The sound-symbolic system of Japanese


Material Information

The sound-symbolic system of Japanese
Physical Description:
vii, 290 leaves : ill., map ; 28 cm.
Hamano, Shoko Saito
Publication Date:


Subjects / Keywords:
Japanese language -- Vocabulary   ( lcsh )
Sound symbolism   ( lcsh )
Japanese language -- Phonology   ( lcsh )
Japans   ( gtt )
Fonologie   ( gtt )
Bijwoorden   ( gtt )
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph. D.)--University of Florida, 1986.
Includes bibliographical references (leaves 286-289).
Statement of Responsibility:
by Shoko Saito Hamano.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 000880456
notis - AEH8275
oclc - 14971436
bcl - 18.87
System ID:

This item is only available as the following downloads:

Full Text







Copyright 1986


Shoko Saito Hamano


I would like to express my gratitude to the members of my committee, Dr. M. J.

Hardman-de-Bautista, Dr. Timothy J. Vance, Dr. D. Gary Miller, and Dr. Theron A. Nunez,

for their guidance and assistance throughout the preparation of this dissertation. I would also

like to express my appreciation to Dr. Norman N. Markel for his support and encouragement.

My deepest appreciation goes to both Dr. Hardman-de-Bautista, my chairperson and general

adviser, and Dr. Vance, my adviser in the specific aspects of this dissertation and Japanese

linguistics. Both of them read the initial incoherent versions of this dissertation with an

admirable patience and kept providing suggestions and moral support. I was truly lucky in

having the two, whose suggestions complemented each other.

My sincere appreciation goes to my friend, Robert Lezzi, who helped me in translating

the Japanesee examples into English. Our mutual frustrations in trying to convey and

identify the right correspondences that sounded natural were very informative.

Special thanks are due to my friend, Ronald Kephart, who helped me in the final stage

of this dissertation by sending me his invaluable advices concerning the fine details of

finishing a dissertation. He also ran around the campus to prepare for my final semester in

Gainesville, when I actually was on the other side of the country teaching.

Finally, I would like to thank my husband Fumio Hamano and my four-year old

daughter Aya. Without their patience and understanding, this dissertation would not have

been possible.


ACKNOWLEDGMENTS .................... ui

ABSTRACT .. .................. vi


1 INTRODUCTION ....................... 1

1.1. Statement of the Problem .. . .1
1.2. Delineation of the Corpus . 6
1.3. Outline . . 7
1.4. Definition of Linguistic Varieties in Japanese 8
1.5. Romanization and Abbreviations of Grammatical Terms 10
1.6. Data Source .. .. .. .12


2.1. Introduction . . 13
2.2. The Grammar of Mimetic Adverbs . .. 13
2.3. The Grammar of Mimetics as Nominal Adjectives .. 30
2.4. Mimetic Words in Noun Phrases . .. 32
2.5. Derivations . . 49


3.1. Accent and Iconicity in Mimetic Adverbs .. 59
3.2. The Iconic Continuum . .... .71
3.3. Accent of Long Vowels . .... .75


4.1. Introduction . . .. .77
4.2. Repetition . . .. 78
4.3. Final Elements /N, Q, 0/ . .. 84
4.4. Vowel Length ......... .... ..... 90
4.5. Diphthongs . . 94
4.6. Vowels . . 97
4.7. Palatalization . . 106
4.8. Voicing of Initial Consonants . ... 106
4.9. Points and Manners of Initial Consonants .. 110
4.10. Summary . . .. 126


5.1. Introduction . . 132
5.2. Repetition . . 133
5.3. Final Elements /N, Q, -ri/ . .. 136
5.4. Medial/Q,N/ . . 137
5.5. Vowel Length .................. 140
5.6. Vowels . . 142
5.7. Palatalization . . .. 163
5.8. Voicing Contrast of Obstruents . .. 163
5.9. Points and Manners of Consonants .... 176
5.10. Summary . . 224


