL
to
Al sold his Volvo to Ed.
EFFERENTIAL
from
Ed bought his Volvo
from Al.
ASSOCIATIVE
in, with, of
He is in debt. He’s a man
with/of many debts.
NONASSOCIATIVE
out of, without, from
He is out of debt. He’s a
man without debts.
TERMINATIVE
into
Al got Ed into debt.
ORIGINATIVE
out of, from
Ed got Al out of debt.
COMITATIVE
with
He made the dinner with
her.
NONCOMITATIVE
without
He made the dinner
without her.
BENEFACTIVE
for
He did it for her.
DIFFERENTIAL
by
He won by a mile.
INSTRUMENTAL
with
He built the fort with tools.
NONINSTRUMENTAL
without
He built the fort without
tools.
PURPOSIVE
for
He learned Italian for fun.
He did it for many reasons.
DELIMITIVE
in the event of, if
In the event of an accident,
call the police.

The definitions for the major nonpositional thematic relations are as follows:
(60) EFFECTIVE (EFC): The animate being, missile, or force typically perceived as bringing
about the action identified in the predicate (variously called AGENTIVE, AGENT, and
FORCE).
BILL killed the bug./The bug was killed BY BILL.
THE BOULDER totaled the car./The car was totaled BY THE BOULDER.
THE TORNADO destroyed the crops./The crops were destroyed BY THE TORNADO.
(61) COMPOSITIONAL (CPS): The material or things out of which something is composed;
the product retains the original ingredients (related to INALIENABLE POSSESSION).
They crafted the vase OUT OF SILVER.
The cabinets are made OF WOOD.
The oak grew OUT OF AN ACORN.
(62) EXPEDIENTIAL (EXP): The means by which something is done.
We went BY CAR.
BY COUNTING TO TEN, she manages not to abuse the kids.
(63) CAUSAL (CAU): The entity which expresses the cause of the action or state identified in
the predicate.
He died FROM SMOKING. He died OF CANCER.
He collapsed FROM THE HEAT.
He acted OUT OF GUILT.
(64) EFFERENTIAL (EFR): The quasi–positional entity from which an action emanates. Often
called SOURCE by others.
John received/bought the car FROM BILL.
BILL gave/sold the car to John.
(65) ORIGINATIVE (ORG): The entity or condition from which something is transformed
(related to ALIENABLE POSSESSION).
He got himself OUT OF DEBT.
He detoxed FROM HEROIN.
281

282
(66) DIFFERENTIAL (DIF): The degree of difference expressed in a contrast.
He won BY A MILE.
They are distinguished BY MANY CHARACTERISTICS.
(67) DELIMITIVE (DEL): Delimitive themes specify the condition(s) from which something
results; hence, they are classified as [+DISJUNCTURAL]. Most delimitive expressions in
English are clauses introduced by if.
His sister will care for his children, IF HE DIES.
IN THE EVENT OF HIS DEATH, his sister will care for his children.
(68) AFFECTIVE (AFC): The animate or inanimate entity directly affected by the state or
action identified in the predicate. Variously called EXPERIENCER or DATIVE by others.
JOHN feels that Bill will win.
John killed BILL./BILL died.
John melted THE ICE./ THE ICE melted.
He put the money IN HIS POCKET.
(69) RESULTATIVE (RES): The product created from some materials and retaining the original
ingredients (related to INALIENABLE POSSESSION).
The wove the straw INTO BASKETS.
THE BASKETS were made out of straw.
She baked (the ingredients into) A CAKE.
The acorn grew INTO AN OAK.
He invented THE TELEPHONE.
(70) REFERENTIAL (REF): The person or thing in reference or relation to which something
occurs.
John will always be a hero TO BILL (AS FAR AS BILL IS CONCERNED).
A term paper is optional FOR UNDERGRADUATES.
He is smart FOR HIS AGE.
Bill is a stickler FOR DETAILS.
(71) CONSECUTIVE (CNS): The result or consequences of an action or state.
He became too sick TO/FOR WORK.
He became so sick THAT HE COULDN’T WORK.
It’s too early FOR DINNER.
He has enough qualifications/qualifies FOR THE JOB.

(72) AFFERENTIAL (AFR): The quasi–positional entity to which an action is directed. Often
called GOAL by others.
283
Bill sold the car TO JOHN./Bill sold JOHN the car.
JOHN received/bought the car from Bill.
He gave a new coat of paint TO THE HOUSE./He gave THE HOUSE a new coat of paint.
(73) TERMINATIVE (TRM): The entity or condition into which something is transformed
(related to ALIENABLE POSSESSION).
He came INTO A FORTUNE./He became WEALTHY.
He got himself INTO DEBT.
He withdrew INTO DRUGS AND ALCOHOL.
(74) BENEFACTIVE (BEN): The entity, usually animate, for whose benefit or on whose behalf
the action or state identified in the predicate occurs (see below, Endnote ?).
Bill bought a gift FOR JOAN/bought JOAN a gift.
JOAN was bought a gift by Bill.
(75) PURPOSIVE (PUR): The entity which specifies the purpose of or reason for the action or
state identified in the predicate.
They trained him FOR THE JOB/TO DO THE JOB.
He entered the competition FOR THE MONEY/TO GET MONEY.
They dug a hole FOR WATER/TO GET WATER.
She jumped FOR JOY.
She did it FOR MANY REASONS.
They rewarded him FOR HIS RESEARCH.
4.15 DIMENSIONAL THEMATIC RELATIONS
In addition to the feature specifications given in the above figures, the RG system recognizes another
set of thematic relations of a DIMENSIONAL nature. These are defined as follows and illustrated
comparatively in Figure Eleven:
(76) COMPONENTIAL1: the whole (holonym) in a part–whole relationship.
a. A ROOM has a ceiling.
b. A HOUSE has a roof.

284
(77) COMPONENTIAL2: the part (meronym) in a part–whole relationship.
a. A CEILING is part of a room.
b. A ROOF is part of a house.
(78) TYPOLOGICAL1: the class (hypernym) in a member–class relationship.
a. The class MAMMAL contains primates.
b. The class BIRD contains robins.
(79) TYPOLOGICAL2: the member (hyponym) in a member–class relationship.
a. PRIMATES are mammals.
b. ROBINS are a type of bird.
Since the relationships may involve either inalienable or alienable characteristics and nonpossession
as well as possession, we have the sixteen relationships specified in Figure Eleven.
+PRX
–DSJ
–CNJ
+PRX
+DSJ
+CNJ
–PRX
–DSJ
–CNJ
–PRX
+DSJ
+CNJ
FIGURE ELEVEN: CLASSIFICATORY THEMATIC RELATIONS
–EXT +EXT
+FST –FST +FST –FST
COMPONENTIAL1 COMPONENTIAL2 TYPOLOGICAL1 TYPOLOGICAL2
An arm is A BODY
PART.
The neck is not part of
THE HEAD.
A flower is part of A
BOUQUET.
Hawaii is not part of
CONTINENTAL
USA.
AN ARM is a body
part.
THE NECK is not part
of the head.
A FLOWER is part of a
bouquet.
HAWAII is not part of
the continental USA.
A penguin is A BIRD. A PENGUIN is a bird.
A penguin is A
NONFLIER.
A PENGUIN is a
nonflier.
John is a SMOKER. JOHN is a smoker.
John is a
NONSMOKER.
JOHN is a nonsmoker.
Notice that any particular characteristic of an object might be expressed in several different ways.
For example, in (70, 80a), black is treated as an inalienable characteristic while in (70, 80b) it is
treated as a class.

(80) a. ATTRIBUTIVE: He is black.
b. TYPOLOGICAL: He is a black.
Similarly, in (81a) poor is treated as an alienable characteristic; in (81b) it is treated as a class.
(81) a. ASSOCIATIVE: He is poor.
b. TYPOLOGICAL: The poor need assistance.
Since the same characteristic can fit more than one of our proposed thematic relations, it might appear
that the system we have described is unnecessarily redundant or complex. But this is not the case.
Different individuals classify and define the same things in numerous, diverse ways. For example,
one can classify United States judges as members of a particular branch of government or as
nonmembers of other branches:
(82) a. Judges are part of the judicial branch of government.
b. The judicial branch of government includes judges.
(83) a. Judges are not part of the legislative branch of government.
b. The legislative branch of government does not include judges.
Further, some speakers may view (82) as an INALIENABLE classification, others as ALIENABLE,
and these particular assignments are relatively stable. Consider the possible assignments when
attempting to classify a word like teenager. Assignments into different cells of the figures above is
often a matter of personal beliefs, habit and experience. For example, in a random poll, it was found
that most students in an introductory linguistics class did not view the neck as part of the head;
however, several did, some with great conviction: “Well, after all, the neck goes with the head in
decapitation,” one student remarked ominously.
Lastly, observe that care must be taken in comparing meanings. For example, the assignment of the
number π to the class of irrational numbers and the description of an individual as a irrational person
do not employ the same meaning of irrational:
(84) a. TYPOLOGICAL: Pi (π) is irrational.
b. CIRCUMSTANTIAL: John is irrational.
The feature opposition [±DIMENSIONAL] is the last semantic feature we will consider. This
opposition, in addition to the others discussed previously, will figure prominently in our discussion
of semantic networks below.
285

286
4.16 PROTOTYPES AND STRENGTH OF ASSOCIATION
As we noted above, the nonpositional thematic relations must be related to the positional ones, if only
to account for the fact that the same set of markers appears for both in most of the world’s languages.
This is not to say that languages never distinguish the two sets. In fact, in English, forms like
here/there/where are never used for nonpositional relations, as the following pairs show.
(85) a. I know where he turned; it was here, into this street.
b. *I know where he turned; it was there, into a frog.
(86) a. I know where that chandelier hung; it was here, over this table.
b. *I know where that chandelier cost; it was here, over two dollars.
(87) a. I know where John decided; it was there, on the boat.
b. *I know where John decided; it was here, on this ring.
The distinction between [+PST] (literal) and [–PST] (metaphorical) uses of markers is not sharp. A
more realistic characterization must recognize a gradation from expressions that are invariably
considered [+PST] by speakers to those that never or rarely are. This same gradation or strength of
association operates in many areas of semantic categorization. For example, using a scale from 4
(most strength of association) to 0 (no strength of association), we might categorize nouns as
referring to a location in space in the following way:
(88) a. PST4: desert, planet
b. PST3: bed, office
c. PST2: eye, lettuce
d. PST1: law, phrase
e. PST0: some, the
This classification asserts that speakers primarily use the word desert to denote a place (i.e., a desert
is a dry sandy place), whereas the word law is rarely used to signify a place (cf. It is stated in the
law), and the word some (for all practical purposes) never is. Exactly how many numbers are needed
in such a scale is an empirical matter. Suppose we begin by proposing a five–point scale from 0 up
to 4. The assignment of nouns and other expressions to specific classes will also depend on
investigation: clearly, all speakers will not agree with the judgments in (88).
Psychologists like Eleanor Rosch have researched such categorizations and noted that speakers
formulate prototypes, that is, typical examples of categories. For example, most American English
speakers feel that the robin is the prototypical bird, so a sentence like A robin is sort of a bird is
generally considered very odd. This contrasts with sentences like A bat is sort of a bird, which seem
acceptable. On the basis of such work on prototypes, one might propose, the following scale of
“birdiness” for American English speakers:

(89) a. BIRD4: robin
A robin is a bird.
*A robin is sort of a bird.
b. BIRD3: chicken
A chicken is a bird.
?A chicken is sort of a bird.
c. BIRD2: bat
*A bat is a bird.
A bat is sort of a bird.
d. BIRD1: kite
*A kite is a bird.
?A kite is sort of a bird.
e. BIRD0: hammer
*A hammer is a bird.
*A hammer is sort of a bird.
Examples like the above show that it is possible to construct prototype scales for a number of
different categories. Suppose we use the concept of prototype to categorize various expressions
containing positional markers in an effort to deal with the complex problem of specifying the
meanings and uses of prepositions. The progression we find along the PST–scale moves from true
positional expressions, those that denote a location ([PST4] and [PST3]), to expressions of condition,
circumstance, or manner ([PST1] and [PST0]), with [PST2] expressions lying somewhere in between.
For example, consider the following phrases, in particular, what constitutes an appropriate response
to the request made:
(90) a. [PST4]: live in Ann Arbor
1. As a location:
REQUEST: Please tell me where he lives.
RESPONSE: He lives in Ann Arbor.
2. As a manner or condition:
REQUEST: Please tell me how he lives.
*RESPONSE: He lives in Ann Arbor.
b. [PST3]: live in a pigsty (=live amid filth)
1. As a location:
REQUEST: Please tell me where he lives.
RESPONSE: He lives in a pigsty.
287

288
2. As a manner or condition:
REQUEST: Please tell me how he lives.
?RESPONSE: He lives in a pigsty.
RESPONSE: He lives like a pig.
c. [PST2]: live in luxury
1. As a location:
REQUEST: Please tell me where he lives.
?RESPONSE: He lives in luxury.
2. As a manner or condition:
REQUEST: Please tell me how he lives.
RESPONSE: He lives in luxury.
RESPONSE: He lives like a king.
d. [PST1]: live on a budget
1. As a location:
REQUEST: Please tell me where he lives.
*RESPONSE: He lives on a budget.
2. As a manner or condition:
REQUEST: Please tell me how he lives.
RESPONSE: He lives on a budget.
e. [PST0]: live on a shoestring
1. As a location:
REQUEST: Please tell me where he lives.
*RESPONSE: He lives on a shoestring.
2. As a manner or condition:
REQUEST: Please tell me how he lives.
RESPONSE: He lives on a shoestring.
These examples are proffered to illustrate a progression from concrete expressions, with maximum
strength of association to location in space, to ever more figurative ones, where the strength of
association to true location diminishes. Notice, for example, that the answer He lives in a pigsty in
(90b1) does not supply the precise information requested, though it is an acceptable response.
The most concrete of the four types are the [PST4] relations. These are most concrete because they
can be verified by a human or machine vision system (He is on the roof). The most figurative of the
four are the [PST0] relations. These are the true idioms, that is, those expressions whose meaning

cannot be derived from the (literal) meaning of their parts (He is on hand). Sometimes the same
phrase can be interpreted at either extreme. For example, He is on the wagon can refer to his actual
position or to his state of sobriety. Somewhere in between these extremes are the [PST3], [PST2],
and [PST1] relations.
[PST3] expressions are not completely metaphorical, because they co–occur with true positional
markers like here, there, where, etc. (I couldn’t possibly live here, in this pigsty or His hopes must
lie somewhere, but I don’t know where). Because they add more information than just position, they
cannot be consider [PST4]; hence, they have slightly less strength of association to position.
The difference between [PST4] and [PST3] expressions is also manifested in the wide variation
possible for the introductory preposition. The verb live in (90a) can be followed by virtually any
positional preposition including, on, under, behind, among, around, and so on. However, the
prepositions used in expressions like (90b) are highly idiosyncratic. In the case of an expression like
rest one’s hopes on, possibly the only other preposition than on that can be used is in; in the case of
live in a pigsty, the preposition seems invariable.
[PST2] and [PST1] expressions have a meaning that is more concrete than fully idiomatic, yet not
really positional. The meaning of such expressions can be deduced from the meanings of their parts,
and, more significantly, the force of the preposition used is still felt to the extent that related
prepositions can be substituted, albeit restrictively (live in/amidst luxury or live on/within/by a
budget). The difference between these two classes is that, while [PST2] expressions might
conceivably be used locatively, [PST1] cannot be used locatively except with great license. [PST1]
expressions are true expressions of circumstance or manner.
To integrate the five–point scale with the features we have been discussing, let us refer to expressions
that are [PST4], [PST3], or [PST2] as [+POSITIONAL], those that are [PST1] or [PST0] as
[–POSITIONAL].
The distinctions presented here are not always clear cut, and the degree to which speakers feel the
literal and/or positional sense of an expression varies. In my own speech, (91a) seems fine, but (91b)
does not.
(91) a. I’ll tell you where he’s living. He’s living in the lap of luxury in New York City.
b. *I’ll tell you where he’s living. He’s living on a shoestring in New York City.
Both of these examples contain idiomatic expressions, but only (91a) seems to be [+POSITIONAL].
Notice that the presence of a clear–cut positional expression like in New York City is not the
determining factor. We have examples like the following which contain more than one locative
expression:
(92) a. He’s living in New York City on a shoestring.
b. He’s living in New York City in a hovel.
289

290
Notice also that the questioning associated with examples like (91b) actually calls for an expression
of manner:
(93) a. QUESTION: How can he afford to live in New York City?
b. RESPONSE: Certainly not on a shoestring.
Further, the borderline status of (91a), somewhere between true positional and true nonpositional,
i.e., [PST2], makes possible the following interchange:
(94) a. QUESTION: How is he living?
b. RESPONSE: In the lap of luxury.
As we have noted in many places in this book, a proposal must arise from the data and be supported
with examples from the data. In the present case, we need to determine how a particular expression
will be classified along the PST–scale, that is, how can we determine if an expression is PST4 or
PST0 or somewhere in between. As a first approximation, let us propose six criteria as a means to
determining where the expression falls on the PST–scale as follows:
(95) [PST4]: live in Ann Arbor
a. Allow positional questions words like where, from what place, etc.: I’ll tell you
where he lives; he lives in Ann Arbor. *Ann Arbor is what he lives in.
b. Allow positional adverbials like there, here, etc.: Does he live right there in Ann
Arbor?
c. Allow positional relatives, e.g., Ann Arbor is the only place in which he can live.
d. Allow free use of positional prepositions: He lives around/outside of/down in/over
by Ann Arbor.
e. Allow free use of directional phrases: He lives just east of Ann Arbor.
f. Do not allow expressions of manner: *The only way to live is in Ann Arbor.
(96) [PST3]: live in a pigsty (=live amid filth)
a. Allow positional questions words like where, from what place, etc.: I’ll tell you
where he lives; he lives in a pigsty.
b. Allow positional adverbials like there, here, etc.: Does he live right there, in that
pigsty?
c. Marginally allow positional relatives, e.g., ?A pigsty is the only place in which he can
live.
d. Do not allow free use of positional prepositions: He lives in/*within/*amid a pigsty.
e. Marginally allow use of directional phrases: ??He lives just east of a pigsty.
f. Often allow expressions of manner: The only way to live is in a pigsty.

(97) [PST2]: live in luxury
291
a. Allow positional questions words like where, from what place, etc.: I’ll tell you
where he lives; he lives in luxury.
b. Do not freely allow positional adverbials like there, here, etc.: *Does he live right
there, I mean in luxury?
c. Do not allow positional relatives, e.g., *Luxury is the only place in which he can live.
d. Allow only limited variety of positional prepositions: live in/amid/amidst/within
luxury; *live out of/around/behind luxury; ?live beyond luxury.
e. Do not allow directional phrases: *He lives just south of luxury.
f. Allow expressions of manner: The only way to live is in luxury. (cf. The only way to
live is luxuriously.)
(98) [PST1]: live on a budget
a. Do not allow positional questions words like where, from what place, etc.: *I’ll tell
you where he lives; he lives on a budget.
b. Do not allow positional adverbials like there, here, etc.: *Does he live right there, I
mean on that budget?
c. Do not allow positional relatives, e.g., *A budget is the only place on which he can
live.
d. Allow only limited variety of positional prepositions: live on/within/by a budget;
*live out of/around/behind a budget. (cf. live according to a budget).
e. Do not allow directional phrases: *He lives just south of a budget.
f. Allow expressions of manner: The only way to live is on a budget.
(99) [PST0]: live on a shoestring
a. Do not allow positional questions words like where, from what place, etc.: *I’ll tell
you where he lives; he lives on a shoestring.
b. Do not allow positional adverbials like there, here, etc.: *Does he live right there, I
mean on that shoestring?
c. Do not allow positional relatives, e.g., *A shoestring is the only place on which he can
live.
d. Do not allow even limited variety of positional prepositions: *live upon a shoestring.
e. Do not allow directional phrases: *He lives just south of a shoestring.
f. Allow expressions of manner: One way I don’t want to live is on a shoestring.
As these examples how, a major problem for semantics is categorizing and specifying all the various
expressions with prepositions that native speakers know. One of the factors that makes this problem
so thorny is the variability that exists from speaker to speaker. In phonology and syntax, many rules
and principles can be generalized across all speakers despite dialectal differences. In semantics, the
data are often elusive, there is considerable variability from speaker to speaker, and it is often
difficult to formulate principles with precision. As we have noted, semantic analysis is typically a

292
wicked problem in the sense mentioned in the Introduction (Page 5). The hypothesis here is that the
problem might be solved using the concept of prototypes. The examples above support that
hypothesis. Attempting to apply and test the hypothesis with other examples of non–idiomatic and
idiomatic uses of specific thematic relations leads to examples like the following:
(100) a. He put the milk into the refrigerator (ILLATIVE – [PST4]).
He put all his eggs into one basket (ILLATIVE – [PST4] or [PST0]).
b. He took the milk out of the refrigerator (ELATIVE – [PST4]).
He took the wind out of my sails (ELATIVE – [PST4] or [PST0]).
c. He bought the book from Bill (EFFERENTIAL – [PST4]).
It’s like taking candy from a baby (EFFERENTIAL – [PST4] or [PST0]).
To be viable, the hypothesis must now be tested against the representation of thematic relations in
as many other examples and languages as possible. Further, it must be integrated with the other
aspects of semantics that we have considered.
4.17 THE PERCEPTUAL BASIS OF SEMANTICS
Our discussion of semantics has centered on the hypothesis that thematic relations such as
EFFECTIVE, AFFECTIVE, ELATIVE, ILLATIVE, etc. are actually categorial labels for
constellations of semantic features. Each one of the semantic features can be precisely defined in
terms of human (or machine) perception. Given this approach, the use of one thematic marker
(preposition, postposition, grammatical case, etc.) for a variety of thematic relations can be attributed
to the presence of identical features in those relations.
This approach overcomes the loss of descriptive adequacy that theories of case relations mentioned
above have shared. In those systems, the common features associated with thematic relations are not
expressible, and it becomes an accident that the same marker is used across relations. Further, as we
saw, there was a loss of explanatory adequacy as well.
The theory outlined and justified above overcomes these deficiencies. In addition to explicating the
cross–categorial uses of thematic markers, the semantic features can serve as the basis of a
knowledge representation system. People use language to communicate, to share information, to ask
questions, to express feelings, and so on. Somehow, grammars must interface with other cognitive
systems to represent the general knowledge that people have and use during everyday
communication. To see how this might be possible, let us look again at the features that define
thematic relations.
As Figures Nine and Ten reveal, different clusters of features define different thematic relations. A
feature cluster such as [+PST, –TMP, –DIM, –EXT, +FST, +PRX, –DSJ, –CNJ] has the following
meanings:

(101) a. +PST POSITIONAL INVOLVING LOCATION
b. –TMP TEMPORAL LOCATION IN SPACE, NOT TIME
c. –DIM NONDIMENSIONAL NOT CONCERNED WITH MEASUREMENT
d. –EXT NONEXTENSIONAL PINPOINTED LOCATION
e. +FST FIRST ORDER LOCATION RELATIVE TO POINT, LINE,
OR SURFACE
f. +PRX PROXIMAL INVOLVING CONTACT
g. –DSJ NONDISJUNCTURAL NOT EMPHASIZING PARTING
h. –CNJ NONCONJUNCTURAL NOT EMPHASIZING JOINING
Thus, the cluster of features in (101) defines the relation LOCATIVE1 (see Figure Nine on Page 277)
realized in English in such phrases as on walls or on Mars as they would occur in sentences like
Pictures are found on walls and Martians are found on Mars.
Changing one or more feature specifications produces different thematic relations as follows:
(102) [+PST, +TMP, +DIM, –EXT, +FST, +PRX, –DSJ, –CNJ] defines expressions like on
Sunday (still LOCATIVE1) as it might occur in a sentence like Church services often occur
on Sunday.
(103) [+PST, –TMP, –DIM, –EXT, –FST, +PRX, –DSJ, –CNJ] defines expressions like in water
(LOCATIVE2), in sentences like Fish are found in water.
(104) [–PST, –TMP, –DIM, –EXT, –FST, +PRX, –DSJ, –CNJ] defines expressions like in Latin
(ATTRIBUTIVE), which underlie sentences like The poem is in Latin.
(105) [–PST, –TMP, +DIM, –EXT, –FST, +PRX, –DSJ, –CNJ] defines COMPONENTIAL2
expressions like A wing is part of a bird which underlie sentences like All birds have wings.
As these examples show, sets of semantic features specify individual thematic relations. On a higher
level of organization, that of knowledge representation, sets of thematic relations can be used to
construct a semantic network.
293

294
4.18 SEMANTIC NETWORKS
Semantic networks consist of nodes and links. For example, they might contain a node FISH–1
(‘cold–blooded aquatic vertebrate’) and FISH–2 (‘attempt to catch FISH–1’). The node FISH–1 is
connected to the node BODY_OF_WATER–1 (‘a body of water’) by a LOCATIVE2 link, thereby
expressing the fact that fish are found in water. FISH–1 is also connected to TROUT–1 by a
TYPOLOGICAL2 link indicating that one type of fish is a trout. Further, since TROUT–1 is a type
of FISH–1, TROUT–1 “inherits” the LOCATIVE2 link to BODY_OF_WATER–1; thus, TROUT–1
are found in BODY_OF_WATER–1.
The difficulty with semantic networks available in the literature is that they lack independent criteria
for specifying what can be a link in a network. We can overcome that deficiency here. Each link in
the network can be defined as a cluster of semantic features, that is, as a thematic relation. For
example, consider (106) and (107), where each arrow indicates a pointer (link) from one node in a
network to another node, and where the thematic relation left of the arrow names the pointer.
NODE ONE POINTER NODE TWO
(106) a. LIVING_THING–1 ––TYP2 ––> ENTITY–1
b. LIVING_THING–1 ––EFC ––> REPRODUCTION–1
c. LIVING_THING–1 ––EFC ––> INGESTION–1
d. LIVING_THING–1 ––AFC ––> DEATH–1
e. LIVING_THING–1 ––AFC ––> GROWTH–1
f. LIVING_THING–1 ––IPS ––> LIFE–1
This list encodes the information that a LIVING_THING–1 is a type of ENTITY–1, that it can be an
agent (EFC) of REPRODUCTION–1 and INGESTION–1, that it can experience (AFC) DEATH–1
and GROWTH–1, and that it is an inalienable possessor (IPS) of LIFE–1.
(107) a. ANIMAL–1 ––TYP2 ––> LIVING_THING–1
b. ANIMAL–1 ––EFC ––> MOVEMENT–1
c. ANIMAL–1 ––AFC ––> SLEEP–1
d. ANIMAL–1 ––AFC ––> SENSATION–1
e. ANIMAL–1 ––IPS ––> SPONTANEITY–1
f. ANIMAL–1 ––APS ––> HOME–1
This list encodes the information that an ANIMAL is a type of LIVING_THING–1 which can be an
agent (EFC) of MOVEMENT–1, which can experience (AFC) SLEEP–1 and SENSATION–1, and
which inalienably possesses (IPS) SPONTANEITY–1 and alienably possesses (APS) HOME–1.
Each link in representations like the above can be made bi-directional so that information can be
retrieved in reverse. For example, the network can proceed from ENTITY–1 to LIVING_THING–1
in (106a) and then from LIVING_THING–1 to ANIMAL–1 in (107a), or it can proceed in reverse.

