09.05.2014 Views

Compound verbs in Japanese: Types and Constraints

Compound verbs in Japanese: Types and Constraints

Compound verbs in Japanese: Types and Constraints

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

<strong>Compound</strong> <strong>verbs</strong> <strong>in</strong> <strong>Japanese</strong>:<br />

<strong>Types</strong> <strong>and</strong> Constra<strong>in</strong>ts<br />

Yo Matsumoto<br />

Kobe University<br />

yomatsum@lit.kobe-u.ac.jp


outl<strong>in</strong>e<br />

Basic properties of compound <strong>verbs</strong> <strong>in</strong> <strong>Japanese</strong><br />

Constra<strong>in</strong>ts on lexical compound <strong>verbs</strong><br />

previous solutions: general constra<strong>in</strong>ts<br />

proposal: constra<strong>in</strong>ts at multiple levels<br />

general <strong>and</strong> type-level constra<strong>in</strong>ts<br />

<strong>in</strong>dividual-level, knowledge-based constra<strong>in</strong>ts


Basic properties<br />

a sequence of two <strong>verbs</strong> as one word<br />

two major types: syntactic <strong>and</strong> lexical<br />

• syntactic: yomi-hazimeru (read-beg<strong>in</strong>)<br />

‘beg<strong>in</strong> to read’<br />

• lexical: osi-taosu (push-topple) ‘topple<br />

by push<strong>in</strong>g, push down’


Kare-wa hon-o yomi-hazime-ta<br />

He-Top book-Acc read-beg<strong>in</strong>-Pst<br />

‘He began to read a book.’<br />

Kare-wa hei-o osi-taosi-ta.<br />

He-Top fence-Acc push-topple-Pst<br />

‘He pushed down the fence.’


comparison with -te complexes<br />

V1s <strong>in</strong> compounds end <strong>in</strong> the Renyookei<br />

form; different from -te complexes<br />

compound:<br />

saka-o<br />

slope-Acc<br />

kake-agat-ta<br />

run-go.up-Pst<br />

-te complex<br />

saka-o<br />

slope-Acc<br />

kakete it-ta<br />

run-TE go-Pst


“lexical <strong>in</strong>tegrity”<br />

Unlike -te complexes, noth<strong>in</strong>g can be<br />

<strong>in</strong>serted between V1 <strong>and</strong> V2 <strong>in</strong> compounds<br />

saka-o kakete-wa it-ta ga...<br />

slope-Acc run-TE-Cont go-Pst but<br />

‘Though he ran (up) the slope, ...’<br />

*saka-o kake-wa agat-ta ga...<br />

slope-Acc run-Con go.up-Pst but<br />

‘Though he ran up the slope, ...’


lexical <strong>in</strong>tegrity of syntactic<br />

compounds<br />

V-te iru (progressive) vs V-hazimeru<br />

Kare-wa hon-o yonde-wa i-ta ga...<br />

he-Top book-Acc read-TE-Cont be-Pst but<br />

‘Though he was read<strong>in</strong>g a book, ...<br />

*Kare-wa hon-o yomi-wa hazime-ta ga...<br />

he-Top book-Acc read-Cont began but


syntactic <strong>and</strong> lexical compounds<br />

[Kare-wa [ ___ hon-o yomi]-hazime-ta]<br />

he-Top book-Acc read-beg<strong>in</strong>-Pst<br />

‘He began to read a book.’<br />

[Kare-wa hei-o osi-taosi-ta]<br />

he-Top fence-Acc push-topple-Pst<br />

‘He pushed the fence down.’


