P-1B-23 複文理解過程における限定形容詞の処理の検討(2001年度 日本基礎心理学会第20回大会優秀発表賞)

The Japanese Psychonomic Society

NII-Electronic Library Service

TheJapanesePsychonomicSociety

lke

kwnese

jbntanai

of

isi,chonomic'

2002,

Vol,

21,

No.

1,

43-44

Science

Prizewinner's

SummaryP-IB-23

Comprehension

of

Japanese

adjective-noun

Daisuke

Hiroshima

sentences

including

phrases

FuJIKI

(inive7sily'

This

study

investigated

comprehension

processes

of

Japanese

sentences which

included

an adjective-noun

_phrase,

Specifically,

we examined whether evaluation of a consistent semantic relation

between

an

adjective

and a noun

in

a sentence of reading

material

was

suspended

until

the

argument structure of

the

sentence was

constructed.

Twenty-three

_participants

read

two

types

of

sentences

:

acceptable

sentences

that

contained a

plausible

adiectjve-noun

_phrase,

and

unaccepta-ble

sentences

that

contained

an

implausible

adjective-noun

_phrase.

In

comparison with

the

acceptable

sentences,

phrase-by-phrase

reading

time

for

the

unacceptable sentences was

_prolonged

not only at

the

_position

of adjective-noun

phrase

but

also

at

the

position

of

the

verb

that

construct-ed

the

propositional

representation ef

the

sentence.

This

result

suggests

that

evaluation

of a semantic relation of an

adjective-noun

_phrase

is

_postponed

until

the

coherent

argument

structure

is

constructed.

Key

words : sentence

comprehension,

adjective-noun

_phrase,

propositional

representation

A

Japanese

sentence such as "gghS[]iJilltzztL7t

(A

police

othcer was afraid of a

_puppy)"

often

causes

a

negative reaction

by

a

reacler,

but

a

sentence

such

as

"mefiigggabSjikigMtLk

(A

cowardly

police

oMcer

was

afraid

of a

_puppy)']

does

not.

The

difference

in

the

reaction

to

these

sentences

depends

on whether

the

attributive

adjective

"cowardly"

modifies

the

noun

[`police

oMcer"

in

the

noun

_phrase

(Murphy,

1990).

According

to

Murphy

(1990),

an

adjective

modifies a noun

by

altering a

schema

in

some

way, and

the

schema can

be

described

as

a

structured

set

of

slots

and

fi11ers.

Within

this

frame-work,

the

reason

fer

the

awkwardness

evoked

by

the

sentence

mentioned

above

can

be

attributed

to

the

idea

that

the

schema of a

police

oficer

may

have

a

PERSONALITY

slot with

brave

as

a

default

value.

However,

in

the

second example,

,the

adjective

`Ccow-ardly"

replaces

the

value of

the

PERSONALITY

slot

with

cowantly.

The

readers

therefore

evaluate "a

'

_Graduate

_school

_{of education,} sity,

1-1'2

Kagamiyama,

Hiroshima,

739"8523

Hiroshima

Univer-Higoshi-hiroshima,

cowardly

_police

oficer"

as one who could

_plausibly

be

afraid ef a

puppy,

Consequently

readers

find

the

sentence "a

cowardly

po]ice

oMcer was

afraid

of

a

puppy"

as

beleivable.

According

to

the

proposal

that

sentence

compre-hension

is

a

_process

of construction of

propositional

representations, readers must

judge

the

_plausibility

of concepts of adjective-noun cembinations as either a subject or object of

the

predicate,

We

therefore

examined

whether evaluation of a consistent

seman-tic

relation

between

an adjective and

a

noun

in

a

sentence

was suspended until

the

argument structure

of

the

sentence

was

constructed.

To

examine

this

possibility

we

compared

the

comprehensien of

two

types

ef

sentences

that

had

the

same syntactic

struc-ture

but

with a

different

semantic

relation

in

the

'

adjective-noun

phrase.

One

sentence was

acceptable

'

and

included

a

_plausible

adjective-noun

phrase

: "ve

Lv)gghSfilMbe'ti-"

v'

(A

gentle

police

oficer

tell

a

joke)".

The

other

sentence

was an unacceptable

ancl

included

an

implausible

adjective-noun

phrase

:

"ev)

Y:bgJi:ta

igES

(A

thick

po]ice

orncer

tell

a

joke)",

We

compared

the

reading

tirnes

at

the

verb where

the

The Japanese Psychonomic Society

NII-Electronic Library Service

TheJapanesePsychonomic Society

44

The

Japanese

Journal

ofPsychonomicScience

Vol,

21,

No.