As a further example, consider the fragment of the network in (108).
(108) a. HUMAN–1 ––TYP2 ––> ANIMAL–1
b. HUMAN–1 ––EFC ––> SPEECH–1
c. HUMAN–1 ––EFC ––> WRITING–1
d. HUMAN–1 ––EFC ––> IDEA–1
e. HUMAN–1 ––EFC ––> TOOL–1
f. HUMAN–1 ––EFC ––> REASON–1
g. HUMAN–1 ––AFC ––> JOY–1
h. HUMAN–1 ––IPS ––> LANGUAGE–1
I. HUMAN–1 ––IPS ––> KINSHIP–1
j. HUMAN–1 ––IPS ––> RACE–1
k. HUMAN–1 ––APS ––> JOB–1
l. HUMAN–1 ––APS ––> SPOUSE–1
m. HUMAN–1 ––APS ––> CLOTHING–1
n. HUMAN–1 ––APS ––> NAME–1
o. HUMAN–1 ––APS ––> TOOL–1
This list encodes the information that a HUMAN–1 is a type of ANIMAL–1 which can be an agent
(EFC) of SPEECH–1, WRITING–1, IDEA–1, TOOL–1, and REASON–1, which can experience
(AFC) JOY–1, which inalienably possesses (IPS) LANGUAGE–1, KINSHIP–1, and RACE–1 and
alienably possesses (APS) JOB–1, SPOUSE–1, CLOTHING–1, NAME–1 and TOOL–1.
The lists in (106) ) (108) are intended to encode information as people do; therefore, there are
apparent redundancies. Notice for example that (108) specifies that humans own (APS) and use
(EFC) tools. It is certainly possible for humans to use tools but not own any, e.g., some of one’s
neighbors. It is also possible for humans to own tools but never use any of them.
One problem with the above examples is that they do not account for variability and deviations from
expected cases. To solve that problem, we can once again use the concept of prototypes. Each one
of the links in the semantic network can contain a strength of association indicating how strong the
link is. All links are assumed by default to have the highest strength of association on the scale from
0 (lowest) to 4 (highest). When a link does not have the highest strength of association, the particular
strength is indicated here in square brackets. For example, a simple network for BIRD–1 (the animal)
might be the following:
(109) a. BIRD–1 ––IPS ––> BILL–1
b. BIRD–1 ––IPS ––> WING–1
c. BIRD–1 ––APS ––> NEST–1
d. BIRD–1 ––EFC ––> FLY–1
e. BIRD–1 ––EFC ––> LAY_EGGS–1
f. WING–1 ––INS ––> FLY–1
g. FLY–1 ––TYP2 ––> LOCOMOTION–1
h. AIR–1 ––LOC2 ––> FLY–1
295

296
The link from BIRD–1 to FLY–1 has the highest strength of association [4] by default, indicating that
the prototypical bird is an agent (EFC) of FLY–1. However, there are some birds that can’t fly. We
might encode that information in the following special link:
(110) PENGUIN–1 ––EFC[0] ––> FLY–1
Thus, although most birds (robins, sparrows, wrens, etc.) will inherit the link indicating that they are
capable of flight, penguins will not because they have a specific link in the network which indicates
that they cannot fly. This method is intended to capture that fact that it is quite exceptional for a bird
not to be able to fly.
4.19 POSTSCRIPT
We began this chapter by noting that semantic analysis presents linguists with some of the thorniest
problems about language they are likely to encounter. The problem of meaning is immense because,
among other things, it is difficult to know where grammar ends and knowledge of the word begins.
Consider, for example, what is involved in the interpretation of the following sentences which are
fairly typical of the kinds of sentences people come across on a regular basis:
(111) a. Wouldn’t you really love to Buick home tonight? (from a TV commercial)
b. They need to adore me, so Christian–Dior me. (from the musical Evita)
c. Do–it–yourself–ers generally have to call in an expert.
d. Last night’s come–from–behind win by Michigan was an upset.
e. “Could the most anticipated New Year’s Eve party in our lifetimes really usher in a
digital nightmare when our wired–up–the–wazoo civilization grinds to a halt.”
(Newsweek, 6/2/97, p. 54).
f. The future isn’t what it used to be. (from the movie “Deterence”)
g. Those who give up liberty for security will lose both and deserve neither. (Benjamin
Franklin)
h. It is better to keep your mouth shut and only appear stupid than to open it and remove
all doubt. (Mark Twain)
i. Thomas Demery, an ex-Birmingham developer and former top federal housing
official, asked executives doing business with his agency to donate to an overseas
charity whose founders solicited supporters of African guerillas who have slaughtered
unarmed civilians, records and interviews show. (Detroit Free Press, 9/5/89)
While the semantic networks described above might serve as a start in representing knowledge, it is
clear from data like the above that much work remains to be done to account for how people
understand ordinary language in ordinary situations.

EXERCISES FOR CHAPTER FOUR
In many ways, linguistics is grammatical detective work. It is knowing what questions to ask, which
results to consider and which to dismiss, and how to arrive at the most generalized overall
description. Remember that the underlying principles of language are logical, not arbitrary. But, as
we have seen again and again, they are also hidden from obvious conscious notice.
1. One of the best clues to determining underlying thematic relations is the prepositions that show
up in paraphrases. For example, consider what the case of the safe is in (i).
(i) The safe contains the jewels.
Determining underlying thematic relation (like determining part of speech) often benefits from
considerations of meaning. However, there are also many pitfalls to relying on meaning alone.
While it is useful to ask oneself what role the phrase the safe plays in (i), this is not the best
way to begin. Of potentially greater reliability are paraphrases. Consider (ii).
(ii) The jewels are contained in the safe.
Since the preposition in shows up in the paraphrase, a good guess is that the N3 the safe in (i)
is LOCATIVE. Of course, we could have a genuine exception here, a purely idiosyncratic and
unpredictable usage. Contrary to popular belief, this is NOT usually the case.
Sometimes paraphrases with a particular verb are not possible. For example, there are no
paraphrases with the verb fill in (iii) that realize the bucket as a prepositional phrase.
(iii) He filled the bucket with water.
On the other hand, we do have verbs like fill that involve the transference of something into
something else. Consider (iv).
(iv) He poured the water into the bucket.
Given (iv), a good guess for the case of the bucket in (iii) is ILLATIVE. This guess receives
support from inject which allows the complement structure of both (iii) and (iv), as we see in
(v).
(v) He injected the patient with it.
He injected it into the patient.
297

298
Using this method, determine the underlying thematic relations for all the underlined phrases
(watch for ambiguities).
a. Bill resembles Bob.
b. Mary knows the answer.
c. John acts weird.
d. Sue told Bill a lie.
e. Mary met a bachelor.
f. Kathy rode the horse.
2. Another clue to determining underlying cases, is the form questioning and pronominalization
take. Positional relations ordinarily are elicited by questions words like, where or when; they
are replaceable by forms like there and then. Using these "tests" along with paraphrases,
determine the underlying relations in the following phrases introduced by on:
a. He put on the hat.
b. He sat on the horse.
c. He got on the plane.
d. He insisted on the gin.
e. They disagreed on the budget.
f. They wrestled on the mat.
3. Using the above clues, determine the uses of to in the following:
a. He objected to her.
b. He submitted to her.
c. He wrote to her.
d. He appeals to her.
e. He went to her.

4. What other tests than those in Exercise One and Two can you use to help determine underlying
thematic relations?
5. Identify the thematic relations for each noun phrase in the following sentences (note
ambiguities):
a. The burglar inserted the key into the lock.
b. The librarian shelved the books without her assistant.
c. The teacher read the story to the students with enthusiasm.
d. The cook from Paris cooked the meal.
e. The cook from Paris broiled the chicken.
f. The campus police are patrolling the corridors of buildings after dark in search of
students.
g. The soccer game takes place at Troy High on Saturday.
h. The president elect is considered an honest man by many voters.
i. The clothes are lying on the bed in disarray.
j. Who put the glue in the toothpaste?
6. In English, thematic relations functioning in subject or direct object positions lose their
prepositions. Compare the following:
a. The gardener sprinkled the flowers with water.
b. The gardener sprinkled the water onto the flowers.
c. The flowers were sprinkled with water by the gardener.
d. The water was sprinkled onto the flowers by the gardener.
In (a) and (b), the subject is EFC; in (c), the subject is AFC; in (d), the subject is ASC. Find
as many different examples of thematic relations serving as subject as you can. Then, do the
same for direct object.
299

300
7. How would CASE GRAMMAR describe the syntax of the following sentences?
a. The boy chopped down the tree with an axe.
b. John traveled from Greece to Rome.
c. Sally set the table for her mother.
d. In July, the harvest was ruined by rain.
e. Bill hates zucchini with a passion.
8. Say as much as you can, in as explicit a form as you can, about the syntax and semantics of
following ambiguous sentences.
a. The robot moved the patient with the nurse.
b. The nurse moved the patient with a broken arm.
9. Identify the thematic relations introduced by the prepositions with and without in the following
(as always, note ambiguities):
a. He made it WITH HER.
b. He made it WITH LOVE.
c. He made it WITHOUT TOOLS.
d. I met several people WITHOUT CHILDREN.
e. A person WITH LONG SOBRIETY is reliable.
f. He went WITH THE CHILDREN.
g. The sidewalk is level WITH THE STREET.
h. He did it WITH GUSTO.
i. He punched her WITH HIS FIST.
j. He is WITHOUT THE NECESSARY FUNDS.

10. The following uses of the ABLATIVE case in Latin have been identified by the labels usually
given by classical grammarians.
a. THE ABLATIVE OF QUALITY:
Vir summo ingenio (ABL) est.
a man of the greatest genius he is
He is a man of the greatest genius.
b. THE ABLATIVE OF ATTENDANT CIRCUMSTANCES:
Nudo corpore (ABL) pugnant.
unprotected body they fight
They fight with their bodies unprotected.
c. THE ABLATIVE OF ACCOMPANIMENT:
Omnibus copiis (ABL) subsequebatur.
with all forces he followed
He followed with all his forces.
d. THE ABLATIVE OF COMPARISON:
Cato est Cicerone (ABL) eloquentior.
Cato is than Cicero more eloquent
Cato is more eloquent than Cicero.
e. THE ABLATIVE OF MANNER:
Mirabili celeritate (ABL) venerunt.
wonderful speed they came
They came with wonderful speed.
f. THE ABLATIVE OF MEANS:
Pugnis (ABL) certant.
with fists they fight
They fight with their fists.
Reclassify each of the above ablatives in our terminology, that is, find the thematic relations
they express. Is there a way to generalize the Latin ablative using the RG feature matrix? If
so, what is the generalization?
301

302
APPENDIX A: SUMMARY OF THEMATIC RELATIONS
1. CASE GRAMMAR (THEMATIC RELATIONS ARE ATOMIC CATEGORIES):
a. AGENTIVE: the case of the typically animate perceived instigator of the action
identified in the verb (often simply called AGENT).
(1) JOHN opened the door.
(2) The door was opened BY JOHN.
(3) THE DEAN sent a note.
(4) A note was sent BY THE DEAN.
(5) There was a note FROM THE DEAN.
b. SOURCE: the case which identifies the origin of the state or action identified in the
verb.
(1) John flew FROM ITALY.
(2) John ran OUT OF THE HOUSE.
c. DATIVE: the case of the usually animate entity affected by the state or action identified
in the verb (often called EXPERIENCER or RECIPIENT).
(1) John gave the book TO BILL.
(2) BILL received the book.
(3) The movie appeals to HIM.
(4) THE SNOW melted.
b. GOAL: the case which identifies the direction of the state or action identified in the
verb.
(1) John flew TO ITALY.
(2) John ran INTO THE HOUSE.

2. RG SEMANTICS:
a. [+POSITIONAL] [–POSITIONAL]
He put it into the vase. He turned it into a vase.
He took it out of a vase. He made it out of a vase.
b. [+POSITIONAL] ([+PST]): having the primary focus on location, orientation, or
movement in space or time.
c. [–POSITIONAL] ([–PST]): not having the primary focus on location, orientation, or
movement in space or time.
d. [+DISJUNCTURAL] ([+DSJ]): emphasizing separation, dissociation, detachment,
withdrawal and the like.
e. [–DISJUNCTURAL] ([–DSJ]): not emphasizing dissociative ideas.
f. [+CONJUNCTURAL] ([+CNJ]): emphasizing union, association, attachment, advance
and the like.
g. [–CONJUNCTURAL] ([–CNJ]): not emphasizing associative ideas.
3. RG SEMANTICS (THEMATIC RELATIONS ARE BUNDLES OF FEATURES):
a. SOURCE: [+PST, +DSJ, –CNJ] from
b. AGENT: [–PST, +DSJ, –CNJ] from, by
d. GOAL: [+PST, –DSJ, +CNJ] to
e. EXPERIENCER: [–PST, –DSJ, +CNJ] to
303

304

CHAPTER FIVE: LANGUAGE, PEOPLE, AND COMPUTERS
5.1 SYNTAX AND COGNITION
As we have seen, each semantic category (thematic relation) and syntactic category (part of speech)
in RG is distinguished from and related to all other such categories by a system of features, so that,
in fact, each category is no more than an abbreviation for a bundle of these features in the way that
a sound such as [p] is an abbreviation in phonology for a bundle of phonological features like
[+CONSONANTAL], [–VOCALIC], [–VOICED], etc. The figures in the preceding chapters
expresses the interface between the categories of RG and some of the more traditional categories.
As we have seen, one of the distinguishing aspects of RG is that its syntactic feature matrix
recognizes two supercategories not found in other theories of syntax. These are the category
ADJUNCT ([–VBL]3 or, more simply, A3), which subsumes all of the traditional categories except
verbs, and the category CHARACTERIZER ([–VBL, –NML]3 or, more simply, C3), which
subsumes all of the remaining traditional categories except nouns. The breakdown is diagramed in
(1) (see (170) in Chapter Three, Page 171).
(1)
[+VBL] includes VERB, AUXILIARY, MODAL, GERUND, INFINITIVE
[–VBL] includes NOUN, PRONOUN, QUANTIFIER, ADJECTIVE, ADVERB,
PREPOSITION, SUBORDINATOR, COORDINATOR,
COMPLEMENTIZER, DETERMINER, DEGREE WORD
[+NML] includes NOUN, PRONOUN, QUANTIFIER, GERUNDS
[–NML] includes VERB, AUXILIARY, MODAL, ADJECTIVE, ADVERB,
PREPOSITION, SUBORDINATOR, COORDINATOR,
COMPLEMENTIZER, DETERMINER, DEGREE WORD
Given this syntactic framework together with the phonological features described in Chapter Two
and the semantic features described in Chapter Four, we have the following breakdown for the major
categories of language:

306
(2) PHONOLOGY (see Figure Five on Page 62):
a. [αCON, –αVOC] — consonants and vowels
(1) [+CON, –VOC] — consonants ([p], [t], [k], etc.)
(2) [–CON, +VOC] — vowels ([a], [e], [i], [o], [u], etc.)
b. [αCON, αVOC] — liquids and glides
(1) [–CON, –VOC] — glides ([w], [y], etc.)
(2) [+CON, +VOC] — liquids ([r], [l], etc.)
(3) SEMANTICS (see Figure Nine on Page 277 and Figure Ten on Page 280):
a. [αDSJ, –αCNJ] — expressions of motion and causation
(1) [+DSJ, –CNJ] — source (extract, exit, take, buy, from, out of, etc.)
(2) [–DSJ, +CNJ] — goal (insert, enter, give, sell, into, to, etc.)
b. [αDSJ, αCNJ] — expressions of location and possession
(1) [–DSJ, –CNJ] — location/possession (live, contain, have, own, with, etc.)
(2) [+DSJ, +CNJ] — nonlocation/nonpossession (lack, differ, without, etc.)
(4) SYNTAX (see Figure Eight on Page 183):
a. [αVBL, –αNML] — verbs and nouns
(1) [+VBL, –NML] — verbs (destroy, predict, refuse, etc.)
(2) [–VBL, +NML] — nouns (destruction, predication, refusal, etc.)
b. [αVBL, αNML] — characterizers and gerunds
(1) [–VBL, –NML] — adjectives (sad), adverbs (sadly), prepositions (to), etc.
(2) [+VBL, +NML] — gerunds (swimming)
The central concept of RG syntax is the concept RESIDENCE which is a specific syntactic or
morphological position that is associated with a specific semantic relation. For example, the category
label DETERMINER or DET simply specifies a group of words that occupy a [+PRH, +X3L, +ENH]
residence, that is, that occur in prehead position ([+PRH]) on the X3 level ([+X3L]) of a category
whose head is a noun ([+ENH]). Every syntactic category (part of speech) in RG is defined in this
way, as a constellation of syntactic features that refer only to positions in phrase structure.
The full range of syntactic features in RG provide a noun phrase like all those many young
philosopher kings of Greece from Rome with the diagram in (5).

(5) all those many young philosopher kings of Greece from Rome
The phrasal architecture of diagrams like (5) is used in RG to specify the relationship between related
structures: expressions of quantity (all), definiteness (those), and the like are always residents of the
X3 level, that is, they are daughters of X3; all descriptive modifiers of the head, e.g., adjectives
(young), restrictive relative clauses, and participial and prepositional phrases (from Rome), are
residents of the X2 level; in X1 level prehead position reside the elements of compounds (philosopher
in philosopher king); and, in X1 level posthead position reside all complements of the head X
(Greece in king of Greece).
As a result of such specifications, RG directly incorporates into the phrasal architecture many
relationships which other grammars handle transformationally. The potential difficulty of
constructing a revealing set of phrase structure rules is solved with the use of the syntactic features
given in Figure Eight (Page 183). All possible phrase structure sequences in English are reducible
to one phrase structure filter (PSF), namely, (6).
307

308
(6) [X n ([+NML]) ([–NML]) X m ([+NML]) ([–NML]) ]
where m # n
(6) can be interpreted as a NODE ADMISSIBILITY CONDITION, that is, a condition on what
nodes can consist of. It asserts that each X level must dominate a lower X level with an equal or
lesser prime value, e. g., X2 over X2, X1, or X but not X2 over X3. Further, (6) asserts that each X
level may have at most one [+NML] and one [–NML] category, in that order, to the right and/or to
the left of the head X.
The schema (6) means that syntactic categories in RG all refer to the basic phrasal structure (7).
(7)
From each X level (m # n), both to the left and to the right of the head, can occur at most two
elements. The first of these two is specified by the feature [+NML]; the second, by the feature
[–NML]. In short, the feature opposition [±NUM] refers to specific linear positions in a hierarchy
of positions, and nothing more. Thus, [+NML] is something that occurs in position α of (8); [–NML]
is something that occurs in position β.
(8)
The feature [+NML] is the major defining feature of the English syntactic category commonly
referred to as Noun. All RG syntactic categories are similarly defined. To take another example, the
feature opposition [±ENH] (in the environment of noun head) refers to either α or β in (9).
(9)
Notice that RG syntax claims that there can be at most five units on any one level. Although RG
phrase structure was worked out solely on the basic of syntactic evidence, that is, autonomously, this

number is, interestingly, the lower end of human short–term memory capacity. The claim in RG is
that the severe restrictions on human short term memory demand that the syntax of human languages
be organized into hierarchical units, and that the definitions of syntactic categories make reference
to those hierarchical units. All twelve of the RG syntactic features are so defined.
In RG syntax, there are, of course, dependencies. Consider for example the diagram (10).
(10)
If X is [+NML], then β is a characterizer, typically, a determiner. If X is [–NML] and δ is [+NML],
then X is a transitive verb and δ is its direct object. These specifications are determined by the RG
syntactic feature matrix (Figure Eight, Page 183). From such specificity, one derives the concept of
residence. For example, one says that a determiner is nothing more than a class of words that resides
in X3 Level Prehead position of a [+NML] category.
In recent years, there has been much research on the problem of writing a computer program which
can understand natural language. Such a program is called a PARSER. In fact, in the summer of
1989, there was a renewed flurry of interest in developing programs for translating texts from one
language into another. The Japanese, in particular, have currently invested considerable resources
in the field of machine translation.
An RG computer parser is based on the concept of residence. There is, therefore, a direct connection
between a theory of syntax, human syntactic ability, and the algorithms necessary to make a parser
work. This connection is revealed in the number of distinct advantages there are to developing a
parser based on a feature–based analysis of language, as opposed to an analysis based on atomic
syntactic and semantic categories.
First, when a parser based on TG or on traditional grammar looks up a word, say, the, in the
dictionary, it usually returns a part–of–speech specification such as [+DETERMINER]. Similarly,
when it looks up available, it returns a specification such as [+ADJECTIVE]. Such categorial
specifications provide very limited help in facilitating further parsing, particularly in determining
what the immediate and upcoming structural possibilities are. The reason for this is that features like
[+DETERMINER] and [+ADJECTIVE] have never been adequately defined in syntactic terms in
either TG or traditional grammar; hence, computerized parsing programs have been hampered by the
lack of an available syntactic feature framework on which to base parsing strategies.
309

310
On the other hand, a parser based on RG receives very useful information in a dictionary lookup.
Determiners, as we have seen from Figure Eight (Page 183), are [+PRH, –PSH, +X3L, +ENH,...],
which means that they occur only in prehead position on the X3 Level of noun phrases; further,
adjectives are generally [+PRH, –PSH, +X2L, +ENH,...], which means that they reside in posthead
position on the X2 Level of noun phrases. An adjective like available is exceptional in being marked
[+PSH]. This information exactly specifies the syntactic structure that should be assigned to a
sequence of words that begins, say, with the available. As a result of such precise syntactic
specifications, a structural analysis of successive words in a sentence can be built and checked with
great speed and accuracy by an RG parser. Therefore, the RG feature system has provided a new
paradigm on which to base a parsing strategy.
A second advantage of a feature system is that so–called cross categorial relationships can be very
succinctly specified. Consider, for example, the word that. Traditional and transformational
grammars analyze this word as either one of the following four categories: determiner, as in That
man is famous; degree word, as in He is that famous; complementizer, as in I know that he is famous;
and, relative pronoun, as in The man that John saw is famous. In RG, the word that receives the
feature specifications in (11).
(11)
VBL NML OPH OCL PRH PSH ENH EVH ECH X1L X2L X3L
REL – – – – + – – + – – – +
CPL – – – – + – – + – – – +
DET – – – – + – + – – – – +
DGR – – – – + – + – – – – +
As (11) indicates, ten of the twelve RG features are the same for all uses of the word that. Further,
the determiner and degree word have identical feature matrices, and so do the relative pronoun and
complementizer. Through this system we can see how one word can come to have such “diverse”
uses. Actually, the RG description of English does not include a category REL at all; its inclusion
in (11) is for comparative purposes only. Similarly, the RG description of English contains no
category demonstrative pronoun, common in traditional descriptions. A sentence such as Look at
that is really look at that [u], where [u] means ‘understood’ thing. The feature specifications in (11)
are represented diagrammatically in (12), where the N3 [much] is an abstract quantifier (before
nouns the word much is overt as in that much bread and that much money; before characterizers, it
is abstract as in that famous and that happily).
As (12) reveals, all instances of that reside in X3 level prehead position. A parser based on this
analysis will naturally function in a highly efficient manner. When the word that is encountered in
a sentence, the parser can assign it ten of the twelve possible RG features without consideration of
context. Conversely, a parser based on the categories of traditional or transformational grammar is
unable to assign any analysis at all to the word that without considering the context in which it is
used.