More on lexical compound <strong>verbs</strong><br />

1157 entries <strong>in</strong> Tagashira & Hoff;<br />

2936 <strong>in</strong> Iwanami Goitaikei;<br />

probably lots more (Who knows?)<br />

productive (<strong>in</strong> some types): new<br />

comb<strong>in</strong>ation is possible<br />

“fixed” comb<strong>in</strong>ations <strong>in</strong>volv<strong>in</strong>g archaic<br />

<strong>verbs</strong> are found


Major types<br />

Co-actions naki-sakebu (cry-shout) ‘cry out’<br />

Means osi-taosu (push-topple) ‘push down’<br />

Manner kake-agaru (run-go.up) ‘run up’<br />

Cause<br />

yake-s<strong>in</strong>u (be.burned-die) ‘die from<br />

be<strong>in</strong>g burned’<br />

Background tabe-nokosu (eat-leave.unconsumed)<br />

‘leave unconsumed <strong>in</strong> the act of eat<strong>in</strong>g’


Note: <strong>in</strong>choative/causative pairs<br />

koware(-ru)<br />

kowas(-u)<br />

taore(-ru)<br />

taos(-u)<br />

ak(-u)<br />

ake(-ru)<br />

‘be.destroyed, break(<strong>in</strong>tr.)’<br />

‘destroy, break (trans.)’<br />

‘fall over’<br />

‘topple’<br />

‘open (<strong>in</strong>transitive)’<br />

‘open (transitive)’


comparison with English<br />

Susan pushed the door {open/down}.<br />

<strong>Compound</strong>s lexically convey mean<strong>in</strong>gs<br />

which are conveyed through verb +<br />

particle/resultative constructions <strong>in</strong><br />

English<br />

<strong>Japanese</strong> osi-akeru/osi-taosu <strong>in</strong>volve<br />

transitive akeru ‘open’ <strong>and</strong> taosu ‘topple’<br />

as the head verb <strong>and</strong> <strong>in</strong> this respect<br />

different from English counterparts.


Comparison with Ch<strong>in</strong>ese<br />

Ta tui-dao-le dashu<br />

he push-fall.down Prf big.tree<br />

‘He pushed down a big tree.’<br />

Unlike Ch<strong>in</strong>ese, <strong>Japanese</strong> has transitive<br />

taosu ‘topple’ as V2; the subject of V1<br />

<strong>and</strong> V2 are shared <strong>in</strong> <strong>Japanese</strong>, but not <strong>in</strong><br />

Ch<strong>in</strong>ese.


Question<br />

Though productive, not all comb<strong>in</strong>ations are<br />

possible; 2,353 <strong>verbs</strong> would produce more<br />

than 5,500,000 comb<strong>in</strong>ations <strong>in</strong> theory.<br />

How are the comb<strong>in</strong>ations of lexical V-V<br />

compound <strong>verbs</strong> constra<strong>in</strong>ed?<br />

Subevents are <strong>in</strong>tegrated <strong>in</strong>to one (complex)<br />

event --> may suggest possible patterns <strong>in</strong><br />

which complex conceptual materials are<br />

<strong>in</strong>tegrated <strong>in</strong>to one whole <strong>in</strong> language.


Previous accounts:<br />

General solutions (only)<br />

transitivity match (traditional)<br />

“transitivity harmony” (Kageyama 1993)<br />

subject shar<strong>in</strong>g (Yumoto 1996, 2005,<br />

Matsumoto 1998)


Transitivity Match<br />

Transitive + Transitive<br />

tataki-kowasu (strike-destroy)<br />

Intransitive + Intransitive<br />

kake-agaru (run-go.up)<br />

*Transitive + Intransitive<br />

*tataki-kowareru (strike-be.destroyed)<br />

*Intransitive + Transitive<br />

*kake-ageru (run-make.go.up)


ut...<br />

Some transitive-<strong>in</strong>transitive comb<strong>in</strong>ations<br />

are found<br />

moti-aruku (have-walk) ‘carry around’


transitivity harmony<br />

Two types of <strong>in</strong>transitive <strong>verbs</strong><br />

• unaccusative (roughly, nonagentive changeof-state<br />

<strong>verbs</strong>)<br />

• unergative (roughly, agentive action <strong>verbs</strong>)<br />

New comb<strong>in</strong>atory generalizations (Kageyama 1993)<br />

OK: Transitive/unergative + Transitive/unergative<br />

OK: Unaccusative + Unaccusative<br />

bad: *Transitive/unegative + Unaccusative<br />

Issue of aruki-tsukareru (walk-get.tired) solved!


