!

"17, 2007 6
What do English Speakers Know about gera-gera and yota-yota:
A Cross-linguistic Investigation of Mimetic Words for Laughing and
Walking

IWASAKI Noriko*
David P. VINSON**
Gabriella VIGLIOCCO***

Key words: Onomatopoeia, mimetics, sound symbolism, iconicity

Summary
The relation between word form and meaning is considered arbitrary; however, Japanese mimetic
words, giseigo and gitaigo, are exceptions. For giseigowords mimicking voices
, there is a direct
resemblance
iconicity
between the sound of the word and the sound it refers to; for gitaigo
words that mimic manners/statesthere is a symbolic relationship
sound symbolism
between the
sound and the manner/state to which the word refers. While native speakers intuitively recognize
these relationships, it is questionable whether speakers of other languages are able to access the meaning of Japanese mimetic words from their sounds. In the current study, we asked native English
speakers with no prior experience with the Japanese language to listen to Japanese mimetic words for
laughinggiseigoand for walkinggitaigo
, and rate each word
s meaning on semantic differential
scalese.g.GRACEFUL-VULGAR
laughing
,GRACEFUL-CLUMSY
walking

. We compared English and Japanese speakersratings and found that English speakers construed many of the
features of laughing in a similar manner as Japanese native speakerse.g., words containing /a/ were
rated as more amused, cheerful, nice and pleasant laughs
. They differed only with regard to a few
sound-meaning relationships of an evaluative naturee.g., words for laughing containing /u/ were
rated as more feminine and graceful, and those containing /e/ were rated as less graceful and unpleasant
. In contrast, for the words referring to walking, English speakersratings differed greatly from
native Japanese speakers
. Native Japanese speakers rated words beginning with voiced consonants as
referring to a big person walking with big strides, and words beginning with voiceless consonants as
more even-paced, feminine and formal walking; English speakers were sensitive only to the relation
between voiced consonants and a big person walking. Hence, some sound-meaning associations were
language-specific. This study also confirmed the more conventional and lexicalized nature of the mimetic words of manner.

Assistant Professor, University of California, Davis

Research Fellow, University College London

Professor, University College London

53

54

1Introduction
Mimetic wordsgiongo/giseigowords that mimic sounds/voicesand gitaigowords that mimic
states or mannersare indispensable in vivid and precise descriptions and narratives in Japanesee.g.,
Kakehi 2001. They are commonly used in spoken language as well as in written language such as in
magazines and literaturesee, for example, Schourup 1993, Yamaguchi 1986, and are thus essential
items for learners of Japanese as a second languageJSLto learnMakino and Tsutsui 1986. Yet,
the acquisition of mimetic words seems problematic for JSL learnerse.g., Yamaguchi 1986, Hamano
1998.
Banana Yoshimoto, for example, uses mimetic words frequently in works such as Kitchen. Examples
12below show the use of the mimetic words kusu-kusu and zoro-zoro by Yoshimoto1a, 2aand
the English translations by Megan Backus1b, 2b. Despite natural and lucid translation into English,
native Japanese speakers may feel that the English translations do not sufficiently convey the senses expressed by the mimetic words.

1 a

"

... dakara anata no koto mo hito goto to wa omoenai no ne. Danzyo no ai ka doo ka hosyoo dekinai kedo.Kusu-kusu okaasan wa waratta 1
Yoshimoto 1991, 29
b
... So it follows that Yuuichi feels close to you. I can
t guarantee it
s romantic, though
Mom shook with laughter.translation by Megan Backus 1993, 1819
2 aTalking about a bus

Yoshimoto 1991, 52

Me no mae ni nagarete kite yukkuri tomari, hitobito wa narande zoro-zoro norikomu.

bIt seemed to float to a bus stop before my eyes, and people lined up, got on, one by one.
translation by Megan Backus 1993, 33.
Kakehi et al.1996define kusu-kusu, used in1, asthe manner of laughing to oneself
720,
and zoro-zoro, used in2, asthe manner in which a number ofusually livingthings follow one after another or gather together
1302. Kakehi et al.
s dictionary is the most comprehensive dictionary
written in English, yet native Japanese speakers may feel that these wordsand their soundsconvey
more than these definitions imply.
1

Following Hamano1998
, we represent Japanese words phonemicallye.g., t refers totsfollowed by /u/
and totfollowed by /i/ as well ast

, using Hepburn Romanization.

What do English speakers know about gera-gera and yota-yota

55

A question arises as to whether JSL learners can decipher any aspects of the manner of laughing or
the manner of entering a bus simply by the sounds of these mimetic words. In other words, can JSL
learners sense any relationships between the sounds of these mimetic words and their meanings upon
hearing or reading them? Although the relationship between wordssounds and their meanings is generally considered to be arbitrary in linguistic theorySaussure 1916, Japanese speakers may intuitively
feel that there is some resemblance between the sound of mimetic words and their meanings. Some linguists argue that this sound-meaning correspondence that native Japanese speakers sense is primarily
due to associations among Japanese words and meanings that they have learned over time, associations
which, therefore, are culture-specificJorden 1982. Others, however, argue in favor of a soundmeaning relationship in mimetic words in the Japanese language and have analyzed this relatione.g.,
Hamano 1998, Tamori & Schourup 1999.
The sound forms of mimetic words may resemble the meanings of the wordsi.e., iconicityor they
may symbolize their meaningsi.e., a sound symbolismpresumably on the basis of iconicity in some
cases. Some scholars have suggested that there exists universal sound symbolism. For example, the
vowels /i/ and /a/ are broadly associated with smallness and largeness respectivelyJespersen 1928, Sapir 1929, Newman 1933.2 While some researchers dispute the universality of sound symbolisme.g.,
Taylor & Taylor 1962, others find some commonalities of sound-meaning relations across speakers of
different languagessee, for example, Miron 1961, who compared English and Japanese speakers. If
sound symbolism is indeed universal to a certain extent, and if Japanese mimetic wordsat least partiallyreflect this, then JSL learners may sense and appreciate certain semantic dimensions of previously unheard or unseen Japanese mimetic words. If so, it is important to distinguish between those semantic dimensions of Japanese mimetic words that are readily perceived by learners of JSL and those
that are noti.e., associations built over time within Japanese culture.
The current study investigates whether and to what extent English speakers with no prior knowledge
of Japanese can appreciate the imagery of different aspects of the mimetic words for laughing and walking. The former is representative of giseigowords that mimic voicesand the latter representative of
gitaigowords that mimic mannersalthough it should be noted that many mimetic words seem to refer to both sounds and manners at the same time, evoking both auditory and visual imageryfor example, dosun-dosun refers to noises created by heavy people and of walking and floor-shaking at the same
time. We chose these two types of mimetic words because these two are presumably on the opposing

The association of /a/ with largeness and that of /i/ with smallness may not be entirely universal. According to
Diffloth1994
, in BahnarMon-Khmer language of Vietnam
, high vowels such as /i/ signify largeness and
low vowels such as /a/ smallness.