(12) I know that (CPL) that (DET) man that (REL) John saw seems that (DGR) famous.
A third advantage of the RG feature system is that it is completely nontransformational and allows
for the direct expression of alternative constructions such as those discussed above, e.g., available
books versus books available. RG syntactic constraints are stated with reference to surface structure,
so that parsing is further facilitated by the complete elimination of the concept of syntactic derivation
through transformation. The RG paradigm directly generates and interprets all sentences without
reference to deep structures such as those found in every version of TG.
The direct application of RG to computer parsing has had a salutary effect on the problem of
adequately characterizing the grammar of English. In many cases, parsing difficulties have lead to
a better understanding of syntax. Constraints on language structure are directly related to parsibility,
311

312
the ability of people or machines to understand language. For example, constraints on the internal
structure of English noun phrases appear to be directly related to the problem of locating the head of
the noun phrase. To see this, consider again the word that in the following:
(13) a. A teacher of that language is hard to find.
b. *A that language teacher is hard to find.
(14) a. A teacher that clever is hard to find.
b. *A that clever teacher is hard to find.
(15) a. How clever (of) a teacher is she?
b. That clever (of) a teacher is hard to find.
c. *Very clever (of) a teacher is hard to find.
These are not peculiar or disconnected data, and the explanation for the constructions is clear with
an RG model. A prehead modifier in an N3 cannot have its own determiner unless it is to the left of
the word of (we will see why this is so shortly). Since determiners and degree words share identical
sets of RG features (see (11)), the source of the ungrammaticality of (13b) with DETERMINER–that
and (14b) with DEGREE–that is the same. They violate this general phrase structure restriction on
prehead modifiers in N3.
On the other hand, we do have (15a) and (15b), which contain adjective phrases with their own
degree modifiers to the left of of. Actually, these adjective phrases are quantificational: how clever
is, abstractly, how [much] clever; that clever is that [much] clever. The occurrence of such a phrase
in such a position in other theories of English syntax is an oddity. This is not the case in an RG. The
X3 level prehead residence of all categories is the home of expressions of quantification. This
explains the presence of of in (15a) and (15b), and it rules out (15c) because very clever lacks a
quantifier.
The same logic applied in (13) and (14) explains the loss of the complementizer in (16a) as compared
to its presence in (17b):
(16) CPL obligatorily absent:
a. I am an I’d–rather–do–it–myself person.
I am [ N3 [ C3 an ] [ N1 [ V3 I’d rather–do–it–myself ] [ N0 person] ] ]
b. *I am a that–I’d–rather–do–it–myself person.
(17) CPL obligatorily present:
a. *I’d rather do it myself ought to be clear.
b. That I’d rather do it myself ought to be clear.

Given the above data, the RG analysis seems justifiable: all uses of that are essentially the same
word.
The above generalizations are important because they link the study of syntax directly to the kinds
of routines a computer parser contains, as well as to human psychology and biology. The claim of
RG is that all N3 with non–quantifier heads, that is heads which are [+NML, +OCL], must have a
determiner (the, those, Ø, etc.) as a PERMANENT RESIDENT (one not omissible) in X3 level
prehead position. This claim can be supported independently and autonomously by examining
English syntax. In turn, this claim from syntactic investigation can be related to the nature of the
human language apparatus. The determiner in an English N3 acts like a cuing device with the
message here comes a noun phrase. Between this cuing device and the head, there can be no other
determiners; hence, we have neither (13b) nor (14b). This syntactic constraint makes it possible to
write an algorithm that unambiguously finds the head of N3, a fact discovered during the writing of
the first RG parser. In short, the description of the English N3 in RG has a direct relationship with
computer parsing and with sentence comprehension.
It is interesting to note that the determiner of N3 parallels the wh–phrase of V3 interrogatives: (i)
both are prehead X3 level adjuncts ([+VBL, +PRH, +X3L]); (ii) both are permanent residents; (iii)
repetitions of both are impossible, because repetitions produce self–embedded constructions like (18)
and (19); both act like cuing devices.
(18) *[ N3 that i [ N3 that j [ language j ] ] teacher i ]
(19) *[ V3 where i [ V3 what j [ did he put [e] j [e] i ] ] ]
The RG approach to grammar has uncovered some provocative connections between human language
and human biology. As we have seen, the generality of the RG syntactic feature matrix makes
possible the reduction of all phrases and clauses to variations on a single schema (3). In itself, this
is a significant result. But the schema does not exist in theoretical isolation. We have seen that it
allows at most five units per phrase, which is the lower end of human short term memory capacity.
From this relationship between grammatical and psychological unit, one can formulate hypotheses
about the relationship between grammatical constructs and the nature of man, for example, the
hypothesis that syntactic organizing principles like verb agreement derive from constraints on human
memory and human perceptual abilities. Consider, for example, the case of so–called free word order
languages like Classical Latin, which has the syntactic rules in (20) and the rules of concord in (21):
(20) a. The subject is in the nominative case.
b. The direct object is in the accusative case.
(21) a. A finite verb and its subject must agree in person and number.
b. A noun and adjective must agree in case, number, and gender.
313

314
Given (20) and (21), the sentences in (22) could, theoretically, be uttered with the individual words
in any order (NS = nominative singular, NP = nominative plural, AS = accusative singular, AP =
accusative plural, 3S = third person singular, 3P = third person plural).
(22) a. Custos fidelis consulem veterem ducit.
NS NS AS AS 3S
guard trusty consul old is leading
The trusty guard is leading the old consul.
b. Custodes fideles consules veteres ducunt.
NP/AP NP/AP NP/AP NP/AP 3P
The trusty guards are leading the old consuls.
The old guards are leading the trusty consuls.
The trusty consuls are leading the old guards.
The old consuls are leading the trusty guards.
Notice that the rules of syntax (20) and concord (21) completely disambiguate (22a). However, (22b)
is presumably four–ways ambiguous, because the es–ending is either nominative or accusative plural.
Needless to say, one rarely comes upon sentences like (22b) in the literature, though ambiguities are
found as in any language. The point here is that so–called free word order languages like Classical
Latin must have some devices like (20) and (21) to hold together the integrity of the clausal unit.
Latin, like all other languages, is constrained because of human neurological limitations: humans are
not clairvoyant (usually), and they operate with severe constraints on short term memory. Quite
naturally, languages have invented a number of devices to handle these limitations. Concord is a
means of binding phrases together when word order constraints are relaxed. In such cases, all the
elements of a phrase need not be under one node. Since English is relatively poor in inflectional
endings, it has very severe restrictions on word order. Therefore, most constructions in English are
continuous, that is, all the elements in the construction occur successively. But there are a few
constructions that are discontinuous, such as the b examples in the following pairs:
(23) a. The wind blew open the barn door.
b. The wind blew the barn door open.
(24) a. John called up a former girl friend.
b. John called a former girl friend up.
Such constructions are under severe limitations of short term memory. If the direct object is too
internally complex, the sentences become unacceptable:

(25) a. John called a former girl friend from Texas up.
b. ?John called a former girl friend he used to date when he lived in Texas up.
c. *John called a former girl friend he used to date when he was working as a rattle
snake defanger in Texas up.
Much more is known today about the nature of the human language apparatus than was known in
1957, the year of Chomsky’s Syntactic Structures and Skinner’s Verbal Behavior. For example, a
number of psychological processes can be directly connected to human linguistic abilities; in
particular, man’s ability to deal with unstructured expressions is very limited. The structuring
processes important in language include linear ordering, hierarchical ordering, and cuing. Thus,
many studies indicate quite clearly that the justification of phrase structures must assume a very high
priority in grammatical description. In the grammatical model we have described, this is exactly the
case: the model attempts to draw a connection between human syntactic ability and human memory
capacity.
The objective of RG research is to provide a biological basis for grammatical description. In
particular, the model draws a connection between human syntactic ability and human memory
capacity. Let us now examine this more carefully focusing on the description and interpretation of
English noun phrases (N3) and sentences (V3).
As we have seen, all nouns and verbs in English are specified by the following features:
(26) CLASS FEATURES EXAMPLES
Common Verbs: [+VBL, –NML, +OCL] go, eat
Auxiliary Verbs: [+VBL, –NML, –OCL] may, can
Common Nouns: [–VBL, +NML, +OCL] cake, toy
Quantifiers and Pronouns: [–VBL, +NML, –OCL] some, she
Notice that verbs and nouns fall into two groups distinguished by the feature opposition [+OCL]
meaning ‘open class word.’ Such common verbs and nouns must have an X3 level prehead
characterizer, which is called a BINDING RESIDENT (BR). The BR itself is really an abbreviation
for the cluster [–VBL, –NML, –OCL, –ECH, +X3L, –X2L, –X1L, +PRH]. This includes the
categories COMPLEMENTIZER, DETERMINER, CASE (POSSESSIVE), and TENSE. The
domain of a BR, which includes all nodes it C–commands, is called a NEIGHBORHOOD (see
Section 3.8, Page 195 ff.). In common nouns, the BR is a determiner or POS (the possessive case);
in main verbs, it is tense (PRESENT, PAST, FUTURE, CONDITIONAL, IMPERATIVE). We have
representations like the following:
315

316
(27) N3:
(28) V3:
a. that vote:
b. John’s vote (POS = POSSESSIVE):
c. all the votes:
a. Vote! (IMP = IMPERATIVE):
b. John voted (PST = PAST TENSE):

In V3, complementizers are also a BR. In most indirect statements in English, a complementizer BR
is optional; it is a FREE RESIDENT:
(29) John thinks that Sue went.
John thinks Sue went.
In indirect questions in English, a complementizer BR is obligatory; it is a PERMANENT
RESIDENT:
(30) John wonders if Sue went.
*John wonders Sue went.
317

318
Notice that each BR, whether a free or a permanent resident, occupies a fixed position in underlying
(abstract) syntax when it does occur. If a BR, in addition, occupies a fixed position in superficial
syntax, then it acts as a BARRIER to reference, that is, it does not allow categories on either “side”
of it to refer to each other.
English contains a number of different categories called ANAPHORS which refer to other phrases.
For example, in John takes care of himself, the phrase himself refers to the phrase John. There are
conditions on such reference. For example, in John told Bill to take care of himself, himself can only
refer to Bill, not to John. The element [e], which we discussed in Chapter Three, is also an anaphor.
As we have seen, it must be bound to a referent (see Section 3.8, Page 195 ff.). Sometimes this
reference is blocked by a barrier BR. Thus, one can ask a question like What did John say that Sue
bought, when the sentence has the free resident that. However, one cannot ask a question like *What
did John wonder if Sue bought, when the sentence contains the permanent resident if.
Those BR that are barriers include DET, POS and CPL. The position of each of these categories in
both underlying and superficial syntax is fixed (when they occur). Also, each of these is a
HEAD–ONLY CATEGORY, that is, a category that tolerates no prehead or posthead adjuncts on
any of its own internal levels. Their structure is simply [C3 C0].
The one BR that is never a barrier to reference is TNS. Its position in underlying syntax is fixed
([–ENH, +EVH, –ECH, +X3L, –X2L, –X1L, +PRH, –PSH]), but its realization in superficial syntax
is variable: sometimes it appears on the modal, sometimes on the main verb, sometimes on the
perfective have, etc. Also, TNS can have a rather complex internal structure of its own that might
include modals, perfective, negatives, emphatics, etc. Apparently, for a BR to be a barrier, its
position in both underlying and superficial syntax must be fixed and simple. We will explore the
reason for this below.
A barrier BR establishes one of two types of RESIDENTIAL DOMAINS, which we can specify
explicitly by extending the C–command relation (see Section 3.8, Page 195 ff.) as follows:
(31) a. C–COMMAND: a category α C–commands another category β if the first branching
category Z above α dominates β.
b. L–COMMAND: a category α L–commands another category β if α C–commands
β and if β is an anaphor to the left of the head of Z.
c. R–COMMAND: a category α R–commands another category β if α C–commands β
and if β is an anaphor to the right of the head of Z.
Notice that these two new command relations simply extend the original definition and make
reference to the left–to–right (linear) position of constituents. This extension is crucial to describe
the restrictions on anaphoric processes in English, that is, on the rules which determine what
anaphors can refer to in sentences. Given the new relations, we can make the following
generalizations:

(32) a. If a barrier is a free resident (optional), its residential domain includes all items it
L–commands.
b. If a barrier is a permanent resident (obligatory), its residential domain includes all
items it either L–commands or R–commands.
c. A binding relationship cannot be formed with an empty category that falls within a
residential domain.
In each of the following, observe, in particular, that questioning an item commanded by the barrier
BR is impossible (the b example is ungrammatical), and that the barrier BR is obligatory (the c
example is ungrammatical):
(33) [Bill’s vote against gun control]; common noun head vote:
a. Mary criticized Bill’s vote against gun control.
b. *What i did Mary criticize Bill’s vote against [N3 e ] I?
c. *Mary criticized vote against gun control.
d. Mary criticized the vote against gun control.
(34) Indirect question:
a. Mary wondered if Bill voted against gun control.
b. *What i did Mary wonder if Bill voted against [N3 e ] I?
c. *Mary wondered Bill voted against gun control.
d. *Who i did Mary wonder if [N3 e ] I voted against gun control?
e. *Who i did Mary wonder [N3 e ] I voted against gun control?
(35) Indirect Statement with obligatory barrier BR:
a. Mary chortled that Bill voted against gun control.
b. *What i did Mary chortle that Bill voted against [N3 e ] I?
c. *Mary chortled Bill voted against gun control.
d. *Who i did Mary chortle that [N3 e ] I voted against gun control?
e. *Who i did Mary chortle [N3 e ] I voted against gun control?
In contrast to (33) – (35), observe that the data fall out differently when the barrier BR is a free
resident; it is only when the barrier BR is present and L–commands the empty category that
ungrammaticality results:
319

320
(36) Indirect Statement with optional barrier BR:
a. Mary thought that Bill voted against gun control.
b. What i did Mary think that Bill voted against [N3 e ] I?
c. Mary thought Bill voted against gun control.
d. *Who i did Mary think that [N3 e ] I voted against gun control?
e. Who i did Mary think [N3 e ] I voted against gun control?
Sometimes it appears as though there is no BR in an N3 with a common noun head, specifically,
when the head is a plural count noun (votes) or a mass noun (water). However, in such cases, the null
determiner Ø is present:
(37)
There are a number of reasons for positing this particular empty category. First, it allows us to
distinguish between nouns that ordinarily can have determiners (all common nouns) and those that
ordinarily can’t:
(38) COMMON, COUNT NOUN:
a. He wants those votes
b. He wants Ø votes
c. He wants a vote
d. *He wants vote
(39) COMMON, MASS NOUN:
a. He wants that freedom.
b. He wants Ø freedom.
c. *He wants freedom.
(40) NONCOMMON NOUNS (PROPER NOUNS, QUANTIFIERS, PRONOUNS):
a. *He wants the John Smith.
(but: Will the John Smith from Troy please stand?)
b. *He wants the some.
c. *He wants the them.

Given the Ø–analysis, we say that an N3 with a common noun head has an obligatory BR, whereas
one with a noncommon noun head does not.
A second reason for the Ø–analysis is that, with it, the following data are parallel to (33) – (36):
(41) [ Ø votes against gun control]; common noun head vote:
a. Mary criticized Ø votes against gun control.
b. *What i did Mary criticize Ø votes against [N3 e ] I?
c. *Mary criticized votes against gun control.
d. Mary criticized that vote against gun control.
Third, if we adopt an analysis with Ø, we can generalize the insertion of of in all the following:
(42) a. A number of the systems are complicated.
b. A number of those systems are complicated.
c. A number of Ø systems are complicated.
Given this description, we can now explore what the function of a barrier BR is. In both V3 and N3,
the barrier BR, as the obligatory element, signals the beginning of the phrase. Therefore, we might
suppose that, for parsing purposes, it acts like a cuing device. We see this most clearly in subject
complements where the complementizer can never be left out:
(43) a. That Bill voted against gun control is odd.
b. * Bill voted against gun control is odd.
c. *What i that Bill voted against [N3 e ] I is odd.
Leaving out the complementizer in (43b) results in a miscue: it appears as though the clause Bill
voted against gun control is the main clause of the sentence.
In N3, the major parsing problem is locating the head. Consider the following observing the number
on the verb:
(44) a. The school shows play here. (Subject is shows; the verb play is plural.)
b. The school show plays here. (Subject is show; the verb plays is singular).
c. The school shows plays here. (Subject is school; the verb shows is singular.)
d. The schools show plays here. (Subject is schools; the verb show is plural.)
As these examples attest, finding the head of an N3 is not always straightforward, particularly when
a compound noun interpretation like school show(s) or student teacher(s) is possible. But the
language does provide a number of constraints that facilitate the correct location of the head.
321

322
First, the only noun in a compound noun phrase that can have its own BR is the head:
(45) a. a teacher of Sanskrit
a Sanskrit teacher
b. a teacher of that language
*a that language teacher
a language teacher
c. Ø teachers of that language
*Ø that language teachers
Ø language teachers
d. a teacher that clever (is hard to find)
*a that clever teacher (is hard to find)
a clever teacher (is hard to find)
Second, the only noun in a compound noun phrase that can be plural is the head:
(46) a. a teacher of Ø languages
*a Ø languages teacher
a language teacher
b. a wall of four feet
*a four feet wall
a four foot wall
*a wall of four foot
Third, any item occurring to the left of the BR must be separated from the BR by the word of:
(47) a. A number of the systems are complicated.
b. A number of Ø systems are complicated.
c. *the number Ø systems ...
d. The number systems are complicated.
(here number systems is a compound noun)
Fourth, if no overt marker is present separating a head noun from its complement, e.g., against in a
vote against gun control, then of must be inserted to signal the right boundary of the head:
(48) a. The student letters are here.
The student of letters is here.
b. His love stories are charming.
His love of stories is charming.
Thus, if the head noun is a singular count noun, we know what the left boundary of the head is,
because in such cases the BR is overt (cf. (38c)). If the head is a plural count noun, the BR need not
be overt, that is, Ø is possible (cf. (38b)). But, in such a case, we know that the plural noun must be
the head because only the head can be plural in a compound noun sequence.

Summarizing, we have the following N3 constraints:
(49) The N3 Constraints:
323
a. All N3 with a common noun head must have a BR as a permanent resident. This BR
is either a determiner or the possessive case.
b. Items to the left of the BR are separated from the BR by of.
c. Items to the right of the head are separated from the head by of if no overt
characterizer is present.
d. Any noun occurring between the BR and the head cannot have any of its own X3
Level elements:
(1) Such nouns show no number or case distinctions of their own.
(2) Such nouns have no determiners of their own.
This conspiracy of restrictions works very well. There are some minor exceptions. The quantifier
all does not have to be followed by of:
(50) They counted all (of) the votes.
There are a few sporadic N3 where a plural (usually irregular) is not the head:
(51) a. I met the women delegates.
b. *I met the ladies delegates.
c. I met the lady delegates.
Despite these minor glitches, the N3 Constraint is well established for English. We may now ask
why a language should have such a constraint. Since the N3 constraint was discovered during the
search for an algorithm for locating the head of an N3, it has always seemed natural to conclude that
the reason for the constraint is just that, namely, to help locate the head of an N3. This conclusion
concurs with the RG hypothesis that syntax and memory are linked. Let us now investigate this
connection.
No matter how large and internally complex an N3 becomes, we have seen that there are never more
than five units on any level; that N3 always precede C3 on each level, both to the left of the head and
to the right of the head; that of separates quantifiers and determiners; that of separates the head from
following posthead modifiers. Thus, the essential structure of every N3 is (52) with area X
specifying the bounds of the N3 Constraints.

324
(52) N3
+))))0)))))))3)))),
...of.......DET...........N0........of...
.)))0))))-
8 8 X 8 8
anchor cue head anchor
The area marked “X” in (52) is the area in which three of the noun phrase constraints operate. Within
X, there can be only one determiner and only singular nouns. Since any violation of these constraints
makes it impossible to locate the head unambiguously, the suggestion is that the constraints are not
arbitrary; rather, they exist to help speakers locate the head of a noun phrase.
Notice that, with the constraints, ambiguity disappears. First, the word of acts like anchor points
delineating the boundaries of the head. Second, if the head is singular, the noun phrase must be cued
by an overt determiner; we don’t have sentences like *he is carrying book. Third, if the head is
plural, the determiner need not be overt (he is carrying books), but in such cases one knows that the
plural noun must be the head, because the only allowable plural between the DET and the head is the
head. In short, the constraints appear to result from cognitive limitations. Humans are not
clairvoyant and need structural principles to locate the heads of phrases in languages like English.
The N3 Constraints are such principles.
Given this analysis, we have an explanation for why the past tense has become regular in English for
all persons and number, but the present tense has not. Compare (53) and (54).
(53) Regular Past Tense; played for all persons and number: He/They played.
(54) Irregular Present Tense: He plays versus They play.
The basic problem is that singular and plural noun inflection is identical to verb inflection:
(55) a. play: either a singular noun or a plural present
b. plays: either a plural noun or a singular present
There are many nouns in English that also function as verbs. On the other hand, there are only a few
past tense verbs that are also nouns, and they are all irregular, e.g., cut and hit. Further, locating the
head of an English noun phrase is not a trivial matter. There are many opportunities for ambiguity.
All these potential ambiguities are eliminated by the rule of subject verb agreement. Quite simply,
without the inflectional differences between singular and plural count nouns and between singular
and plural present verbs, speakers would be unable to locate the head of a noun phrase. Thus, the
present tense of verbs cannot become regular
We have already noted in a number of places that there is substantial psycholinguistic evidence
supporting the analysis of language into chunks of constituent structure. As we have seen, RG syntax

educes all phrases and clauses to variations on one structural schema (6) repeated here for
convenience:
(56) [X n
([+NML]) ([–NML]) X m ([+NML]) ([–NML]) ] where m # n
(56) is, of course, a realization of (6). Thus, the structure of even gargantuan N3 (see Chapter Three,
Page 179) reduces to five chunks of constituents, the lower end of human short term memory
capacity.
We also suggested above that the BR acts like a cuing device signaling the obligatory left boundary
(beginning) of an N3. There has been considerable psycholinguistic interest in the role of cuing. In
particular, studies focusing on the word that ) what we are calling a BR ) indicate that sentence
comprehension is facilitated when the BR is overt.
There are two broad categories that can serve as a BR in English syntax: the determiners (DET and
CPL) and the inflections (POS and TNS). Of these, only TNS lacks a fixed position in both
underlying and superficial syntax, as we have seen. And, of these, only TNS does not serve as a
barrier. If, in fact, the BR is a cuing device, this seems appropriate. It is hardly likely that something
that serves as a cue to upcoming structures would itself have an unreliable realization.
To this, we may add various studies on anchors: experimental results indicate that subjects have
better access to structure at the beginning (front anchoring) and end (end anchoring) of strings of
elements. One might view the two occurrences of of in (52) as anchor points defining the obligatory
bounds of an N3. It is well–known that front–anchoring is very important in linear ordering and that
hierarchical structure is very important in the processing of long lists. These findings are in direct
accord with the model of phrase structure proposed in this book and provide a link between syntactic
description and the so–called special human faculty of language.
5.2 OVERVIEW OF ANAPHORA
Given the above description of English syntax, let us now turn to a more in–depth analysis of
reference to see how our model can account for grammatical relationships in a revealing way.
English has the following types of anaphors (items which refer to phrases in a sentence):
(57) a. Personal Pronouns (he, him, his; she, her, hers; they, them, their, theirs, etc.)
b. Reflexive Pronouns (himself, herself, themselves, etc.)
c. Reciprocal Pronouns (each other, etc.)
d. Floating Quantifiers (all, both, each, etc.)
e. The Empty Category [e].
The basic problem is to provide a description of English which will account for the items that
anaphors can and cannot refer to. For example, consider the following:
325

326
(58) a. Ann said that Sue hurt her. (her = Ann not Sue)
b. Ann said that Sue hurt herself. (herself = Sue not Ann)
We need a description of English which will account the way native speakers understand the above
sentences. Specifically, native speakers know that the personal pronoun her in Ann said that Sue hurt
her cannot refer to Sue, whereas herself in Ann said that Sue hurt herself must refer to Sue. The use
of pronouns like her and herself is very common in everyday speech, and native speakers have no
trouble determining what they can and cannot refer to. Since the number of possible sentences in
English and every other language is infinite, this means that speakers must know some principles for
determining the possible referents for anaphors. Our objective is to specify those principles.
Among the anaphors listed above is the empty category [e] which always has a sentence internal
referent. For example, a sentence like This book, the city must burn is represented as follows (see
(241) on Page 196):
(59)
The [e] in the above diagram refers to the phrase this book, a sentence internal element. This empty
category must be distinguished from the empty category [u] (=[understood]) which is not an anaphor
because it has no sentence internal referent. For example, a sentence like Let’s eat means Let us eat
something. We represent the meaning as “Let us eat [u]” where the referent for “[u]” is understood
and not interpreted in relation to some other noun phrase in the sentence in the way [e] is.
As we have seen throughout this book, syntactic principles are based on phrase structure. In regard
to anaphors, we will make use of the following structural relationships described in Chapter Three:

(60) a. C–command: a category α C–commands another category β, if the first branching
category Z above α dominates β.
b. Neighborhood: includes all items C–commanded by the same TNS characterizer.
To see how these relationships work, consider the following diagram which will serve as the basis
for our entire discussion of anaphors:
(61)
In (61), the following relationships obtain:
(62) a. N3 w C–commands N3 x, N3 y, and N3 z.
b. N3 x does not C–command N3 w, but does C–command both N3 y and N3 z.
c. N3 y does not C–command N3 w or N3 x, but does C–command and N3 z.
d. N3 z does not C–command either N3 w or N3 x or N3 y.
(63) a. N3 w and N3 x are in the same neighborhood.
b. N3 y and N3 z are in the same neighborhood.
c. N3 w and N3 y are not in the same neighborhood.
d. N3 w and N3 z are not in the same neighborhood.
e. N3 x and N3 y are not in the same neighborhood.
f. N3 x and N3 z are not in the same neighborhood.
327

