Innovations
Transform your PDFs into Flipbooks and boost your revenue!
Leverage SEO-optimized Flipbooks, powerful backlinks, and multimedia content to professionally showcase your products and significantly increase your reach.
CONFERENCE ON LANGUAGE, LEARNING, & CULTURE<br />
<strong>Innovations</strong><br />
APRIL<br />
11-12,<br />
2014<br />
at the Intersection of Language,<br />
Learning, and Culture<br />
11200 Waples Mill Road #360, Fairfax, VA 22030, USA<br />
Tel: (703) 591-7042 Fax: (703) 537-0950 Email: info@viu.edu Web: www.viu.edu
ONLINE.VIU.EDU<br />
TEST DRIVE A COURSE AT ONLINE.VIU.EDU<br />
HAVE LIMITED<br />
TIME TO STUDY?<br />
GET YOUR DEGREE ONLINE.<br />
STUDY ANYWHERE, ANYTIME!<br />
FLEXIBLE<br />
8-WEEK<br />
SCHEDULE<br />
AFFORDABLE<br />
TUITION<br />
SCHOLARSHIP<br />
OPPORTUNITIES<br />
MOST TEXTBOOKS<br />
FOR NO ADDITIONAL<br />
CHARGE<br />
WEB: ONLINE.VIU.EDU<br />
EMAIL:<br />
PHONE: +1.800.514.6848
I want to express my heartfelt thanks to those of you who daily dedicate yourselves to improving<br />
the lives of students by providing quality and innovative classroom experiences. Whether you<br />
are on the delivery end of instruction, the research end, or both, you should know that your<br />
daily actions promote classrooms and research topics that have the ability to impact lives and<br />
promote positive social change. At our inaugural conference, we explored many topics and<br />
issued related language, learning, and culture.<br />
The conference theme and strands came about as a result of the need to highlight current<br />
successes in the field and to begin a conversation about moving forward together. The 2014<br />
theme, “<strong>Innovations</strong> at the Intersection of Language, Learning, and Culture,” framed educational<br />
experiences as ideally meeting the needs of diverse students, their communities, and the various<br />
stakeholders who can benefit from high-quality, socially responsive language training programs.<br />
Many thanks to our speakers and presenters who speakers and presenters explored topics of<br />
pedagogy and research with relevancy to the daily lives of teaching professionals and learners.<br />
An additional thanks to those in the School of Education, the School of Language Studies, and<br />
the larger VIU learning community who contributed to the success of CLLC 2014!<br />
Very best regards,<br />
Kevin J. Martin<br />
Associate Dean, School of Education, Virginia International University
ACKNOWLEDGEMENTS
TABLE OF CONTENTS<br />
Schedule at a Glance pg. 6<br />
Conference Strands pg. 7<br />
Conference Honorees pg. 8<br />
Plenary Session #1 pg. 9<br />
Schedule of Events pg. 10<br />
Plenary Session #2 pg. 12<br />
Plenary Session #3 pg. 13<br />
Proceedings from CLLC 2014 pgs. 14 - 34<br />
Speaker Contact Information pg. 35
SCHEDULE AT A GLANCE<br />
6
CONFERENCE STRANDS<br />
7
CONFERENCE HONOREES<br />
8
MORNING PLENARY<br />
9
SCHEDULE OF EVENTS<br />
10
MIDDAY PLENARY<br />
12
EVENING PLENARY<br />
13
PROCEEDINGS<br />
FROM CLLC 2014<br />
14
A CONTRASTIVE RHETORICAL STUDY OF ARGUMENTATIVE ESSAYS IN<br />
ENGLISH BY TUNISIAN EFL STUDENTS AND AMERICAN STUDENTS<br />
SALWA ENNEIFER<br />
L’Institut Supérieur des Langues de Tunis « ISLT ». Tunisia.<br />
salwa2083@hotmail.fr<br />
TOPIC AREA<br />
The present study intends to examine the rhetorical structure of argumentative essays written<br />
by some American and Tunisian students in English. It seeks to investigate whether Tunisians, as<br />
native Arabic speakers, write different essays from those written by American native speakers of<br />
English in terms of rhetoric. The study also aims at identifying whether these differences, if they<br />
exist, stem from Tunisian students’ first language. The researcher assumes that the transference of<br />
Arabic rhetorical structures, if it exists, changes the quality of the argument in English.<br />
It is motivated by the prevalence attributed to contrastive rhetoric (CR) as an ‘international’ project<br />
aiming at building “an international databank available for other international researchers” (Connor,<br />
1996: 173). It endeavors to test the validity of the contrastive rhetoric hypothesis; at least in the<br />
Tunisian context.<br />
CR was first introduced by Robert Kaplan (1966). CR hypothesis claimed that non-native speakers<br />
of English write different texts than those produced by their native English speaking counterparts.<br />
The mastery of syntactic structures, lexicon or spelling cannot by themselves make an EFL student<br />
use English as a native English speaker does. The difference lies in the fact that non-native English<br />
speaking learners have internalised the rhetorical structures of their mother tongues. Therefore CR, as<br />
a theory, aims to study and identify these differences.<br />
Some previous studies have dealt with the difficulties experienced by Arab EFL students when writing<br />
essays in English, such as Ouaouicha (1986) or Ismail (2010). They showed the differences and/or<br />
similarities between Arab students’ essays written in English and English native speakers’ essays. For<br />
instance, Ouaouicha identified differences only in terms of audience awareness and emotional<br />
appeals for persuasion. Americans used both aspects significantly more often than did Moroccans.<br />
Ismail (2010) also suggests that no significant differences exist between US and Arab advanced<br />
writers. He assumes that within- group variation is more important than between- group variation in<br />
the rhetorical performance of his participants.<br />
Nevertheless, there is still a gap in the studies involving Tunisians as a particular population. No<br />
previous studies have studied Tunisians. Moreover, previous studies results cannot be generalised to<br />
the whole Arab population. The current research endeavors to fill this gap.<br />
This article, then, is an investigation of the structure of argumentative essays written by some<br />
American and Tunisian students in English. It studies the rhetoric in essays by postgraduate Tunisian<br />
students of English written in Arabic and in English, in order to find out whether there is interference<br />
between the rhetorical structures of Arabic to English. Then, it compares the rhetoric of English<br />
argumentative essays written by postgraduate Tunisian students of English with American native<br />
speakers’ English argumentative essays to identify similarities and differences. The current article<br />
starts by presenting the hypotheses tested, the methodology and approach used and it ends with<br />
the results and implications of the study.<br />
15
HYPOTHESES<br />
The hypotheses to be tested are:<br />
1. Analytic variables selected to study argumentative essays in this study, namely: argument<br />
superstructure, reasoning, persuasive appeals and persuasive adaptiveness are valid, as they can<br />
predict the holistic score.<br />
The current study also hypothesized that, if the contrastive rhetoric claims presented first by<br />
Kaplan (1966) are correct, the analysis of essays in the current study will confirm that:<br />
2. Significant differences exist between argumentative essays written by postgraduate Tunisian<br />
students of English and American native speaking students’ essays written in English, in terms of<br />
selected analytic variables.<br />
3. Postgraduate Tunisian students of English produce similar argumentative essays in English and<br />
Arabic, in terms of selected analytic variables.<br />
4. Differences between argumentative essays written by postgraduate Tunisian students of English<br />
written in English and those written by American native speaking students in English are due to<br />
Arabic language rhetoric transfer.<br />
3. METHODOLOGY<br />
3.1. CORPUS<br />
The corpus for this study includes three sets of argumentative essays, thus: essays written in<br />
English by American native English speakers (AEE); essays written in English by postgraduate Tunisian<br />
students of English (TEE); and essays written in Arabic by postgraduate Tunisian students of English (TEA).<br />