56

ends of the iconicity continuum: mimetic words for laughing being the most iconic, and the mimetic
words for walking being close to the least iconic. Because the former mimic laughing sounds which are
produced by the same organs as those used to articulate words, the sounds of these words are likely to
resemble the quality of voices and manners of laughing that the words refer to. In contrast, mimetic
words for walking refer to sounds made by footsteps and to manners that can be perceived visually or
by proprioception. There can only be an indirect resemblance between the wordssounds and meanings
in a more abstract, symbolic sense, especially in cases of mimetic words of manner. Hence, it is likely
that mimetic words for laughing possess iconic relationships that learners can take advantage of, but less
likely that mimetic words for walking do.
Specifically, we examine which semantic dimensions of these two types of Japanese mimetic words
native English speakers with no prior knowledge of the Japanese language can decipher. This is to avoid
two confounding factorsi.e., prior knowledge of any mimetic words and proficiency levels of learners
and to limit the potential influence of the first language to one languageEnglish, in this case, which
possesses only a limited repertoire of mimetic words. English speakers with no prior knowledge of
Japanese should be able to decipher more aspects of the mimetic words of laughter than of the mimetic
words for the manner of walking. We use existing lexicalized Japanese mimetic words in order to assess
the extent to which these words are associated with their meanings by conventionalized language and
cultural experience, versus the extent to which their sound-meaning relationships reflect iconicity and
sound symbolism.

2Previous literature

2-1Sound-meaning relationships in mimetic words of laughing and walking.

Hirose1981compared English and Japanese words that describe laughing and walkingas well as
other semantic domains. While English verbs are synthetic in that different manners of laughing and
walking are expressed by single verbse.g., giggle, chuckle, snigger for laughing; stroll , lumber, toddle, swagger for walking, the Japanese equivalents of these verbs are compositional as they are comprised of combinations of a generic verbwaraulaughor arukuwalkand mimetic words, such as
kusu-kusu waraugiggleand bura-bura arukustroll .He divides mimetic words for laughter and
those for walking into sub-groups: mimetic words for laughter into loud laughtere.g., hahaha, keta-

Q is used to refer to the first consonant of geminatesdouble consonants

, following Hamano1998
.

What do English speakers know about gera-gera and yota-yota

57

ketaand quiet, suppressed laughtere.g., uhuQ 3, kusu-kusu, and the mimetic words for walking into
three:1those that capture auditory aspects of walking such as dosin-dosin, dosun-dosun, and petapeta;2those that express visual aspects such as yoro-yoro, yota-yota, syanari-syanari, and yoti-yoti;
and3those expressing affective aspectsi.e., how people feelsuch as uro-uro. Hirose argues that
sequences -V1C2V2first vowel, second consonant, second vowelin C1V1C2V2 are thecorrectelements that are associated with meanings, rather than individual phonemes. Among such sequences he
discusses are -erareferring to lowbred, vulgar tasteand -urareferring to unsteadiness and swaying.
The former occurs in some mimetic words for laughinge.g., kera-kera, gera-gera, and the latter in
some mimetic words for walkinge.g., hura-hura, bura-bura.
Kakehi1986, who also compared Japanese and English mimetic words, argues that while English
counterparts reflect some sound symbolism, Japanese mimetic words do so more systematically and extensively. Hamano1998made a comprehensive study of this systematic and extensive sound symbolism. She distinguishes between CVConsonant-Vowel-based words such as pi, pii, piQ , pipiQ and
CVCV-based words such as pisi, pisiQ , pisi-pisi, considering the former more iconic and the latter
more conventionalized and lexicalized. According to her, /e/ has evaluative meaningsi.e., inappropriateness and vulgarity, and the other four Japanese vowels, /a/, /i/, /u/, and /o/ indicate shape of an object or the size of an affected area. The vowel /a/ signifies a large area, /i/ straightness and tenseness, /u/
smallness and protrusion, /o/ a small area, modesty, and inconspicuousness. In CVCV-based mimetic
words, these aspects of meaning are reflected only in the first vowel. Voiced and voiceless consonants
contrast with each other in that voiced consonants signify massiveness and heaviness while their voiceless counterparts signify the opposite. The first consonant in CVCV sequence is related to tactile properties, and the second consonant is related to movement. Table 1 summarizes the consonant meanings that
Hamano discusses. In addition, palatalization in such sequences as kyaQ -kyaQ , and yoro-yoro are associated with unsteadiness and childishnessHamano 1994, 1998. A number of other systematic soundmeaning associations have been described in the literature: moraic nasal signifying echoing, voiced consonantse.g., /g/, /d/referring to loudness, geminates referring to forceful and rigorous action, and a
final -ri referring to completion or quiet endinge.g., Hamano 1998, Hirose 1981, Tamori & Schourup
1999.
Tamori & Schourup1999examine Hamano
s1986analyses of sound symbolism found in Japanese mimetic words and compare English and Japanese sound symbolism. They argue for common
sound symbolism across the two languages, finding that words for similar referents share similar segments: e.g., words referring to sounds made by water contain bilabial stops and sibilants both in English
e.g., splash, sprinkle, spray, spatterand in Japanesee.g., batyaQ , basya, potyari. They also found

58
Table 1

Soundconsonants
-Meaning relationships according to Hamano1998
.

First consonants in CVCV

Second consonants in CVCV

/p/ taut surface; light, small, fine

explosion, breaking, decisiveness
/b/ taut surface; heavy, large, coarse
/t/

lack of surface tension, subduedness; light, small, fine

/d/

lack of surface tension, subduedness; heavy, large,

coarse

/k/ hard surface; light, small, fine

hitting of a surface, coming into close contact, complete agreement

opening, breaking up, swelling, expanding, puffing

out, emission from inside, surfacing, in-out movement.

/g/ hard surface; heavy, large, coarse

/s/

non-viscous body, quietness; light, small, fine

/z/

non-viscous body, quietness; heavy, large, coarse

soft contact, friction

/h/ weakness, softness, unreliability, indeterminateness

breath

/m/ murkiness

/n/ viscosity, stickiness, sliminess, sluggishness

bending, elasticity, unreliability, lack of force, weakness

/y/ leisurely motion, swinging motion, unreliable motion

sound from many sources, haziness, childishness

/w/ human noise, emotional upheaval

softness, faintness, haziness

/r/

rolling, fluid movement

Note: Blanks indicate that the given consonant never or rarely occurs in the position

similar sound symbolism in the closest equivalents of vowels in the two languages: /a/ referring to flatness, /o/ roundedness, /u/ small protrusion, /e/ to vulgar or negative referentse.g., belch, fleck, mess,
squelch, and fret in English. Tamori & Schourup
s1999study is based on existing words in the two
languages; so it is difficult to tell whether conventionalized sound-meaning relationships happen to correspond between the two languages in the specific cases investigated or if there exists a more general
level of sound symbolism such that speakers of English speakers can comprehend Japanese mimetic
wordsor vice versa. Some researchers discussed below have investigated how speakers of different
languages comprehend Japanese mimetic words.
2-2Experimental studies on sound symbolism in Japanese mimetic words.
Several experimental studies have examined whether non-native speakers of Japanese were able to
guess the meaning of mimetic words. Ihara and Iwahara1938asked four Japanese and three Chinese
speakers to give short sentences using each of 418 Japanese mimetic words. The data for the Japanese

