1P093 校正読みにおける誤字の音韻と位置の効果(2003年度 日本基礎心理学会第22回大会優秀発表賞)

The Japanese Psychonomic Society

NII-Electronic Library Service

The JapanesePsychonomic Society

7vae

,laPangse

Jou"nalo.f lteychonomicScience

200.1･,Vol.23,No,1,103-le4

Summary

ofAwarded

PresentationIP093

The

effectsof

_phonology

and

misprint

on

Japanese

_proofreading

location

MichikoAsANo

and

Kazuhiko

7:heUitiversily

of

7bkyo'YOKOSAWA

The infiuencesof _phonology at the_pitchaccent leveland misprint

lecation

in

a sentence on

Japanese

proofreading were investigated.The _participantssilcntly read short

Japanese

sentences

and

detected

misprints

in

them.

We

found

that

more misprint detectionerrors were made when

the misprint was the sarne _pitch-accent

homophone

as thatof a word thatwould fitthe sentence meaning, than when themisprint was a different_pitch-accenthomophone or a non-homophone.

This

result suggests that phonologica}

jnformation

at the pitch-accent levelisused

in

Japanese

proofreading,

The

interaction

of thc phonology and the misprint

locatien

had

an

infiucnce

on misprint detectionspeed, This indicatesthatthecontextual

information

of a sentence

is

essential

formisprint detection,and the homophony and the locationof the misprint jnfiucnccsthespeed of

context comprehension.

Key words: hornophone, Kanji._{phonology, proofreading,}sentence processing

I)uring proofreading, some misprints can

be

de-tectedeasily and others -'ith diMculty. The

charac-teristicsof _{pr()ofrcading}errors may contribute toa

betterunderstanding of thereading proces$,

Some research

has

revealed that phonological

in-forrnation of misprints isactivated during silent

proofrcading

(e,g.

Morita

&

Tamaoka, 2002).

How-ever, previous proofreading studies

have

not taken

account of pitch accent, which conveys important

phonological

iniormation

in

the

Japanese

language.

Inthisstudy, we

investigaLed

the

infiuence

of

pho-nology at the _pitch accent levelon

Japane$e

proof-reading. We alse investigated the effect of the

loca-tionof a rnisprint ina sentence on

Japanese

proof-rcading, The

location

of themisprint could

influence

thespeed of misprint detectionbecause

it

could

influ-ence the speed of context comprehension, which is

esscntial

for

rnisprint

detection.

Method

Participants

Eighteen

Japanese

undergraduate students _participated

in

thisexperiment,

Stimuli Sixty short

Japanese

$entences

(12-14

*

_Department

_of

_Psychology,

_Graduate

_School

_of

Humanities

and

Sociology,

The

University

of

Tokyo, 7-3-1 Hongo, Bunkyo-ku,

Tokyo

O033

charactcrs in length) were used as stimuli. Each

$cntence contained one misprint, which was a

two-Kanji eharacter noun, There were threeconditions

involving thehomophony of themisprint:

(a}

a same

pitch-accent homophone condition, inwhich the rn

is-prints were the same pitch-accent

homophones

as

the correct words

(sharing

the same consenants, vowels, and _pitchaccents). For example, ""FYFLgelt

ali/koushi!e:

L'(iglJlk.''

(the

letter$

in

italics

in/1

represent higher _pitchcs,and those not

in

italics

lower

pitchcs,

Correct

word: `$Efii'

/koushif);

(b)

a

different

pitch-accent homophone condition, in

which themisprints were hornophonic with the

cor-rect words but a differentpitch accent

(e.g.

`msL]t'

1

koushil);

and

(c)

a non-homophene condition, in

which themisprints were not the

homophone

ofthe

correct words

(e.g.

`its'

/bibouf).

There

were also

threeconditions forthelocationof the misprints

in

each sentence:

_(a)

a top condition, inwhich the

mis-printwas the firstword of thesentence;

(b)

a middle condition, inwhich the misprint was

in

thernlddle

11

3

of the sentence; and

(c>

a bottom condition, in which the misprint was

in

the

Iast

1/3

of the

sen-tence. Sixty correct sentences were addcd as fi11ers.