ut...<br />

Some unergative-unaccusative pairs are<br />

found <strong>in</strong> certa<strong>in</strong> semantic types<br />

(Matsumoto 1996)<br />

aruki-tukareru (walk-get.tired) ‘get tired<br />

from walk<strong>in</strong>g’


subject shar<strong>in</strong>g<br />

Subjects of V1 <strong>and</strong> V2 must be the same<br />

entity (Yumoto 1996, Matsumoto 1998)<br />

aruki-tukareru (walk-get.tired): subject<br />

shared<br />

*tataki-kowareru (strike-be.destroyed):<br />

subject unshared


V-V Nouns<br />

V-V nouns do not respect Subject-shar<strong>in</strong>g<br />

(Matsumoto 1998)<br />

V: omoi-dasu (th<strong>in</strong>k-make.go.out) ‘recall’<br />

N: omoi-de (th<strong>in</strong>k-go.out) ‘remembrance’


But...<br />

Subject-shar<strong>in</strong>g alone is too general <strong>and</strong><br />

says too little.<br />

tataki-kowareru (strike-be.destroyed)<br />

would be OK <strong>in</strong> the read<strong>in</strong>g where the<br />

striker is destroyed.<br />

Such compounds are ruled out because<br />

of queer semantics, but queerness must<br />

be separately accounted for.


General solutions II<br />

semantic constra<strong>in</strong>ts based on time <strong>and</strong><br />

causation (Matsumoto 1996: Ch. 10)<br />

Subevents of a complex event<br />

(represented by V1 <strong>and</strong> V2) must be<br />

temporally coextensive or causally<br />

related


But...<br />

Such constra<strong>in</strong>ts may be a necessary<br />

condition for a complex event to be<br />

expressed <strong>in</strong> a s<strong>in</strong>gle verb, they are too<br />

general to restrict possible lexical<br />

compounds.<br />

kake-agaru ‘run-go.up’<br />

*tati-yomu ‘st<strong>and</strong>-read’<br />

cf. tati-yomi suru (st<strong>and</strong>-read do)


The <strong>in</strong>adequacy of these general<br />

approaches was that they looked for<br />

general constra<strong>in</strong>ts without consider<strong>in</strong>g<br />

properties of different types of<br />

compounds


Solution<br />

While general constra<strong>in</strong>ts do exist,<br />

specific constra<strong>in</strong>ts on different types of<br />

lexical compounds must be recognized;<br />

some even at the level of <strong>in</strong>dividual<br />

items, too.<br />

--> Constra<strong>in</strong>ts are stated at multiple<br />

levels of generality (cf. Booij 2010,<br />

Goldberg 1995)


V1-V2<br />

general<br />

constra<strong>in</strong>ts<br />

V1cause-V2<br />

V1means-V2<br />

V1manner-V2<br />

type-level<br />

constra<strong>in</strong>ts<br />

tataki-kowasu<br />

osi-taosu<br />

<strong>in</strong>dividuallevel<br />

constra<strong>in</strong>ts


Type-level constra<strong>in</strong>ts<br />

Lexical compounds must be <strong>in</strong>terpreted <strong>in</strong><br />

a restricted set of semantic schemas. If they<br />

do not match any of such schemas, they are<br />

ruled out as “impossible compounds.”


major schemas<br />

Co-actions (Pair compounds <strong>in</strong> M96)<br />

Means<br />

Manner<br />

Cause<br />

Background (Not found <strong>in</strong> M96)


Note: represent<strong>in</strong>g verb semantics<br />

Bill broke the vase.<br />

[x CAUSES [Y BECOME Z]]<br />

CAUSATION<br />

CHANGE<br />

(STATE,<br />

LOCATION)


Co-action compound<strong>in</strong>g<br />

hikari-kagayaku (sh<strong>in</strong>e-sh<strong>in</strong>e.brightly) ‘sh<strong>in</strong>e brightly’<br />

naki-sakebu (cry-shout) ‘cry out’<br />

tobi-haneru (jump-leap) ‘jump up’<br />

V1 <strong>and</strong> V2 represent actions<br />

Actions of V1 <strong>and</strong> V2 must be the same action<br />

(described <strong>in</strong> different ways)<br />

V1 <strong>and</strong> V2 share their actors.