What do English speakers know about gera-gera and yota-yota

59

showed that only 61 words were used to express the same meanings across the four speakers, suggesting
the flexibility of sound symbolism. The data for the Chinese indicated that symbolic sounds do not have
the same meaning in Chinese culture and that sound-meaning relationships are established as a custom
of the linguistic community. Likewise, Frei1970 cited in Hirose 1981investigated whether French
speakers were able to guess the meaning of Japanese mimetic words by asking four French speakers to
write down the meanings of words uttered by a native Japanese speaker. The French speakers were
given isolated words in the first experiment and then words with contexts given in French in a second
experiment. Frei found that French speakers also had difficulty guessing the meaning of the Japanese
words, but found that French speakers could guess slightly better when contexts were given, and better
with auditory mimetic words.
Oda2000argued that the sensations experienced by the movement of articulatory organs mediate
the connection between the linguistic sounds and meanings of mimetic words, and supported his hypothesis with experimental results. He provided mimetic words related to tactile sensations and movements to two groups of participantsprimarily English speakersand asked both groups to perform two
tasks: to choose a mimetic wordout of three words listedappropriate for a given English definition
accompanied by a drawing; and to match two definitions accompanied by drawings with two minimallypaired mimetic words. Participants in one group were asked to concentrate on sound images after hearing a native speaker read the words; those in the other group were asked to pronounce the words after a
native speaker read them. Those who pronounced the words made more correct responses, providing
evidence that articulation facilitated the connection between the sound and its meaning.
These previous studies show that speakers of other languages could not easily guess the meaning of
Japanese words, but that their performance improves in some situations. However, because these studies
did not investigate any specific semantic dimensions, they cannot explain which semantic dimensions
are associated with which phonological qualities of the Japanese mimetic words among speakers of
other languages. To answer this question, Iwasaki et al.2005compared Japanese speakersratings
and English speakersratings of 20 semantic dimensions of giongo, mimetic words for sounds. They
found that English speakers without prior Japanese study experience rated the meaning of giongo similarly to Japanese speakers for several semantic dimensions: whether the word refers to sounds that are
continuous, high-pitched, sharp, and deep. At the same time, English speakersratings diverged from
Japanese speakersratings for certain semantic dimensions, most notably the pleasantness and beauty of
the sounds to which the words refer.

60

2-3Acquisition of Japanese mimetic words

To our knowledge, no studies address JSL acquisition of mimetic words except for Yang1984,
who concluded that those who have more experience with Japanese culturee.g., those who had lived in
Japan, had Japanese relatives, had been exposed to Japanese TV programs, etcwere better at choosing
appropriate mimetic words. Unfortunately, proficiency in Japanese was not assessed, so it is impossible
to rule out the possibility that thiscultural experience effectmay instead simply be related to greater
Japanese proficiency.
More studies can be found on children learning Japanese as a first language. Toda et al.1990
found that Japanese mothers used more onomatopoetic words with their 3-month-old infants than US
mothers did, and that learning mimetic words appeared to pose no difficulty for Japanese children.
Ishiguro1993states that children first create their own idiosyncratic mimetic words and then gradually start using lexicalized mimetic words. He also notes that children first learn auditory mimetic
words, citing Okubo1967, who found that children started using sound mimetic words around the
ages of 23, and only laterages 45began using mimetic words of manner. Harlofsky1998confirmed through the testing of 3 to 6yearold children that the occurrence of auditory mimetic words
preceded that of visual mimetic words.
These findings are compatible with Frei
s1970finding that French speakers were able to guess the
meanings of auditory mimetic words better and Iwasaki et al
s2005finding concerning auditory mimetic words that Japanese and English speakersratings were correlated for a number of words and semantic dimensions. This compatibility suggests that JSL learners may find it easier to learn auditory mimetic words than visual mimetic words. It is important, however, to first examine which aspectsi.e.,
relationship between sound and semantic dimensionof Japanese mimetic words are more iconic and
are readily perceived by speakers of other languages in a manner similar to Japanese speakers, and how
auditory mimetic words differ from visual mimetic words. The current study specifically addresses differences between mimetic words that are primarily auditorylaughingand those of the manner of
walking that are more non-auditory and/or visual. Moreover, this study examines specific soundmeaning relationshipse.g., the relation between the respective vowel and its meaning.

What do English speakers know about gera-gera and yota-yota

61

3Method

3-1Materials and Procedures

We adopted Iwasaki et al
s2005method and created two questionnaires: one containing the 24 mimetic words for laughing listed in Appendix A and the other containing the 28 mimetic words for manner of walking listed in Appendix B. The selection of mimetic words for laughing does not include mimetic words for smilinge.g., nikkori, niyari, niko-niko, as these words refer to the visual aspect of
smiling rather than auditory aspects and are thus not giseigowords describing voices. The 24 mimetic
words for laughing consist of 17 CV-based words and 7 CVCV-based words. Of the 24 words for
laughter, 13 refer to loud laughtere.g., ahaha, hahaha, haQhaQ , waQhaQha, kera-kera, gera-gera,

Table 2

Semantic dimensions of laughing used in the questionnaire in two languages

In English

In Japanese

LoudSoftin volume

Pleasant voiceUnpleasant voice

GracefulVulgar

Beautiful voiceUgly voice

EnergeticNo energy

GoodBad

Mouth wide openMouth shut

High pitchedLow pitched

PurposefulInvoluntary

Resonant voiceNon-resonant voice

ContinuousMomentary

AmusedNot amused

FormalInformal

CheerfulNot cheerful

Many peopleOne person

FeminineMasculine

UnrestrainedRestrained

ExcitedCalm

Young childAdult

Nice/kindNasty/mean

62
Table 3

Semantic dimensions of walking used in the questionnaire

In English

In Japanese

NoisyQuiet

Big stridesSmall strides

Echoing noiseNon-echoing noise

FastSlow

RecreationalNon-recreational

Wet surfaceground, floorDry surfaceground, floor

Hard sole shoesSoft sole shoes

PurposefulAimless

Many peopleOne person

SteadyUnsteady

FormalInformal

Big personSmall person

Even pacedUneven paced

FeminineMasculine

EnergeticNo energy

GracefulClumsy

Long distanceShort distance

High self imageLow self image

GoodBad

Hard surfaceground, floorSoft surfaceground, floor

Young childAdult

keta-keta, keQkeQ , geta-geta, ohoho, kara-kara, koro-koro, kyaQkyaQ and 11 to quiet suppressed
laughteruhuhu, huhuQ , puQ , huhuhu, uhuQ , kusu-kusu, kusuQ , ehehe, hehehe, hihihi, heQheQhe.
The English definitions of these words from Kakehi et al.1996are given in Appendix A.
The 28 mimetic words for walking in the present study are largely CVCV-basedexcept for one
word, taQtaQ because most common mimetic words for walking are CVCV-based. They consist of 8
words that capture auditory aspects of walkingdosi-dosi, dosin-dosin, dosun-dosun, dota-bata, batabata, kotu-kotu, katu-katu, peta-peta, and 20 that capture visual and/or affective aspectsyoro-yoro,
yota-yota, yoti-yoti, noQsi-noQsi, syanari-syanari, teku-teku, bura-bura, hura-hura, sorori-sorori,
noso-noso, zoro-zoro, suta-suta, taQtaQ , tuka-tuka, toko-toko, tyoko-tyoko, uro-tyoro, hyoko-hyoko,
uro-uro, tobo-tobo; see Appendix B for English definitions. Although these words can be classified