The researcher used the LOCNESS corpus for the American students’ essays. LOCNESS is one of<br />
the available corpora of the ‘Centre for English Corpus Linguistics’ in the ‘Université Catholique de<br />
Louvain’ by Sylvianne Granger. For this study, the researcher has selected only twenty seven essays,<br />
all written at Indiana University in Indianapolis. The latter are timed argumentative essays written in<br />
March 1995 by twenty seven English native speakers. The age of the writers ranged from 22 to 48. The<br />
topics of the essays were:<br />
1. Money is the root of all evil<br />
2. Crime does not pay<br />
3. A man/woman’s financial reward should be commensurate with their contribution to<br />
the society in which they live<br />
4. Feminists have done more harm to the cause of women than good<br />
The Tunisian essays are written by Tunisian students who were studying English at first year Masters’<br />
degree level at the ‘Faculty of Letters and Humanities of Kairouan’ in Tunisia during the academic<br />
year 2011-12. Students at this particular level of education were chosen because it was assumed<br />
that they have less linguistic difficulties; such as grammatical errors, than undergraduate students.<br />
The essays were timed essays and the students answered one of the questions from the LOCNESS’<br />
Indiana University topic list already presented. Each Tunisian student was required to answer one of<br />
the four questions.<br />
The choice of Indiana University essays among all the essays in LOCNESS is explained by the fact that<br />
these essays were timed essays. Also, the age of the students was the closest to the age of first year<br />
Masters’ degree students. The American and Tunisian essays needed to share a maximum number of
similarities in their characteristics in order to be comparable.The same assignments already presented<br />
were translated into Arabic and administered to the same postgraduate Tunisian students of English.<br />
Each was required to answer one of the four questions, in Arabic. In order to obtain an equally<br />
distributed corpus, containing equal numbers of both Tunisian and American students’ essays, the<br />
Tunisian students sample also includes twenty seven essays written in Arabic and twenty seven essays<br />
written in English written by the same Tunisian students.<br />
3.2. APPROACH<br />
Ulla Connor (1990) presents a comprehensive model to assess persuasive writing. The model is multidimensional<br />
because it considers both linguistic and discourse-level features of essays and assesses a<br />
specific kind of discourse by developing analytical scales. In Connor’s article (1990: 70), six variables<br />
are used to analyse persuasive student writing. Two variables are concerned with the study and<br />
analysis of syntax and coherence in students’ argumentative essays. Both variables are not included<br />
in the following study, as it is basically and only concerned with the rhetorical differences between<br />
Tunisian and American students’ argumentative essays.<br />
Four variables of argumentative/persuasive texts are used to evaluate persuasive features in the<br />
sample essays, namely: the superstructure of the argument, the strength of logical reasoning<br />
involved; the use of persuasive appeals; and the adaptation of the text to one’s audience. The<br />
following table presents the independent variables and the scales used to describe them.<br />
INDEPENDENT VARIABLES SCALES<br />
Argument Superstructure Situation, problem, solution, evaluation<br />
Reasoning Claim, data, and warrant scale (Connor & Lauer, 1988)<br />
Persuasive appeals<br />
Rational, affective, and credibility appeals scale<br />
Persuasive adaptiveness Audience awareness and persuasive adaptiveness scale<br />
Table 1: Independent variables and scales for analysing rhetoric in students’ argumentative essays<br />
For more details concerning the independent variables used in this study and presented in the table<br />
above, the reader can consult Connor (1990).<br />
In addition to the independent variables presented in the above table, the corpus’ essays are also<br />
holistically rated. The overall writing quality of each essay in this study is rated using a 0-5 point holistic<br />
scale. The following section presents the analytical techniques.<br />
3.3. TECHNIQUES USED IN CORPUS ANALYSIS<br />
3.3.1. RATING ESSAYS<br />
Both the holistic and analytic aspects of the corpus’ essays were scored in this research by two<br />
independent raters. Once all the corpus’ essays were rated, the researcher performed statistical<br />
tests in order to obtain quantitative results.<br />
3.3.2. QUANTITATIVE ANALYSIS<br />
All statistical analyses in the current study were performed via SPSS 19. In order to study the<br />
raters’ agreement, an intra-class correlation coefficient was calculated to measure if there was<br />
an agreement between the scores each rater assigned to each essay, using each variable.<br />
Furthermore, a multinomial logistical regression was performed in order to study the relationship<br />
between the holistic score and analytic measures. The purpose of multinomial logistic regression<br />
analysis is to check the validity of the selected variables used to study argumentative essays. If the<br />
selected variables are valid, they can predict the overall writing quality and hence the holistic score.<br />
If the analytic variables are valid, the focus will be on testing the contrastive rhetoric hypothesis; or
they can be used to identify whether significant rhetorical differences exist between postgraduate<br />
Tunisian students of English argumentative essays and American native speaking students’ essays<br />
written in English, using MANOVA as a statistical tool. If a difference was found, the researcher<br />
investigated whether it stemmed from the Tunisian students native language; that is Arabic.<br />
4. RESULTS<br />
4.1. RESULTS OF INTRA- CLASS CORRELATION<br />
The results of intra-class correlations revealed that there was a high level of agreement between<br />
both raters. The results of intra class correlation are summarised in table 2.<br />
VARIABLE<br />
INTRA CLASS CORRELATION COEFFICIENT<br />
Situation 1.00<br />
Problem 1.00<br />
Solution .96<br />
Evaluation .81<br />
Claim .70<br />
Data .83<br />
Warrant .83<br />
Rational appeals .79<br />
Credibility appeals .83<br />
Affective appeals .83<br />
Persuasive adaptiveness .86<br />
Overall mark .84<br />
Table 2: Interrater Agreement for all variables<br />
As shown in the table the agreement rate between raters ranges from 70% to 100 % agreement.<br />
Hence, the researcher was confident that the scores she would use for later statistical analyses would<br />
reliably measure the rhetorical performance of the participants.<br />
4.2. RESULTS OF MULTINOMIAL LOGISTIC REGRESSION<br />
In order to identify predictors of overall score, a multinomial logistic regression was made. It included<br />
four covariates or variables, namely: argument superstructure; added Toulmin; persuasive appeals;<br />
and persuasive adaptiveness. It should be mentioned here that the researcher calculated the<br />
mean of individual claim, data, and warrant scores to obtain only one score or; an added Toulmin‘s<br />
informal reasoning score. The purpose is to facilitate later statistical analysis. The resulting added<br />
Toulmin score for each participant was a potential value of one to three.<br />
The researcher also included grouping factors in the multinomial logistic regression model. The results<br />
of the regression analysis are summarized in table 3.<br />
MODEL EFFECT(S) MODEL FITTING CRITERIA EFFECT SELECTION TESTS PSEUDO R-SQUARE<br />
AIC -2 Log Chi df Sig Nagelkerke<br />
likelihood Square<br />
Step 0 Intercept 155.38 137.38<br />
Group<br />
Step 1 Added 108.78 84.78 52.60 3 .000<br />
Toulmin<br />
Step 2 Persuasive 95.09 65.09 19.69 3 .000<br />
appeals<br />
Step 3 Argument<br />
superstructure<br />
93.04 57.04 8.06 3 .045 .82