Co-action schema<br />

V1<br />

V2<br />

action<br />

identity


Means compound<strong>in</strong>g<br />

tataki-kowasu (strike-destroy) ‘destroy by strik<strong>in</strong>g’<br />

keri-ageru (kick-make.go.up) ‘kick up (<strong>in</strong> the air)’<br />

very productive<br />

V2 represents causation + change of state<br />

V1 represents action<br />

V1 is the means by which V2 is performed<br />

V1 <strong>and</strong> V2 share their subjects, <strong>and</strong> usually<br />

their objects


V1<br />

V2<br />

Means


Cause compound<strong>in</strong>g<br />

obore-s<strong>in</strong>u (be.drowned-die) ‘be drowned (to<br />

death)’<br />

aruki-tukareru (walk-get.tired) ‘get tired from<br />

walk<strong>in</strong>g’<br />

V2 must be nonagentive<br />

V1 can be agantive or nonagentive (<strong>in</strong><br />

violation of transitivity harmony)<br />

Undergoers/actors of V1 must be<br />

identical with the undergoer of V2


Cause schema<br />

V1<br />

V2<br />

process-type neutral


Manner compound<strong>in</strong>g<br />

kake-oriru (run-go.down) ‘run down’<br />

mai-agaru (dance-go.up) ‘soar up’<br />

V1 represents an action<br />

V2 represents motion (or similar activity)<br />

V1 represents the manner <strong>in</strong> which V2 is<br />

carried out.


Manner schema<br />

motion<br />

Manner<br />

action


nonexist<strong>in</strong>g schemas<br />

Non-subject-shar<strong>in</strong>g<br />

Nonsubject-shar<strong>in</strong>g cause-result schema<br />

Subject-shar<strong>in</strong>g<br />

• sequential actions<br />

• circumstantial situations


schema excluded <strong>in</strong> <strong>Japanese</strong><br />

V1<br />

V2<br />

cf. tui-dao (push-fall) <strong>in</strong> Ch<strong>in</strong>ese


*tataki-kowareru (strike-be.destroyed)<br />

*tataki-s<strong>in</strong>u (strike-die)<br />

There are restricted exceptions. They are not<br />

the direct results of compound<strong>in</strong>g but are<br />

derived from the means compounds.<br />

uti-agaru (hit-go.up) ‘go up due to shoot<strong>in</strong>g’<br />

< uti-ageru (hit-make.go.up) ‘shoot ... up’


Subj-shar<strong>in</strong>g cases excluded<br />

Circumstantial situations<br />

V-V nouns + suru permitted<br />

tati-guwi suru (st<strong>and</strong>-eat do) ‘eat while<br />

stad<strong>in</strong>g’<br />

tati-yomi suru (st<strong>and</strong>-read do) ‘read while<br />

st<strong>and</strong><strong>in</strong>g’<br />

V-V <strong>verbs</strong> are excluded<br />

*tati-kuwu (st<strong>and</strong>-eat), *tati-yomu (st<strong>and</strong>-read)


Subj-shar<strong>in</strong>g cases excluded<br />

sequential actions<br />

hiki-nige suru (hit-flee do) ‘hit <strong>and</strong> run’<br />

*hiki-nigeru (hit-flee)<br />

multiple (repeated) actions<br />

agari-sagari suru (go.up-go.down) ‘go up <strong>and</strong><br />

down repeatedly’,<br />

iki-ki suru (go-come do) ‘come <strong>and</strong> go<br />

repeatedly’<br />

*agari-sagaru, *iki-kuru


subject-shar<strong>in</strong>g<br />

as a general schema<br />

subject shar<strong>in</strong>g is a property of a “superschema,”<br />

i.e., a generalization of all<br />

major permitted compound<strong>in</strong>g schemas.<br />

Not all subject-shar<strong>in</strong>g schemas are<br />

permitted <strong>in</strong> compounds (hence we need<br />

a list of permitted ones), but all major<br />

schemas permitted <strong>in</strong>volve subjectshar<strong>in</strong>g<br />

(hence we need a generalization).