6.1. Palatalization in Mimetic Words . ... 228
6.2. The Bilabial Stops /p, b/ in the Second Syllable 245

7 CONCLUDING REMARKS .... ............ 248

7.1. Comparisons of Monosyllabic and Bisyllabic Adverbs .. 248
7.2. Prospects for Further Studies . .. 249







REFERENCES ...................... 286

BIOGRAPHICAL SKETCH ................. 290

Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy



Shoko Saito Hamano

May 1986

Chairperson: M. J. Hardman-de-Bautista
Major Department: Anthropology

The sound-symbolic system of Japanese, commonly called giongo/giseego/gitaigo

'mimetic words,' is one aspect least understood about Japanese in spite of its systematic

nature and significance in daily communication. Very little systematic research has been

done on it, and textbooks for non-native speakers include only the least sound-symbolic few

of the system which could be readily checked in conventional dictionaries. In attempting to

remedy the situation, this dissertation demonstrates that Japanese mimetic words indeed

constitute a sound-symbolic system which has implications not only to the study of sound

symbolism in general but also to the historical sound change of Japanese.

The dissertation first deals with the formal characteristics of the sound-symbolic

system. The syntactical, morphological, and phonotactic properties are described. The

accentual properties of the system are next described, leading to the characterization of the

system as a continuum between the more iconic and the less iconic.

The sound symbolism of both monosyllabic and bisyllabic mimetic adverbs is treated.

Concrete evidence is presented to demonstrate that there indeed are systematic ties between

the semantic and phonological components of these mimetic words. For instance, the

semantics of the voiceless bilabial stop in monosyllabic mimetic adverbs is broken down to

'abrupt movement, stretched-out surface or line, light/small/fine' corresponding to its

phonological features of 'explosive, bilabial, voiceless.' Functional differences between

positions are also clarified. The first consonant, for instance, is demonstrated to denote

'texture,' while the second denotes 'movement'

The dissertation furthermore demonstrates two aspects of the mimetic phonology

which are related to the historical sound change of Japanese, namely palatalization and

bilabial stops in the second syllable. Based upon the distributional characteristics of

palatalized syllables, it is concluded that palatalization in the sound-symbolic system has an

origin different from that of palatalization in Sino-Japanese loan words. Also, based upon

the distributional and semantic characteristics of bilabial stops, it is suggested that the voiced

bilabial stop in the second syllable, or in intervocalic position, in mimetic words derives from

the voiceless bilabial stop.


1.1. Statement of the Problem

Japanese has an extensive set of lexical items commonly called

giongo/giseego/gitaigo "mimetic words." The majority of these words are never entered

into ordinary Japanese dictionaries or textbooks or grammars, and, for the limited few that

are entered, only their least mimetic meanings are given. Hence, the more expressive of

these mimetic words escape the acquaintance of beginning students of Japanese. For

native speakers of Japanese, on the other hand, their expressive meanings are immediately

understood, and most of the mimetic forms are easily identifiable as such because of their

distinct form.

Phonologically speaking, a large number of mimetic words in Standard Japanese

have the shape of C1V1C2V2- C1V1C2V2 or C1V1N-C1V1N (where N stands for a long

nasal). Semantically speaking, many of them indicate manners of action, psychological

conditions, and sounds accompanying certain actions; their sounds are perceived as

imitative of or implying some physical properties. In other words, they are

sound-symbolic. Syntactically speaking, most of them function as adverbs; some of them

are used as modifiers in noun-phrases.

The extensiveness of the mimetic system is related to the semantic

underdifferentiation of Japanese verbs. Sato (1977) notes that Japanese has a very small

number of verbal and nominal roots in comparison to other languages. Hirose (1981)

contrasts, for example, how the semantic domain of "walking" is treated in Japanese and

English. This will serve to illustrate the point.


In order to specify the manner of "walking" in English, the most common method is

to use different verbs other than the core verb "walk." These include "waddle, toddle,

totter, wobble, stagger, reel, stride, strut, swagger, lumber, stomp, tramp, trudge, plot,

wander, ramble, stroll, amble, saunter, hike, and shuffle." The use of these various verbs

adds to the expressiveness and explicitness of the utterances that contain them.