1

propositional

representation of

the

sentence was

con-structecl.

If

judgment

of

the

consistency of

the

adje6tive-noun

_phrase

was suspendecl

the

reading

time

of unacceptable

sentences,

compared

with

the

reading

time

of acceptable

sentences,

would

be

pro-longed

at

the

verb.

Methods

M'aterial

The

reading material was

Japanese

sen･

tences

consisting of an adjective-noun

_phrase,

a

bare

noun

_phrase,

and a verb.

The

adjective-noun

_phrase

was either

the

subject or

the

object of

the

sentence.

A

verb

is

usually

placed

at

the

final

pesition

in

Japanese

sentences.

The

word order was

"`subject-object-predicator(verb)".

'In

the

_present

experiment

com-plex

sentences were used so

that

the

reading

_process

at

the

position

of verb could

be

determined,

'

A

com-plex

sentence consists of acomplement

sentence

(that

is,

a

sentence

of

reading

material)

followed

by

amain

sentence.

For

example,

the

reading

material

sentence

"eeLL)grrbgfi

_thigS'

_5"

was

followed

by

the

main

sentence "tkmaSAvhSeeV]k".

Two

lists

of

the

experimental material were created.

Each

list

con-sisted

of

40

reading material

sentences

based

on

a

2

(acceptability:

acceptable

vs.

unacceptable)

×

2

(position

of

the

adjective-noun

_phrase

:

subject

vs.

'object)

factorial

clesign

(10

sentences

per

condition), and

20

fi11er

sentences

(half

of

them

were

ungrammat-ical).

Procedure

In

every

trial,

a

sentence

was

present-ed

_{phrase-by-phrase:a}

_phrase

appeared

and

remained

on

the

screen

until a

following

phrase

was

added

to

it

when

the

participant

pressed

the

space

bar

of

the

keyboard.

A

microcomputer

measurecl

the

'

'

interval

from

the

appearance of a

_phrase

to

the

key

'

press

for

the

next

_phrase.

Each

main sentence

ended

in

a verb and

the

_participants

made

a

response

_to

indicate

whether'the sentence made sense.

Twenty

three

students

_{participated.}

Results

and

Discussion

Tablel

lists

the

mean reading

times

for

the

adjective-noun

_phrases,

'the

bare

noun

_phrases,

and

the

verbs

of

the

reading

material

sentences,

A

2

×

2

Table

1

The

mean reading

times

(in

millisecond>

for

the

adjective-noun

phrases,

bare

noun

phrases

and

verbs.

Accepta-

adj.-n

bility

Position

Mean

Rading

Times

(SD)

SubjectObject

Predicator

Subject

Accep-table

Object

104S(376)

vaLL]gtrht

adj.-n

707(275)

kJtrt"

neun

735(138)

zsce

noun

918(323)

egv]dive

adj,-n

844(208)

zscig

noun1052<372)

iuthigMig

adj,-n

638(113)

:5

verb674(160)

fi6

verb

Subject

UIaabCfeeP'

object

1106(441)

ev]mp.#etht

adj.-n

661(191)

ktrhr

noun

694(171)

S5

verb758(196)

fi6

verb

ANOVA

was calculated

for

each

mean reading

time.

A

marginally reliable main effect of acceptability

(FkubjecL(1,22)=4.74,

P<.05;

Etern(1,18)=3.70,

P

<

_.07)

was observed

for

the

adjective-noun

_phrases.

For

tfie

reading material sentences

the

verb

displayed

a reliab]e rnatn effect of acceptability

(Fkubjeet(1,

22)=

7.75,

_P<.05;

Et..(L18)==11.10,

P<.Ol).

These

results

indicated

that

in

comparison

to

the

acceptable

sentences,

the

phrase-by-phrase

reading

time

of

the

unacceptable

sentences

was

_prolonged

not only at

the

adjective-noun

_phrase

but

also at

the

verb, where a

prbpositional

representation of

the

sentence must

be

constructed.

This

result

supported

the

prediction

that

the

evaluation

of

a

consistent

semantic

relation

in

an

adjective-noun

_phrase

was

suspended

until

the

argu-ment structure of

the

sentence was constructed.

The

findings

suggest

that

readers

depend

on a

grammati-cal category of words and on

the

syntactic

informa-tion

of

a

sentence

to

combine

the

arguments

with

the

predicate.

In

addition

readers

use

the

meaning

of

words and

their

knowledge

of

the

world

to

construct coherent argument structures.

References

Murphy,

G,

L.

1990

Noun

phrase

interpretation

and

conceptual combinatton.

foumal

of

Memory

and