Table 3: Multinomial Logistic Regression<br />
As shown in Table 3, multinomial logistic regression indicates that the predictor variables of the overall<br />
writing scores were added Toulmin, persuasive appeals and argument superstructure. The pseudo<br />
R-Square equals 82%, which means that the four previously identified variables together can predict<br />
82% of the variance in the holistic scores. The regression excluded persuasive adaptiveness as a<br />
predictor of overall scores.<br />
Consequently, the added Toulmin score is the best predictor of the overall score. The second best<br />
predictor of overall mark is the persuasive appeals variable, which includes credibility appeals and<br />
affective appeals. The third predictor of the overall score is argument superstructure. It should be<br />
noted here that persuasive adaptiveness was dismissed as a predictor variable; as the regression<br />
excluded persuasive adaptiveness as a predictor of overall scores. Participants demonstrated a low<br />
performance on the persuasive adaptiveness scale, therefore cross-cultural variations of the use of<br />
persuasive adaptiveness in the corpus essays was insignificant for the current study.<br />
When adding grouping factors into the model, the AIC improved. It reached 155.38 (Table 3) against<br />
167.095 (Table 4). Therefore, grouping had an effect on the improvement of predictability of the<br />
score range. As shown in Table 5, the final model explains score range variations with a significance<br />
equaling p
OBSERVED<br />
PREDICTED<br />
[1,2[ [2,3[ [3,4[ [4,5[ PERCENT CORRECT<br />
[1,2[ 3 2 0 0 60,0%<br />
[2,3[ 2 33 4 0 84,6%<br />
[3,4[ 0 8 22 1 71,0%<br />
[4,5[ 0 0 2 4 66,7%<br />
Overall Percentage 76,5%<br />
Table 6: Classification of predicted overall score ranges<br />
4.3. TESTING THE CONTRASTIVE RHETORIC HYPOTHESIS<br />
The second part of the statistical analysis aimed to identify whether significant rhetorical differences<br />
existed between argumentative essays written by postgraduate Tunisian students of English and essays<br />
written in English by American native speaking students. Therefore, it aimed to test the contrastive rhetoric<br />
hypothesis, and, in so doing, to answer research questions 2, 3 and 4, as previously presented.<br />
In order to answer the following questions, MANOVA is used. MANOVA was conducted with holistic<br />
and analytic measures as the dependent variables, and group membership as the independent<br />
variable. The statistical comparative analysis began by checking whether either grouping essays, on<br />
one hand, or the categorization of essays in terms of TEA, TEE and AEE, on the other, had an impact<br />
on the analytic variables, namely: argument superstructure, added Toulmin, persuasive appeals and<br />
persuasive adaptiveness, as well as overall score.<br />
DEPENDENT VARIABLE (I) GROUPS (J) GROUPS MEAN DIFFERENCE (I-J) STD. ERROR SIG. a<br />
Added Toulmin<br />
Persuasive Appeals<br />
Overall mark<br />
TEA TEE -,086 ,116 ,459<br />
AEE -,469* ,116 ,000<br />
TEE TEA ,086 ,116 ,459<br />
AEE -,383* ,116 ,001<br />
AEE TEA ,469* ,116 ,000<br />
TEE ,383* ,116 ,001<br />
TEA TEE -,130 ,133 ,333<br />
AEE -,296* ,133 ,029<br />
TEE TEA ,130 ,133 ,333<br />
AEE -,167 ,133 ,214<br />
AEE TEA ,296* ,133 ,029<br />
TEE ,167 ,133 ,214<br />
TEA TEE ,093 ,155 ,551<br />
AEE -,648* ,155 ,000<br />
TEE TEA -,093 ,155 ,551<br />
AEE -,741* ,155 ,000<br />
AEE TEA ,648* ,155 ,000<br />
TEE ,741* ,155 ,000
Based on estimated marginal means<br />
a. Adjustment for multiple comparisons: Least Significant Difference (equivalent to no adjustments).<br />
*. The mean difference is significant at the, 05 level.<br />
Table 7: Pairwise comparisons<br />
A significant difference is signaled between TEE and TEA, on one hand, and AEE, on the other, in<br />
terms of added Toulmin. No significant difference exists between TEA and TEE in terms of added<br />
Toulmin (p > 0.05).<br />
AEE and TEA have a significant difference in terms of persuasive appeals. Still, no significant<br />
differences exist between TEA and TEE (p > 0.05) or AEE and TEE (p > 0.05) in terms of persuasive<br />
appeals.<br />
A significant difference exists between TEE and AEE in terms of overall score; (p < 0.05). Also, as shown<br />
in table 7, a significant difference exists between AEE and TEA (p < 0.001). However, no significant<br />
difference exists between TEA and TEE in terms of overall mark: (p > 0.05).<br />
Through the previously stated findings, hypotheses two, three and four for the current study are<br />
validated. As a result, the current study assumes that:<br />
(1) Significant differences exist between the argumentative essays of postgraduate Tunisian<br />
students of English and American native speaking students’ essays written in English, in terms of<br />
added Toulmin and overall score.<br />
(2) Postgraduate Tunisian students of English produced similar argumentative essays in English<br />
and Arabic in terms of added Toulmin, persuasive appeals and overall score.<br />
(3) A significant difference existed between TEA and AEE essays in terms of added Toulmin,<br />
persuasive appeals and overall score.<br />
(4) Tunisian students of English produced similar argumentative essays in English and Arabic<br />
in terms of added Toulmin and persuasive appeals, furthermore differences between<br />
argumentative essays written by postgraduate Tunisian students of English and those written in<br />
English by American native speaking students in English are also signalled in the use of added<br />
Toulmin and persuasive appeals, the researcher claims that the latter differences stem from<br />
Arabic language rhetoric transfer.<br />
5. IMPLICATIONS<br />
The implications for the field of contrastive rhetoric include the validity of the analytic model (Connor,<br />
1990) selected to study rhetorical performance in argumentative essays. It is a valid tool of analysis<br />
for the investigation of various rhetorical aspects of Tunisian and American rhetorical structures and<br />
identifies their similarities and differences.<br />
Furthermore, the current research found that both Tunisian and American groups displayed a<br />
relatively low performance on persuasive adaptiveness measures. One implication of this study is to<br />
enhance training on the effective use of persuasive adaptiveness.<br />
The study revealed that no significant difference existed between TEE and AEE essays of the same<br />
Tunisian students. The same rhetorical techniques were found in both groups. Hence, the writing<br />
problems experienced by EFL Tunisian students are the byproduct of negative Arabic language
transfer. Therefore, the researcher claims that contrastive rhetoric is a valuable resource. It may<br />
explain one of the reasons why ESL/EFL students produce different English texts in comparison to<br />
native English speakers. It may also locate the differences. Hence, course designers and teachers<br />
may use contrastive rhetoric results to improve their teaching methods and focus more on the<br />
difficulties experienced by their students. That way, teaching will satisfy the special needs of each<br />
population or discourse community.<br />
REFERENCES<br />
Connor, U. (1990). Linguistic/Rhetorical Measures for International<br />
Persuasive Student Writing. Research in the Teaching of English, 24 (1), 67-87.<br />
Connor, U. (1996). Contrastive Rhetoric: Cross-Cultural Aspects of<br />
Second Language Writing. Cambridge: Cambridge University Press.<br />
Ismail, S. (2010). Arabic and English Persuasive Writing of Arabs<br />
from a Contrastive Rhetoric Perspective. (Doctoral dissertation) Indiana University of Pennsylvania.<br />
The School of Graduate Studies and Research. Department of English.<br />
Kaplan, R. B. (1966). Cultural thought Patterns in Inter-Cultural<br />
Education. Language Learning, 16, 1- 20.<br />
Ouaouicha, D. (1986). Contrastive Rhetoric and the Structure of<br />
Learner-Produced Argumentative Texts in Arabic and English. (Doctoral dissertation). The University of<br />
Texas at Austin.