In order to achieve a similar effect in Japanese, one uses various mimetic words with

the verb aruku "walk." Hirose includes about seventy such mimetic words in his

discussion. A few of them are listed below.

(1-1) dosa-dosa (to) aruku 'walk with loud noise'

doka-doka (to) aruku 'individuals walk noisily and

kotu-kotu (to) aruku 'walk with the shoes making
the characteristic noise'

saku-saku (to) aruku 'walk in something like soft

yoro-yoro (to) aruku 'wobble'

bura-bura (to) aruku 'stroll'

teku-teku (to) aruku 'hike'

dara-dara (to) aruku 'walk slowly due to lack of

noro-noro (to) aruku 'walk slowly'

zoro-zoro (to) aruku 'mass of people walk'

uro-uro (to) aruku loiter, roam'

It is important to point out that these mimetic words are not idiosyncratically

produced expressions. While there are a few short-lived expressions coined by individuals

or ingeneous expressions used by poets in their poems, the majority of the mimetic words

conform to rigid structural constraints. In the above examples, for instance, a morpheme

of the shape (C1)V1C2V2 is reduplicated before it is proposed to the verb. For mimetic

expressions of this type, a particle of quotation, to/te, may be inserted between the mimetic

word and the verb. When a partially reduplicated mimetic word of the type C1V1Q -C1V1

is used, 1o must appear, as below.

(1-2) saQ-sa to aruku 'walk speedily'

paQ-pa to aruku 'walk speedily'

In spite of the systematic nature and the significance of the Japanese mimetic system,

it is one aspect little understood about Japanese on both sides of the Pacific. Very little

systematic research has been done on it, and textbooks for non-native speakers include

only the least sound-symbolic, most idiomatic items in the system, which could readily be

found in conventional dictionaries. Recent concern about the difficulty of learning

mimetic words by non-native speakers of Japanese has produced a dictionary of

gioneo/gitaigo (Asano 1978), which deals exclusively with mimetic words. This

dictionary contains about 1450 entries and explains their meanings in relation to their use in

specific contexts. Significant as this dictionary may be, it has touched upon only the tip

of the problem; it has not identified the sound-symbolic meanings of the basic phonological

units; nor has it clarified the systematic nature of Japanese mimetic words as a linguistic

structure which is precisely what allows native speakers to take them for granted. The

present dissertation attempts to fill this gap in the study of Japanese grammar.

It is hoped that the rigorous method applied in this study of the Japanese mimetic

system will persuade even the most skeptical that there is sound-symbolism in natural

languages. In what follows, I will briefly describe the circumstances that led to this

general skepticism about sound symbolism in natural languages.

First of all, it will be recalled that a society of linguists in Europe in the 19th century

banned the discussion of the evolution of language by professional linguists, because the

unscientific and simplistic evolutionism in discussions of language evolution rampant at

that time was considered to be harmful to the healthy growth of scientific linguistic

disciplines. Representative of this simplistic evolutionism were the "bow-wow theory,"

the "ding-dong theory," and the like, which traced the origin of language to imitative or

sound-symbolic use of sounds. So simplistically were these "theories" constructed that

iconic properties of language generally seem to have been considered as unworthy of

serious scientific endeavor. Any introductory textbook in descriptive linguistics will verify

this. Furthermore, arbitrariness is the hallmark of the most human of the human

communicative systems -- language -- both within the general vocabulary and within the

grammatical system (cf. Hockett 1960; Hockett and Ascher 1964).

The second cause of the scepticism about sound symbolism has much to do with the

data base with which linguists have been working. In European languages, where modern

linguistics first developed, sound symbolism plays only a negligible role within the

linguistic systems themselves. The expressive role of sounds is largely latent and comes

into focus only when there is a choice between homonymic arbitrary symbols (Markel

1966). Brown (1958) is more pessimistic about the role of sound symbolism in natural

languages. See the following passage from Brown (1958).