Subj-shar<strong>in</strong>g schema<br />

V2<br />

V1


constra<strong>in</strong>ts at the level of<br />

<strong>in</strong>dividual compounds<br />

semantic congruency<br />

Each compound must have a mean<strong>in</strong>g<br />

which is congruent with our general<br />

knowledge concern<strong>in</strong>g what typically or<br />

conventionally happen <strong>in</strong> the world.<br />

lexical economy<br />

Each lexical compound must have a reason<br />

for existence <strong>in</strong> light of lexical economy.


congruency<br />

Congruency <strong>in</strong>volves world knowledge<br />

(Fillmore’s “frame”)<br />

Each verb has a background <strong>in</strong>formation<br />

concern<strong>in</strong>g the likely causes <strong>and</strong> results<br />

of the process or the manners/means by<br />

which the process may be executed.


For example, a causative verb like break has<br />

a set of background <strong>in</strong>formation on how<br />

the change may be achieved; break<strong>in</strong>g can<br />

be achieved by actions <strong>in</strong>volv<strong>in</strong>g forceful<br />

contact (e.g., hitt<strong>in</strong>g) but not by<br />

nonforceful contact (e.g., lick<strong>in</strong>g).<br />

Each compound mean<strong>in</strong>g is checked aga<strong>in</strong>st<br />

such background knowledge. Incongruence<br />

results <strong>in</strong> <strong>in</strong>congruent compounds.


<strong>in</strong>congruent compounds<br />

• tataki-kowasu (strike-destroy) ‘strike to<br />

pieces’ vs. *name-kowasu (lick-destroy)<br />

• aruki-tukareru (walk-get.tired) vs. *arukis<strong>in</strong>u<br />

(walk-die)<br />

Our mental database says that one is likely<br />

to be tired after walk<strong>in</strong>g, but one does not<br />

usually die from it.


Change <strong>in</strong> our knowledge may lead to<br />

new compound<strong>in</strong>g possibilities.<br />

• name-otosu (lick-make.fall) ‘make ... fall<br />

off by lick<strong>in</strong>g’ may become acceptable<br />

after acquir<strong>in</strong>g the knowledge of horse<br />

mothers lick<strong>in</strong>g dirty objects off their<br />

foals’ sk<strong>in</strong>s.


lexical economy<br />

The pr<strong>in</strong>ciple of lexical economy rules out<br />

complex forms as unnecessary if they are<br />

synonymous with irregular/simpler forms<br />

(cf. *goed (=went), *pale red (≒p<strong>in</strong>k)).<br />

<strong>Compound</strong>s are blocked by (functionally)<br />

synonymous simple <strong>verbs</strong> or “more fixed”<br />

compound <strong>verbs</strong>.<br />

*kake-hairu (run-enter) cf. kake-komu<br />

*aruki-agaru (walk-go.up) cf. agaru


In conclusion<br />

<strong>Japanese</strong> lexical compounds do provide<br />

<strong>in</strong>terest<strong>in</strong>g data for the l<strong>in</strong>guistic<br />

patterns of packag<strong>in</strong>g complex mean<strong>in</strong>gs<br />

<strong>in</strong>to one whole.<br />

Constra<strong>in</strong>ts need to be stated at<br />

different levels of generality <strong>in</strong><br />

morphology.


Booij, Geert. 2010. Construction Morphology. Oxford: Oxford University<br />

Press.<br />

(Kageyama, Taro). 1993. <br />

Matsumoto,Yo. 1996. Complex Predicates <strong>in</strong> <strong>Japanese</strong>:A Syntactic <strong>and</strong><br />

Semantic Study of the Notion 'Word'. Stanford: CSLI Publications,<br />

Tokyo: Kurosio Publishers.<br />

(Matsumoto, Yo). 1998.<br />

114: 37~83<br />

Tagashira, Yoshiko, <strong>and</strong> Jean Hoff. 1986. H<strong>and</strong>book of <strong>Japanese</strong> <strong>Compound</strong><br />

Verbs. Tokyo: Hokuseido.<br />

(Yumoto, Yoko). 1996. <br />

105-118. <br />

(Yumoto, Yoko). 2005. ―

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!