THE EFFECTS OF INCREASED EXPOSURE OF THIRD PERSON OBJECT<br />
PRONOUNS ON THEIR AURAL COMPREHENSION IN SECOND<br />
LANGUAGE ACQUISITION<br />
OLIVIA AMZALLAG<br />
University at Albany<br />
Author Note<br />
Olivia Amzallag, Department of Languages, Literatures & Cultures, University at Albany<br />
Correspondence concerning this article should be addressed to Olivia Amzallag<br />
Contact:<br />
obarone@albany.edu<br />
oliviaamzallag@yahoo.com<br />
P.O. Box 42<br />
Chautauqua, NY 14722<br />
518-495-6808<br />
ABSTRACT<br />
This semester-long study sought to understand the impact of increasing exposure to object pronouns<br />
during instruction, on the aural detection of third person object pronouns, in learners of French by<br />
observing two diverse groups of university students enrolled in a high-beginning level French course.<br />
After personal observations on French language teaching and materials showed a tendency to<br />
omit third person object pronouns examined in six current French textbooks commonly used in both<br />
high schools and universities across the United States. This evaluation uncovered evident instances<br />
of avoidance strategies, lowering the instances of naturally occurring pronouns in the instructional<br />
input, and in some cases, modeling unnatural language patterns. Currently, the preliminary results of<br />
this study indicate that the increased exposure to object pronouns appears to have a positive effect<br />
on the aural comprehension of third person object pronouns. This study was designed to be useful in<br />
developing instructional methods and specifically honing French language instruction. Subsequently,<br />
there are several avenues open to future research in the fields of aural comprehension, third person<br />
object pronouns, input frequency and subject retention.<br />
This recent study sought to understand the impact of increased exposure to object pronouns on<br />
the aural comprehension of third person object pronouns by increasing pronominal exposure, in the<br />
hopes of facilitating the aural acquisition of third person object pronouns. The research questions<br />
were twofold: First, what is the effect of increased exposure to object pronouns during instruction, on<br />
the aural comprehension of third person object pronouns, prior to their being taught explicitly; and,<br />
second, what is the effect of avoiding object pronouns during the beginning stages of instruction,<br />
on third person object pronoun comprehension? Understanding the impact of increasing the exposure<br />
to these particular forms during instruction, without bringing direct attention to them grammatically, is<br />
valuable in developing means to facilitate object pronoun instruction and acquisition. Re-introducing<br />
these grammatical forms in their natural settings would provide a more realistic model of language during<br />
instruction, without inadvertently teaching object pronoun avoidance techniques.<br />
This study is rooted in both the field of pedagogy and applied linguistics, specifically drawing upon<br />
research on input frequency, object pronoun acquisition, and auditory comprehension in Second<br />
Language Acquisition (SLA). Presently, very few studies exist involving all three of these aspects of<br />
23
language learning. Furthermore, despite extensive research, no consensus is found regarding how<br />
object pronouns are aurally acquired by second-language learners, although they are difficult<br />
to detect by Anglophone learners. Hence, the research supports using a variety of instructional<br />
practices in the field of SLA where object pronouns are concerned. A brief overview of the current<br />
research reveals studies such as that of Erlam (2003), who measured the ability of high school<br />
students to guess the meaning of pronouns based on their context, and Grüter (2005) who analyzed<br />
the aural sensitivity to pronominal forms in six-year-old children. However, the present study largely<br />
follows the findings of VanPatten (1984) in his study on the manner in which learners of Spanish<br />
interpret pronouns, with a particular attention to word order. He concludes that in cases of low formmeaning<br />
correlation, frequency plays an important role in acquisition. Yet, the study reported here<br />
flows primarily from the work of Wust (2010), with the most recent research on the auditory acquisition<br />
of pronouns. She studies high-intermediate learners of French in a university classroom, explaining<br />
the manner in which learners of French understand and codify object pronouns. Following the<br />
results, Wust suggests that lower-level learners would benefit from exposure to frequent occurrences<br />
of object pronouns, and she makes a case for a need for explicit instruction in order to improve<br />
the aural comprehension of these forms. Moreover, she states that it is precisely because object<br />
pronouns are difficult to acquire that they are ideal targets for instruction and points out the scarcity<br />
of research on this subject.<br />
Personal observations on French language teaching and teaching materials showed a tendency to<br />
omit third person object pronouns. This observation was confirmed by conducting an evaluation of<br />
six current French textbooks commonly used in both high schools and universities across the United<br />
States, which indeed revealed infrequent use of object pronouns prior to their explicit instruction, and<br />
very little in context thereafter.<br />
Additionally, this evaluation uncovered evident instances of avoidance strategies, lowering the<br />
instances of naturally occurring pronouns in the instructional input, and in some cases, modeling<br />
unnatural language patterns. For example one textbook reads “Les jeunes Américains aiment<br />
beaucoup le lait. Mais, en général, les jeunes Français n’aiment pas le lait,” where the repetition<br />
of the phrase “le lait” would not be natural for a French native speaker. In this example, a native<br />
speaker would have most likely used the object pronoun l’. The same text presents a conversation<br />
which reads: “ – Tu vas au restaurant? – Non, je ne vais pas au restaurant; je vais à la bibliothèque,”<br />
in which the word restaurant is repeated in an unnatural way. In this example, a native speaker<br />
would have most likely used the object pronoun y. Furthermore, of the 235 pages prior to the lesson<br />
on object pronouns, this textbook only features seven examples of object pronouns in the third<br />
person. In another similar textbook the same type of avoidance is noted. For example, this text reads<br />
“– Tu aimes l’opéra? – Non, je déteste l’opéra. Et toi?” in which the word opera is repeated in an<br />
unnatural way. In this example, a native speaker would have most likely used the object pronoun l’.<br />
In this second textbook there is a clear lack of object pronoun use. Although there are 16 examples<br />
of third person object pronouns, these instances are all found between pages 110 and 112, of 154<br />
pages, and only in the passages taken from authentic texts such as Le Petit Prince. However, without<br />
counting these instances, there are no object pronouns featured at all in the instructional materials.<br />
Whether intentional or not, these examples clearly demonstrate a lack of modeling of frequently<br />
used French forms in the current beginning-level language teaching materials. Yet, learners often<br />
need to use these well before they have contact with them, and the consequences of this lack of<br />
modeling may render object pronoun use even more difficult for students. Anglophone learners of<br />
French have notable difficulty using third person object pronouns and have a tendency to avoid<br />
them or replace them by another grammatically incorrect form due to their innate nature. French<br />
third person object pronouns are truly renowned for their acquisition difficulty, and this is true for<br />
several structural reasons: their position in a sentence differs than in English as French object pronouns<br />
are usually pre-verbal, they must agree with the noun they are replacing while possessing masculine,<br />
feminine and plural forms, they are short and often contain only one syllable or letter, they can<br />
be contracted and subject to reduction, some forms are homophonic with other grammatical
morphemes as for example le, la, les can be definite articles, they are redundant as the pronoun<br />
replaces a concept introduced prior in the communication, and they are often neglected in speech.<br />
This difficulty creates a series of problems attributed to the usage of object pronouns such as an<br />
overgeneralization of object pronouns into subject pronouns, errors in gender, person, and number,<br />
more frequent comprehension of certain object pronouns and not others, errors in agreement<br />
and verbal structure with the use of double object pronouns, and object pronoun omission or<br />
misplacement, to name a few. Due of the multiplicity of these issues, it is desirable to facilitate object<br />
pronoun acquisition and instruction, especially in the area of auditory sensitivity to these forms.<br />