What do English speakers know about gera-gera and yota-yota

63

into different modalitiesauditory, visual, or affectiveaccording to their primary or dominant interpretation, it is highly plausible that auditory mimetic words evoke certain visual as well as auditory images.
Thus, it is important to capture senses related to other semantic dimensions rather than limiting the semantic dimensions in question.
These mimetic words and semantic dimensions for investigation were selected through the consultation of several dictionaries of mimetic wordse.g., Atoda & Hoshino 1995, Kakehi et al. 1996, Shogakkan 1991. We selected 20 semantic dimensions relevant to the mimetic words for laughing as shown in
Table 2, and 21 semantic dimensions of the mimetic words for manner of walking as shown in Table 3.
For each word, a seven-point semantic differential scale was created for the 20laughingor 21
walkingdimensions. These scales were later converted to 3 to +3 to express polarity, with positive
values corresponding to the label of the scalee.g. an item rated the most loud is given a value of +3
on the loud scale, and one rated the most soft, a value of 3. Japanese speakers were given printed
wordseach word was written both in hiragana and katakana, while English speakers listened to digitized audio recordings of a native Japanese speaker pronouncing each of the words.
3-2Participants
The laughter questionnaire was completed by 12 native Japanese speakers11 females and 1 male,
ranging in age between 18-49, median 29, and the walking questionnaire was also completed by 12 native Japanese speakers11 females and 1 male, ranging in age between 21-53, median 28.5. Three
speakers participated in both. The English versions were completed by native English speakers with no
prior Japanese learning experience. The laughter questionnaire was completed by 15 English speakers
10 females and 5 males, ranging in age between 18-33, median 21, and the walking questionnaire was
completed by 12 English speakers8 females and 4 males, ranging in age between of 18-41, median
21. All participants were paid a small sum for their participation.
3-3Analyses
Two sets of correlation analyses were conducted initially. The ratings were averaged across participants within each language groupJapanese and English, and separate Pearson
s correlation coefficients for each word were calculated between the two groupsmean ratings of the various semantic dimensions. This analysis assesses the extent to which Japanese and English speakers agreed on their ratings of the 20 or 21 semantic dimensions for any given word, and reveals those words for which the

The words that start with vowelse.g., uhuhu, ahahawere excluded from this analysis.

64

two populations agree more than chance would predict. In addition, for mimetic words for laughing, the
average correlation of CV-based wordsN=10and that of CVCV-based wordsN=8were also computed to see if there is greater correlation for CV-based words, which are claimed to be more iconic
than CVCV-based wordsHamano 1998.4
Furthermore, separate correlation coefficients were calculated for each semantic dimension between
the two groupsmean ratings of the various words. This analysis assesses whether Japanese and English
speakers agreed on their application of a given semantic scale across the words in the test set, and reveals which of the scales are commonly applied by the two groups. In addition, the numbers of scales
for which the English-Japanese correlation reached significance for CV-based words and for CVCVbased words were counted.
Further, we also analyzed the effects of certain phonological qualities that have been argued to be
relevant in the mapping between meaning and sound. For words for laughing, we examined the effects
of vowelsi.e., /a/, /e/, /u/ in the V1 positions of C1V1- and C1V1C2V2-based words, onsetsi.e.,
whether the initial sound is a consonant or vowel, and geminatesi.e., whether the word contains double consonants or not. For the vowel comparison we used one-way ANOVA with ratings as the dependent variable and the different vowels as levels of the independent variable. For the other sound
properties we employed t-tests comparing the average ratings across the two languages. For words for
manner of walking, we analyzed the effects of different vowelsi.e., /a/, /o/, /u/by one-way ANOVA
as above, and the effects of voiced/voiceless word-initial consonants by t-test. The voiced/voiceless contrast can be examined only in word-initial consonants because voiced consonants do not occur in C2 positions. In order to ensure that these comparisons generalize across participants and materialsClark
1973, all of these analyses of phonological properties were conducted using participants as a random
factoraveraging across the different words in each condition for each participant, and treating phonological factors such as the voiced/voiceless contrast as within-subjects, and with items as a random
factoraveraging across the individual participants in each language group for each word and treating
phonological factors as between-items. Only those cases for which both subject and item analyses
reach significancealpha=.05are discussed as differences. Moreover, we compared means of English
and Japanese speakersratings for words containing -eralaughingand -urawalkingfor their most
relevant semantic dimensionsaccording to Hirose 1981: Gracefulas opposed to vulgarfor -era,
and Steady and Even-Paced for -ura.

What do English speakers know about gera-gera and yota-yota

Figure 1

65

Average ratings by Japanese and English speakers for kusu-kusu

4Results

4-1Mimetic words for laughter

4-1-1Correlation analyses for words

Japanese and English speakersratings were significantly correlated for half12 of 24of the words
used in this study. The most highly correlated ratings were for kusu-kusur=0.686, p<.001, rated by
both Japanese and English speakers as quiet, restrained, non-resonant laughter produced by a female
adult with her mouth shutsee Figure 1 for average ratings across all the semantic dimensions. This
was followed by kera-kera r=0.679, p<.001, hahaha r=0.664, p<.001, waQhaQha r=0.662, p
<.001, keQkeQker=0.617, p=.002
, eheher=0.575, p=.004, and kara-karar=0.559, p=.005. Ratings for four more words were significantly correlated albeit with lower correlation coefficients: huhuQ
r=0.456, p=.022, hehehe r=0.449, p=.024, uhuQ r=0.399, p=.041, and keta-keta r=0.388, p
=.045. The words for which English and Japanese speakersratings were correlated include a mixture
of both loud and quiet laughter. It is also interesting to note that the highest correlations were found for
the CVCV-based mimetic words kusu-kusu and kera-kera, words which are less iconic/more lexicalized
according to Hamano1998. The overall average correlation for CV-based words was r=+0.113 and

66

that for CVCV-based words was r=0.069, although this difference did not reach statistical significance.

4-1-2Correlation analyses for semantic dimensions

As expected from the above correlation patterns, there were also semantic dimensions for which Japanese and English speakersratings were very highly correlated. The most significantly correlated was
Loud r=0.78, p<.001, followed by Mouth-Wide-Open r=0.72, p<.001, Continuous r=0.68, p
<.001, Resonant-Voicer=0.63, p<.001. Only these four were significantly correlated across languages; the next highest correlation failed to reach significanceMany-People, r=.315, p=.067. Interestingly, there were also two semantic dimensions for which English and Japanese speakersratings
were significantly negatively correlated: Beautiful-Voicer=0.345, p=.049and Gracefulr=0.389, p
=.030. In other words, English speakersimpressions of the beauty and grace of the mimetic words for
laughing were the opposite of the meanings that Japanese speakers attribute to these words.
When CV-based words and CVCV-based words were considered separately, significant correlations
were found for three dimensions of CV-based wordsContinuous, Loud, Mouth-Wide-Open, and four
of CVCV-based wordsResonant-Voice, Mouth-Wide-Open, Good, and Nice. However, for two of the
four dimensionsGood, Nice, the correlation was negative. Thus, there is more agreement between
English and Japanese speakers ratings for CV-based words.