328
As a start, we recognize the following principles:
(64) PRINCIPLE 1: EVERY ANAPHOR MUST AGREE WITH ITS REFERENT IN
PERSON, NUMBER AND GENDER.
a. Bob likes himself. (himself agrees with Bob)
b. *Bob likes myself. (no agreement in person)
c. *Bob likes themselves. (no agreement in number)
d. *Bob likes herself. (no agreement in gender)
(65) PRINCIPLE 2: A PERSONAL PRONOUN CANNOT BE IN THE SAME
NEIGHBORHOOD AS ITS REFERENT.
a. [ V3 Bob told him many lies ] (him … Bob)
b. [ V3 Bob told Sue [ V3 that Ann likes him ] ] (him = Bob)
(66) PRINCIPLE 3: A REFLEXIVE PRONOUN MUST BE IN THE SAME
NEIGHBORHOOD AS ITS REFERENT.
a. [ V3 Bob told himself many lies ] (himself = Bob)
b. *[ V3 Bob told Sue [ V3 that Ann likes himself ] ] (himself … Bob)
c. [ V3 People tell themselves lies ] (themselves = people)
(67) PRINCIPLE 4: A RECIPROCAL PRONOUN MUST BE IN THE SAME
NEIGHBORHOOD AS ITS REFERENT.
a. [ V3 People tell each other lies ] (each other = people)
b. *[ V3 People have told me [ V3 that Sue told each other lies ] ]
(68) PRINCIPLE 5: A FLOATING QUANTIFIER MUST BE IN THE SAME
NEIGHBORHOOD AS ITS REFERENT.
a. [ V3 Both Bob and Ted told lies ] (both = Bob and Ted)
b. [ V3 Bob and Ted both told lies ] (both = Bob and Ted)
c. [ V3 the two men said [ V3 that the two women both told lies ] ] (both = the two women
not the two men)
Some examples which illustrate the above principles are as follows (the subscripted N3 refer to the
basic diagram in (61):

(69) a. Bob told himself that Ann liked him.
N3 w PST tell N3 x that N3 y PST like N3 z
Bob told himself that Ann liked him
329
The N3 above himself is N3 x, and the N3 above Bob is N3 w. N3 x and N3 w are in the
same neighborhood. Therefore, himself must refer to Bob. The N3 above him is N3 z,
and the N3 above Bob is N3 w. N3 z and N3 w are not in the same neighborhood.
Therefore, him can refer to Bob.
b. *Bob told Sue that Ann liked himself.
N3 w PST tell N3 x that N3 y PST like N3 z
*Bob told Sue that Ann liked himself
The N3 above himself is N3 z, and the N3 above Bob is N3 w. N3 z and N3 w are not in
the same neighborhood. Therefore, himself cannot refer to Bob. The sentence is
ungrammatical because the only N3 that himself can refer to is the one dominating
Ann; this intended reference involves a conflict in gender.
c. The two men told Sue that the two women both liked them.
N3 w PST tell N3 x that N3 y PST like N3 z
the two men told Sue that the two women both liked them
The word both must refer to the two women because it is in the same neighborhood
as the two women. The personal pronoun them cannot refer to the two women because
N3 y and N3 z are in the same neighborhood; them can only refer to the two men
d. Bob told the two men that the two women liked each other.
N3 w PST tell N3 x that N3 y PST like N3 z
Bob told the two men that the two women liked each other
The reciprocal pronoun each other must refer to the two women because N3 y and N3 z
are in the same neighborhood; each other cannot refer to the two men
Given the above basic principles, let us now look at the anaphors individually. Beginning with
personal pronouns, we propose the following:

330
(70) PRINCIPLE 6: A PERSONAL PRONOUN CAN REFER TO ANY N3 OUTSIDE ITS
OWN NEIGHBORHOOD UNLESS IT BOTH PRECEDES AND C–COMMANDS
THAT N3.
We see this principle illustrated in the following examples:
(71) a. He told Sue that Ted liked Ann. (he … Ted)
N3 w PST tell N3 x that N3 y PST like N3 z
he told Sue that Ted liked Ann
N3 w precedes and C–commands N3 y; therefore, a personal pronoun in N3 w slot cannot
refer to a noun in N3 y slot.
b. Bob told Sue that he liked Ann. (he = Bob)
N3 w PST tell N3 x that N3 y PST like N3 z
Bob told Sue that he liked Ann
N3 y neither precedes nor C–commands N3 w; the two phrases are in different
neighborhoods; therefore, a personal pronoun in N3 y slot can refer to a noun in N3 w
slot.
c. Bob told Sue that Ted liked her. (her = Sue)
N3 w PST tell N3 x that N3 y PST like N3 z
Bob told Sue that Ted liked her
N3 z neither precedes nor C–commands N3 x; the two phrases are in different
neighborhoods; therefore, a personal pronoun in N3 z slot can refer to a noun in N3 x
slot.
d. Bob told her that Ted liked Sue. (her … Sue)
N3 w PST tell N3 x that N3 y PST like N3 z
Bob told her that Ted liked Sue
N3 x precedes and C–commands N3 z; therefore, a personal pronoun in N3 x slot cannot
refer to a noun in N3 z slot.

e. Bob told Ann that Sue liked her. (her = Ann; her =/ Sue)
N3 w PST tell N3 x that N3 y PST like N3 z
Bob tell Ann that Sue liked her
331
N3 z neither precedes nor C–commands N3 x, and the two phrases are in different
neighborhoods, so a personal pronoun in the N3 z slot can refer to a noun in the N3 x
slot, and her can refer to Ann. But, since Sue and her are in the same neighborhood,
her cannot refer to Sue even though they agree in person, number, and gender.
Turning to reflexive and reciprocal pronouns, we have the following principle and illustrative
examples:
(72) PRINCIPLE 7: A REFLEXIVE OR RECIPROCAL PRONOUN MUST BE
R–COMMANDED BY ITS REFERENT IN ITS OWN NEIGHBORHOOD.
R–command: a category α R–commands another category β if the first branching category
Z above α dominates β and if β is an anaphor to the right of the head of Z.
a. Bob told himself that Sue liked him. (himself = Bob)
N3 w PST tell N3 x that N3 y PST like N3 z
Bob told himself that Sue liked him
The first branching category above N3 w is the top V3. The top V3 dominates N3 x. The
head of the top V3 is the V0 tell. N3 x is to the right of tell. Thus, N3 w R–commands
N3 x; further, both phrases are in the same neighborhood, so himself can refer to Bob.
Since N3 z neither precedes nor C–commands N3 w, and both phrases are in different
neighborhoods, a personal pronoun in the N3 z slot can refer to a noun in the N3 w slot.
b. *Bob told Ann that Sue liked himself. (himself … Bob)
N3 w PST tell N3 x that N3 y PST like N3 z
Bob told Ann that Sue liked himself
N3 w R–commands N3 z but the two phrases are not in the same neighborhood;
therefore, a reflexive pronoun in N3 z slot cannot refer to a noun in N3 w slot.

332
c. Bob told Ann that Sue liked herself. (herself = Sue; herself =/ Ann)
N3 w PST tell N3 x that N3 y PST like N3 z
Bob told Ann that Sue liked herself
N3 y R–commands N3 z and both phrases are in the same neighborhood; therefore, a
reflexive pronoun in N3 z slot can refer to a noun in N3 y slot, and herself can refer to
Sue. But, even though N3 x R–commands N3 z, herself cannot refer to Ann because the
two phrases are in different neighborhoods.
d. *Herself told Ann that Sue liked Kay.
*Kay told Ann that herself liked Sue.
N3 w PST tell N3 x that N3 y PST like N3 z
*herself told Ann that Sue liked Kay
*Kay told Ann that herself liked Sue
Neither sentence is possible because the reflexives do not have any R–commanding
referent in their own neighborhoods.
e. The two men told Sue that the two women liked each other.
N3 w PST tell N3 x that N3 y PST like N3 z
the two men told Sue that the two women liked each other
The phrase each other must refer to the two women because it is in the same
neighborhood as the two women.
f. *The two men told Sue that Ann liked each other.
N3 w PST tell N3 x that N3 y PST like N3 z
the two men told Sue that Ann liked each other
The phrase each other has no possible referent in its own neighborhood so the
sentence is ungrammatical.
.
Turning to floating quantifiers, consider the following principle and illustrative diagram:
(73) PRINCIPLE 8: A FLOATING QUANTIFIER MUST BE L–COMMANDED BY ITS
REFERENT IN ITS OWN NEIGHBORHOOD.
L–command: a category α L–commands another category β if the first branching category
Z above α dominates β and if β is an anaphor to the left of the head of Z.

(74)
The behavior of floating quantifiers is illustrated in the following examples:
(75) a. The men will all tell the women to buy it.
N3 w will FUT N3 x tell N3 y to buy N3 z
the men will all tell the women to buy it
333
The first branching category above N3 w is the top V3. The top V3 dominates N3 x. The
head of the top V3 is the V0 tell. N3 x is to the left of tell. Thus, N3 w L–commands
N3 x, so all can refer to the men.
b. The men will tell all the women to buy it.
N3 w will FUT N3 x tell N3 y to buy N3 z
the men will tell all the women to buy it
The first branching category above N3 w is the top V3, and the head of the top V3 is
the V0 tell. Since all occurs to the right of tell, it cannot refer back to the men. All
is inside the N3 y, so all quantifiers women. We know that all is inside N3 y because
it cannot be separated from the rest of the elements within N3 y:
[1] All the women were told to buy it.
[2] *The women were told all to buy it.

334
To the principles above, we add the following two principles for formally determining the subject and
object of a verb:
(76) PRINCIPLE 9: SUBJECTS MUST C–COMMAND AND PRECEDE THEIR VERBS
AND NOUNS.
(77) PRINCIPLE 10: OBJECTS MUST C–COMMAND AND FOLLOW THEIR VERBS
AND NOUNS.
These principles are illustrated in the following diagram and examples:
(78)
a. The girls will all want to dress themselves. (no N3 y)
The girls will all want to dress each other. (no N3 y)
N3 w will FUT N3 x want N3 y to dress N3 z
the girls will all want to dress themselves
The first branching category above N3 w is the top V3.
The top V3 dominates N3 x. The head of the top V3 is the V0 want. N3 x is to the left
of want. Thus, N3 w L–commands N3 x, so all can refer to the girls.
The top V3 dominates N3 z. The head of the top V3 is the V0 want. N3 z is to the right
of want. Thus, N3 w R–commands N3 z, so themselves can refer to the girls.

Also: N3 w is the subject of both want and dress since it C–commands and precedes
both verbs; N3 z is the object of dress since it C–commands and follows dress.
335
The verb want has no object because there is no N3 which C–commands and follows
it. Because there is no object, there is nothing between want and to, so they contract
to wanna.
b. The girls will all want the boys to dress themselves.
The girls will all want the boys to dress each other.
N3 w will FUT N3 x want N3 y to dress N3 z
the girls will all want the boys to dress themselves
N3 w is the subject of want since it C–commands and precedes want; N3 y is the subject
of dress since it C–commands and precedes dress. N3 y is the object of want since it
C–commands and follows want; N3 z is the object of dress since it C–commands and
follows dress. As above, the referent for all is the girls; however, the referent for
themselves is the boys: the first branching category above N3 y is the V2, V2
dominates N3 z, and N3 z contains an anaphor to the right of the head of V2.
Lastly, we consider the following principle for the interpretation of [e], the empty anaphor:
(79) PRINCIPLE 11: [e] CAN BE ANYWHERE EXCEPT IN A RESIDENTIAL
DOMAIN. THUS, [e] CANNOT BE C–COMMANDED BY A PERMANENT
RESIDENT OR L–COMMANDED BY A FREE RESIDENT.
a. Binding residents and barriers.
[1] Binding resident (BR) – the C3 dominated by the lowest recursion of X3 (DET
or POS in N3; TNS or CPL in V3).
[2] Barrier BR – a C3 whose position is constant (DET, POS, CPL) – acts as a
cuing device to structure.
b. Residents and residential domains.
[1] Permanent residents – those that are obligatory (e.g. whether in I wonder
whether he went). The domain of a permanent resident is all items it
C–commands.
[2] Free residents – those that are optional (e.g. that in He said (that) she knew me).
The domain of a free resident is all items it L–commands.

336
c. Empty category condition (ECC): An empty anaphor [X3 e] within some immediate
neighborhood must be bound to a V3 Level phrase of the same type in the extended
neighborhood.
d. Wh–binding condition (WHBC): all phrases in the extended neighborhood must be
bound to an [e]. Usually, but not always, such phrases contain a WH–word, e.g., who,
which, at what time, etc.
Principle 11 is illustrated in the following examples discussed previously:
(80) [Bill’s vote against gun control]; common noun head vote:
a. Mary criticized Bill’s vote against gun control.
b. *What i did Mary criticize Bill’s vote against [N3 e ] I?
c. *Mary criticized vote against gun control.
d. Mary criticized the vote against gun control.
(81) Indirect question:
a. Mary wondered if Bill voted against gun control.
b. *What i did Mary wonder if Bill voted against [N3 e ] I?
c. *Mary wondered Bill voted against gun control.
d. *Who i did Mary wonder if [N3 e ] I voted against gun control?
e. *Who i did Mary wonder [N3 e ] I voted against gun control?
(82) Indirect Statement with obligatory barrier BR:
a. Mary chortled that Bill voted against gun control.
b. *What i did Mary chortle that Bill voted against [N3 e ] I?
c. *Mary chortled Bill voted against gun control.
d. *Who i did Mary chortle that [N3 e ] I voted against gun control?
e. *Who i did Mary chortle [N3 e ] I voted against gun control?
(83) Indirect Statement with optional barrier BR:
a. Mary thought that Bill voted against gun control.
b. What i did Mary think that Bill voted against [N3 e ] I?
c. Mary thought Bill voted against gun control.
d. *Who i did Mary think that [N3 e ] I voted against gun control?
e. Who i did Mary think [N3 e ] I voted against gun control?

EXERCISES FOR CHAPTER FIVE
1. Account for the reference of all of the anaphors in the following examples:
a. The little girls i want to dress themselves i.
b. The little girls i say that they i dressed each other i.
c. The little girls i [e] i that [e] i dressed themselves i are cute.
d. The little girls i have all i dressed alike.
e. The little girls i all i want to dress each other i.
2. Determine the subjects of each use of swim in the following:
a. John wants to swim in the ocean.
b. John wants you to swim in the ocean.
c. It is forbidden to swim in the ocean.
d. It is forbidden for you to swim in the ocean.
e. You are forbidden to swim in the ocean.
3. Explain the ungrammaticality of the following:
a. *What did John put the coats in the closet?
b. *Which books does John think is good?
c. *John wants her to drive himself to the party.
d. *Each other like(s) the women.
e. *The women who is next to himself like John.
4. What principle(s) account for the following sets of data?
a. That John hit something is clear.
*What i is that John hit [e] i clear?
What i is it clear that John hit [e] i?
b. John denied the claim that his car hit the tree.
*What i did John deny the claim that his car hit [e] i?
What i did John deny that his car hit [e] i?
337

338

SUPPLEMENT ONE: THE NATURE OF HUMAN LANGUAGE
I. FUNDAMENTAL EMPIRICAL PROBLEMS FOR LINGUISTICS.
A. The GRAMMATICAL CHARACTERIZATION PROBLEM: to account for the
LINGUISTIC COMPETENCE of native speakers, that is, the unconscious knowledge
they have of their native language.
Crucial Question: What is the nature of the unconscious knowledge that all native speakers
have?
1. PHONOLOGY.
a. (1) [t]: wait, write, vote, suit,
(2) [d]: wade, ride, load, feed
(3) [D]: waiter, writer, voter, suitor;
wader, rider, loader, feeder
b. (1) [ ]: ploud
(2) [ ]: plouder
(3) [ ]: plout
2. MORPHOLOGY.
a. (1) [t]: kicked, stopped, laughed
(2) [d]: plunged, crammed, loved
(3) [cd]: hunted, raided, relented
b. (1) [ ]: goob
(2) [ ]: nork
(3) [ ]: frist
3. SEMANTICS.
a. (1) in: John ran in the street.
(2) to: John ran to the street.
(3) into: John ran into the street.
(4) on: John ran on the street.
(5) from: John ran from the street.
(6) out of: John ran out of the street.
b. (1) John fell...
(2) John went...
(3) John emerged...

340
4. SYNTAX.
Judgements: ?Questionable; *No good; **Garbage
a. Sentence Adverbs: [ultimately]
The student will read the book.
(1) Ultimately the student will read the book.
(2) ** The ultimately student will read the book.
(3) The student ultimately will read the book.
(4) The student will ultimately read the book.
(5) * The student will read ultimately the book.
(6) ** The student will read the ultimately book.
(7) The student will read the book, ultimately.
(8) ? The student will read the book ultimately.
b. Manner Adverbs: [thoroughly]
The student will read the book.
(1) * Thoroughly the student will read the book.
(2) ** The thoroughly student will read the book.
(3) * The student thoroughly will read the book.
(4) The student will thoroughly read the book.
(5) * The student will read thoroughly the book.
(6) ** The student will read the thoroughly book.
(7) The student will read the book thoroughly.
c. Miscellaneous Adverbs: [hardly]
The student will read the book.
(1) * Hardly the student will read the book.
(2) ** The hardly student will read the book.
(3) ? The student hardly will read the book.
(4) The student will hardly read the book.
(5) * The student will read hardly the book.
(6) ** The student will read the hardly book.
(7) * The student will read the book hardly.
Crucial Fact: The knowledge that native speakers have of their native language is
substantially uniform and largely unconscious.

B. The GRAMMATICAL REALIZATION PROBLEM: to account for the LINGUISTIC
PERFORMANCE of native speakers, that is, the way in which they acquire and use their
native language.
Crucial Question: How do children acquire their native language?
Crucial Question: How do native speakers produce and understand their native language?
1. Comparison of native and second language acquisition.
NATIVE LANGUAGE SECOND LANGUAGE
Data fragmentary Data detailed and specific
Data haphazard Data organized
Data often defective Data fully grammatical
Speech sole input Speech and writing inputs
Unconscious Conscious and deliberate
Novel sentences Patterns and drills
Almost no instruction Formal instruction by experts
2. Characteristics of native language acquisition.
a. Rapid: Native language acquisition is completed in a remarkably short period for all
normal children.
b. Comprehension proceeds ahead of production (Kevin, age 14 months):
(1) Production: approximately 10 words
mommy, daddy, cookie, pops, balloon, coca-cola, truck; some; up, down, sit
(2) Comprehension: approximately 100 words and phrases
341
helicopter, plane, Volkswagon; Burt, Bambi, Donald Duck; cheese, milk, hot
dog; teeth, toes, belly button
up in high chair; go play piano; give daddy some; don’t spill it; don’t throw it
on the floor; let’s go get changed; brush your hair; etc.
c. Not dependent on peripheral sensory mechanisms; language develops even in cases
of blindness, deafness, and mutism.

342
d. Not dependent on intelligence; language develops except in cases of severe
intellectual disability.
e. Creative.
(1) Expressions with no adult models.
Allgone itchy. (after being scratched)
I do it Kevin's-self.
Do I have a jack sized bed?
open = sharpen or inflate
dirty = unpeeled
big water = lake
(2) Overgeneralization.
throwed for threw
mans for men
(3) Imitation without comprehension.
f. Predictable.
I do it twelve times. (it = go to sleep)
I'm having a nervous breakdown.
(1) Children make the same kinds of mistakes.
He goed there.
Him goed there.
I goed the truck.
(2) Children use the same strategy.
g. Uniform: the stages of acquisition.
(1) Primitive sentences.
Noun truck
Modifier + Noun pretty baby
Noun + Noun daddy coat
Verb + Noun see truck
Verb + Noun + Noun see baby shoe

(2) The differentiation of word classes.
Pivot (modifier) and Open (noun) classes.
(a) Time One - around 24-27 months
Pivot One Open
that horsie
a truck
my car
pretty baby
more milk
343
Pivot One includes determiners (the, a), demonstratives (this, that),
possessives (my), adjectives (pretty, big), and quantifiers (more).
(b) Time Two - about three months later
(Dem) + (Det) + Pivot Two + Open
that a my truck
that the big truck
* that my a truck
* my a that truck
Pivot Two includes possessives, quantifiers, and adjectives.
(c) Time Three - about three months later
(Det) + (Adj) + Open
a big cup
*a my cup
*a more cup
Pivot Three consists only of possessives and quantifiers.

344
(3) Relevance of word order.
(a) Time Two
* your that truck
Yours is that truck.
(b) Time Three
* the more milk
The more milk you drink the bigger you'll grow.
(4) Relevance of imitation.
(a) that doed
(b) allgone shoe
(c) big a truck
(5) Relevance of memorization.
The number of pivot/open combinations recorded in one study in successive
months was: 14, 24, 54, 89, 350, 1400, 2500+
h. Dependent on maturation.
(1) Immediate memory limitations: the magic number 7 ± 2.
(2) Cognitive limitations.
Crucial Fact: Language acquisition proceeds in children unconsciously with almost the
complete lack of instruction.
3. Characteristics of the normal use of language.
a. Rapidity of the speech process.
(1) Slurring.
(2) Voice Identification.
b. Simultaneous integration of all components of grammar.
c. One time exposure to sentences.
Crucial Fact: Ordinary use of language does not require paper and pencil analysis

II. THEORIES OF LANGUAGE.
A. Myths.
1. Parental instruction.
2. The need/pleasure hypothesis.
B. Philosophical/Psychological Theories.
1. EMPIRICISM (BEHAVIORISM, NURTURISM).
John Locke (1632-1704); George Berkeley (1685-1753); David Hume (1711-1776)
345
The empiricist approach claims that human beings are born with very little in the way
of instinctive behavior, that is, they are born with very little in the way of innate
mechanisms for acquiring knowledge about language or anything else. Most of what
they comes to know is a product of their experiences with the physical and social
world. Human beings learn by imitating, associating, generalizing, and so on.
2. RATIONALISM (MENTALISM, NATIVISM).
Rene Descartes (1596-1650); Baruch Spinoza (1632-1677); Gottfried Wilhelm von
Leibniz (1646-1716)
The rationalist approach claims that human beings are born with a rich and detailed
system of innate mechanisms which determine the way in which they acquire
knowledge and interpret the data of everyday experience.
C. Medical evidence.
1. The relationship between sex and maturation.
2. Interlocking of speech and motor milestones.
3. Critical period for language acquisition -
CEREBRAL LATERALIZATION
a. Cerebral lateralization - the asymmetric specialization of the brain for language. In
most humans, language is under the dominant control of the left hemisphere.

346
b. Arguments for cerebral lateralization.
(1) Electrical stimulation of the brain.
(2) Anesthetization of the brain (injection of amobarbital into the carotid artery).
(3) Hemispherectomy.
(4) Aphasia studies.
c. Arguments for the critical period.
(1) Necessity for exposure before puberty.
(2) Language in those who are intellectually disabled.
(3) The effects of congenital deafness.
(4) The effects of sudden acquired deafness.
(5) The nature of second language acquisition.
D. Levels of adequacy.
1. OBSERVATIONAL ADEQUACY.
2. DESCRIPTIVE ADEQUACY.
3. EXPLANATORY ADEQUACY.

III. The theory of LINGUISTIC UNIVERSALS.
Linguistic universals are potential characteristics of natural human languages. The complete
list of possible universals is called CORE GRAMMAR. Linguistic universals derive from
the capacities and limitations of the human language apparatus.
A. SUBSTANTIVE UNIVERSALS: stipulations on the features which can occur in a
natural language.
1. Phonology (see pages 58-9 in LA for the interpretation of symbols in square brackets).
a. [±NASAL].
(1) [+NASAL]: [m], [n], [õ]
Air leaves vocal apparatus through the nose.
(2) [–NASAL]: [p], [b]. [t], [d], [k], [g]
b. [±VOICED].
Air leaves vocal apparatus through the mouth.
(1) [+VOICED]: [b], [m], [d], [n], [g], [õ]
Vocal cords are vibrating.
(2) [–VOICED]: [p], [t], [k]
c. *[±AURAL].
Vocal cords are not vibrating.
Air leaves vocal apparatus through the ears
347

348
2. Syntax.
a. [±PREHEAD]; [±POSTHEAD].
PREHEAD HEAD POSTHEAD
all those new books about syntax available
(1) [+PREHEAD, –POSTHEAD]: those and new
(a) those new books
(b) *books new those, *books those new
(2) [–PREHEAD, +POSTHEAD]: about syntax
(a) those books about syntax
(b) *those about syntax books
(3) [+PREHEAD, +POSTHEAD]: available
(a) those available books
(b) those books available
b. *[±EVEN-NUMBERED-WORD].
3. Semantics.
a. [±DISJUNCTIVE].
(1) [+DISJUNCTIVE]: emphasizing parting
(2) [–DISJUNCTIVE]: not emphasizing parting
b. [±CONJUNCTIVE].
(1) [+CONJUNCTIVE]: emphasizing joining
(2) [–CONJUNCTIVE]: not emphasizing joining
c. DISJUNCTIVE CONJUNCTIVE
(1) from + –
(2) to – +
(3) in – –
(4) out of + +
f. *[±INSCRIBED-CIRCLE > x].