Therefore, the following study enabled us to better understand the impact of the reintroduction of<br />
object pronouns where they would feature naturally in the instructional materials, to be able to study<br />
the lack of modeling of object pronouns during instruction, and also study the learner techniques<br />
used to avoid their use, anticipating that these results would facilitate the instruction and mastery of<br />
these difficultly acquired forms.<br />
This study sought to understand the impact of increasing exposure to object pronouns during<br />
instruction, on the aural detection of third person object pronouns, in learners of French by observing<br />
two groups of university students in all levels of study, freshmen through seniors, enrolled in a variety of<br />
majors, in a semester-long, high-beginning level French course.<br />
These participants consisted two groups—a 16 person experimental group, of whom 4 had declared<br />
minors in French, and a 15 person control group, of whom 1 had declared a minor in French, and no<br />
students majored in French. Both groups of French university students, statistically evenly distributed,<br />
were studied under two classroom models and received the same quality, quantity and type of<br />
instruction. However, the experimental group received increased exposure to object pronouns,<br />
featured where they would naturally occur, both orally and in written instruction, for the first 10 weeks<br />
of a 16-week course. Following the first 10 weeks of modified instruction, the two groups received an<br />
instructional lesson on the subject of third person object pronouns, and then an identical curriculum<br />
for the remaining six weeks of the course. All participants took a placement test, which was used as<br />
a general benchmark for student ability level, and a way to administer the first auditory task in the<br />
first week of classes. The placement test also assured that the participants enrolled in the two groups<br />
were at the correct linguistic level for the course.<br />
The teaching materials were taken from the textbook Chez Nous – Branché sur le monde<br />
francophone, written by Valdman, Pons and Scullen (year) and published by Pearson. Teaching<br />
methods followed a communicative approach based on textbook content, and were representative<br />
of a standard curriculum taught in most universities and high schools. The experimental group met for<br />
1 hour and 50 minutes twice a week, and the control group met for 55 minutes four times a week. All<br />
instructional materials for the control group were taken from the course textbook, and throughout the<br />
instruction in the experimental group all types of object pronouns were incorporated both orally and<br />
in written materials for the first 10 weeks of a 16-week course.<br />
The documents used in the experimental group as compared to those used in the control group<br />
differed very little. For example, the textbook is rather repetitive and avoids using an object pronoun<br />
in the following homework reading activity therefore it was modified to incorporate a pronoun where<br />
it would naturally be featured: the control group receives the input “Ils achètent leurs sandwichs<br />
surtout dans les chaines spécialisées et pas nécessairement dans des cafés,” while the experimental<br />
group received the same information although featuring a pronoun: “Ils les achètent surtout dans<br />
des chaînes spécialisées et pas nécessairement dans des cafés.” We also added a question at<br />
the end of the reading to incite the student to think about the structure of the form in context: “Où<br />
achetez-vous vos sandwichs d’habitude?” Another example is taken from a set of instructions,<br />
demonstrating a postposition of the pronoun in the input; the control group sees: “Pour comprendre<br />
son histoire, considérez les questions suivantes” as the textbook reads, and the experimental group<br />
receives: “Lisez-le et répondez aux questions qui suivent,” changing neither the subject nor meaning
of the text, but slightly adjusting the structure and vocabulary to accommodate an object pronoun.<br />
Following this type of input increase for 10 weeks, the two groups received explicit instruction on<br />
object pronouns during the 11th week of the course, and ultimately identical instruction, without any<br />
variance between the two groups, for the last six weeks of the course. Although this study focused<br />
particularly on object pronouns of the third person, it must be noted that the input throughout the<br />
course featured pronouns of all persons—first, second, and third of both the singular and the plural, of<br />
both genders, for the experimental group, and could be found after the verb, as in this last example<br />
in the imperative.<br />
The testing tools throughout the study consisted of four aural comprehension tasks, embedded<br />
into the coursework, identified as Tasks 1 through 4, conceived to evaluate the student’s aural<br />
comprehension of object pronouns. Task 1 served as a benchmark in order to measure the rate<br />
of improvement of individual participants and groups. The students listened to a text read by the<br />
instructor, and based on their understanding of it, answered multiple-choice content questions.<br />
Additional qualitative information was drawn from the departmental student questionnaire,<br />
and collected in the form of a teacher’s journal. Observations were drawn from in-class student<br />
interaction, homework, quizzes & assignments.<br />
The statistical results of the testing portion of the research were analyzed using the Public/social/<br />
private partnership (referred to generally as PSPP) free software application and all data collected<br />
was analyzed for trends and variations in student performance. According to the average trends,<br />
one observes that at the beginning of the study, in Task 1, the control group was definitively stronger<br />
in aural comprehension, with a score of 17.79% higher than the experimental group. As well, although<br />
the control group does not demonstrate much improvement from Task 1 to Task 2, their performance<br />
increased by 13.33% in Task 3, following the lesson on object pronouns. Nevertheless, as the<br />
experimental group began with generally weak aural comprehension, they show a notable increase<br />
in comprehension between Task 1 and Task 2, with an improvement of 28.57%; although this group<br />
did not improve significantly between Task 2 and Task 3 after having received the lesson on object<br />
pronouns, with a simple increase in comprehension of 5.65%. Yet, these two groups improved at a<br />
relatively equal rhythm following the lesson on object pronouns, as the variance between groups<br />
between these two tasks is not statistically significant. Finally, the improvement was only maintained<br />
in the experimental group, with a decrease of only 0.29% between Tasks 3 and 4, whereas the<br />
control group shows a decrease of 5.71% between these same tasks, indicating a possible loss of<br />
retention. Therefore, it is evident that the improvement between the first and last tasks is much more<br />
pronounced in the experimental group, with more than double the scores of the control group,<br />
with 33.93% improvement over 12.39% improvement overall. Consequently, having increased the<br />
exposure to object pronouns seems to have had a statistically significant impact on the aural<br />
comprehension of concepts requiring the understanding of a third person object pronoun.<br />
Moreover, each listening task was followed by a question hoping to provide insight into participant<br />
reactions to the tasks: “Please write a few sentences sharing your thoughts about this listening task<br />
as a whole. Please share anything that comes to mind, how you felt, level of difficulty, particular<br />
vocabulary or grammatical structures that you found challenging, etc.…” These comments were<br />
sorted into three categories: negative outlooks were labeled difficult, the ambiguous comments<br />
were labeled accordingly, and positive outlooks were labeled easy/fair based on the two dominant<br />
terms used by students in these comments. From this classification, one observes that the increase<br />
in positive comments in the experimental group coincides with the same period that the test scores<br />
indicate an improvement in comprehension. Following Task 2 the experimental group touts 5<br />
comments in the difficult category, 10 ambiguous, 1 easy/fair and 1 blank. This is in contrast to the<br />
control group who shows 9 comments in the difficult category, 5 in the ambiguous, none in the easy/<br />
fair category and 1 blank; thus highlighting the discrepancy in student comprehension. Additionally,<br />
comments specific to object pronouns were only found among students in the control group:<br />
“Favorite one so far. I like the use of lui,” and, “It was difficult for me to pick up the object nouns.”<br />
There were no comments of this type found in those of the students in the experimental group, which<br />
could indicate that these forms were more natural to them.