Procedure Thc experiment consisted of one

block

of

120

trials.

The

participantswerc required to

con-The Japanese Psychonomic Society

NII-Electronic Library Service

The JapanesePsychonomic Society

104 The

_Japanese

_Journal

of Psychnomic

Science

VoL23, No. 1

tained misprints by reading the sentence silently,

and to respond by _pressing a key as _quickly as

possible.

To

rnake sure thattheparticipants

compre-hended

the sentences which they read an easy

com-prehension testwas added at theend of each triaL

Results

and

Discussion

Responses

to

trials

in which errors were made in

thesentence comprehension test

(3.29%

of the total

tria]s)were excluded

frorn

the analysis.

Error rates The rnean error rates are shown

in

Figure 1(a), The anaiysis revealed that the main effect of

homophony

was significant,

F(2,

34)=5.72,

P<.Ol.

Tukey's'HSD testrevealed thatmore errors

were made in detecting same _pitch-accent

homo-phone misprints than

in

detecting

different

pitch-accent or non-homophone misprints, both _ps<.05.

This result indicated

that

_participants used

phono-logical

information

atthe pitch-accent

level.

Correct response latencies The mean correct

re-sponse latenciesare shown inFigure 1(b).The rnain

cffect of the misprint

lecation

was revealed as

sig-nificant,

F(2,

34)=5.41,P<.Ol,

by

the analysis.

Tuk-ey's HSD testshowed thatthe_participantstook more

time _todetectmisprints when _they were atthe topof

the sentences than when they were

in

the middle,

P<,O1, Inaddition, theinteractionbetween

homoph-ony and the

location

was significant,

F(4,

68)=4.12,

P<.O05.

Tukey's HSD tcstrevcaled thatthe misprint

locationeffect was significant only inthe different

pitch-accentand the non-homophone conditions,

In

these two conditions the response latencieswere

]ongerwhen the misprints were at the topof a

sen-tence than when the misprints were

in

the middle of a sentence,

_P<,O05

and

_p<.OOI,

respectively.

These re$ults suggest thatcontextual information was used formisprint detection,and the locationof

the misprint

had

an

infiuence

on obtaining the

con-textual information. When a differentpitch-accent

or non-homophone misprint was at the topof a

sen-tence,the phonology of themisprint

interfered

with

the comprehension of thecontext, and misprint

de-tection was

delayed,

On

_the other hand, when the

misprint was

in

the middle of asentence the

preced-Agevm"oskLoLL ¢ =NoE

n op E

w h o c o v re e co = o a co m L "

8

t

8

E

2

Figure 10 8

6

4

2 o 18001700l6001500140013001200 Sarne-pitch-accent

(a)g,itf,fh?,r,e,n,t,-t

h.N,Op"h-,,, Homophony DSame-pitch-accent

elnDifterent-pitch-accent

Top

Middle

Bottom

(b)

Location

1.

{a)

The mean error rates

(b)

The

meancorrectresponse

latencies

Error

bars

represent the standard error of the mean.

ing

and

following

context was available and the

misprint was detected rapidly, Inaddition, a

mis-printwith thesame pitch-accent

homophone,

which

"soundcd"

correct, helpedtheparticipants to

compre-hend the

"correct"

sentence context regardLess of the

location

of the misprint.

Therefore

a

location

effect

was not observed inthc same _pitch-accent

homo-phone condition.

This

study rcvealed the use of phonology of

pitch-accent

level

and contextual

information

in

Japanese

proofreading, The use of _pitch accent in[ormation

might contribute toeMcient hemophone

discrimina-tion

in

Japanese

reading.

References

Morita,

A.

&

Tarnaeka,

K.

2002 Phonolegical

in-volvement

in

the processing of

Japanese

at the ]exicaland sentence levels.

Reading

and

VV)'iting,