B. FORMAL UNIVERSALS: stipulations on the organization of grammars and on the form
and functioning of grammatical rules.
1. The feature system of language must permit cross-categorial reference:
a. [+NASAL]: m, n, õ
b. [–NASAL]: p, b, t, d, k, g
c. [+ANTERIOR]: m, p, b, n, t, d
d. [–ANTERIOR]: õ, k, g
e. [+CORONAL]: t, d, n
f. [–CORONAL]: m, p, b, õ, k, g
2. The feature system must allow precise identification.
a. [+NASAL, +VOICED, +ANTERIOR, –CORONAL]: m
b. [+NASAL, +VOICED, +ANTERIOR, +CORONAL]: n
c. [+NASAL, +VOICED, –ANTERIOR, +CORONAL]: ñ
d. [+NASAL, +VOICED, –ANTERIOR, –CORONAL]: õ
3. The rules of language are structure dependent; they cannot involve counting.
a. Sentence Adverbs: [ultimately]
The student will read the book.
(1) Ultimately the student will read the book.
(2) The student ultimately will read the book.
(3) The student will ultimately read the book.
(4) The student will read the book, ultimately.
b. Units.
349

350
c. Justification of units.
(1) REFERENCE.
(a) Who will read the book?
(b) The student said that he will read the book.
(c) What will the student do?
(2) OMISSION.
(a) The student said he would read the book if he had to________.
(b) This student will read the book, and that one will ________ too.
(3) PLACEMENT.
(a) Ultimately the student will read the book.
(b) The student ultimately will read the book.
(c) The student will ultimately read the book.
(d) The student will read the book, ultimately.
d. Immediate memory limitations: chunking.
(1) Telephone numbers.
(2) Word lists versus structured sentences.
(a) the little boy is playing in the backyard
(b) fence girl under long the looking is a tall
C. Relating grammatical characterization to grammatical realization; attaining explanatory
adequacy.
1. PHONOLOGY.
a. (1) [m]: impossible, imbalance
(2) [n]: intangible, indiscrete
(3) [õ]: incorrigible, ingratitude
b. (1) [m], [p], [b] are [+ANTERIOR, –CORONAL]
(2) [n], [t], [d] are [+ANTERIOR, +CORONAL]
(3) [õ], [k], [g] are [–ANTERIOR, –CORONAL]
c. ASSIMILATE the nasal to the following consonant.

2. MORPHOLOGY.
a. (1) [t]: kicked, stopped, laughed
(2) [d]: plunged, crammed, loved
(3) [cd]: hunted, raided, relented
b. (1) [p], [t], [k], [f], [s], etc. are [–VOICED]
(2) [b], [d], [g], [v], [z], etc. are [+VOICED]
c. (1) use [t] after [–VOICED] sounds
(2) use [d] after [+VOICED] sounds
(3) use [cd] after [t] and [d]
(a) *huntt/*huntd/hunted
(b) hurted/goed/ketchuped
3. SEMANTICS.
a. (1) into: John ran into the street.
(2) from: John ran from the street .
b. (1) run, go: [±DISJUNCTIVE, ±CONJUNCTIVE]
(2) emerge: [+DISJUNCTIVE, –CONJUNCTIVE]
(3) enter: [–DISJUNCTIVE, +CONJUNCTIVE]
4. SYNTAX.
a. Data on Sentence Adverbs: [ultimately]
b. Sentences contain three basic units.
351

352
c. Syntactic Representations.
(1) Labelling of Units.
(a) Traditional Grammar.
SUBJECT = Person or thing performing the action of the verb.
OBJECT = Person or thing affected by the action of the verb.
(b) Modern Structural Grammar.
S = SENTENCE DET = DETERMINER
VP = VERB PHRASE MOD = MODAL
NP = NOUN PHRASE NEG = NEGATIVE
V = VERB N = NOUN
(2) Structure versus Function.
(a) The student will burn the money.
(b) The student will earn the money.
(c) The student will have the money.
(3) Phrase Structure Rules ( Parentheses indicate optional elements).
(a) S ÷ NP + AUX + VP
(b) VP ÷ V + (NP)
(c) NP ÷ (DET) + N
(d) AUX ÷ MOD + (NEG)

D. The generality of linguistics universals.
Comparison of Language and Vision.
1. Ambiguity in Isolation.
a. Linguistic.
353
(1) This baby is too young to stand up by himself.
(2) This baby is too young to be stood up.
(3) This baby is too young either to stand up by himself or to be stood up by
someone else.
(4) This baby is too young to stand up.
b. Visual. Necker Cube:
c. General characteristics.
(1) Interpretation is unconscious.
(2) Interpretation is involuntary.

354
2. Ambiguity in Context.
a. Linguistic.
(1) The farmer looked over the fence.
(2) The farmer looked under the fence.
(3) The farmer looked under the fence, and the farmer looked over the fence.
b. Visual.
c. General characteristics.
(1) Interpretation is unconscious.
(2) Interpretation is involuntary.
(3) Interpretation is uniform.

3. Ellipsis.
a. Linguistic.
(1) John likes apples, Sue likes oranges, and Bill likes pears.
(2) John likes apples, Sue, oranges, and Bill, pears.
(3) *John, apples, Sue, oranges, and Bill likes pears.
b. Visual.
c. General characteristics.
(1) Sophisticated pattern matching.
(2) Closure (Gestalt psychology).
355

356
4. Contextual sensitivity.
a. Linguistic.
(1) The chickens are ready to eat.
(2) The chickens are ready to lay eggs.
(3) The chickens are ready to be plucked.
(4) The chickens are ready to cook.
b. Visual.
c. General characteristics.
(1) Interpretation is unconscious.
(2) Contextual information automatically processed.

5. Complexity.
a. Linguistic.
(1) People who believe that the world is flat are nuts.
(2) Bill said that to him that the world is flat is idiotic.
(3) Bill said that that the world is flat is ridiculous.
(4) That that the world is flat is idiotic is surprising to Bill.
(5) Who is that that the world is flat is idiotic surprising to?
(6) Who is it surprising to that it is idiotic that the world is flat?
b. Visual
357

358
c. General characteristics.
(1) GRAMMATICAL/UNGRAMMATICAL
(a) Grammatical:
She is mixing up her children.
She is mixing her children up.
(b) Ungrammatical:
*She her children mixing up is.
(2) ACCEPTABLE/UNACCEPTABLE
(a) Acceptable:
He passed up the chance to look up the name.
(b) Unacceptable:
*He passed the chance to look the name up up.
(3) Important points:
(a) Acceptability is a technical linguistic concept that has nothing to do with
appropriateness.
(b) Acceptability is only an issue when the question of brain capacities
arises.
(c) None of the following sentences involve acceptability:
[1] John mixed up the recipe.
[2] John screwed up the recipe.
[3] John fucked up the recipe.
(4) Examples:
(a) Grammatical but Unacceptable: They attributed the painting [that was
stolen from the Louvre last night despite the intensive security throughout
the building] to Picasso.
(b) Ungrammatical but Acceptable: They attributed to Picasso the painting
[that was stolen from the Louvre last night despite the intensive security
throughout the building].
(c) Ungrammatical and Unacceptable: They attributed to Picasso [a
Spanish-born French painter, sculptor, graphic artist and ceramist,
considered by many to be the greatest artist of the 20th century] the
painting [that was stolen from the Louvre last night despite the intensive
security throughout the building].

6. Incongruity.
a. Linguistic.
(1) Bill convinced Mary that the world is round, but she didn't believe him.
359
(2) Now that the surgeon has removed all of his vital organs, John feels better than
he ever has.
b. Visual.
c. General characteristics.
(1) Automatic pattern matching.
(2) Integration of linguistic and pragmatic facts.

360
7. Rules.
a. Linguistic.
(1) John saw the man walking toward the railroad station? (3 ways ambiguous)
(2) What did John see the man walking toward? (unambiguous)
b. Visual: Kopfermann Cubes.
c. General characteristics.
(1) The interpretations are rule governed.
(2) The initial interpretation is automatic, unconscious and involuntary.

E. Linguistic universals and human mental abilities.
1. Memory Limitations
a. John called up a girl
b. John called a girl up.
c. John called up a girl he used to go with.
d. ?John called a girl he used to go with up.
e. John called up a girl he used to go with back in Idaho before he left for college.
f. *John called a girl he used to go with back in Idaho before he left for college up.
2. Complexity Limitations.
361
a. MULTIPLE BRANCHING CONSTRUCTION: one with no internal structure,
that is, no internal dependency relationships. Multiple branching constructions simply
involve a string of concatenated elements.
(1) I saw John, Bill, Mary, Sue,...
(2) [John] + [Bill] + [Mary] + [Sue] + ...
b. RIGHT BRANCHING CONSTRUCTION: one whose internal dependency
relationships branch to the right, that is, the most deeply embedded constituent is the
one furthest to the right.
(1) This is the cat that saw the rat that ate the cheese...
(2) [this is [the cat that caught [the rat that ate the cheese] ] ]
c. LEFT BRANCHING CONSTRUCTION: one whose internal dependency
relationships branch to the left, that is, the most deeply embedded constituent is the
one furthest to the left.
(1) Mary's brother's wife's dog is named Spot.
(2) [ [ [ [Mary’s] brother’s] wife’s] dog]

362
d. NESTED CONSTRUCTION: phrase A and phrase B form a nested construction
if phrase B falls totally within phrase A , and if there is some nonnull element to to
the left and to the right of phrase B within A.
(1) John called Sue up yesterday.
(2) John [called [Sue] up] yesterday.
e. CENTER (SELF)-EMBEDDED CONSTRUCTION: phrase B is self-embedded
in A if B is nested in A and, furthermore, B is a phrase of the same type as A.
(1) (a) Something upset Ann.
(b) That something annoyed Sue upset Ann.
(c) That that something pleased John annoyed Sue upset Ann
(2) (a) The man killed the cat.
(b) The man that the woman loves killed the cat.
(c) The man that the woman that the boy annoyed loves killed the cat.
(3) (a) When the cat died, the boy cried.
(b) Because when the cat died the boy cried, the girl left.
(c) Even though because when the cat died a boy cried the girl left, the
doctor was not concerned.
(4) (a) When the man saw the woman who grabbed the dog because the boy said
that it bit the girl who was sick, he laughed.
(b) Als der Mann die Frau die den Hund weil der Knabe dass er das Mädchen
das krank war biss sagte packte sah, lachte er.
(c) When the man the woman who the dog because the boy that it the girl
that sick was bit said grabbed saw, laughed he.
(5) It upset Ann that it annoyed Sue that something pleased John. (=1c)
(6) The cat was killed by the man who is loved by the woman who was annoyed by
the boy. (=2c)
(7) The doctor was not concerned even though the girl left because a boy cried
when the cat died. (=3c)
(8) Als der Mann die Frau sah, die den Hund packte, weil der Knabe sagte, dass er
das Mädchen das krank war biss, lachte er. (=4b)

IV. TYPES OF LANGUAGES.
A. Classification.
1. Natural Languages.
a. Animal Languages.
b. Human Languages.
(1) Vocal/auditory.
(2) American Sign Language
2. Artificial Languages.
a. Non-natural Human Languages.
b. Computer Languages.
(1) Programming languages: BASIC, LISP, Pascal.
(2) Word processors: Applewriter, Word Perfect.
B. Characteristics.
1. Animal Languages.
a. Type One.
363
(1) Description: the system consists of a fixed number of signals each one of which
is associated with a specific range of eliciting conditions or internal states.
(2) Example: Howling Monkeys
(a) voluminous roar ‘defense of group territory’
(b) rapid grunting (male) ‘response to a disturbance’
(c) terrible bark (female) ‘response to a disturbance’
(d) purr ‘mother/young relationship’
(e) series of cries (usually three) ‘a falling infant’
(3) Example: Honey Bees
(a) round dance: food is less than 20 feet away
(b) sickle dance: food is from 20 to 60 feet away
(c) tail-wagging dance: food is more than 60 feet away.

364
b. Type Two.
(1) Description: the system consists of a fixed number of linguistic dimensions
each one of which is associated with a nonlinguistic dimension such that
selection of a point along one indicates a corresponding point along the other.
(2) Example: Honey Bees - the tail-wagging dance.
2. Human Language.
The direction of the food source is also conveyed by this dance, the sun being
used as a reference point. A tail-wagging dance proceeding on the vertical
honeycomb upward means that the food is in the direction of the sun; if
downward, it means the food is in the direction opposite the sun; if 60 degrees
left of straight up, it means the food is 60 degrees to the left of the sun; etc.
The system consists of an INFINITE NUMBER OF SIGNALS each one of which
involves the pairing of a superficial sound structure with an underlying semantic
structure. There is NO ONE-TO-ONE CORRESPONDENCE BETWEEN SOUND
AND MEANING.
a. AMBIGUITY (one sound, more than one meaning).
(1) Lexical level.
(a) After making the right turn at the corner,...
(b) After making the left turn at the corner,...
(2) Phrasal level.
(a) He looked over the fence.
(b) He hit the lady with the hat.
(3) Sentential level.
(a) Visiting relatives can be boring.
(b) The chickens are ready to eat.
(4) Examples from the speech of children.
(a) Milk.
(b) Mommy sock.

. PARAPHRASE (one meaning, more then one sound).
(1) Adult speech.
(a) The nurse gave the shots to the students.
(b) The nurse gave the students the shots.
(c) The shots were given to the students by the nurse.
(d) The students were given the shots by the nurse.
(2) Examples from the speech of children.
(a) Get it baby up.
(b) Get it me up.
(c) Get baby up.
c. Disparity in meaning.
(1) (a) John is eager to please.
(b) John is easy to please.
(2) (a) Bill appeared to John to behave himself.
(b) Bill appealed to John to behave himself.
d. Cultural constraints.
(1) I'm very sorry to hear that your wife
(a) passed away.
(b) died.
(c) kicked the bucket.
(d) went to heaven.
(2) Would you mind
(a) leaving a specimen in that container.
(b) leaving a urine specimen in that container.
(c) going peepee in that paper cup.
(d) taking a leak.
3. Computer languages: word processors.
a. Based on mathematical operations like counting, e.g., go to the fifth page of a
document.
b. Memories are flawless and expandable.
c. Equipped with special operations: SEARCH, COPY, CUT, PASTE, etc.
365

366
d. Controlled by magic words that are ad hoc.
e. Anything can be a unit for the purposes of movement, insertion, deletion, or
substitution of text.
(1) INPUT: The student will read the book.
(2) OUTPUT:
(a) Movement: will between read and the
The student read Will the book.
(b) Insertion: s' teacher after student
The students' teacher will read the book.
(c) Deletion: h, stu, and ent
Ted will read the book.
(d) Substitution: sha for wi
The student shall read the book.
f. Based on the computer's capacities.
V. THE EVOLUTION OF HUMAN LANGUAGE.
A. Theories of the evolution of human language.
1. CONTINUITY THEORY I: a straight line evolution of language with only
QUANTITATIVE differences.
a. Links linguistic competence with intelligence.
b. Can't measure intelligence across species.
2. CONTINUITY THEORY II: a straight line evolution of language with
QUALITATIVE differences, with a step-wise accumulation of skills.
a. Ignores PHYLOGENETIC relationships.
b. Subhuman primates should have the closest languages to human languages.
3. DISCONTINUITY THEORY: human language is species specific to man, like bipedal
gait. It is the result of anatomical and neurological specializations.
a. INNATENESS HYPOTHESIS.
b. UNIVERSAL (CORE )GRAMMAR.

B. Innovations in the STRUCTURE of the language apparatus.
1. Neurological.
1,2,3: Primary sensory cortex
7: Secondary sensory area
4,6: Primary motor cortex
9: Motor area for the hand (Exner's center)
44: Motor area for the face (Broca's area)
40: Supramarginal gyrus
39: Angular gyrus
17: Primary visual area
18,19: Secondary visual area
41,42: Primary auditory area
22: Secondary auditory area (Wernieke's area)
367

368
a. Development of the supramarginal and angular gyri.
b. LIMBIC and NON-LIMBIC stimuli.
(1) Limbic stimuli are concerned with survival: eating, drinking, sleeping,
reproduction, self-preservation, etc.
(2) Non-limbic stimuli are not concerned with survival: seeing, hearing,
speaking, etc.
c. Asymmetry of the temporal lobes.
d. Increase in the relative size of the primary auditory area.
2. Vocal
a. Lowering of the larynx.
(1) Optimally efficient system (the tilde "˜" above the vowel means that the
vowel is nasal):
m õ b g ˜e õ e o
CONSONANT + + + + – – – –
NASAL + + – – + + – –
ANTERIOR + – + – + – + –
(2) Human languages about 50% efficient.
(3) Neanderthal Man couldn’t produce: [i], [u], [a], [k], [g], [õ]
b. Development of the alveolar ridge.
c. Increased prominence in the muscles of the lips and complexity of the muscles
associated with articulation.
3. Auditory: slight anatomical changes, reasons unknown

C. Innovations in the FUNCTIONING of the language apparatus.
1. Neurological.
a. Cerebral lateralization.
b. Localization of language to specific areas.
c. Cross-modal associations.
2. Vocal.
a. The use of the tongue and pharyngeal wall to modify phonations.
b. The use of the nasal and oral separation.
3. Auditory.
a. Ear most sensitive to sounds within frequency range of 250-4000 cycles per
second--range of normal speech.
b. 50-65 decibel range for normal conversation.
VI. LIMITATIONS ON THE FORM OF NATURAL LANGUAGE.
A. Ambiguity and Paraphrase.
B. Sentence complexity and memory.
C. Rules not mathematically based. Rules based on structure.
D. Inability of the brain to utilize an analytical procedure while it is already in the process
of performing that procedure.
1. That that John surprised Sue worried Fred.
2. The man that the woman that the dog bit loves left.
E. Empty Categories ([e]) and Contraction.
1. a. They have won the game.
b. They’ve won the game.
2. a. They could have won the game.
b. Could i they [e] i have won the game?
c. *Could they’ve won the game?
369

370

SUPPLEMENT TWO: GRAMMAR REVIEW
SYNTACTIC CATEGORIES (THE PARTS OF SPEECH)
NOUN: Nouns are identifiable on the basis of several criteria. First, they can occur after definite
articles, e.g., the book, after indefinite articles, e.g., a book. Second, noun are inflected to show
number distinctions, e.g., one book versus two books, and to indicate possession, e.g., the book’s
cover. Third, nouns fulfill a number of specific functions in sentences. For instance, they can be
used as the subject of a verb, e.g., The book ends sadly, as the object of a verb, e.g., John likes that
book, and as the object of a preposition, e.g., There are no pictures in the book. Nouns are variously
subcategorized as follows:
PROPER NOUNS are those which refer to specific individuals, places, or things, e.g., John,
Rome, and English; COMMON NOUNS are non–specific in reference, e.g., man, city, and
language.
CONCRETE NOUNS are those which refer to tangible objects, e.g., cow, fence, and painting;
ABSTRACT NOUNS name intangible things, e.g., admiration, gravity, and painting. Notice
that some nouns have both concrete and abstract references.
COUNT NOUNS are those which are numerable, e.g., tree, leg, and linguist; MASS NOUNS
are not numerable, e.g., butter, water, and furniture. Notice that one does not ordinarily say
butters, waters, and furnitures (but, the waters of the Nile).
. COLLECTIVE NOUNS are those which refer to groups of items, e.g., herd, crowd, and mob.
PRONOUN: A pronoun is a word which is used in place of a noun phrase, e.g., he, she, who, myself,
etc. The various subclasses of pronouns are as follows:
PERSONAL PRONOUNS refer to the speaker, the hearer, or the person or thing spoken
about, e.g., I, me, you, and him.
REFLEXIVE PRONOUNS are object pronouns which refer back to the subject of the clause
in which they occur, e.g., myself, itself, and themselves.
INTENSIVE PRONOUNS emphasize a preceding noun or pronoun, e.g. myself, itself, and
themselves. The form of reflexive and intensive pronouns is the same. In I myself saw myself
in the mirror, the first myself is intensive, the second is reflexive.
INDEFINITE PRONOUNS refer to indefinite persons and things, e.g., someone, anybody,
and everything.
POSSESSIVE PRONOUNS are those which take the place of possessive nouns, e.g., her, his,
and its.

372
DEMONSTRATIVE PRONOUNS are used to designate particular persons or things, e.g.,
this, that, and these.
RELATIVE PRONOUNS are those which introduce relative clauses, e.g., who, whom, and
which.
INTERROGATIVE PRONOUNS are used in place of nouns in questions, e.g., who, which
and what.
RECIPROCAL PRONOUNS express mutual actions or relationships, e.g., each other and
one another.
VERB: A verb is a word which serves to express an action, an occurrence, or a state of being, e.g.,
shot in The inspector shot the bandit, blow up in The school blew up, be in That girl is clever. Verbs
are identifiable by a variety of inflectional distinctions, including tense, person, number, and voice.
The subclasses of verbs are as follows:
TRANSITIVE VERBS are those which take objects, e.g., break in John broke the vase, and
destroy in John destroyed the vase.
INTRANSITIVE VERBS are those which do not take objects, e.g., break in The vase broke,
and disappear in The vase disappeared.
COPULATIVE VERBS are those which link the subject with a nominal or adjectival
complement, e.g., appear in John appears healthy, and be in John was the king for two years.
ADJECTIVE: An adjective is a word which qualifies, limits, or otherwise modifies the meaning of
a noun, e.g. pretty, little, and white. In English, adjectives occur in two principal positions, either
before the nouns they modify, e.g., the rebellious student, or after verbs like be, e.g., the student is
rebellious. The two positions are called, respectively, the ATTRIBUTIVE POSITION and the
PREDICATE POSITION. Adjectives are inflected to show one of three degrees of comparison:
the POSITIVE DEGREE, the COMPARATIVE DEGREE, and the SUPERLATIVE DEGREE.
The comparative degree is used when two items are being compared; the superlative is used when
three or more items are being compared. For example: John is a good boy (positive degree); John
is a better boy than Bill (comparative degree); John is the best boy in the class (superlative degree).
ADVERB: An adverb is a word which modifies a verb or an adjective, e.g., quickly in John ran
quickly, and unbelievably in John is unbelievably clever. Like adjectives, adverbs are compared in
the positive, comparative, and superlative degree, e.g., Mary cleans her house thoroughly all the time
(positive), Mary cleans her house more thoroughly on weekdays than on weekends (comparative),
and Mary cleans her house (the) most thoroughly before her mother–in–law comes to visit
(superlative).

CONJUNCTION: A conjunction is a word which connects one grammatical category or
grammatical construction with another. There are two basic types of conjunctions:
COORDINATING and SUBORDINATING. A coordinating conjunction connects elements in
such a way as to indicate approximately equal order or rank amongst the elements, e.g., John and
Harry, on the table or on the desk, and Sit down and eat your spinach. A subordinating conjunction
on the other hand, connects elements in such a way as to place one in a lower order or rank with
respect to the other, e.g., Sit down, while you are eating your spinach.
PREPOSITION: A preposition is a word which introduces a phrase ending in a noun, or the
equivalent of a noun, and which typically serves as an adverbial or adjectival modifier, e.g., John ran
into the yard, and John is in the yard.
SYNTACTIC CONSTRUCTIONS
SENTENCE: A sentence is a structured expression consisting minimally of two main constituents,
a SUBJECT, which specifies the topic of the expression, and a PREDICATE, which specifies what
is asserted of the topic. There are four principal types of sentences: DECLARATIVE, those which
make statements, e.g., John wrote Helen a letter; INTERROGATIVE, those which ask questions,
e.g., Did John write Helen a letter; IMPERATIVE, those which give a command or make a request,
e.g., Please, write Helen a letter; and EXCLAMATORY, those which express strong or sudden
feeling, e.g., What a beautiful letter John wrote Helen!
CLAUSE: Like sentences, clauses are also expressions which contain a subject and a predicate. All
clauses must contain a tensed verb. Clauses are of two types: INDEPENDENT and DEPENDENT
(or SUBORDINATE). An independent clause is one which can stand alone; hence, a sentence is an
independent clause. A dependent clause is one which occurs in conjunction with an independent
clause. When a sentence consists of only one independent clause, it is called a SIMPLE sentence;
when it consists of two or more independent clauses and no dependent clauses, it is called a
COMPOUND sentence; and when it contains one or more dependent clauses, it is called a
COMPLEX sentence.
PHRASE: A phrase is any group of words which serves as a unit within a clause. Phrases are
usually named for their most distinctive element, called the HEAD, e.g., noun phrase, verb phrase,
prepositional phrase, and so on.
373

374
MORPHOLOGY
Most grammars divide the study of sentence structure into two areas, MORPHOLOGY, which deals
with the internal structure of words, and SYNTAX, which deals with the way words are combined
to form sentences. In morphology, the basic units under investigation are the morphemes of a
language. By definition, a MORPHEME is a minimal unit of meaning, that is, a meaningful
sequence of sounds which is not divisible into smaller meaningful units, e.g., ball. It is important not
to equate morphemes with either words or syllables; in both theory and practice, there is no necessary
coincidence between a morpheme and either of these units. For example, the English noun hands is
one word consisting of one syllable, but having two morphemes: hand, which defines a specific body
part, and s which indicates plural. Conversely, finger has two syllables but is only one morpheme.
One cannot divide the word finger into smaller meaningful units: the sequences fing, inger, ger, etc.
are all without content. Even the sequence fin, which does have meaning in English, is not part of
the morphological structure of finger, because its meaning is unrelated to the meaning of the whole
word.
Morphemes are variously classified by linguists as BOUND or FREE and as ROOTS or AFFIXES.
A BOUND MORPHEME is one which cannot occur as an independent word, e.g., the un– meaning
‘not’ in untrustworthy, unappreciative, unnecessary, etc. A FREE MORPHEME is one which can
stand alone, e.g., the morpheme trust. In English, bound morphemes constitute a relatively small
class of items, while the number of free morphemes is very large. In other languages, the situation
is often quite different. For example, Classical Greek has very few free morphemes, but a great
number of bound ones. In Vietnamese, almost all of the morphemes are classified as free.
A ROOT is a morpheme which constitutes the core element of a word or group of words of closely
related meaning. For example, the morpheme trust serves as the root in all of the following words:
trusted, untrustworthy, mistrustful, entrust, and so on. An AFFIX is a bound morpheme added to
the beginning of a word (in which case it is called a PREFIX) or to the end of a word (in which case
it is called a SUFFIX). Some common affixes in English are un–, mis–, con–, ex–, –ly, –ing, –ness,
and –er. That part of a word to which affixes are added is called a STEM. In English, the stem and
the root of a word are usually the same string; in other languages, they differ.
Affixes that are added to roots to indicate grammatical relationships are known as INFLECTIONAL
ENDINGS, and the process of combining these endings with roots is called INFLECTION. The
–s used to indicate the plural of English nouns is an example of an inflectional ending, as is the –ed
used to indicate past tense. Comparatively speaking, English is very poor in inflectional endings.
Other languages have highly developed inflectional systems. For example, in Latin, Russian, and
Finnish, the form of a noun changes depending on whether it is used as the subject of a verb or as the
object. In English, nouns are not inflected for this particular contrast, although some pronouns are,
e.g., I/me, she/her, he/him, we/us, they/them, who/whom. When applied to SUBSTANTIVES (nouns,
pronouns, and adjectives) inflection is referred to as DECLENSION; when applied to verbs, it is
called CONJUGATION.