Currently, the preliminary results of this study indicate that the increased exposure to object pronouns<br />
appears to have a positive effect on the aural comprehension of third person object pronouns.<br />
Furthermore, there is evidence to suggest that the lack of contact with these pronominal forms<br />
may indicate a loss of subject retention following instruction, although there is insufficient data to<br />
be able to draw firm conclusions. The pending categorizing and trending the information noted<br />
in the instructor journal, derived from student interactions, classwork and homework may uncover<br />
additional information pertinent to these research questions and qualitative results will be discussed in<br />
conjunction to the statistical results and converted into objective data sets when possible.<br />
Thusly the results contribute to the larger body of research on the instruction and acquisition of the<br />
French pronominal system, and may also prove applicable to other languages and/or grammatical<br />
concepts. This study was designed to be useful in developing instructional methods and honing<br />
specifically French language instruction, and subsequently, there are several avenues open to future<br />
research in the fields of aural comprehension, third person object pronouns, input frequency and<br />
subject retention.<br />
BIBLIOGRAPHY:<br />
Erlam, R. (2003). Evaluating the Relative Effectiveness of Structured-Input and Output-Based<br />
Instruction in Foreign Language Learning: Results from an Experimental Study. Studies in Second<br />
Language Acquisition 25(4): 559-82.<br />
Erlam, R. (2003). The Effects of Deductive and Inductive Instruction on the Acquisition of Direct Object<br />
Pronouns in French as a Second Language. The Modern Language Journal 87(2): 242–260.<br />
Grüter, T. (2005). Comprehension and production of French object clitics by child second language<br />
learners and children with specific language impairment. Applied Psycholinguistics, 26(3): 363-391.<br />
Valdman, Pons and Scullen. (2010). Chez Nous – Branché sur le monde francophone. 4e<br />
édition. Toronto, Canada: Pearson Canada Inc., 2010.<br />
VanPatten, B. (1984). Learners’ comprehension of clitic pronouns: More evidence for a word order<br />
strategy. Hispanic Linguistics, 1: 57-67.<br />
Wust, Valerie A. (2009). A la recherche des clitiques perdus: The dictogloss as a measure of the<br />
comprehension of y and en by L2 learners of French. The Canadian Modern Language Review, 65(3):<br />
471-499.<br />
Wust, Valerie. (2010). L2 French Learners’ Processing of Object Clitics: Data from the Classroom: L2<br />
Journal 2(1): 45-72.<br />
Wust, Valerie A. (2010). Pronominalization in French: Bridging the Gap between Research and<br />
Practice. The Journal of Linguistics and Language Teaching, 1(1): 75-110.
INCOMPLETE ATTAINMENT IN L2 AND HERITAGE GRAMMARS:<br />
AN ERROR ANALYSIS<br />
TERESA LEE<br />
University of Virginia<br />
Email: teresa902lee@yahoo.com<br />
INTRODUCTION<br />
Many studies in heritage language acquisition have noted that heritage speakers exhibit nonnative-like<br />
attainment of the heritage grammar, as second language (L2) learners do in their<br />
L2 development (Lee, 2011; Montrul, 2008; Polinsky, 2008; to just name a few). In many cases,<br />
linguistic deficits found in heritage speakers are not random and similar to those that are observed<br />
with L2 learners (Montrul, 2005; Montrul & Ionin, 2012). Systematic deficits are observed in various<br />
grammatical aspects (e.g., verbal and nominal agreement, use of reflexive pronouns, article use,<br />
etc) and they are very prominent in complex syntactic structures including relative clauses (Kim,<br />
2008; Kim, Montrul, & Yoon, 2010; Laleko & Polinsky, 2013; Montrul, 2004, 2005). In general, deficits<br />
are manifested either in the form of conversion on the corresponding features of the dominant<br />
language or in a hybrid form between the features of the heritage language and those of the<br />
dominant language. The primary goal of this paper is to examine the similarities and differences in<br />
the knowledge of L2 learners and heritage speakers of Korean in hopes to contribute to a better<br />
understanding of the nature of linguistic deficits observed in L2 and heritage language acquisition.<br />
PREVIOUS STUDIES<br />
Korean relative clauses differ from their English counterparts in two major respects. One difference<br />
has to do with the fact that Korean utilizes case markers to indicate grammatical relations. The other<br />
difference involves the branching direction. English relative clauses are post-nominal and the relative<br />
clause follows the head (the woman [RC who reads the book]). On the other hand, Korean relative<br />
clauses are pre-nominal and the relative clause precedes the head ([RC chayk-ul ilk-nun] yeca ‘the<br />
woman who reads a book’). This difference is important to note because it might cause trouble for<br />
the learner if the branching direction of the L2 or heritage language is different from that of his or her<br />
first language. As reported by several studies, the learner might rely on the properties of the first or<br />
dominant language (i.e., English) when interpreting relative clauses in an L2 or heritage language,<br />
(Kanno, 2007; Kim, 2008; Lee-Ellis, 2011; O’Grady, Lee, & Choo, 2003). In this regard, examining errors<br />
(especially those involving incorrect word order) can provide useful insight into the nature of linguistic<br />
deficits observed in L2 and heritage grammars.<br />
Several studies on the acquisition of Korean (and Japanese) relative clauses have noted that<br />
head errors reflect the influence from English (Kanno, 2007; Kim, 2008; O’Grady et al., 2003). Head<br />
errors involve the erroneous selection of the first noun in a relative clause as the head, appropriate<br />
for English, but not for Korean. Given this, head errors imply that the learner had trouble correctly<br />
locating the relative clause and the head in Korean. In Korean, subject relatives have the structure<br />
of [RC O V] S and direct object relatives have the structure of [RC S V] O. Lee-Ellis (2011) reported<br />
that some of her participants relied on the English basic word order (SVO), producing deviant forms<br />
that did not conform to the word order of Korean relative clauses. As noted earlier, this type of<br />
error implies that the learner is affected by his or her knowledge of the dominant language in<br />
producing Korean relative clauses.<br />
Previous studies have also reported another error pattern that involves the Korean basic word order<br />
(SOV) and the use of the bound noun kes, which is not required in the head external relative clauses<br />
28
(those with a gap). In their oral production of L2 learners of Korean, Jeon and Kim (2007) reported a<br />
high rate of this error type. Lee-Ellis (2011) also reported a similar type of word order error in her oral<br />
production study with Korean heritage speakers. As in the case of Jeon and Kim’s (2007) study, this<br />
error type was observed in non-target-like relative clauses involving the [RC S O V] kes pattern. In<br />
other words, heritage speakers as well as L2 learners are prone to producing a hybrid form of some<br />
sort, which do not conform to the word order of Korean relative clauses.<br />
THE PRESENT STUDY<br />
RESEARCH QUESTIONS<br />
In general, L2 learners start to learn Korean in a classroom setting after puberty whereas heritage<br />
speakers are exposed to Korean in the home since birth. Although heritage speakers are exposed to<br />
Korean early in childhood, their dominant language subsequently switches to English once schooling<br />
starts at around age 4 or 5 (Montrul, 2008). Then, the input that they receive in the heritage language<br />
and its use tend to be limited to the home, which affects their knowledge of the heritage language<br />
later in life, resulting in deficits (Montrul, 2004, 2005).<br />
Previous studies have noted that linguistic deficits tend to be more evident with L2 and heritage<br />
learners with low proficiency (Ionin & Montrul, 2010; Schwartz & Sprouse, 1996). Just as a look into<br />
the role of proficiency level can be useful, examining the role of accuracy level can be as insightful<br />
in probing further into the systematic differences that may exist between L2 and heritage learners.<br />
The following hypotheses are formulated with respect to the acquisition of Korean relative clauses.<br />
It is hypothesized that L2 learners may produce errors including word order errors more often than<br />
heritage speakers if age of acquisition plays a role. Also, both L2 learners and heritage speakers<br />
with low accuracy would produce word order errors more often than those with higher accuracy if<br />
accuracy level is a factor.<br />
PARTICIPANTS<br />
The experiment included 26 L2 learners and 11 heritage speakers (early simultaneous bilinguals) of<br />
Korean. Participants were enrolled in the second-semester Korean language class at a four-year<br />
university on the East Coast of the United States at the time of testing, and all of them had completed<br />
the first-semester beginning course. Their age ranged from 17 to 22 years old and the mean age was<br />
19 years old. The L2 learners in the present study started to learn Korean in college whereas all early<br />
simultaneous bilinguals were exposed to both English (from one parent) and Korean (from the other<br />
parent) in the home during early childhood. Yet, all participants were English-dominant in proficiency<br />
and use. Some of the early simultaneous bilinguals had prior knowledge of the Korean alphabet, but<br />
most of them could barely write simple sentences in Korean and their oral skills were mostly limited to<br />
simple greetings.<br />
EXPERIMENTAL METHODS<br />
The experiment employed two tasks involving written production and reading comprehension,<br />
respectively. In the written production task, participants were asked to translate English sentences<br />
containing relative clauses into Korean, and participants were asked to do the opposite in the<br />
reading comprehension task. In both tasks, relative clauses were embedded in a full sentence and<br />
modified the subject of the copular sentence. Each task included subject and direct object relatives<br />
(e.g., The woman [RC who is eating dinner] is my sister & The cola [RC that my friend drinks] is very<br />
delicious). It should be noted that the production task was administered prior to the comprehension<br />
task in order to avoid a possibility of priming effects. Relative clauses were introduced at the outset of<br />
the 2nd-semester beginning Korean course, and the experiment was conducted within several days<br />
after the introduction of relative clauses in class.