The failure to detect any effects of phonetic symbolism in situations
where it is not specifically called for by the experimenter [by the use of
nonsense words] suggests that we do not usually expect speech to represent or
imitate. (138)

To be sure, a number of linguists have pointed out interesting correlations between

certain English sounds and meanings (Householder 1946, Bolinger 1950, Marchand 1959,

Wescott 1971a; 1971b). The problem here, however, is that the data base of these studies

is not well-defined: the described correlations come from diverse subcategories of the

language, and, within the selected lexical sets, there is little more in common other than the

posited sound-symbolic correlation that would tie them together as a meaningful subset;

counterexamples, on the other hand, are grossly ignored. Hence, there always remains a

criticism that the data were selectively skewed in favor of sound symbolism.

Recent researchers of sound symbolism in both European and non-Indo-European

languages are more aware of this problem of skewing of data and attempt to define the data

base more explicitly, but the problem remains because formal criteria that would

exclusively select sound-symbolic forms do not exist in these languages, either (Joseph

1985, Nichols 1985). Likewise, although studies of African languages brought the study

of sound symbolsim to the attention of modem linguists (Samarin 1970b), and there were

attempts to tie formal classes to sound symbolic forms, these attempts failed (Courtenay


The third factor concerns the elusive nature of sound-symbolic meanings; that is,

elusive from the point of view of the outsiders (Samarin 1967, 1970a). Accurate

translation is always extremely difficult to achieve, but the semantics of expressive speech

above all seems to be the most difficult to translate into another language, defying either

concise approximation or full explanation even when the native speakers know exactly how

to use and interpret the respective forms. This is why, although some East Asian

languages such as Korean and Japanese are reported to possess very systematic sound

symbolism, the systems have not been treated fully by outside linguists.

The situation was unfortunate for the study of the Japanese sound-symbolic system.

Japanese and Japanese speakers were used in a number of experiments to represent

"exotic" languages and speakers, when the discussion of sound symbolism reached a peak

in the 1960s. See Maltzman, Morrisett and Brooks (1956), Blackbill and Little (1957),

and Taylor and Taylor (1962). However, no one mentioned giongo/giseego/ gitaigo, even

when it was stated, as in Taylor and Taylor (1962), that sound symbolism is

culture/language-specific. The omission does not seem to have been the fault of the

psychologists alone. Ultan (1978) uses grammars of natural languages including that of

Japanese for cross-linguistic comparison of size-sound symbolism, but he does not find

any evidence of size-sound symbolism in the Japanese grammar. In Chapters 4 and 5, it

will be demonstrated that the vowels are sound-symbolic of size and shape in the Japanese

mimetic system.

1.2. Delineation of the Corpus

Traditionally, Japanese sound symbolism has been defined semantically, two major

subcategories being words imitative of sounds and words imitative of modes respectively.

Such semantic definitions, however, are not rigorous enough to enable us to delineate the

mimetic system at the beginning of the research.

On the other hand, an attempt to define Japanese mimetic words solely in terms of

forms/grammatical functions/ distributions proves to be futile, because their formal

characteristics cross over the boundary into the regular stratum of Japanese. For instance,

although reduplication is a salient property of mimetic words, not all mimetic morphemes

can be reduplicated. Nor is reduplication limited to mimetic words; a kind of plural or

distributive derivation is accomplished by reduplication (Martin 1952). The infixation of

intensifier /Q/ or IN/ and the suffixation of /-ri/ and /-N/ spread over more than the domain

of mimetic words.

Historically speaking, mimetic words have not been entirely distinct from the rest of

the grammar. Some words that are now considered as sound-symbolic by some people

were originally non-sound-symbolic. For instance, according to Otsubo (1982), simi-zimi

'keenly, heartily' derived from the verb simu 'to soak into,' and noNbiri 'leisurely'

derived from the verb nobu 'to stretch.' Conversely, some words which are commonly

considered as part of the ordinary stratum originated in sound symbolism. In the

compound verb, awate-hutameku, the latter element derived from mimetic huta plus a

derivational suffix -meku.