SOME INFLECTIONAL CATEGORIES
PERSON: the inflectional category in pronouns and verbs that refers to the three–way distinction
between the speaker (FIRST PERSON), e.g., I am tall, the hearer (SECOND PERSON), e.g., You
are tall, and someone or something else (THIRD PERSON), e.g., He (she, it) is tall.
NUMBER: an inflectional category in substantives and verbs that refers to the distinction between
singular, e.g., the house IS big, and plural, e.g., the houses ARE big.
TENSE: an inflectional category in verbs which indicates distinctions in the TIME (PRESENT,
PAST, FUTURE) and the ASPECT (PROGRESSIVE, PERFECTIVE) of an action or state. For
example, the verb phrase is looking is in the present progressive tense, that is, it indicates an action
which is going on in the present; the verb phrase has looked is in the present perfective tense, that
is, it indicates an action that was completed in the past. Strictly speaking, tense is only partially an
inflectional category in English, since it is signaled both by independent words, e.g., forms of the
verbs be and have and by endings, e.g., –ing and –ed.
VOICE: an inflectional category in verbs that refers to the distinction between ACTIVE and
PASSIVE. An active verb is one whose subject is viewed as performing the action it defines, e.g.,
John killed the tigers; a passive verb is one whose subject is viewed as undergoing the action it
defines, e.g., the tigers were killed by John.
MODE/MOOD: an inflectional category in verbs that refers to the distinction between
INDICATIVE, CONDITIONAL, and IMPERATIVE. The indicative mode expresses ordinary
statements (he stopped); the conditional mode expresses conditions (he would stop); the imperative
mode expresses commands (stop!).
CASE: an inflectional category which indicates the relationship of substantives to other words in a
sentence. In English, the opposition he/him/his is a case opposition: he is used to indicate the subject
(NOMINATIVE CASE), him to indicate the object (OBJECTIVE or ACCUSATIVE CASE), and
his to indicate the possessor (POSSESSIVE or GENITIVE).
GENDER: an inflectional category in substantives that refers to the tripartite distinction between
MASCULINE, FEMININE, and NEUTER. In English, only the third person singular pronouns
show gender distinctions, e.g., he, she, and it; moreover, the distinction is based on sex. In other
languages, e.g., German, gender is a grammatical category of substantives and bears no relationship
to sex.
375

376
SUPPLEMENT THREE (PART I): TYPOLOGICAL CLASSIFICATION
ANALYTIC LANGUAGES (ISOLATING LANGUAGES): words consist of single
morphemes; most words consist only of a root. Mandarin Chinese, Vietnamese, Cantonese,
Cambodian. Examples in Mandarin adapted from Norman J., Chinese, Cambridge, 1988:
1. Ta chī fàn le.
he eat food past
‘He ate the food.’
2. Ta chī le fàn.
he eat past food
‘He ate the food.’
3. Júzi w4 chī le.
orange I eat past
‘I ate the orange.’
AGGLUTINATING LANGUAGES: words consist of a stem and one or more clearly
identifiable affixes. Finnish, Hungarian, Estonian, Swahili, Turkish. Examples in Estonian
adapted from Oinas, F., Basic Course in Estonian, Indiana University, 1966:
1. Ta on kohvikus.
he is coffee–house–in
‘He is in the coffee house.’
2. Lähme kohvikusse.
go–we coffee–house–into
‘Let us go into the coffee house.’
3. Ma tulen uuest kohvikust.
I come new–from coffee–house–from
‘I am coming out of the new coffee house.’
4. Cases:
nominative kohvik the coffee house
genitive kohviku of the coffee house
partitive kohvikut the coffee house
illative kohvikusse into the coffee house
inessive kohvikus in the coffee house
elative kohvikust out of the coffee house
allative kohvikule to the coffee house
adessive kohvikul by the coffee house
ablative kohvikult from the coffee house
translative kohvikuks for the coffee house
essive kohvikuna as the coffee house
terminative kohvikuni up to the coffee house
comitative kohvikuga with the coffee house
abessive kohvikuta without the coffee house

SUPPLEMENT THREE (PART II): TYPOLOGICAL CLASSIFICATION
INFLECTIONAL LANGUAGES (FUSIONAL LANGUAGES): words consist of stem and
affixes which often mark several grammatical categories simultaneously. Greek, Latin, Sanskrit,
Russian. Examples in Latin (NomSg=nominative singular, NomPl=nominative plural,
AccSg=accusative singular, AccPl=accusative plural, 3Sg=third person singular, 3Pl=third person
plural):
1. Custos fidelis consulem veterem ducit.
NomSg NomSg AccSg AccSg 3Sg
guard trusty consul old is leading
‘The trusty guard is leading the old consul.’
2. Custodes fideles consules veteres ducunt.
NomPl/AccPl NomPl/AccPl NomPl/AccPl NomPl/AccPl 3Pl
‘The trusty guards are leading the old consuls.’
‘The old guards are leading the trusty consuls.’
‘The trusty consuls are leading the old guards.’
‘The old consuls are leading the trusty guards.’
POLYSYNTHETIC LANGUAGES: words consist of long strings of stems and affixes, which
may translate as an entire English sentence. American Indian languages. Examples in Ayacucho
Quechua adapted from Parker, G., Ayacucho Quechua Grammar and Dictionary, Mouton, 1969:
Verbs can be inflected for both actor and object in different persons and number.
1. riku–yki ‘I see you.’
2. riku–yki–čik ‘I see you all.’
3. riku–yki–ku ‘We see you.’
riku: ‘see’
yki: first person singular actor and second person singular object
čik: marks object as plural
ku: marks actor as plural
4. riku–wanki ‘You see me.’
5. riku–wanki–čik ‘You all see me.’
6. riku–wanki–ku ‘We see you.’
wanki: second person singular actor and first person singular object
čik: marks actor as plural
ku: marks object as plural
377

378
GERMANIC:
SUPPLEMENT FOUR (PART I): HISTORICAL CLASSIFICATION
LEXICON: INDO–EUROPEAN
love (n) mother father brother break (v) three five hundred heart
ENGLISH: love mother father brother break three five hundred heart
MID ENGLISH: love moder fader brother breke thre fīf hund(red) herte
OLD ENGLISH: lufu mōdor fœder brōþor brecan þrī fīf(e) hund heorte
GERMAN: liebe mutter vater bruder brechen drei fünf hundert herz
DUTCH: liefde moeder vader broeder breken drie vijf honderd hart
GREEK:
ANCIENT GREEK: erōs mētēr patēr adelphos hrēgnūmi treis pente hekaton kardia
CELTIC:
IRISH: searc māthair athair brāthir brisim trī cóic cēt croidhe
WELSCH: serch mam tad brawd torri tri pump cant calon
ITALIC:
LATIN: amor māter pater frater frangere trēs quinque centum cor
rumpere
ROMANCE:
SPANISH: amor madre padre hermano romper tres cinco ciento corazón
PORTUGUESE: amor mãe pai irmão quebrar três cinco cem coração
ITALIAN: amore madre padre fratello rompere tre cinque cento cuore
FRENCH: amour mère père frère rompre trois cinq cent coeur
RUMANIAN: amor mamă tată frate frînge trei cinci sută inimă
rupe
BALTO–SLAVIC:
RUSSIAN: ljubov' mat' otec brat lomat' tri pyat' sto serdce
CHURCH SLAVIC: ljuby mati ot0c0 brat lomiti tr0je petь sßto srßd0ce
INDO–IRANIAN:
SANSKRIT: kāma– mātar– pitar– bhrātar– bhañj– trayas páñca çatam hr;d–
HINDI: muhabbat mátá báp bháí torná tīn pãc sau dil
AVESTAN: kan– mātar– pitar– brātar– sčand– ›rāyō panča satcm zcrcd–

ALTAIC:
SUPPLEMENT FOUR (PART II): HISTORICAL CLASSIFICATION
LEXICON: NON–INDO–EUROPEAN
love (n) mother father brother break (v) three five hundred heart
JAPANESE: ai haha chichi kyôdai kowareru san go hyaku shinzô
SINO–TIBETAN:
CHINESE: ai mßchīn fùchin syūngdì nùngpwò sān wß băi syīn
URALIC:
FINNISH: rakkaus äiti isä veli rikkoa kolme viisi sata sydän
AUSTRONESIAN:
SAMOAN: alofa tinā tamā uso 1 talepe 4 tolu lima selau fatu
HAWAIIAN: aloha makuahine makua kua'ana 2 wāhi 4 kolu lima hanele pu'uwai
AFRO–ASIATIC:
CHADIC:
HAUSA: so 5 ūwa ōba wā 3 kária úku biál darí zúts-ia
SEMITIC:
ARABIC: h;ubb §imm §ab(u) §ax yiksir talaate xamse miyye §alb
UTO–AZTECAN:
COMANCHE kamaku–ru– 5 pia ahpu–§ samohpu–§ tahparu– pihihtu– makekitu–§ pihi
NOTES:
1 uso = brother of a man; tuagane = brother of a woman
2 kua'ana = older sibling of the same sex; kaina = younger sibling of the same sex; kuāne = sibling of a female;
etc.
3 wā = older male sibling; kāne = younger male sibling
4 There are many verbs for break depending on the nature of the thing broken, e.g., hard things, brittle things,
things that break easily, bones, etc.
5 There is no noun meaning love; these are verb forms.
379

380
SUPPLEMENT FIVE (PART I): LANGUAGES OF AFRICA AND THE MID EAST*
Family Branch Group Representative Languages Principal Locations Number of Speakers
Afro–Asiatic
(241)
Chadic Hausa N Nigeria; Niger; Cameroon 39,000,000
Semitic Hebrew Isreal 5,000,000
Tigrinya S Eritrea; N Ethiopia 4,000,000
Amharic Ethopia 20,000,000
Arabic Egypt; Iraq; Syria; Morocco 230,000,000
Hamitic Berber Tamazight W Algeria; N Morocco 3,000,000
Kabyle W Kabylia; N Algeria 3,000,000
Shilha W Algeria; S Morocco 3,000,000
Cushitic Somali Somalia; Kenya; Ethiopia; Djibouti 4,000,000
Galla (Oromo) W Ethopia; N Kenya 9,000,000
Omotic Wolaytta SE Ethopia 2,000,000
Khoisan (31) Nama (Hottentot) Namibia 146,000
Niger–Congo
(1032)
Nilo–Sarahan
(138)
Sandawe Tanzania 70,000
Mande Mende Sierra Leone 2,000,000
West Atlantic Fula Senegal; Nigeria; Cameroon 13,000,000
Wolof Senegal 7,000,000
Kwa Akan Ghana; Côte d’Ivoire 7,000,000
Yoruba SW Nigeria; Benin 20,000,000
Igbo (Ibo) Nigeria 17,000,000
Ewe S Togo; SE Ghana 3,000,000
Benue–Congo Efik SE Nigeria 6,000,000
Tiv SE Nigeria; Cameroon 2.000,000
Bantu Swahili Tanzania; Kenya; Zaire; Uganda 49,000,000
Zulu S Africa; Lesotho 9,000,000
Xhosa S Africa 8,000,000
Gur (Voltaic) More Burkina Faso 4,000,000
Adamawa–Ubangian Sango Central African Republic 4,000,000
Nilo–Hamitic Maasai Kenya; Tanzania 689,000
Nilotic Luo Kenya; Nyanza; Tanzania 4,000,000
* There is much scholarly debate on the classification of languages. The various subdivisions in Figure Four I, II, III, IV, and V reflect only
major distinctions generally recognized. The approximate number of languages in each family is in parentheses (data from Ruhlen 1987).
Statistics for individual languages with more than 1 million speakers are from The World Almanac and Book of Facts, 1997, Mahwah, NJ:
World Almanac Books. Statistics for individual languages with less than 1 million speakers are from Grimes 1992.

SUPPLEMENT FIVE (PART II): AMERICAN INDIAN LANGUAGES
Family Branch Group Representative Languages Principal Locations Number of Speakers
Algonquian (15) Cree Canada; USA 46,700
Andean–Equatorial
(163)
Ojibwa Canada; USA 51,000
Arapaho Wyoming; W Oklahoma 1,500
Blackfoot S Alberta; Montana 9,000
Andean Quechua Peru; Equador; Bolivia; Argentina 8,000,000
Aymara Bolivia; Peru 2,000,000
Equatorial Guarani Paraguay 4,000,000
Eskimo–Aleut (9) Eskimo Inuit Canadian Arctic; Greenland 68,500
Yupik Alaska; Siberia 17,000
Aleut Aleut Aleutian Islands 700
Ge–Pano–Carib (117) Carib Northern South America 100,000
Hokan (28) Yuman Diegueño Baja California less than 400
Tequistlatecan Tequistlatec Mexico 5,000
Iroquoian (7) Mohawk N New York 3,000
Cherokee E Oklahoma; W North Carolina 22,500
Caddoan (4) Pawnee N Central Oklahoma 200
Na–Dené (34) Tlingit SE Alaska 2,000
Athapaskan Navajo New Mexico; Arizona; Utah 130,000
Apache Oklahoma; New Mexico; Arizona 14,300
Chipewyan N Alberta; Saskatchewan 4,000
Oto–Manguean (17) Otomian Zapotec S Mexico 500,000
Mixtecan Mixtec S Mexico 250,000
Penutian (68) Nez Perce N Idaho 500
Mayan Quiche Guatemala 885,000
Mosan (27) Squamish SW British Columbia 20
Siouan–Yuchi (11) Dakota (Sioux) S Manitoba; Nebraska; Minnesota 19,000
Crow S Montana 5,500
Winnebago Wisconsin; E Nebraska 1,500
Uto–Aztecan (25) Hopi NE Arizona 5,000
Shoshonean Comanche W Oklahoma 500
Takic Luiseño S California 100
Sonoran Huichol Mexico 12,500
Aztecan Nahuatl (Aztec) Southern Mexico 1,000,000
381

382
SUPPLEMENT FIVE (PART III): INDO–EUROPEAN LANGUAGES
Family Branch Group Representative Languages Principal Locations Number of Speakers
Indo–European
(144)
Albanian Albanian Albania; Yugoslavia 5,000,000
Anatolian Hittite Ancient Asia Minor Extinct
Armenian Armenian Armenia; Iran; Syria; Lebanon 5,000,000
Balto–Slavic Baltic Lithuanian Lithuania 3,000,000
Latvian Latvia 2,000,000
Slavic Russian Russia 280,000,000
Serbo–Croatian Serbia; Croatia; Bosnia 20,000,000
Czech Czechoslovakia 12,000,000
Polish Poland 44,000,000
Celtic Brythonic Welsh Wales 580,000
Goidelic Irish Ireland 120,000
Germanic East Gothic Eastern Europe Extinct
North Danish Denmark 5,000,000
Norwegian Norway 5,000,000
Swedish Sweden; Finland 9,000,000
West Dutch Netherlands; Belgium 21,000,000
English England, America 487,000,000
German Germany 124,000,000
Greek Greek Greece 12,000,000
Indo–Iranian Indic Sanskrit Ancient India Extinct
Bengali Bangladesh; India 204,000,000
Hindi India 457,000,000
Urdu Pakistan; India 104,000,000
Punjabi India; Pakistan 95,000,000
Iranian Pashto Afghanistan; Pakistan 21,000,000
Persian Iran; Afghanistan 35,000,000
Italic Latin Roman Empire Extinct
Romance French France 125,000,000
Romanian Romania; Moldova 26,000,000
Italian Italy 62,000,000
Portuguese Portugal; Brazil 186,000,000
Spanish Spain; Central and South America 401,000,000

SUPPLEMENT FIVE (PART IV): LANGUAGES OF EAST ASIA & THE PACIFIC
Family Branch Group Representative Languages Principal Locations Number of Speakers
Australian (170) Pama–Nyungan Walbiri Australia 3,000
Austro–Asiatic
(155)
Dyirbal Australia 40 to 50
Mon–Khmer Khmer Kampuchea; Vietnam; Thailand 9,000,000
Vietnamese Vietnam 63,000,000
Nung Vietnam; China 2,000,000
Munda Santali E India; Nepal 5,000,000
Nicobarese Central Nicobarese Nicobar Islands 5,000
Austronesian (959) Formosan Atayal Taiwan 41,000
Western Malay–Indonesian Indonesia; Thailand 164,000,000
Toba Batak Indonesia 4,000,000
Javanese Java 64,000,000
Sundanese Java 26,000,000
Madurese Madura 10,000,000
Tagalog Philippines 56,000,000
Cebuano Philippines 12,000,000
Malagasy Madagascar 12,000,000
Oceanic Samoan Samoa 328,000
Maori New Zealand 100,000
Tahitian Tahiti 125,000
Hawaiian Hawaii 2,000
Indo–Pacific (731) Motu Papua New Guinea 15,000
Sino–Tibetan (258) Tibeto–Burman Tibetan SW China; N India; Nepal 5,000,000
Burmese Mayanmar (Burma) 31,000,000
Newari Nepal 500,000
Sinitic Mandarin Mandarin China 999,000,000
Wu Wu Chiang Jiang delta, China 65,000,000
Min Taiwanese Taiwan 50,000,000
Yue Cantonese Guangdong province, China 70,000,000
Hakka Hakka SE China 34,000,000
Tai (57) Thai Thailand 52,000,000
Shan E Mayanmar (Burma) 3,000,000
Lao Thailand; Laos 4,000,000
383

384
SUPPLEMENT FIVE (PART V): OTHER LANGUAGES OF EUROPE AND ASIA
Family Branch Group Representative Languages Principal Locations Number of Speakers
Altaic (63) Turkic Turkish Turkey 61,000,000
Azerbaijani Azerbaijan; Iran; Turkey 15,000,000
Tartar Russia 8,000,000
Uzbek Uzbekistan 14,000,000
Mongolian Mongol Mongolian People’s Republic 6,000,000
Manchu–Tungus Evenki Russia 24,000
Manchu Northeast China less than 1,000
Japanese Japanese Japan 126,000,000
Korean Korean North Korea; South Korea 76,000,000
Caucasian (38) Kartvelian Georgian Georgia; Azerbaijan; Turkey 4,000,000
Svan Georgia 35,000
Northwest Kabardian Northwest Caucasus; Russia 370,00
Northeast Dagestanian Avar Northeast Caucasus 483,000
Lezgian (Lezgi) Russia; Azerbaijan 383,000
Nakh Chechen Chechnia 945,000
Dravidian (28) Telugu Southeast India 75,000,000
Tamil India; Sri Lanka 73,000,000
Kannada S India 45,000,000
Malayalam S India 36,000,000
Paleosiberian (5) Chukchi Northeast Siberia 10,900
Uralic (24) Finno–Ugric Finnic Finnish Finland 6,000,000
Estonian Estonia 1,000,000
Ugric Hungarian Hungary 14,000,000
Samoyedic Nenets Arctic Russia 27,000

SUPPLEMENT SIX: LANGUAGES AND LANGUAGE GROUPS
(Roman numerals in square brackets refer to the Parts in Supplement Five)
Adamawa–Ubangian [I]
Afro–Asiatic [I]
Akan [I]
Albanian [III]
Aleut [II]
Algonquian [II]
Altaic [V]
Amharic [I]
Anatolian [III]
Andean [II]
Andean–Equatorial [II]
Apache [II]
Arabic [I]
Arapaho [II]
Armenian [III]
Atayal [IV]
Athapaskan [II]
Australian [IV]
Austronesian [IV]
Austro–Asiatic [IV]
Avar [V]
Aymara [II]
Azerbaijani [V]
Aztec [II]
Aztecan [II]
Baltic [III]
Balto–Slavic [III]
Bantu [I]
Bengali [III]
Benue–Congo [I]
Berber [I]
Blackfoot [II]
Brythonic [III]
Burmese [IV]
Caddoan [II]
Cantonese [IV]
Carib [II]
Caucasian [V]
Cebuano [IV]
Celtic [III]
Central Nicobarese [IV]
Chadic [I]
Chechen [V]
Cherokee [II]
Chipewyan [II]
Chukchi [V]
Comanche [II]
Cree [II]
Crow [II]
Cushitic [I]
Czech [III]
Dagestanian [V]
Dakota [II]
Danish [III]
Diegueño [II]
Dravidian [V]
Dutch [III]
Dyirbal [IV]
East Germanic [III]
Efik [I]
English [III]
Equatorial [II]
Eskimo [II]
Eskimo–Aleut [II]
Estonian [V]
Evenki [V]
Ewe [I]
Finnic [V]
Finnish [V]
Finno–Ugric [V]
Formosan [IV]
French [III]
Fula [I]
Galla [I]
Georgian [V]
German [III]
Germanic [III]
Ge–Pano–Carib [II]
Goidelic [III]
Gothic [III]
Greek [III]
Guarani [II]
Gur [I]
Hakka [IV]
Hamitic [I]
Hausa [I]
Hawaiian [IV]
Hebrew [I]
Hindi [III]
Hittite [III]
Hokan [II]
Hopi [II]
Hottentot [I]
Huichol [II]
Hungarian [V]
Ibo [I]
Igbo [I]
Indic [III]
Indo–European [III]
Indo–Iranian [III]
Indo–Pacific [IV]
Inuit [II]
Iranian [III]
Irish [III]
Iroquoian [II]
Italian [III]
Italic [III]
Japanese [V]
Javanese [IV]
Kabardian [V]
Kabyle [I]
Kannada [V]
Kartvelian [V]
Khmer [IV]
Khoisan [I]
Korean [V]
Kwa [I]
Lao [IV]
Latin [III]
Latvian [III]
Lezgi [V]
Lezgian [V]
Lithuanian [III]
Luiseño [II]
Luo [I]
Maasai [I]
385

386
Madurese [IV]
Malagasy [IV]
Malayalam [V]
Malay–Indonesian [IV]
Manchu [V]
Manchu–Tungus [V]
Mandarin [IV]
Mande [I]
Maori [IV]
Mayan [II]
Mende [I]
Min [IV]
Mixtec [II]
Mixtecan [II]
Mohawk [II]
Mongol [V]
Mongolian [V]
Mon–Khmer [IV]
More [I]
Mosan [II]
Motu [IV]
Munda [IV]
Nahuatl [II]
Nakh [V]
Nama [I]
Navajo [II]
Na–Dené [II]
Nenets [V]
Newari [IV]
Nez Perce [II]
Nicobarese [IV]
Niger–Congo [I]
Nilotic [I]
Nilo–Hamitic [I]
Nilo–Sarahan [I]
North Germanic [III]
Northeast Caucasian [V]
Northwest Caucasian [V]
Norwegian [III]
Nung [IV]
Oceanic [IV]
Ojibwa [II]
Omotic [I]
Oromo [I]
Otomian [II]
Oto–Manguean [II]
Paleosiberian [V]
Pama–Nyungan [IV]
Pashto [III]
Pawnee [II]
Penutian [II]
Persian [III]
Polish [III]
Portuguese [III]
Punjabi [III]
Quechua [II]
Quiche [II]
Romance [III]
Romanian [III]
Russian [III]
Samoan [IV]
Samoyedic [V]
Sandawe [I]
Sango [I]
Sanskrit [III]
Santali [IV]
Semitic [I]
Serbo–Croatian [III]
Shan [IV]
Shilha [I]
Shoshonean [II]
Sinitic [IV]
Sino–Tibetan [IV]
Siouan–Yuchi [II]
Sioux [II]
Slavic [III]
Somali [I]
Sonoran [II]
Spanish [III]
Squamish [II]
Sundanese [IV]
Svan [V]
Swahili [I]
Swedish [III]
Tagalog [IV]
Tahitian [IV]
Tai [IV]
Taiwanese [IV]
Takic [II]
Tamazight [I]
Tamil [V]
Tartar [V]
Telugu [V]
Tequistlatec [II]
Tequistlatecan [II]
Thai [IV]
Tibetan [IV]
Tibeto–Burman [IV]
Tigrinya [I]
Tiv [I]
Tlingit [II]
Toba Batak [IV]
Turkic [V]
Turkish [V]
Ugric [V]
Uralic [V]
Urdu [III]
Uto–Aztecan [II]
Uzbek [V]
Vietnamese [IV]
Voltaic [I]
Walbiri [IV]
Welsh [III]
West Germanic [III]
West Atlantic [I]
Western Austronesian [IV]
Winnebago [II]
Wolaytta [I]
Wolof [I]
Wu [IV]
Xhosa [I]
Yoruba [I]
Yue [IV]
Yuman [II]
Yupik [II]
Zapotec [II]
Zulu [I]

SUPPLEMENT SEVEN: SOME ALPHABETS
ROMAN GREEK CYRILLIC HEBREW ARABIC
A a Α α alpha А а a א 0aleph,0alef ا \ 0alif
B b Β β beta Б б be ב bēth ب ` bā
C c Γ γ gamma В в ve ג gimel ت d tā
D d Δ δ delta Г г ge ד dāleth ث h thā
E e Ε ε epsilon Д д de ה hē ج l jim
F f Ζ ζ zeta Е е je ו vav, waw ح p hā
G g Η η eta Ё ё jo ז zayin خ t khā
H h Θ θ theta Ж ж že ח h;eth د v dāi
I i Ι ι iota З з ze ט t;eth ذ x dhāi
J j Κ κ kappa И и i י yod, yodh ر z rā
K k Λ λ lambda Й й í krátkcjc ך, כ kāph ز | zāy
L l Μ μ mu К к ka ל lamedh س sin
M m Ν ν nu Л л el ם, מ mēm ش … shin
N n Ξ ξ xi М м em ןנ nūn ص ‰ s;ād
O o Ο ο omicron Н н en ס samekh ض d;ād
P p Π π pi О о o ע 0ayin ط ‘ t;ā
Q q Ρ ρ rho П п pe ף, פ pē ظ • z;ā
R r Σ σ, ς sigma Р р er ץ, צ sade, s;adhe ع 0yn
S s Τ τ tau С с es ק qōph غ ghayn
T t Υ υ upsilon Т т te ר rēsh ف ¡ fā
U u Φ φ phi У у u ש sin ق ¥ qāf
V v Χ χ chi Ф ф ef ש shin ك © kaf
W w Ψ ψ psi Х х xa ת tāv, tāw ل − lam
X x Ω ω omega Ц ц ce م ± mim
Y y Ч ч čB a ن μ nun
Z z Ш ш ša ﻩ ¹ hā
387
Щ щ šB čB a و ½ wāw
Ъ ъ tvB órdi–j znák ي Á yā
Ы ы i–
Ь ь mB áxkcj znák
Э э e
Ю ю ju
Я я ja

388

SUPPLEMENT EIGHT: SAMPLE PARSES FROM THE LANGTECH PARSER
SAMPLE PARSES ILLUSTRATING BASIC SENTENCE PATTERNS
(1) INTRANSITIVE:
the senator resigned
The clause “the senator resigned” is a statement.
The phrase “the senator” is a noun phrase and the subject of the verb “resigned.”
The word “resigned” is a past tense verb.
The word “resigned” is the main verb of “the senator resigned.”
The word “senator” is a singular count noun.
The word “senator” is the head of the noun phrase “the senator.”
The word “the” is a determiner specifying the noun “senator.”
389

390
(2) TRANSITIVE (DIRECT OBJECT COMPLEMENT ON V1):
the senator admired the president
The clause “the senator admired the president” is a statement.
The phrase “the president” is a noun phrase and the direct object of the verb “admired.”
The phrase “the senator” is a noun phrase and the subject of the verb “admired.”
The word “admired” is a past tense verb.
The word “admired” is the main verb of “the senator admired the president.”
The word “president” is a singular count noun.
The word “president” is the head of the noun phrase “the president.”
The word “senator” is a singular count noun.
The word “senator” is the head of the noun phrase “the senator.”
The word “the” is a determiner specifying the noun “president.”
The word “the” is a determiner specifying the noun “senator.”