DATA CODING<br />
As for the comprehension (Korean-to-English translation) task, each sentence was checked for<br />
completion and the accuracy of the relative clause. Each participant was given six test sentences,<br />
and all participants were successful in translating all six of them, yielding a total of 222 relative<br />
clauses. A sentence was assigned a score of 0 if it involved major errors that changed the structure<br />
of the test sentence (e.g., the lunch that I ate for [RC cemsim-ul mek-nun] salam ‘the person who is<br />
eating lunch’). Otherwise, a sentence was assigned a score of<br />
1. Some sentences included lexical errors or minor grammatical errors including tense errors, but they<br />
were considered acceptable.<br />
The coding procedure for the production (English-to-Korean translation) task required a longer list<br />
of coding categories, due to the fact that more types of errors were observed in this task. Each<br />
participant was given six test sentences, and all participants were successful in translating all six<br />
of them, yielding a total of 222 relative clauses. Each response was checked for completion, and<br />
sentences with no relative clauses were considered inaccurate. Then, each relative clause was<br />
examined for accuracy and coded for error types if any existed.<br />
The most common type of error involved the production of incorrect word order in the relative<br />
clause. In Korean, subject and direct object relatives have the structure of [RC O V] S and [RC S V]<br />
O, respectively. Relative clauses which did not conform to these respective word order patterns were<br />
coded as word order errors. An example is given in (1). 1<br />
(1) Correct Incorrect<br />
a. [RC yonghwa-lul po-nun] salam *[ RC salam-i yonghwa-lul po-nun]<br />
movie-ACC watch-PRES person person-NOM movie-ACC watch-PRES<br />
‘the person who is watching a movie’<br />
Another common type of error involved the use of incorrect case markers on the noun (argument)<br />
that appears inside the relative clause and the head noun. Errors in this category included the use<br />
of a subject marker on the (direct object) argument inside the relative clause (e.g., [RC nolay-*ka/-<br />
lul tut-nun] namca ‘the man who is listening to a song’) and use of a topic marker on the (subject)<br />
argument inside the relative clause in place of a subject marker (e.g., [RC chinkwu-*nun/-ka masi-enun]<br />
cola ‘the cola a friend drinks’).<br />
Another error type involved repetition of the head, which alters the word order of the relative clause.<br />
An example of a head repetition error is given in (2).<br />
(2) *[ RC i yeca-ka newspaper ilk-nun] salam-un cey rwummeytu-yeyo.<br />
this woman-NOM newspaper read-PRES person-TOPIC my roommate-sentence.ending<br />
‘*the person who this woman is reading the newspaper.’<br />
Other types of errors included missing arguments of the relative clause, repetition of the verb of<br />
the relative clause, failure to mark the verb with a relativizer, and missing verbs. Relative clauses<br />
were coded as acceptable if none of these errors were observed. Also, there were errors that were<br />
not penalized: lexical errors (e.g., salam ‘person’ for yeca ‘woman’) and morphological errors in<br />
case markers or adnominal verbal suffixes that carry the tense of an embedded clause and signal<br />
relativization (e.g., ilk-*un (read-past) for ilk-nun (read-present).2
RESULTS<br />
THE COMPREHENSION TASK<br />
After coding was completed, data were quantified and statistical analysis was performed on the<br />
number of target-like relative clauses. According to a two-way repeated measures ANOVA, there<br />
was no main effect of Group and Sentence Type and no interaction effect between the two. The<br />
results showed that the comprehension task was fairly easy for both the L2 and heritage learners,<br />
which suggests that both groups had good receptive knowledge of Korean relative clauses. Figure 1<br />
shows the accuracy rates of target-like relative clauses by group.<br />
Figure 1. Comprehension accuracy rates of each group.<br />
THE PRODUCTION TASK<br />
After coding was completed, data were quantified and statistical analysis was performed on the<br />
number of target-like relative clauses. 3 As for the between-group comparison, the L2 group was<br />
more successful in producing target-like relative clauses than the early simultaneous bilinguals.<br />
Figure 2 presents the accuracy rates of target-like relative clauses by group.<br />
Figure 2. Production accuracy rates of each group.<br />
A two-way repeated measures ANOVA was performed on the number of target-like relative<br />
clauses with Group and Sentence Type as variables. There was a significant main effect of Group<br />
(F(1,35)=16.447, p=0.00027), which suggests that the L2 learners performed differently from the early
simultaneous bilinguals, the former group being significantly more accurate at producing targetlike<br />
relative clauses than the latter group. The main effect of Sentence Type was also significant<br />
(F(1,35)=16.447, p=0.00027), but there was no interaction effect between the two variables<br />
(F(1,35)=1.382, p=0.24732).<br />
Turning now to errors produced by each group, major error types included production of incorrect<br />
word order, repetition of the head, and use of incorrect case markers. Table 1 lists the total number of<br />
errors by group in these three categories.<br />
Table 1. Rates of major error types by group.<br />
GROUP L2 LEARNERS (N=26) EARLY SIMULTANEOUS (N=11)<br />
Total count 156 66<br />
Incorrect word 19 (12.2%) 26 (39.4%)<br />
order<br />
Head repetition 3 (1.9%) 4 (6.1%)<br />
Incorrect case<br />
markers<br />
13 (8.3%) 13 (19.7%)<br />
The early simultaneous bilinguals (39%) produced relative clauses with incorrect word order more<br />
frequently than the L2 learners (12%). Error rates indicate that the L2 learners also had trouble<br />
producing relative clauses with appropriate word order, but not as much as the early simultaneous<br />
bilinguals did. A similar trend was observed with errors involving repetition of the head and the use of<br />
incorrect case markers.<br />
In order to probe further into the differences in the linguistic deficits between L2 learners and heritage<br />
speakers, each group was divided into three accuracy levels (high, intermediate, and low), based<br />
on their individual accuracy test scores.4 No early simultaneous bilingual scored high enough to be<br />
placed into the high accuracy level and statistical analysis was performed only on the intermediate<br />
and low accuracy levels. Error rates at each accuracy level suggest that both the L2 and heritage<br />
learners with low accuracy produced word order errors more often than those with intermediate<br />
accuracy. Table 2 lists the number of word order errors by accuracy level.<br />
Table 2. Word order error rates by accuracy level.<br />
ACCURACY L2 LEARNERS (N=26) EARLY SIMULTANEOUS (N=11)<br />
n Total Errors Mean SD n Totals Error Mean SD<br />
High 1 8 1 n/a<br />
Intermediate 14 84 5 0.36 0.63 3 18 4 1.33 1.53<br />
Low 11 66 16 1.45 1.57 8 48 26 3.25 2.12<br />
(SD: Standard Deviation)<br />
A two-way repeated measures ANOVA was performed on the number of word order errors with<br />
Group and Accuracy Level as variables. There was a significant main effect of Group (F(1,32)=5.991,<br />
p=0.02005), which suggests that the performance of the L2 learners significantly differed from that of<br />
the early simultaneous bilinguals. Given the difference in the error rates, this result suggests that the<br />
L2 learners at both the intermediate and the low accuracy levels performed better than the early<br />
simultaneous bilinguals, producing fewer errors. There was also a significant main effect of Accuracy<br />
Level (F(1,32)=7.085, p=0.01206), which suggests that the learners at the intermediate and low<br />
accuracy levels differed from each other, those with intermediate accuracy producing significantly<br />
fewer word order errors than those with low accuracy. There was no interaction effect (F(1,32)=0.524,<br />
p=0.47463).