One way to circumvent the problems was found, however. Once noticed, it is

deceptively common-sensical. Hashimoto (1950) hypothesizes that proto-Japanese had /p/

word-initially and word-medially. According to his view, the word-initial /p/ remained

only in mimetic words; elsewhere, it was replaced by /h/. As a result, except for

loanwords from European languages, it is only in sound-symbolic expressions that one

finds initial /p/. For the sake of convenience, I shall call these forms "p-initial mimetic

words." Since loanwords of European origin are treated basically as nominals, their

adverbial forms are invariably accompanied by ni, a form of the copula da; they are almost

never accompanied by the quotative particle Io. One rare exception is an English

onomatopoetic expression 'tick tock,' which was incorporated in to Japanese as ti'ku-taku

IQ, because of its structural similarity to Japanese bisyllabic mimetic adverbs. Note that the

exception proves the rule in this case; it is precisely because the form is mimetic in English

that it is assimilated into the Japanese mimetic system.

The above means that p-initial adverbs which are used with the quotative particle lo

are all mimetic. In other words, p-inital mimetic words are mimetic par excellence.

Moreover, out of 1450 or so entries in Asano's dictionary of mimetic words (1978), about

240, or one sixth, are p-initial. Their proportion is exceedingly large in comparison to the

proportion of other forms. It is inconceivable that the defining characteristics of p-initial

mimetic forms are irrelevant synchronically. We can expect that they are central to the

mimetic system. For this reason, I use p-initial forms as the starting point of every

analysis in this dissertation, and general patterns observed in p-initial adverbs are used as


1.3. Outline

Chapter 2 is an overview of the Japanese sound-symbolic phenomena. It delimits

the core of the research, the sound-symbolic system, in terms of its formal characteristics.

In addition, the borderline phenomena are described briefly to provide a glimpse of the

powerful influence that the mimetic system has on the ordinary stratum of Japanese.

The question of continuity between idealized categories is a recurrent issue in the

study of the mimetic system. Chapter 3 deals with the mimetic system as a scaled

continuum. The more iconic and the less iconic within the system are contrasted in relation

to accentual characteristics.

Chapter 4 and Chapter 5 describe the sound symbolism of monosyllabic and

bisyllabic mimetic adverbs respectively. The morphological and phonological components

are broken down to sound-symbolically relevant units.

Chapter 6 considers two aspects of the mimetic phonology which have bearing on

historical sound changes in Japanese.

Chapter 7 summarizes the features shared by monosyllabic and bisyllabic mimetic

adverbs and suggests some implications of this study for studies of sound symbolism in


1.4. Definition of Linguistic Varieties in Japanese

Standard Japanese refers to a variety of the Tokyo dialect in the typically

middle-class region of Tokyo and its vicinity, of which the writer is a native speaker. This

variety is the instrument in much of printed literature. The data for this dissertation were

acquired largely from Standard Japanese sources. Other areas of Japan to which

references are made in this dissertation are summarized in the following map.


1 Aomori
2 Ivate
3 Akita
4 Ishikava
5 Tokyo (Edo)
6 Shiga
7 Nara
8 Kyoto
9 Osaka
10 Ehime
11 Kochi
12 Iki
13 Okinawa

Figure 1-1: Map of Japan

The history of Japanese is traditionally divided into Proto-Japanese, Old Japanese,

Middle-Old Japanese, Middle Japanese, Recent Japanese, and Modem Japanese (Sato


Old Japanese is the language of the oldest existing written records from an era

roughly between A.D. 593 and A.D. 794. These written records best document the

language of the ruling aristocratic class in the old capitals of Osaka and Nara.

Proto-Japanese is reconstructed on the basis of this historically documented language.

Middle-Old Japanese refers to the language in the Kyoto area between A.D. 794 and

A.D. 1086. This language is recorded in a variety of documents.

Middle Japanese refers to the language between 1086 and 1603. The language

reflects an expanding area of communication and a rapidly changing social structure. An

increasing body of colloquial speech is recorded in the documents.