(3) COPULATIVE (PREDICATE COMPLEMENT ON V2):
the senator became the president
The clause “the senator became the president” is a statement.
The phrase “the president” is a noun phrase describing the subject “senator.”
The phrase “the senator” is a noun phrase and the subject of the verb “became.”
The word “became” is a past tense verb.
The word “became” is the main verb of “the senator became the president.”
The word “president” is a singular count noun.
The word “president” is the head of the noun phrase “the president.”
The word “senator” is a singular count noun.
The word “senator” is the head of the noun phrase “the senator.”
The word “the” is a determiner specifying the noun “senator.”
The word “the” is a determiner specifying the noun “president.”
391

392
(4) COPULATIVE (PREDICATE COMPLEMENT ON V2):
the senator turned red
The clause “the senator turned red” is a statement.
The phrase “red” is an adjective phrase describing the subject “senator.”
The phrase “the senator” is a noun phrase and the subject of the verb “turned.”
The word “red” is an adjective.
The word “senator” is a singular count noun.
The word “senator” is the head of the noun phrase “the senator.”
The word “the” is a determiner specifying the noun “senator.”
The word “turned” is a past tense verb.
The word “turned” is the main verb of “the senator turned red.”

(5) INTRANSITIVE WITH A MANNER ADVERB:
the senator resigned angrily
The clause “the senator resigned angrily” is a statement.
The phrase “angrily” is an adverb phrase modifying the verb “resigned.”
The phrase “the senator” is a noun phrase and the subject of the verb “resigned.”
The word “angrily” is a manner adverb.
The word “resigned” is a past tense verb.
The word “resigned” is the main verb of “the senator resigned angrily.”
The word “senator” is a singular count noun.
The word “senator” is the head of the noun phrase “the senator.”
The word “the” is a determiner specifying the noun “senator.”
393

394
(6) INTRANSITIVE WITH PREDICATE ADJECTIVE:
the senator resigned angry
The clause “the senator resigned angry” is a statement.
The phrase “angry” is a predicate adjective referring back to the subject “senator.”
The phrase “the senator” is a noun phrase and the subject of the verb “resigned.”
The word “angry” is an adjective.
The word “resigned” is a past tense verb.
The word “resigned” is the main verb of “the senator resigned angry.”
The word “senator” is a singular count noun.
The word “senator” is the head of the noun phrase “the senator.”
The word “the” is a determiner specifying the noun “senator.”

(7) TRANSITIVE (DIRECT OBJECT + MANNER ADVERB):
fred left the country quickly
The clause “fred left the country quickly” is a statement.
The phrase “fred” is a noun phrase and the subject of the verb “left.”
The phrase “quickly” is an adverb phrase modifying the verb “left.”
The phrase “the country” is a noun phrase and the direct object of the verb “left.”
The word “country” is a singular count noun.
The word “country” is the head of the noun phrase “the country.”
The word “fred” is a proper noun.
The word “left” is a past tense verb.
The word “left” is the main verb of “fred left the country quickly.”
The word “quickly” is a manner adverb.
The word “the” is a determiner specifying the noun “country.”
395

396
(8) TRANSITIVE (DIRECT OBJECT + MANNER ADVERB):
fred left the country happily
The clause “fred left the country happily” is a statement.
The phrase “fred” is a noun phrase and the subject of the verb “left.”
The phrase “happily” is an adverb phrase modifying the verb “left.”
The phrase “the country” is a noun phrase and the direct object of the verb “left.”
The word “country” is a singular count noun.
The word “country” is the head of the noun phrase “the country.”
The word “fred” is a proper noun.
The word “happily” is a manner adverb.
The word “left” is a past tense verb.
The word “left” is the main verb of “fred left the country happily.”
The word “the” is a determiner specifying the noun “country.”

(9) fred left the country happy
a. TRANSITIVE (DIRECT OBJECT + OBJECTIVE COMPLEMENT):
The clause “fred left the country happy” is a statement.
The phrase “fred” is a noun phrase and the subject of the verb “left.”
The phrase “happy” is an adjective phrase describing “country.”
The phrase “the country” is a noun phrase and the direct object of the verb “left.”
The word “country” is the head of the noun phrase “the country.”
The word “country” is a singular count noun.
The word “fred” is a proper noun.
The word “happy” is an adjective.
The word “left” is a past tense verb.
The word “left” is the main verb of “fred left the country happy.”
The word “the” is a determiner specifying the noun “country.”
397

398
b. TRANSITIVE (DIRECT OBJECT + SUBJECTIVE COMPLEMENT):
The clause “fred left the country happy” is a statement.
The phrase “fred” is a noun phrase and the subject of the verb “left.”
The phrase “happy” is a predicate adjective referring back to the subject “fred.”
The phrase “the country” is a noun phrase and the direct object of the verb “left.”
The word “country” is the head of the noun phrase “the country.”
The word “country” is a singular count noun.
The word “fred” is a proper noun.
The word “happy” is an adjective.
The word “left” is a past tense verb.
The word “left” is the main verb of “fred left the country happy.”
The word “the” is a determiner specifying the noun “country.”

(10) TRANSITIVE (DIRECT OBJECT + INDIRECT OBJECT):
she gave them to us
The clause “she gave them to us” is a statement.
The phrase “she” is a noun phrase and the subject of the verb “gave.”
The phrase “them” is a noun phrase and the direct object of the verb “gave.”
The phrase “to us” is a prepositional phrase and the indirect object of the verb “gave.”
The phrase “us” is a noun phrase and the object of the preposition “to.”
The word “gave” is a past tense verb.
The word “gave” is the main verb of “she gave them to us.”
The word “she” is a third person singular feminine subject pronoun.
The word “them” is a third person plural object pronoun.
The word “to” is a preposition introducing the prepositional phrase “to us"
The word “us” is a first person plural object pronoun.
399

400
(11) TRANSITIVE (DIRECT OBJECT + PREPOSITIONAL PHRASE):
she put them near us
The clause “she put them near us” is a statement.
The phrase “near us” is a prepositional phrase and a complement of the verb “put.”
The phrase “she” is a noun phrase and the subject of the verb “put.”
The phrase “them” is a noun phrase and the direct object of the verb “put.”
The phrase “us” is a noun phrase and the object of the preposition “near.”
The word “near” is a preposition introducing the prepositional phrase “near us"
The word “put” is a past tense verb.
The word “put” is the main verb of “she put them near us.”
The word “she” is a third person singular feminine subject pronoun.
The word “them” is a third person plural object pronoun.
The word “us” is a first person plural object pronoun.

(12) TRANSITIVE (DOUBLE OBJECT):
she gave us them
The clause “she gave us them” is a statement.
The phrase “she” is a noun phrase and the subject of the verb “gave.”
The phrase “them” is a noun phrase and the direct object of the verb “gave.”
The phrase “us” is a noun phrase and the indirect object of the verb “gave.”
The word “gave” is a past tense verb.
The word “gave” is the main verb of “she gave us them.”
The word “she” is a third person singular feminine subject pronoun.
The word “them” is a third person plural object pronoun.
The word “us” is a first person plural object pronoun.
401

402
(13) TRANSITIVE (DOUBLE OBJECT):
they elected a democrat the president
The clause “they elected a democrat the president” is a statement.
The phrase “a democrat” is a noun phrase and the direct object of the verb “elected.”
The phrase “the president” is a noun phrase and the object complement of the verb “elected.”
The phrase “they” is a noun phrase and the subject of the verb “elected.”
The word “a” is a determiner specifying the noun “democrat.”
The word “democrat” is a singular count noun.
The word “democrat” is the head of the noun phrase “a democrat.”
The word “elected” is a past tense verb.
The word “elected” is the main verb of “they elected a democrat the president.”
The word “president” is a singular count noun.
The word “president” is the head of the noun phrase “the president.”
The word “the” is a determiner specifying the noun “president.”
The word “they” is a third person plural subject pronoun.

(14) TRANSITIVE (SENTENTIAL COMPLEMENT; INDIRECT STATEMENT):
I know that she went
The clause “I know that she went” is a statement.
The clause “she went” is an indirect statement.
The clause “that she went” is a complement clause of the verb “know.”
The phrase “I” is a noun phrase and the subject of the verb “know.”
The phrase “she” is a noun phrase and the subject of the verb “went.”
The word “I” is a first person singular subject pronoun.
The word “know” is a present tense verb.
The word “know” is the main verb of “I know that she went.”
The word “she” is a third person singular feminine subject pronoun.
The word “that” is a complementizer specifying the complement clause “she went.”
The word “went” is a past tense verb.
The word “went” is the main verb of “that she went.”
403

404
(15) TRANSITIVE (SENTENTIAL COMPLEMENT; INDIRECT STATEMENT):
I know she went
The clause “I know she went” is a statement.
The clause “she went” is an indirect statement.
The phrase “I” is a noun phrase and the subject of the verb “know.”
The phrase “she went” is a complement clause of the verb “know.”
The word “I” is a first person singular subject pronoun.
The word “know” is a present tense verb.
The word “know” is the main verb of “I know she went.”
The word “Ø” is a complementizer specifying the complement clause “she went.”
The word “she” is a third person singular feminine subject pronoun.
The word “went” is a past tense verb.
The word “went” is the main verb of “she went.”

(16) TRANSITIVE (SENTENTIAL COMPLEMENT; INDIRECT STATEMENT):
I know that man is mortal
The clause “I know that man is mortal” is a statement.
The clause “man is mortal” is an indirect statement.
The clause “that man is mortal” is a complement clause of the verb “know.”
The phrase “I” is a noun phrase and the subject of the verb “know.”
The phrase “man” is a noun phrase and the subject of the verb “Is.”
The phrase “mortal” is a predicate adjective referring back to the subject “man.”
The word “I” is a first person singular subject pronoun.
The word “Is” is a third person singular present tense verb.
The word “Is” is the main verb of “man is mortal.”
The word “know” is a present tense verb.
The word “know” is the main verb of “I know that man is mortal.”
The word “man” is the head of the noun phrase “man.”
The word “mortal” is an adjective.
The word “Ø” is an abstract empty determiner specifying the noun “man.”
The word “that” is a complementizer specifying the complement clause “man is mortal.”
405

406
The clause “I know that man is mortal” is a statement.
The clause “that man is mortal” is a complement clause of the verb “know.”
The clause “that man is mortal” is an indirect statement.
The phrase “I” is a noun phrase and the subject of the verb “know.”
The phrase “mortal” is a predicate adjective referring back to the subject “man.”
The phrase “that man” is a noun phrase and the subject of the verb “Is.”
The word “I” is a first person singular subject pronoun.
The word “Is” is a third person singular present tense verb.
The word “Is” is the main verb of “that man is mortal.”
The word “know” is a present tense verb.
The word “know” is the main verb of “I know that man is mortal.”
The word “man” is the head of the noun phrase “that man.”
The word “mortal” is an adjective.
The word “Ø” is a complementizer specifying the complement clause “that man is mortal.”
The word “that” is a determiner specifying the noun “man.”

(17) TRANSITIVE (SENTENTIAL COMPLEMENT; INDIRECT QUESTION):
I know whether she went
The clause “I know whether she went” is a statement.
The clause “whether she went” is an indirect question.
The phrase “I” is a noun phrase and the subject of the verb “know.”
The phrase “she” is a noun phrase and the subject of the verb “went.”
The word “I” is a first person singular subject pronoun.
The word “know” is a present tense verb.
The word “know” is the main verb of “I know whether she went.”
The word “she” is a third person singular feminine subject pronoun.
The word “went” is a past tense verb.
The word “went” is the main verb of “whether she went.”
The word “whether” is an interrogative pronoun.
407

408
(18) TRANSITIVE (SENTENTIAL COMPLEMENT; INDIRECT QUESTION):
I know where she went
The clause “I know where she went” is a statement.
The clause “where she went” is an indirect question.
The phrase “I” is a noun phrase and the subject of the verb “know.”
The phrase “she” is a noun phrase and the subject of the verb “went.”
The phrase “[e]” is the referent of “where” and the empty modifier phrase of the verb “went.”
The word “I” is a first person singular subject pronoun.
The word “know” is a present tense verb.
The word “know” is the main verb of “I know where she went.”
The word “she” is a third person singular feminine subject pronoun.
The word “went” is a past tense verb.
The word “went” is the main verb of “where she went.”
The word “where” is an interrogative pronoun.

(19) TRANSITIVE (SENTENTIAL COMPLEMENT; INDIRECT QUESTION):
I know who went
The clause “I know who went” is a statement.
The clause “who went” is an indirect question.
The phrase “I” is a noun phrase and the subject of the verb “know.”
The phrase “[e]” is the referent of “who” and the empty subject noun phrase of the verb “went.”
The word “I” is a first person singular subject pronoun.
The word “know” is a present tense verb.
The word “know” is the main verb of “I know who went.”
The word “went” is a past tense verb.
The word “went” is the main verb of “who went.”
The word “who” is an interrogative pronoun.
409

410
(20) TRANSITIVE (SENTENTIAL COMPLEMENT; INDIRECT QUESTION):
I know who he admires
The clause “I know who he admires” is a statement.
The clause “who he admires” is an indirect question.
The phrase “I” is a noun phrase and the subject of the verb “know.”
The phrase “[e]” is an abstract empty noun phrase and the direct object of the verb “admires.”
The word “admires” is a third person singular present tense verb.
The word “admires” is the main verb of “who he admires.”
The word “he” is a third person singular masculine subject pronoun.
The word “I” is a first person singular subject pronoun.
The word “know” is a present tense verb.
The word “know” is the main verb of “I know who he admires.”
The word “who” is an interrogative pronoun.

(21) TRANSITIVE (DIRECT OBJECT + S COMPLEMENT; INDIRECT STATEMENT):
he told me that she went
The clause “he told me that she went” is a statement.
The clause “that she went” is a complement clause of the verb “told.”
The clause “she went” is an indirect statement.
The phrase “he” is a noun phrase and the subject of the verb “told.”
The phrase “me” is a noun phrase and the direct object of the verb “told.”
The phrase “she” is a noun phrase and the subject of the verb “went.”
The word “he” is a third person singular masculine subject pronoun.
The word “me” is a first person singular object pronoun.
The word “she” is a third person singular feminine subject pronoun.
The word “that” is a complementizer specifying the complement clause “she went.”
The word “told” is a past tense verb.
The word “told” is the main verb of “he told me that she went.”
The word “went” is a past tense verb.
The word “went” is the main verb of “that she went.”
411

412
(22) TRANSITIVE (DIRECT OBJECT + S COMPLEMENT; INDIRECT STATEMENT):
he told me she went
The clause “he told me she went” is a statement.
The clause “that she went” is a complement clause of the verb “told.”
The clause “she went” is an indirect statement.
The phrase “he” is a noun phrase and the subject of the verb “told.”
The phrase “me” is a noun phrase and the direct object of the verb “told.”
The word “he” is a third person singular masculine subject pronoun.
The word “me” is a first person singular object pronoun.
The word “Ø” is a complementizer specifying the complement clause “she went.”
The word “she” is a third person singular feminine subject pronoun.
The word “told” is a past tense verb.
The word “told” is the main verb of “he told me she went.”
The word “went” is a past tense verb.
The word “went” is the main verb of “that she went.”

(23) TRANSITIVE (DIRECT OBJECT + S COMPLEMENT; INDIRECT QUESTION):
he told me whether she went
The clause “he told me whether she went” is a statement.
The clause “whether she went” is a complement clause of the verb “told.”
The clause “whether she went” is an indirect question.
The phrase “he” is a noun phrase and the subject of the verb “told.”
The phrase “me” is a noun phrase and the direct object of the verb “told.”
The phrase “she” is a noun phrase and the subject of the verb “went.”
The word “he” is a third person singular masculine subject pronoun.
The word “me” is a first person singular object pronoun.
The word “she” is a third person singular feminine subject pronoun.
The word “told” is a past tense verb.
The word “told” is the main verb of “he told me whether she went.”
The word “went” is a past tense verb.
The word “went” is the main verb of “whether she went.”
The word “whether” is an interrogative pronoun.
413

414
(24) TRANSITIVE (DIRECT OBJECT + S COMPLEMENT; INDIRECT QUESTION):
he told me where she went
The clause “he told me where she went” is a statement.
The clause “where she went” is a complement clause of the verb “told.”
The clause “where she went” is an indirect question.
The phrase “he” is a noun phrase and the subject of the verb “told.”
The phrase “me” is a noun phrase and the direct object of the verb “told.”
The phrase “she” is a noun phrase and the subject of the verb “went.”
The phrase “[e]” is the referent of “where” and the empty modifier phrase of the verb “went.”
The word “he” is a third person singular masculine subject pronoun.
The word “me” is a first person singular object pronoun.
The word “she” is a third person singular feminine subject pronoun.
The word “told” is a past tense verb.
The word “told” is the main verb of “he told me where she went.”
The word “went” is a past tense verb.
The word “went” is the main verb of “where she went.”
The word “where” is an interrogative pronoun.

(25) TRANSITIVE (DIRECT OBJECT + S COMPLEMENT; INDIRECT QUESTION):
he told me who [e] went
The clause “he told me who went” is a statement.
The clause “who went” is an indirect question.
The phrase “me” is a noun phrase and the direct object of the verb “told.”
The phrase “he” is a noun phrase and the subject of the verb “told.”
The phrase “[e]” is the referent of “who” and the empty subject noun phrase of the verb “went.”
The word “me” is a first person singular object pronoun.
The word “he” is a third person singular masculine subject pronoun.
The word “told” is a past tense verb.
The word “told” is the main verb of “he told me who went.”
The word “went” is a past tense verb.
The word “went” is the main verb of “who went.”
The word “who” is an interrogative pronoun.
415

416
(26) TRANSITIVE (DIRECT OBJECT + S COMPLEMENT; INDIRECT QUESTION):
he told me who she admires
The clause “he told me who she admires” is a statement.
The clause “who she admires” is an indirect question.
The phrase “me” is a noun phrase and the direct object of the verb “told.”
The phrase “he” is a noun phrase and the subject of the verb “told.”
The phrase “[e]” is an abstract empty noun phrase and the direct object of the verb “admires.”
The word “admires” is a third person singular present tense verb.
The word “admires” is the main verb of “who she admires.”
The word “he” is a third person singular masculine subject pronoun.
The word “me” is a first person singular object pronoun.
The word “she” is a third person singular feminine subject pronoun.
The word “told” is a past tense verb.
The word “told” is the main verb of “he told me who she admires.”
The word “who” is an interrogative pronoun.

(27) TRANSITIVE (PREPOSITIONAL PHRASE OBJECT):
he relies on her advice
The clause “he relies on her advice” is a statement.
The phrase “he” is a noun phrase and the subject of the verb “relies.”
The phrase “her advice” is a noun phrase and the object of the preposition “on.”
The phrase “her” is a possessive phrase specifying “advice.”
The phrase “on her advice” is a prepositional phrase and a complement of the verb “relies.”
The word “advice” is a singular mass noun.
The word “advice” is the head of the noun phrase “her advice.”
The word “he” is a third person singular masculine subject pronoun.
The word “her” is a third person singular feminine possessive pronoun.
The word “on” is a preposition introducing the prepositional phrase “on her advice.”
The word “POS” is a determiner specifying the noun “advice.”
The word “relies” is a third person singular present tense verb.
The word “relies” is the main verb of “he relies on her advice.”
417

418
(28) INTRANSITIVE (PREPOSITIONAL PHRASE MODIFIER):
he went on her advice
The clause “he went on her advice” is a statement.
The phrase “he” is a noun phrase and the subject of the verb “went.”
The phrase “her advice” is a noun phrase and the object of the preposition “on.”
The phrase “her” is a possessive phrase specifying “advice.”
The phrase “on her advice” is a prepositional phrase and a modifier of the verb “went.”
The word “advice” is a singular mass noun.
The word “advice” is the head of the noun phrase “her advice.”
The word “he” is a third person singular masculine subject pronoun.
The word “her” is a third person singular feminine possessive pronoun.
The word “on” is a preposition introducing the prepositional phrase “on her advice.”
The word “POS” is a determiner specifying the noun “advice.”
The word “went” is a past tense verb.
The word “went” is the main verb of “he went on her advice.”

(29) TRANSITIVE (PREPOSITIONAL PHRASE OBJECT):
she sat on the hat
The clause “she sat on the hat” is a statement.
The phrase “on the hat” is a prepositional phrase and a complement of the verb “sat.”
The phrase “she” is a noun phrase and the subject of the verb “sat.”
The phrase “the hat” is a noun phrase and the object of the preposition “on.”
The word “hat” is a singular count noun.
The word “hat” is the head of the noun phrase “the hat.”
The word “on” is a preposition introducing the prepositional phrase “on the hat.”
The word “sat” is the main verb of “she sat on the hat.”
The word “sat” is a past tense verb.
The word “she” is a third person singular feminine subject pronoun.
The word “the” is a determiner specifying the noun “hat.”
419

420
(30) TRANSITIVE (PARTICLE + DIRECT OBJECT):
she put on the hat
The clause “she put on the hat” is a statement.
The phrase “she” is a noun phrase and the subject of the verb “put.”
The phrase “the hat” is a noun phrase and the direct object of the verb “put.”
The word “hat” is a singular count noun.
The word “hat” is the head of the noun phrase “the hat.”
The word “on” is a particle (intransitive preposition).
The word “put” is the main verb of “she put on the hat.”
The word “put” is a past tense verb.
The word “she” is a third person singular feminine subject pronoun.
The word “the” is a determiner specifying the noun “hat.”

(31) TRANSITIVE (DIRECT OBJECT + PARTICLE):
she put the hat on
The clause “she put the hat on” is a statement.
The phrase “she” is a noun phrase and the subject of the verb “put.”
The phrase “the hat” is a noun phrase and the direct object of the verb “put.”
The word “hat” is a singular count noun.
The word “hat” is the head of the noun phrase “the hat.”
The word “on” is a particle (intransitive preposition).
The word “put” is a past tense verb.
The word “put” is the main verb of “she put the hat on.”
The word “she” is a third person singular feminine subject pronoun.
The word “the” is a determiner specifying the noun “hat.”
421

422
(32) TRANSITIVE (DIRECT OBJECT + PARTICLE):
she put the hat right on
The clause “she put the hat right on” is a statement.
The phrase “right” is an adverb phrase modifying the particle “on.”
The phrase “she” is a noun phrase and the subject of the verb “put.”
The phrase “the hat” is a noun phrase and the direct object of the verb “put.”
The word “hat” is a singular count noun.
The word “hat” is the head of the noun phrase “the hat.”
The word “on” is a particle (intransitive preposition).
The word “put” is a past tense verb.
The word “put” is the main verb of “she put the hat right on.”
The word “she” is a third person singular feminine subject pronoun.
The word “right” is an adverb.
The word “the” is a determiner specifying the noun “hat.”