4-1-3Effects of Vowels

In order to assess the effects of different vowels, the mimetic words were divided into three groups:
those that contain /a/, which is iconic of laughing with one
s mouth wide openN=6, those that contain /e/, which presumably indicates a vulgar, negative type of laughingN=8, and those that contain
/u/N=7. One-way ANOVA by subjects and items revealed significant effects that are broadly similar
between Japanese speakersratings and English speakersratings of seven of 20 semantic dimensions:
Amused, Cheerful, Energetic, Excited, Loud, Mouth-Wide-Open and Resonant-Voiceall of which appear to be characteristic of /a/ words and uncharacteristic of /u/ words in both languages. Table 4 summarizes the significant differences between different vowels by semantic dimension and language group.
Overall, English speakers exhibited more variability in their responses, so fewer fine-grained differences were observed in English speakersratings. English speakers almost always rated words containing /a/ toward one end of a scale, and words containing /u/ toward the other endwith words containing
/e/ numerically intermediate, but hardly ever significantly different from the other two vowel conditionswhile Japanese speakersratings varied considerably depending upon the semantic dimension.
For two of these dimensions, Amused and Cheerfullabeled asSAMEin Table 4, the effects of

What do English speakers know about gera-gera and yota-yota

Table 4

Mimetic words for Laughing: Effects of Vowels between English and Japanese Speakers

Amused
Beautiful Voice
Cheerful
Continuous
Energy
Excited
Feminine
Formal
Good
Graceful
High Pitched
Loud
Many People
Nice
Pleasant
Purposeful
Resonant
Unrestrained
Mouth Wide Open
Young

67

Speakers

/a/
N=6

/e/
N=8

/u/
N=7

English

0.96

0.91

0.04

Japanese

1.76

1.05

0.69

English

0.42

0.22

0.63

Japanese

0.13

0.80

0.85

English

0.90

0.73

0.54

Japanese

2.08

0.54

0.56

English

0.08

0.48

1.82

Japanese

0.92

0.68

1.38

English

0.98

0.69

0.66

Japanese

1.71

0.95

0.19

English

0.88

0.58

0.88

Japanese

0.96

0.16

0.80

English

0.75

0.68

0.13

Japanese

0.31

0.52

2.02

English

0.08

0.50

0.79

Japanese

0.93

1.80

0.01

English

1.00

0.27

0.68

Japanese

1.03

0.58

0.43

English

0.27

0.02

0.82

Japanese

0.40

1.53

0.77

English

0.40

0.56

1.45

Japanese

0.68

0.09

0.71

English

0.60

0.27

1.29

Japanese

1.31

0.50

1.93

English

1.05

1.05

1.64

Japanese

0.33

0.83

1.85

English

0.85

0.42

0.50

Japanese

1.04

0.42

0.64

English

0.83

0.09

0.70

Japanese

1.00

0.76

0.41

English

0.54

0.47

0.18

Japanese

0.94

0.29

0.87

English

0.75

0.20

0.57

Japanese

1.38

0.38

0.86

English

0.46

0.53

0.46

Japanese

1.56

0.44

0.62

English

1.50

0.36

0.95

Japanese

1.63

0.55

2.16

English

1.29

1.06

1.32

Japanese

0.81

0.80

1.48

Cross-linguistic
similarity
SAME
J. ONLY
SAME
DIFF

Significant
differences
a>u
a>u
u>e
a>u
a>u
e>u
a>u
a>u

SAME
J. more sensitive

a>e>u

SAME
J. more sensitive

a>e>u

J. ONLY
J. ONLY
DIFF
DIFF
E. ONLY
SAME
J. more sensitive
J. ONLY
DIFF
DIFF
J. ONLY
SAME
J. more sensitive
J. ONLY
SAME
J. more sensitive
n/a

a>u
u>a, e
u>a>e
a>u
a>e
a>u
u>e
a, e>u
a>u
a>e>u
a>e, u
a>u
a>e
a>u
a>e
e>a, u
a>u
a>e>u
a>u
a>u
a>e>u
-

Differences are listed as significant only if p<.05 both by subjects and items. If a vowel condition is not listed, it does not differ significantly from the other vowels.

68

the vowels were the same in both languages: /a/ words were perceived as more Amused and Cheerful
than /u/ words. For five additional dimensionsEnergetic, Excited, Loud, Resonant, Mouth-Wide-Open,
labeled asSAME:J. more sensitivein Table 4, response patterns were broadly similar, but with
Japanese speakers exhibiting a finer level of contrast than English speakersin each case, both English
and Japanese speakers rated /a/ words higher than /u/ words, while Japanese speakers also significantly
differentiated /e/ words, placing them between /a/ and /u/ on the scales. These results demonstrate a
substantial degree of cross-linguistic similarity for mimetic words for laughing, despite the lack of correlation between the two language groups for many of these dimensions when analyzed at an item-specific
level.
There were also strikingly different effects of vowels between English and Japanese speakers concerning the dimensions of Continuous, Good, Graceful, Nice and Pleasantlabeled asDIFFin Table 4.
For Continuous, both groups rated /u/ words the lowest, but Japanese speakers considered /a/ words to
be the most Continuous while English speakers rated /e/ words as most Continuous. The three related
dimensions of Good, Nice and Pleasant exhibited similar patterns among them: English and Japanese
speakers agreed in rating /a/ words high on these scales, but differed in words given low ratings. English speakers gave lower Good/Nice/Pleasant ratings to words containing /u/ while Japanese speakers
gave lower ratings to words containing /e/. Finally, completely different patterns were observed for
Graceful. English speakers rated /a/ words as high and /u/ words as low, while Japanese speakers gave
high ratings to /u/ words and low ratings to /e/ words. These differences in ratings, with words containing /e/ given low ratings by Japanese speakers on a number of evaluative dimensions, might be attributed to differences in the implications of /e/inappropriateness/vulgarity in Japanese.
Because English speakers exhibited greater variability in their responses, it is not surprising that differences between vowels were observed only among Japanese for a number of dimensions: BeautifulVoice, Feminine, Formal, Many-People, Purposeful and Unrestrained. Words containing /u/ were more
Beautiful-Voice, Feminine and Formal while words containing /e/ were consistently lowest on these
scalesagain consistent with the link between /e/ and vulgarity in Japanese. This pattern was reversed
for Purposeful/e/ rated highest, /a/ and /u/ lowest. For Many-People and Unrestrained, words containing /a/ were rated highest, and words containing /u/ lowest.
There was only one dimension, High-Pitched, for which English speakers rated words containing different vowels significantly differently while Japanese speakers did not. English speakers rated words
containing /a/ and /e/ high on the High-Pitched scale, and words containing /u/ low on the scale. English
speakers may have relied more upon the acoustic characteristics of the words for this scalethe only
one overtly referring to acoustic dimensionsas a consequence of hearing the words rather than reading

What do English speakers know about gera-gera and yota-yota

69

them. Finally, different vowels did not have any effect on the ratings of the laugher
s ageYoungChildfor either language group.