DISCUSSION<br />
The goal of the present study was to examine the similarities and differences in the knowledge of<br />
L2 learners and heritage speakers of Korean. In particular, it aimed to see what type of systematic<br />
difference might exist in the nature of linguistic deficits between L2 learners and early simultaneous<br />
bilinguals. Given the differences in their acquisitional path, it was hypothesized that L2 learners might<br />
exhibit a more severe degree of deficits than early simultaneous bilinguals if age of acquisition was a<br />
factor.<br />
The results of the production task were interesting in that the early simultaneous bilinguals did not<br />
outperform the L2 learners, despite their early exposure to Korean. Rather, the results indicated<br />
otherwise, disconfirming the role of age of acquisition. The L2 learners produced target-like relative<br />
clauses 80 percent of the time whereas the early simultaneous bilinguals did so only 43 percent of the<br />
time. Also, they produced significantly fewer word order errors, compared to the early simultaneous<br />
bilinguals. The L2 learners made 22 errors of the 156 tokens total and the early simultaneous bilinguals<br />
made 30 errors of the 66 tokens total.<br />
These findings seem to suggest that the L2 learners exhibited a lesser degree of deficits in their<br />
knowledge of Korean than the early simultaneous bilinguals. Yet, a close examination of word order<br />
errors made by each group revealed that there was a slight difference in the nature of deficits<br />
between the L2 learners and the early simultaneous bilinguals. The overall error analysis showed<br />
that both groups produced Korean relative clauses, utilizing the English word order (SVO), as similar<br />
observations were made by previous studies (Jeon & Kim, 2007; Lee-Ellis, 2011). In the present study,<br />
interesting to note is that the L2 learners (six instances) made such errors slightly more often than the<br />
early simultaneous bilinguals (three instances), which suggests that the L2 learners might be more<br />
prone to relying on the features of the dominant language in producing Korean relative clauses.<br />
In conclusion, the results of the production task showed that there was a negative correlation<br />
between accuracy level and the degree of linguistic deficits. The L2 learners with low accuracy<br />
exhibited a stronger degree of deficits than those with intermediate accuracy, producing word order<br />
errors more often. The same was true for the heritage group, which suggests that linguistic deficits are<br />
likely to be more prominent with learners at a lower accuracy level. This study has also shown that<br />
both the L2 and heritage learners were influenced by their knowledge of the dominant language.<br />
However, the results did not confirm the advantage of heritage speakers over L2 learners in that the<br />
latter group produced target-like relative clauses more often than the former group. Although the<br />
findings of the present study need to be corroborated with a larger sample size and more tokens of<br />
test items, they provide useful insight into the nature of linguistic deficits that exist in L2 and heritage<br />
grammars.<br />
NOTES<br />
1. The following abbreviations are used in this paper: NOM (nominative), ACC (accusative), TOPIC<br />
(topic), and PRES (present).<br />
2. Case markers in Korean involve allomorphic variations. There are two allomorphs (-i and -ka) for a<br />
subject marker and two allomorphs (-ul and -lul) for an object marker. Allomorphic variations are also<br />
found with some of the verbal suffixes (relativizers): -n/-un (past), -nun (present), and -ul/-lul (future or<br />
unrealized).<br />
3. Five adult native speakers of Korean (graduate students at the same institution) also participated<br />
in this task, and they correctly produced target relative clauses all the time.<br />
4. It was a listening test, which involved eight tokens of Korean relative clauses (e.g., the man who<br />
sees the woman & the man the woman sees). The high accuracy level included the participants<br />
who were accurate on at least seven tokens, and those at the intermediate accuracy level were<br />
correct on six to four tokens. The low accuracy level included those who were accurate on three to<br />
one token and those who failed to give any correct responses.
REFERENCES<br />
Jeon, K.S., & Kim, H. (2007). Development of relativization in Korean as a foreign language. Studies in<br />
Second Language Acquisition, 29, 253‒276.<br />
Kanno, K. (2007). Factors affecting the processing of Japanese relative clauses by L2 learners. Studies<br />
in Second Language Acquisition, 29, 197‒218.<br />
Kim, H. (2008). Heritage and non-heritage learners of Korean: Sentence processing differences and<br />
their pedagogical implications. In K. Kondo-Brown & Brown, J. (Eds.), Teaching Chinese, Japanese,<br />
and Korean Heritage Students: Curriculum needs, Materials, and Assessment (pp.99‒134). Lawrence<br />
Erlbaum Associates.<br />
Kim, J., Montrul, S., & Yoon, J. (2010). Dominant language influence in acquisition and attrition of<br />
binding: Interpretation of the Korean reflexive caki. Bilingualism: Language and Cognition,13, 73‒84.<br />
Laleko, O., & Polinsky, M. (2013). Marking topic or marking case: A comparative investigation of<br />
heritage Japanese and heritage Korean. The Heritage Language Journal, 10, 40‒64.<br />
Lee, T. (2011). Grammatical knowledge of Korean heritage speakers: Early vs. late bilinguals. Linguistic<br />
Approaches to Bilingualism, 1, 149‒174.<br />
Lee-Ellis, S. (2011). The elicited production of Korean relative clauses by heritage speakers. Studies in<br />
Second Language Acquisition, 33, 57‒89.<br />
Ionin, T., & Montrul, S. (2010). The role of L1 transfer in the interpretation of articles with definite plurals<br />
in L2 English. Language Learning, 60, 877‒925.<br />
Montrul, S. (2004). Subject and object expression in Spanish heritage speakers: A case of morphosyntactic<br />
convergence. Bilingualism: Language and Cognition, 7, 125‒142.<br />
Montrul, S. (2005). Second language acquisition and first language loss in adult early bilinguals.<br />
Exploring some differences and similarities. Second Language research, 21, 199‒249.<br />
Montrul, S. (2008). Incomplete acquisition in bilingualism. Re-examining the age factor. Amsterdam:<br />
John Benjamins.<br />
Montrul, S., & Ionin, T. (2012). Dominant language transfer in Spanish heritage speakers and second<br />
language learners in the interpretation of definite articles. The Modern Language Journal, 96, 70-91.<br />
O’Grady, W., Lee, M., & Choo, M. (2003). A subject-object asymmetry in the acquisition of relative<br />
clauses in Korean as a second language. Studies in Second Language Acquisition, 25, 433‒448.<br />
Polinsky, M. (2008). Russian gender under incomplete acquisition. The Heritage Language Journal, 6,<br />
40‒71.<br />
Schwartz, B., & Sprouse, R. (1996). L2 cognitive states and the full transfer/full access hypothesis.<br />
Second Language Research, 12, 40‒72.
SPEAKER CONTACT INFORMATION<br />
35
Introducing CLLC 2015!<br />
“Next-Gen Assessment”<br />
The School of Education atVirginia International University will be hosting<br />
the Conference on Language, Learning, and Culture (CLLC) on April 10-11, 9-11, 2015,<br />
with the goal of sharing best prac tices and emerging trends in assessment .
CONFERENCE ON LANGUAGE,<br />
LEARNING, & CULTURE<br />
THE SCHOOL OF EDUCATION OFFERS CONVENIENT STUDY OPTIONS TO FIT YOUR BUSY LIFESTYLE. STUDY WITH US EITHER ON-GROUND OR ONLINE!<br />
Master of Science in Applied Linguistics | Master of Education | Master of Arts in TESOL