(33) TRANSITIVE (COMPLEMENTARY INFINITIVE):
she asked to go
The clause “she asked to go” is a statement.
The phrase “she” is a noun phrase and the subject of the verb “asked.”
The phrase “to go” is an infinitive complement of the verb “asked.”
The word “asked” is a past tense verb.
The word “asked” is the main verb of “she asked to go.”
The word “go” is an infinitive verb.
The word “she” is a third person singular feminine subject pronoun.
The word “to” is the infinitive marker introducing the infinitive phrase “to go.”
423

424
(34) TRANSITIVE (DIRECT OBJECT + COMPLEMENTARY INFINITIVE):
she asked him to go
The clause “she asked him to go” is a statement.
The phrase “him” is a noun phrase and subject of the infinitive verb “go.”
The phrase “him” is a noun phrase and the direct object of the verb “asked.”
The phrase “she” is a noun phrase and the subject of the verb “asked.”
The phrase “to go” is an infinitive complement of the verb “asked.”
The word “asked” is a past tense verb.
The word “asked” is the main verb of “she asked him to go.”
The word “go” is an infinitive verb.
The word “him” is a third person singular masculine object pronoun.
The word “she” is a third person singular feminine subject pronoun.
The word “to” is the infinitive marker introducing the infinitive phrase “to go.”

(35) TRANSITIVE (PREPOSITIONAL OBJECT WITH INFINITIVE):
she prayed for him to win
The clause “she prayed for him to win” is a statement.
The phrase “for him to win” is a prepositional phrase and a complement of the verb “prayed.”
The phrase “him” is a noun phrase and the subject of the verb “win.”
The phrase “him” is a noun phrase and the object of the preposition “for.”
The phrase “she” is a noun phrase and the subject of the verb “prayed.”
The word “for” is a preposition introducing the prepositional phrase “for him to win.”
The word “him” is a third person singular masculine object pronoun.
The word “prayed” is a past tense verb.
The word “prayed” is the main verb of “she prayed for him to win.”
The word “she” is a third person singular feminine subject pronoun.
The word “to” is the infinitive marker introducing the infinitive phrase “to win.”
The word “win” is an infinitive verb.
425

426
(36) TRANSITIVE (PREPOSITIONAL OBJECT WITH INFINITIVE):
she relied on him to finish
The clause “she relied on him to finish” is a statement.
The phrase “him” is a noun phrase and the object of the preposition “on.”
The phrase “him” is a noun phrase and the subject of the verb “finish.”
The phrase “on him to finish” is a prepositional phrase and a complement of the verb “relied.”
The phrase “she” is a noun phrase and the subject of the verb “asked.”
The word “finish” is an infinitive verb.
The word “him” is a third person singular masculine object pronoun.
The word “on” is a preposition introducing the prepositional phrase “on him to finish.”
The word “relied” is a past tense verb.
The word “relied” is the main verb of “she relied on him to finish.”
The word “she” is a third person singular feminine subject pronoun.
The word “to” is the infinitive marker introducing the infinitive phrase “to finish.”
What are the TG equivalents of these two trees?

SUPPLEMENT NINE: SAMPLE PARSES FROM THE LANGTECH PARSER
SAMPLE PARSES ILLUSTRATING NOUN PHRASES
(1) The Spanish American history teacher resigned.
history teacher who is Spanish and American teacher of American history who is Spanish
history teacher who is Spanish American teacher of Spanish American history
427

428
(2) The kitchen appliance factory guarantee expired.
(3) All the men’s money disappeared.

(4) All those many nosey house guests with excessive demands are a pain in the neck.
(5) The quick brown fox jumped over the lazy dog.
429

430
(6) The soprano from a small town in the south of France collapsed.

SUPPLEMENT TEN: SAMPLE PARSES FROM THE LANGTECH PARSER
(1) Is this the violin Kevin Played?
EMPTY CATEGORIES: Ø, [u], and [e]
a. This violin is the violin.
this violin PRS is the violin
b. This is the violin.
this [u] PRS is the violin
c. Is this the violin?
is i PRS j this [u] [e] j [e] i the violin
d. Is this the violin which Kevin played?
is i PRS j this [u] [e] j [e] i the violin k which k Ø Kevin played [e] k
e. Is this the violin that Kevin played?
is i PRS j this [u] [e] j [e] i the violin k [e] k that Kevin played [e] k
f. Is this the violin Kevin played?
is i PRS j this [u] [e] j [e] i the violin k [e] k Ø Kevin played [e] k
(2) X3 Representaton.
431

432
(2) We believe John to be completely dominated by his wife.
An [e] is coindexed with its closest C-commanding N3.

(3) Saturn’s rings are believed to be completely dominated by its moons.
An [e] is coindexed with its closest C-commanding N3.
433

434
SUPPLEMENT ELEVEN: SAMPLE PARSES FROM THE LANGTECH PARSER
SAMPLE PARSES ILLUSTRATING PARTICIPLES
The internal structure of all –ing and –ed/en forms are as follows:
(1) Progressive Participles (-ing) and Derived Adjectives (DA1).
a. Participle (the lady charming the man): [ C3 [ C0 [ C0 -ing] [ V3 [ V0 charming]]]]
b. Adjective (a very charming lady): [ C3 [ C0 [ V3 [ V0 charming]] [ C0 DA1]]]

(2) Passive Participles (-en) and Derived Adjectives (DA2).
a. Participle (charmed by her flattery): [ C3 [ C0 [ C0 -en] [ V3 [ V0 charmed]]]]
b. Derived Adjective (a very charmed life): [ C3 [ C0 [ V3 [ V0 charmed]] [ C0 DA2]]]
435

436
(3) Gerundial Nominals (-ing) and Derived Nominals (DN).
a. -ing (charming candy from a baby): [ N3 [ N0 [ N0 -ing] [ V3 [ V0 charming]]]]
b. DN: (such charming of snakes): [ N3 [ N0 [ V3 [ V0 charming]] [ N0 DN]]]

(4) Summary of Verbal Structures.
a. Verb resides in posthead position.
[ C3 [ C0 [ C0 -ing] [ V3 [ V0 charming]] ]]
[ C3 [ C0 [ C0 -en] [ V3 [ V0 charmed]] ]]
[ N3 [ N0 [ N0 -ing] [ V3 [ V0 charming]] ]]
b. Verb resides in prehead position.
[ C3 [ C0 [ V3 [ V0 charming]] [ C0 DA1] ]]
[ C3 [ C0 [ V3 [ V0 charmed]] [ C0 DA2] ]]
[ N3 [ N0 [ V3 [ V0 charming]] [ N0 DN] ]]
Notice that the four possible ing–forms in English have the feature specification [–VBL] (they are
subjuncts); two are [–NML] (the derived adjective and the present participle); two are [+NML] (the
derived nominal and the gerundial nominal). Further, the two members of each pair are mirror
images of each other. Thus, all are subjuncts with a V3 embedded on their X0 level.
The nominal nature of gerunds is accounted for by embedding that V3 into an N3, and the adjectival
nature of participles is accounted for by embedding that V3 into a C3.
437

438
SAMPLE PARSES FROM THE LANGTECH PARSER
(5) TRANSITIVE (DIRECT OBJECT + BARE INFINITIVE VERB):
i saw him cheer
The clause "i saw him cheer" is a statement.
The phrase "cheer" is the bare infinitive complement of the verb "saw".
The phrase "him" is a noun phrase and the direct object of the verb "saw".
The phrase "i" is a noun phrase and the subject of the verb "saw".
The word "cheer" is a bare infinitive verb.
The word "him" is a third person singular masculine object pronoun.
The word "i" is a first person singular subject pronoun.
The word "saw" is a past tense verb.
The word "saw" is the main verb of "i saw him cheer".

(6) TRANSITIVE (DIRECT OBJECT + PASSIVE PARTICIPLE):
i saw him cheered
The clause "i saw him cheered" is a statement.
The phrase "cheered" is a passive participle phrase describing "him".
The phrase "him" is a noun phrase and the direct object of the verb "saw".
The phrase "i" is a noun phrase and the subject of the verb "saw".
The phrase "[e]" is an empty noun phrase and the direct object of the verb "cheered".
The word "cheered" is a passive participle.
The word "him" is a third person singular masculine object pronoun.
The word "i" is a first person singular subject pronoun.
The word "saw" is a past tense verb.
The word "saw" is the main verb of "i saw him cheered".
The word "[-ed]" is the abstract marker for the passive participle.
439

440
(7) TRANSITIVE (DIRECT OBJECT + PROGRESSIVE PARTICIPLE):
i saw him cheering
The clause "i saw him cheering" is a statement.
The phrase "cheering" is a progressive participle phrase describing "him".
The phrase "him" is a noun phrase and the direct object of the verb "saw".
The phrase "i" is a noun phrase and the subject of the verb "saw".
The word "cheering" is a progressive participle.
The word "him" is a third person singular masculine object pronoun.
The word "i" is a first person singular subject pronoun.
The word "saw" is a past tense verb.
The word "saw" is the main verb of "i saw him cheering".
The word "[-ing]" is the abstract marker for the progressive participle.
The clause "i saw him cheering" is a statement.
The phrase "cheering" is a progressive participle phrase describing "i".
The phrase "him" is a noun phrase and the direct object of the verb "saw".
The phrase "i" is a noun phrase and the subject of the verb "saw".
The word "cheering" is a progressive participle.
The word "him" is a third person singular masculine object pronoun.
The word "i" is a first person singular subject pronoun.
The word "saw" is a past tense verb.
The word "saw" is the main verb of "i saw him cheering".
The word "[-ing]" is the abstract marker for the progressive participle.

(8) AMBIGUITY:
i heard the tenor singing
The clause "i heard the tenor singing" is a statement.
The phrase "i" is a noun phrase and the subject of the verb "heard".
The phrase "the tenor singing" is a noun phrase and the direct object of the verb "heard".
The word "heard" is a past tense verb.
The word "heard" is the main verb of "i heard the tenor singing".
The word "i" is a first person singular subject pronoun.
The word "singing" is a singular count noun.
The word "singing" is the head of the noun phrase "the tenor singing".
The word "tenor" is a noun forming a compound noun with "singing".
The word "tenor" is a singular count noun.
The word "tenor" is the head of the noun phrase "tenor".
The word "the" is a determiner specifying the noun "singing".
The clause "i heard the tenor singing" is a statement.
The phrase "i" is a noun phrase and the subject of the verb "heard".
The phrase "singing" is a progressive participle phrase describing "tenor".
The phrase "the tenor singing" is a noun phrase and the direct object of the verb "heard".
The word "heard" is a past tense verb.
The word "heard" is the main verb of "i heard the tenor singing".
The word "i" is a first person singular subject pronoun.
The word "singing" is a progressive participle.
The word "tenor" is a singular count noun.
The word "tenor" is the head of the noun phrase "the tenor singing".
The word "the" is a determiner specifying the noun "tenor".
The word "[-ing]" is the abstract marker for the progressive participle.
441

442
The clause "i heard the tenor singing" is a statement.
The phrase "i" is a noun phrase and the subject of the verb "heard".
The phrase "singing" is a progressive participle phrase describing "tenor".
The phrase "the tenor" is a noun phrase and the direct object of the verb "heard".
The word "heard" is a past tense verb.
The word "heard" is the main verb of "i heard the tenor singing".
The word "i" is a first person singular subject pronoun.
The word "singing" is a progressive participle.
The word "tenor" is a singular count noun.
The word "tenor" is the head of the noun phrase "the tenor".
The word "the" is a determiner specifying the noun "tenor".
The word "[-ing]" is the abstract marker for the progressive participle.
The clause "i heard the tenor singing" is a statement.
The phrase "i" is a noun phrase and the subject of the verb "heard".
The phrase "singing" is a progressive participle phrase describing "i".
The phrase "the tenor" is a noun phrase and the direct object of the verb "heard".
The word "heard" is a past tense verb.
The word "heard" is the main verb of "i heard the tenor singing".
The word "i" is a first person singular subject pronoun.
The word "singing" is a progressive participle.
The word "tenor" is a singular count noun.
The word "tenor" is the head of the noun phrase "the tenor".
The word "the" is a determiner specifying the noun "tenor".
The word "[-ing]" is the abstract marker for the progressive participle.

INDEX
Abstract Noun ..............................................................371
Acceptable .................................................................358
Accidental Gap ..............................................................76
Accompanier ...............................................................261
Accusative .................................................................375
Acoustic Phonetics ............................................................55
Active.....................................................................375
Adjective ........................................................53, 54, 182, 372
Adjunct....................................................................148
Adverb ....................................................................372
Affix......................................................................374
Affricate ....................................................................61
Agent .....................................................................260
Agentive...................................................................261
Agglutinating Language ......................................................376
Allomorph ..................................................................78
Allophone ...................................................................70
Alphabet...................................................................387
Alternative Element ..........................................................135
Alveolar ................................................................. 59-61
Ambiguity .............................................................257, 364
Analytic Language ...........................................................376
Anaphor ...............................................................196, 318
Anaphora ..................................................................325
Antepenult ...............................................................69, 89
Anterior .................................................................60, 63
Antonym ..................................................................256
Argumentation ...............................................................30
Articulatory Phonetics .........................................................55
Aspect ....................................................................375
Aspiration...................................................................69
Assimilate .................................................................350
Assimilation .................................................................81
Attributive Position ..........................................................372
Auditory Phonetics ...........................................................55
Auxiliary ..................................................................182
Back .................................................................... 57-59
Barrier ....................................................................318
Barriers....................................................................335
Behaviorism ................................................................345
Bilabial..................................................................59, 60
Binding Relation ............................................................319
Binding Resident ........................................................315, 335
Bound Morpheme ...........................................................374

444
Brace .....................................................................135
C-command ........................................................195, 318, 327
Case ......................................................................375
Case Grammar ..............................................................260
Case Relation ...............................................................261
Central .....................................................................57
Cerebral Lateralization .......................................................345
Characterizer ...............................................................171
Chunk .....................................................................126
Clause..................................................................36, 373
Co-occurrence Restriction .....................................................147
Cognition ..................................................................305
Collective Noun .............................................................371
Command..............................................................195, 203
Common Noun..............................................................371
Common Nouns .............................................................137
Comparative Degree .........................................................372
Complement................................................................156
Complementizer.............................................................182
Concrete Noun ..............................................................371
Conditional.............................................................136, 375
Conjugation ................................................................374
Conjunction ................................................................373
Connotation ................................................................255
Consonant ..................................................................59
Consonantal .................................................................63
Continuant ..................................................................63
Continuity Theory I ..........................................................366
Continuity Theory II .........................................................366
Contraction Block Condition ...................................................200
Coordinating ...............................................................182
Copulative Verb .............................................................372
Core Grammar ...........................................................34, 347
Coronal..................................................................60, 63
Count Noun ................................................................371
Creativity ...................................................................28
Curly Bracket...............................................................135
C–command ................................................................318
Dative.....................................................................261
Declarative .................................................................373
Declension .................................................................374
Deep Structure ..............................................................193
Degree Word ...............................................................182
Demonstrative Pronoun .......................................................372

Denotation .................................................................255
Descriptive Adequacy .....................................................38, 132
Descriptive Grammar..........................................................28
Determiner .................................................................182
Dimensional ................................................................283
Diphthong ..................................................................59
Direct .....................................................................127
Direct Object ............................................................53, 127
Discontinuity Theory .........................................................366
Distinctive Feature.........................................................58, 71
Empiricism.................................................................345
Empty Categories....................................................192, 197, 325
Empty Category Condition ................................................199, 336
Endocentricity ..........................................................145, 157
Exclamatory ................................................................373
Experiencer ................................................................261
Explanatory Adequacy.....................................................38, 132
Feminine ..................................................................375
First Person ............................................................118, 375
Flap .......................................................................62
Floating Quantifier...........................................................325
Formal Universal .........................................................34, 349
Free Morpheme .............................................................374
Free Resident ...........................................................316, 335
Free Variation ...............................................................78
Fricative ....................................................................61
Front.......................................................................57
Functional Category..........................................................133
Fusional Language ...........................................................377
Future .....................................................................136
Gender ....................................................................375
Genitive ...................................................................375
Glide....................................................................59, 62
Glottal .....................................................................60
Goal ......................................................................262
Grammar ...................................................................28
Grammatical.............................................................23, 358
Grammatical Characterization Problem........................................25, 339
Grammatical Realization Problem............................................25, 341
Head ..................................................................152, 373
Head-Only Category .........................................................318
Head–Only .................................................................146
Hierarchical Order ...........................................................124
Hierarchy ..................................................................133
445

446
High .......................................................................58
Historical Classification.......................................................378
Holonymy .................................................................256
Homonym .................................................................256
Homorganic .................................................................81
Hypernymy ................................................................256
Hyponymy .................................................................256
Idioms ....................................................................256
Imperative .........................................................136, 373, 375
Indefinite Pronoun ...........................................................371
Indicative ..................................................................375
Indo–European ..............................................................378
Inflection ..................................................................374
Inflectional Ending ..........................................................374
Inflectional Language ........................................................377
Instrument .............................................................260, 261
Instrumental ................................................................261
Intensive Pronoun ...........................................................371
Inter-dental..................................................................60
Interdental ..................................................................61
Interrogative................................................................373
Interrogative Pronoun ........................................................372
Intransitive Verb ........................................................192, 372
Isolating Language ...........................................................376
L-command ............................................................318, 332
Labeled Brackets ........................................................154, 184
Labio-dental .................................................................60
Labiodental .................................................................61
Language Variation ...........................................................43
Lateral .....................................................................62
Lax ........................................................................59
Left Branching Construction ...................................................361
Levels of Adequacy ...........................................................38
Lexical Ambiguities .........................................................161
Lexical Component ...........................................................86
Lexicon ....................................................................86
Linear Order................................................................124
Linguistic Competence .................................................24, 25, 339
Linguistic Performance ....................................................25, 341
Linguistic Structure ...........................................................30
Linguistic Universal .......................................................34, 347
Liquid......................................................................62
Locative ...................................................................262
Long Term Memory ..........................................................126

Low .......................................................................58
LTM ......................................................................126
L–Command ...............................................................318
Manner ....................................................................182
Manner Adverb .............................................................131
Manner of Articulation ........................................................60
Marked ...............................................................37, 81, 85
Masculine..................................................................375
Mass Noun .................................................................371
Maximal Projection ..........................................................145
Mentalism .................................................................345
Meronymy .................................................................256
Mid........................................................................57
Minimal Pair ................................................................72
Modal .................................................................136, 182
Mode .................................................................135, 375
Mood .................................................................135, 375
Morpheme ..............................................................77, 374
Morphology .............................................................77, 374
Morphophonemic .............................................................78
Morphophonological Rule ......................................................78
Multiple Branching Construction ...............................................361
Nasal ...................................................................62, 63
Native Language .............................................................23
Native Language Acquisition ..................................................341
Native Speaker...............................................................23
Native Speaker Intuition .......................................................32
Nativism...................................................................345
Natural Class ................................................................71
Neighborhood ..........................................................315, 327
Nested Construction..........................................................362
Neuter.....................................................................375
Node Admissibility Condition ..................................................308
Nominative ................................................................375
Nonaccidental Gap ............................................................76
Nonanterior ..............................................................60, 63
Nonback ....................................................................59
Nonconsonantal ..............................................................63
Noncontinuant ...............................................................63
Noncoronal...............................................................60, 63
Nondistinctive Feature .........................................................71
Nonhigh ....................................................................58
Nonlow.....................................................................58
Nonnasal ...................................................................63
447

448
Nonpositional Thematic Relation ...................................279, 283, 286, 292
Nonround ...................................................................59
Nonsibilant..................................................................63
Nonsonorant .................................................................63
Nonstative .................................................................278
Nontense ...................................................................59
Nonvocalic ..................................................................63
Nonvoiced ..................................................................63
Noun............................................................53, 54, 182, 371
Number ...................................................................375
Nurturism ..................................................................345
Object.....................................................................171
Object Formation ............................................................263
Objective ..................................................................261
Objective ..................................................................375
Observational Adequacy ...................................................38, 132
Obstruent ...................................................................63
Omission .......................................................31, 126, 131, 350
Optional Element ............................................................135
Palatal...................................................................60, 61
Paraphrase .............................................................257, 364
Parentheses.................................................................135
Parser .....................................................................309
Parts of Speech ..............................................................371
Passive ....................................................................375
Past.......................................................................136
Patient ....................................................................261
Penult ...................................................................69, 89
Perfective ..................................................................375
Permanent Resident ..................................................313, 317, 335
Person.....................................................................375
Personal Pronoun ............................................................371
Personal Pronoun ...........................................................325
Phoneme ....................................................................70
Phonetic Symbol .............................................................56
Phonetician ..................................................................55
Phonetics ...................................................................55
Phonological Component .......................................................85
Phonological Rule ........................................................73, 108
Phonologist .................................................................55
Phonology ...............................................................55, 69
Phonotactic Rule .............................................................76
Phonotactics .................................................................76
Phrase.........................................................124, 125, 152, 373

Phrase Structure Filter ........................................................182
Phrase Structure Grammar.....................................................143
Pitch .......................................................................67
Place of Articulation ..........................................................60
Placement .......................................................31, 126, 131, 350
Polysynthetic Language .......................................................377
Positional ..................................................................273
Positional Thematic Relation...................................................273
Positive Degree .............................................................372
Possessive .................................................................262
Possessive .................................................................375
Possessive Pronoun ......................................................119, 371
Pragmatics .................................................................259
Predicate...............................................................127, 373
Predicate Position ...........................................................372
Prefix .....................................................................374
Preposition .............................................................182, 373
Prescriptive Grammar .........................................................28
Present ....................................................................136
Progressive.................................................................375
Pronoun ...............................................................182, 371
Proper Noun................................................................371
Proper Nouns ...............................................................137
Quantifier ..................................................................182
R-command ............................................................318, 331
Rationalism ................................................................345
Recipient ..................................................................261
Reciprocal Pronoun ......................................................325, 372
Recoverability of Deletion Condition ............................................202
Recursive ...............................................................28, 174
Redundancy Rule .............................................................85
Reference ...........................................................31, 125, 350
Reflexive Pronoun ...................................................118, 325, 371
Relative Pronoun ............................................................372
Release .....................................................................71
Residence ..............................................................172, 306
Residential Domain ..................................................318, 319, 335
Residential Grammar .....................................................170, 194
Right Branching Construction ..................................................361
Root ......................................................................374
Round ...................................................................58, 59
Rule Formalism ..............................................................89
R–Command ...............................................................318
Second Language Acquisition ..................................................341
449

450
Second Person ..........................................................118, 375
Semantic Function ...........................................................260
Semantic Network .......................................................294, 296
Semantics ..................................................................255
Semivowel ..................................................................62
Sentence ...............................................................127, 373
Sentence Adverb .........................................................30, 131
Short Term Memory .........................................................126
Sibilant .....................................................................63
Slash–dash Notation ......................................................73, 111
Sonorant ....................................................................63
Source ....................................................................262
Specifier ...................................................................149
Spelling ....................................................................87
Stative ....................................................................278
Stem ......................................................................374
STM ......................................................................126
Stop ....................................................................59, 63
Stress ......................................................................67
Structural Ambiguities ........................................................162
Structural Category ..........................................................133
Structural Grammar ......................................................132, 143
Subject .........................................................53, 127, 171, 373
Subject Formation ...........................................................263
Subordinating ...............................................................182
Substantive.................................................................374
Substantive Universal .....................................................34, 347
Suffix .....................................................................374
Superlative Degree...........................................................372
Suprasegmental ..............................................................67
Surface Structure ............................................................193
Synonym ..................................................................256
Syntactic Categories .........................................................371
Syntactic Construction........................................................373
Syntactic Feature ............................................................147
Syntactic Function ...........................................................260
Syntactic Unit ..............................................................152
Syntax ............................................................117, 305, 374
Tense ...........................................................58, 59, 135, 375
Thematic Relation .......................................................261, 302
Theme ....................................................................261
Third Person................................................................375
Time ......................................................................135
Topicalization ..............................................................191

Trace .....................................................................192
Traditional Grammar .........................................................127
Transformational Grammar ................................133, 137, 145, 147, 191, 193
Transformational Rule ........................................................192
Transitive Verb .........................................................192, 372
Tree Diagram ...............................................................127
Typological Classification .....................................................376
Ultima ..................................................................69, 89
Unacceptable ...............................................................358
Ungrammatical...........................................................23, 358
Universal Grammar .....................................................28, 32, 34
Unmarked.............................................................37, 81, 85
Velar....................................................................59, 60
Verb ........................................................53, 54, 127, 182, 372
Verb Phrase ............................................................127, 133
Vocalic .....................................................................63
Voice .....................................................................375
Voiced ...............................................................55, 61, 63
Voiceless ................................................................61, 63
Voicing.....................................................................55
Vowel..................55, 117, 120, 121, 123, 125, 126, 138, 142, 203, 206, 214, 217, 218
Vowel Harmony..............................................................65
Writing .....................................................................29
X-Bar Syntax ...................................................152, 153, 157, 159
451

452
NOTES

NOTES
453

454
NOTES

NOTES
455

456
NOTES

NOTES
457

458
NOTES