4-1-4Effects of Geminates /Q/

Because geminates are considered to indicate abrupt, short, forceful movementse.g., Hamano 1998,
we expected Japanese speakers to rate 9 words containing Qe.g., huhuQ as less Continuous than 15
words that do not contain Q. This was indeed the case. The mean Continuous rating for words containing Q was -0.630 as compared to 0.639 for words that do not contain Q. The difference was significant;
subjects 12
t
=4.112, p=.002; items 22
t
=2.344, p=0.028. However, such an effect was not observed among English speakersboth |t|<1, p>.4, for whom there were no dimensions for which words
containing Q significantly differed from those which did not. English speakers do not appear to be sensitive to this phonological property of Japanese words.

4-1-5Effects of Onsets

The words were divided into two groups depending on whether the word-initial segments are vowels
N=5or consonantsN=19. The former group includes ahaha, ehehe, uhuhu, uhuQ and ohoho.
Japanese speakers rated words beginning with vowels as significantly more Beautiful-Voiceratings for
vowel onsets=.92, consonant onsets=.20, Femininevowel onsets=1.78, consonant onsets=.16, and
Formal vowel onsets=.22, consonant onsets=1.15 than the words that start with consonants:
Beautiful-Voicesubjects 11
t
=2.505, p=.029; items 22
t
=2.365, p=.027; Femininesubjects 11
t
=
3.008, p=.012; items 22
t
=2.274, p=.033; Formal,subjects 11
t
=2.251, p=.046; items 22
t
=2.848,
p=.009. In contrast, English speakers did not rate these two types of words significantly differently for
any dimension.

4-1-6-V1C2V2 sequence -era

The mean Graceful ratings for kera-kera and for gera-gera for Japanese speakers were 0.75 and
2.17, respectively, and those for English speakers were 0.25 and 1.25, respectively. To assess
whether these ratings tend toward Vulgarthe opposite end of the Graceful scaleone-sample t-test was
conducted against a value of zerohalfway between Graceful and Vulgar on the scale. These tests
were significant both for Japanese speakerssubjects 11
t
=4.901, p<.001and for English speakerst
14=2.613, p=.021; both groups felt that the words referred to vulgar rather than graceful laughing
although this effect was more pronounced for Japanese speakers. Thus, the sound symbolism in this
case may exhibit both culturally learned conventional aspects and some commonality between sound-

70

Figure 2

Average ratings by Japanese and English speakers for toko-toko.

meaning association for English and Japanese speakers.

These results indicate a certain degree of correspondence between speakers of the two languages for
mimetic words for laughing, although not to the extent of correspondence for mimetic words that directly refer to soundIwasaki, et al. 2005. We now turn to mimetic words for manner of walking, for
which we expect to see less correspondence as the meanings of such words are less iconic and less directly linked to articulatory mechanisms.
4-2Mimetic words for walking
4-2-1Correlation analyses for words

As expected, there were far fewer mimetic words for walking for which Japanese and English speakersratings were significantly correlated, compared to mimetic words for laughter. We found significant
correlations between English and Japanese speakersratings for only seven of 28 words used in the
questionnaire. The most highly correlated of these was toko-tokor=0.69, p<.001, rated by both Japanese and English speakers as referring to a small person wearing hard-soled shoes walking quickly and
energetically with small strides on a hard, dry surface in an informal mannersee Figure 2 for average
ratings across all the semantic dimensions. This was followed by kotu-kotur=0.67, p<.001, katukatur=0.65, p<.001, dota-batar=0.62, p=.002, taQ -taQr=0.56, p=.005
, hyoko-hyokor=0.47,

What do English speakers know about gera-gera and yota-yota

71

p=.018, and tuka-tukar=0.43, p=.029. Among these, significant correlations are observed not only
for some mimetic words for walking that are closely related to the auditory domainkotu-kotu, katukatu, dota-batabut also some that are much more visual in naturetoko-toko, taQ -taQ , hyoko-hyoko,
tuka-tuka.

4-2-2Correlation analyses for semantic dimensions

As expected from the few significant correlations for words, English and Japanese speakersratings
were significantly correlated for only a small number of semantic dimensions, and with substantially
lower correlation coefficients than for dimensions of laughing: Hard-Solesr=0.37, p=.026and WetSurfacer=0.32, p=.048. There were also some weak positive correlationsr>=0.20for Hard-Surface,
Big-Stride, and Steady, but these were not statistically significant. Interestingly, there were also weak
negative correlations that approached statistical significance for two dimensions: Noisy and ManyPeople.

4-2-3Effects of Vowels

In contrast to the mimetic words for laughing, few effects of vowels were observed, indicating that
the sound-meaning relationships are not as robust for these primarily visual manner mimetic words.
Among English speakers, the effects of different vowels were found only for Big-Stride. English speakers regarded mimetic words containing /u/ as higher on the Big-Stride scale than /a/average rating for
/u/=.36, /a/=1.04; subjects F
2, 22=3.504, p=.048, items F
2, 23=4.067, p=.031, a result that contrasted with the generally claimed sound symbolism of /a/ signifying largeness. Interestingly, there were
no such effects observed among Japanese speakers.
Among Japanese speakers, different vowels had significant effects for two semantic dimensions:
Gracefulsubjects F2, 22=4.111, p=.030, items F
2, 23=3.426, p=.050and High-Self-Imagesubjects F2, 22=3.984, p=.033, items F
2, 23=4.153, p=.029. Japanese speakers rated words containing
/a/ significantly higher than words containing /o/ on Graceful/a/=.19; /o/=1.45and High-Self-Image
/a/=.56; /o/=1.76dimensions.

4-2-4Voiced vs. Voiceless Consonants

The mean ratings of words beginning with voiceless consonantsN=12and those with voiced consonantsN=7for each semantic dimension were compared by t-test. For only one semantic dimension
Big-Person, was there a tendency toward a comparable similar effect of voiced/voiceless contrasts
across languages: speakers of both languages rated words starting with voiced consonants as referring to

72
Table 5

Average ratings for dimensions of walking by Japanese speakers for words beginning with
voiced vs. unvoiced consonants, for which voiced and unvoiced are significantly different.
Dimension

Voiced onset

Voiceless onset

Big-Stride

1.20

1.16

Even-Paced

0.35

1.25

Feminine

1.33

0.81

Formal

2.05

0.30

Good

1.55

0.07

Graceful

1.38

0.11

High-Self-Image

2.19

0.13

Noisy

1.89

.38

the walking of a Big-Person than words starting with voiceless consonantsfor English, voiced=.25, unvoiced=.75; for Japanese, voiced=1.28, unvoiced=.76, but this comparison failed to reach significance in the analysis of English subjectsEnglish subjects 11
t
=1.957, p=.076; items 17
t
=2.491,
p=.023; Japanese subjects 11
t
=5.400, p<.001; items 17
t
=4.619, p<.001.
For Japanese speakers, voiced/voiceless consonants also had a significant effect on 10 additional dimensions as illustrated in Table 5. On one hand, Japanese speakers rated words starting with voiced
consonants as significantly more Big-Stridesubjects 11
t
=4.751, p<.001, items 17
t
=4.619, p<.001,
and Noisysubjects 11
t
=3.903, p=.003, items 17
t
=3.114, p=.006. On the other hand, they rated
words starting with voiceless consonants as more Even-Pacedsubjects 11
t
=3.171, p=.009, items t
17=2.744, p=.014, Femininesubjects 11
t
=2.835, p=.016, items 17
t
=4.887, p<.001, Formal
subjects 11
t
=2.601, p=.025, items 17
t
=2.925, p=.009, Goodsubjects 11
t
=2.272, p=.044, items
17
t
=4.076, p=.001
, Gracefulsubjects 11
t
=3.222, p=.008, items 17
t
=2.867, p=.011, and HighSelf-Imagesubjects 11
t
=2.660, p=.022, items 17
t
=2.963, p=.009.

4-2-5-V1C2V2 sequence -ura

The mean ratings for hura-hura and for bura-bura for Japanese speakers were 1.42 and 2.83, respectively, for Steady and were 1.17 and 2.58, respectively, for Even-Paced. One-sample t-test by
subjects showed these were significantly less than zeroSteady: 11
t
=6.403, p<.001, Even-Paced t
11=4.727, p<.001. This contrasts sharply with English speakersratings of 0.71 and 0.00 for Steady
and 0.57 and 0.29 for Even-Paced which did not differ from zeroSteady: 11
t
=1.518, p=.157; EvenPaced: |t|<1. Japanese speakers clearly felt that the words referred to unsteady, uneven-paced walking

What do English speakers know about gera-gera and yota-yota

73

while English speakers did not have such a construal. Unlike -era in mimetic words for laughing, the
meaning of -ura in mimetic words for walking seems largely language-specific.

5Discussion
As expected, English speakers with no prior knowledge of Japanese related the sounds of Japanese
mimetic words to their meanings in a manner similar to native Japanese speakers to a greater extent for
mimetic words for laughing than they did for mimetic words for walking. This confirms the more iconic
status of the sound/voice mimetic words, supporting Iwasaki et al.
s2005findings that English and
Japanese speakersresponses to sound mimetic words correlated for a number of words and semantic
dimensions, and with Frei
s1970findings that French speakersresponses yielded higher correct responses for auditory mimetic words. At the same time, there were a number of sound-meaning associations that only Japanese speakers clearly demonstrated. The former findings indicate the universality or
at least some commonalities of sound symbolism to which both English and Japanese speakers are sensitive, and that mimetic words for laughing are more iconic than mimetic words for walking. The latter
indicates language-specific sound symbolism in the Japanese language that may be due to language and
cultural experience.
In the correlation analyses of English and Japanese speakersratings of semantic dimensions for respective words for laughing, we found that CV-based words did not always result in higher agreement
between English and Japanese speakers than CVCV-based words. Although some trends were consistent
with the more iconic status of CV-words suggested by Hamano1998
e.g., a slightly higher correlation for CV-words for laughing than for CVCV-words, and slightly more dimensions where correlations
were significant, the findings were not conclusive.
English speakers attributed similar meanings to mimetic words for laughing with regard to a number
of semantic dimensions which were strongly correlated between the two languages. However, there were
also two semantic dimensions for which English and Japanese speakerssense was opposite: BeautifulVoice and Graceful. These findings reflect Iwasaki et al
s.2005observation concerning sound mimetic words: English and Japanese speakersratings were correlated for many semantic dimensions, but
not for evaluative dimensions such as beauty or pleasantness. It appears that the beauty of linguistic
sounds is different in different linguistic communities.
In particular, the Japanese speakersassociation between the vowel /e/ and vulgar laughingtending
toward low ratings on a substantial number of scales like Beautiful-Voice, Feminine, Formal Good,
Graceful, Nice and Pleasantappears to be specific to Japanese language and culture. This is in appar-

74

ent contradiction to the claim by Tamori and Schourup1999that /e/ is also related to vulgar and
negative referents in English, because such a clear pattern was not observed for ratings by English
speakers who tended instead to give lower ratings to words including /u/. Likewise, Japanese speakers
association of the vowel /u/ to Beautiful-Voice, Graceful and Formal laughter also appears to be
language-specific. Other language-specific associations for mimetic words of laughing found in the current study were found in the effect of /Q/ and presence/absence of such consonants as /h/ and /k/ at the
beginning of words. Only Japanese speakers associated words that contained /Q/ as less Continuous and
words that started with vowels as more Beautiful-Voice, Feminine and Formal. However, it should be
underscored that there are also overwhelming commonalities in English and Japanese speakersassociations of /a/ to many other semantic dimensionsAmused, Cheerful, Energetic, Excited, Good, Loud,
Nice, Pleasant-Voice, Resonant-Voice and Mouth-Wide-Open.
With regard to mimetic words for walking, there was hardly any sound symbolism in common across
English and Japanese speakers. Only for two semantic dimensions, Hard-Sole and Wet-Surface, were
similar ratings observed between English and Japanese speakers. English and Japanese speakersratings
for -ura also differed greatly. Despite the lack of common sound symbolism across the two groups, data
for Japanese speakers show that there exist robust language-specific sound-meaning associations in Japanese speakersminds regarding both different vowels and voiced/voiceless contrasts of consonants.
Japanese speakers rated words containing /a/ as the most Graceful and High-Self-Image and words containing /o/ as the least Graceful and High-Self-Image. They associated voiced/voiceless contrasts of consonants with a number of semantic dimensions, by attributing Big-Stride and Noisy to voiced consonants and Even-Paced, Feminine, Formal, Good, Graceful and High-Self-Image to voiceless consonants.

6Conclusions and pedagogical implications

The current study confirmed the existence of sound symbolism in Japanese mimetic words to which
both English and Japanese speakers are sensitive, and that such sound symbolism is far more evident for
mimetic words for laughing that primarily mimic the human voice than for mimetic words for walking
that primarily mimic thevisualmanners of walking. This study also revealed aspects of sound symbolism common across English and Japanese speakers and aspects that are specific to Japanese. The
greatest commonalities were found in vowel-meaning associations among the mimetic words for laughing except for an important difference regarding /e/, whose association to vulgarity is believed to be
specific to the Japanese language. English speakers learning JSL must learn the meaning associated with
/e/ to develop reading and listening comprehension of the nuances of different types of mimetic words

What do English speakers know about gera-gera and yota-yota

75

for laughter.
Commonalities were not found in associations of voiced/voiceless contrasts of consonants to meanings among the mimetic words for walking. This is another area where JSL learnerswhose first language is Englishmay benefit greatly from formal instruction; however, the exact nature of effective instruction needs to be clarified in future research focusing on JSL learnersacquisition of mimetic
words. Such questions as to how JSL learners of different proficiency levels and from different first language backgrounds comprehend the meaning of mimetic words or produce Japanese mimetic words
need to be investigated in the future.

References
Clark, Herbert H. 1973. The language-as-fixed-effect fallacy: A critique of language statistics in psychological research. Journal of Verbal Learning and Verbal Behavior 12: 335359.
Diffloth, Gerard. 1994. i: big, a: small. In Sound symbolism, eds. L. Hinton, J. Nichols, and J.J. Ohala, 107115.
Cambridge: Cambridge University Press.
Frei, Henri. 1970. Cinquante onomatopes japonaises, in Mlanges Marcel Cohen, ed. D. Cohen, 359367. The
Hague: Mouton.
Hamano, Shoko. 1986. The sound-symbolic system of Japanese. Ph.D. diss., University of Florida.
Hamano, Shoko. 1994. Palatalization in Japanese sound symbolism. In Sound Symbolism, eds. L. Hinton, J. Nichols,
and J.J. Ohala, 148157 Cambridge: Cambridge University Press.
Hamano, Shoko. 1998. Sound -symbolic system of Japanese. Tokyo: Kurosio Publishers.
Harlofsky, William J. 1998. The acquisition of Japanese iconic expressions: Giongo versus gitaigo. Journal of Japanese Linguistics & Education 5: 18.
Hirose, Masayoshi. 1981. Japanese and English Contrastive Lexicology: The Role of JapaneseMimetic Adverbs
.
Ph.D. diss., University of California, Berkeley.
Jespersen, Otto. 1928. Language, Its Nature, Development, and Origin. London: George Allen & Unwin.
Kakehi, Hisao, Ikuhiro Tamori, and Lawrence Schourup. 1996. Dictionary of iconic expressions in Japanese. Berlin:
Mouton de Gruyter.
Makino, Seiichi and Michio Tsutsui. 1986. A dictionary of basic Japanese grammar. Japan: The Japan Times.
Miron, Murray S. 1961. A cross-linguistic investigation of phonetic symbolism. Journal of Abnormal and Social Psychology 62, 3: 623630.
Newman, Stanley S. 1933. Further experiments in phonetic symbolism. The American Journal of Psychology 45, 1:
5375.
Oda, Hiromi. 2000. An embodied semantic mechanism for mimetic words in Japanese. Ph.D. diss., Indiana University.
Sapir, Edward. 1929. A study in phonetic symbolism. Journal of Experimental Psychology 12: 225239.
Saussure, Ferdinand de. 1916. Cours de linguistique gnrale. Paris: Payot.
Taylor, Insup Kim, and Maurice M Taylor. 1962. Phonetic symbolism in four unrelated languages. Canadian Journal
of Psychology 16, 4: 344356.
Toda, Sueko, Alan Fogel, and Masatoshi Kawai. 1990. Maternal speech to three-month-old infants in the United
States and Japan. Journal of Child Language 17, 279294.
Yang, Edith. 1984. Sapir-Whorf revised: The relationship between Japanese onomatopoetic wordsgiseigo and gi-

76

taigoand the Japanese Culture. Ph.D. diss., University of San Francisco.

Yoshimoto, Banana. 1993. Kitchen. In KitchenM. Backus, Trans.
, 1105. New York: Washington Square Press.
Original work published 1988
.
1995
1993
2262633
1938
The Japanese Journal of Psychology 13411428
2005
4 233
246
H1982
111140
1967
1986
57
3946
2001
3092836
1991
1999
1986 K.G 345360
1991
1993
7710

!
!
#

#
"

"$!

"

"

"

$!

$!

$!

#
"

!
#
#

"

"

"
"
!

"

"

"

"

What do English speakers know about gera-gera and yota-yota

77

Appendix A
Definitions of mimetic words of laughing from Kakehi et al.1996
1

ahaha

The sound of loud laughter.

hahaha

The sound of cheerful laughter.

haQhaQ

The sound of short burst of laughter, often indicating confidence or pride.

ehehe

The sound of soft laughterusually embarrassed or lascivious laughter

.

hehehe

The sound of subdued laughter, often made to cover embarrassment, an error, etc.

heQheQheQ

hihihi

waQhaQha

The sound of subdued laughter connoting lewd or obscene thoughts.

The sound of weird, uncanny laughter.
The sound of loud, unrestrained laughter.

kera-kera

The sound of high-pitched, unrestrained laughter.

10

gera-gera

The sound of noisy laughter.

11

keta-keta

The sound of loud, unrestrained laughter.

12

keQkeQke

The sound of peculiar laughter.

13

geta-geta

The sound of loud, uncouth and unrestrained laughter.

14

kyaQkyaQ

The sound of repeated short screaming or excited, high-pitched laughing sounds, such as
those made by young children.

15

ohohoho

The sound of suppressed laughter.

16

uhuhu

The sound of laughing, or chuckling to oneself softly.

17

huhuQ

The sound of light laughter.

18

puQ

19

huhuhu

20

uhuQ

21

kusu-kusu

The sound made when air escapes suddenly from between the lips, as when stifling laughter.
The sound of light laughter.
The sound of laughing, or chuckling to oneself briefly and softly.
The manner of laughing to oneself.

22

kusuQ

23

kara-kara

The manner of laughing loudly but without malice.

The manner of laughing once in embarrassment.

24

koro-koro

The merry laughter of a child or young girl.

Notes: The Romanization adopted in Appendix A and B follows Hamano1998and thus is different from Kakehi
et al.
s1996
. Bold face indicates a greater association with either sound or manner.

78

Appendix B
Definitions of mimetic words for walking from Kakehi et al.1996
1

syanari-syanari The manner of walking with an affected gait.

suta-suta

The manner of walking briskly and purposefully.

taQtaQto

tuka-tuka

The manner of walking towards someone briskly and with determination.

toko-toko

The manner of walking, trotting, etc., with quick, short steps.

tobo-tobo

The manner of walking wearily.

The manner of walking or running with quick steps.

teku-teku

The manner of walking steadily for a considerable distance.

yoro-yoro

The manner of walking or moving unsteadily

The manner of walking with heavy, faltering steps.

yota-yota

10

noQsi-noQsi

11

Bura-bura

The manner of walking around leisurely.

12

tyoko-tyoko

The manner of moving about with short, rapid, movements. This form sometimes connotes
restlessness and nervousness.

13

Noso-noso

The manner of moving slowing and lethargically.

The manner of walking heavily.

14

urotyoro

The manner of moving about restlessly and without aim, and often as bothersome.

15

uro-uro

The manner of moving about aimlessly.

16

hura-hura

The manner of swaying gently or moving from side to side unsteadily or rising into the sky
without much propulsive force.

17

hyoko-hyoko

The manner of doing something with short and irregular or unsteady movements, often implying a sense of precariousness or danger.

18

yoti-yoti

19

sorori-sorori

The manner of moving with unsteady, often short, stepsespecially of children

.

20

zoro-zoro

The manner in which a number ofusually livingthings follow one after another or gather
together.

21

dosi-dosi

A sound of a large person or animal walking or stamping heavily; the sound made when a
think, heavy, and relatively soft object strikes or is struck by something.

22

dosin-dosin

23

bata-bata

The pattering sound of quick footsteps.

24

katu-katu

The sound of heels striking a hard floor.

25

kotu-kotu

A repeated rapping or tapping sound, such as that made by hard-soled shoes.

26

dosun-dosun

A loud resonant sound made by a heavy object repeatedly falling or striking something with
great force.

27

dota-bata

A noisy sound made by someone coming, going, or moving about boisterously and quickly.

28

peta-peta

A light pattering, often wet sound made when two flat surfaces come in contact.

The manner of moving slowly.

Loud, resonant sounds as of a very large person or animal walking or stamping heavily.