Recent Japanese refers to the language between 1603 and 1867. The transfer of the

political structure from the Osaka-Kyoto area to Edo (Tokyo) occurred in this era, with an

ensuing merger of the Osaka-Kyoto linguistic traits into the Edo language. Modem

Japanese refers to the language from 1867 to the present

1.5. Romanization and Abbreviations of Grammatical Terms

The romanization I use in this dissertation for the examples is for the most part the

one that Jorden (1963) uses, but I depart from her practice in one important way. While

Jorden uses /pp, tt, kk, ss/ for geminate clusters, I use /Q/ to represent the first elements of

all these geminate clusters because their sound symbolic value is constant. For a similar

functional reason, I use /N/ instead of In/ to represent the mora nasal. For Japanese names

in the main text and the translation, on the other hand, I use the conventional Hepburn

romanization. For the Japanese names in the references, I use the Hepburn romanization,

unless the authors or their publishers have romanized their names otherwise in print In

the latter case, I use their spellings. For the titles of the papers and books, however, I use

Jorden's romanization.

As explained in 3.1, phonemic pitch fall, or word accent of mimetic forms is

marked by /'; intonational pitch fall is marked by /"/. Accent or intonation is marked only

where it is relevant to the discussion.

In the examples, grammatical terms are abbreviated in the following way.

(1-3) Adv adverb

Cont continuative

Cop copula

Emph emphatic

Gen genitive

Ger gerund

Hon honorific

Neg negative

Obj object

Pass passive

Past past

Pol polite

Pot potential

Pres present

Ques question

Quot quotative

Sbj subject

Top topic


1.6. Data Sources

The data for this study came from a wide variety of sources such as ordinary

conversation, TV and radio broadcasting, personal letters, advertisement, literary sources,

magazines, cookbooks, and children's books. All the translations were rendered by me

with the help of a native English speaker. Any errors in judgment are my own



2.1. Introduction

In this chapter, an overview of the Japanese sound-symbolic phenomena in terms of

their formal characteristics is presented. Finer details of morpho-phonological

characteristics of the sound-symbolic system will be discussed later in Chapters 4 and 5.

The core of the Japanese sound-symbolism consists of mimetic adverbs and mimetic

"nominal adjectives." We might refer to this core as the Japanese sound-symbolic system.

The elements of this system are commonly known as giongo/gitaigo/giseego 'mimetic

words.' The first four sections of this chapter deal with this core. Around this core, we

find other sound-symbolic phenomena as appear in nouns, verbs, and adjectives.

Although we find certain systematic characteristics in these borderline sound-symbolic

phenomena, the formation of these borderline sound-symbolic words is more haphazard

than in the core, and it is often only with a high degree of uncertainty that we identify the

components as sound-symbolic. Nevertheless, these borderline phenomena form an

essential part of the sound-symbolic continuum and warrant our attention. Section 2.5 will

briefly describe these borderline phenomena.

2.2. The Grammar of Mimetic Adverbs

2.2.1. Syntax With Verbs and Adjectives

Of the two classes of mimetic words in the sound-symbolic system, if we were to

discuss just one, we would have to do it with mimetic adverbs. Almost all unambiguously

mimetic segmental strings can be used as adverbs provided that they are accompanied by

the accent patterns of mimetic adverbs. (See Chapter 3.) Most of these mimetic adverbs

modify verbs rather than adjectives. See the following examples.

(2-1) a. Tikaku ni ki -ta kara to iQ -te
neighborhood to come-Past because Quot say-and

musume to beta-beta tukia -u to
daughter with keep company-Pres Quot

i -u taipu no haha de wa ari-mas-eN.
say-Pres type Gen mother Cop Top be -Pol-Neg

'(My mother) is not the type of mother who carries on

the (mother-daughter) relationship in an

overly affectionate way even when she lives in her

daughter's neighborhood. = My mother isn't the type who

totally depends on me for company when she lives in

my neighborhood.'

b. Sorede GoN wa ie no uraguti kara koOsori
so Gon Top house Gen back door through

naka e hairi-masi-ta.
inside to enter-Pol -Past

'So, Gon stealthily entered the house through the back


The mimetic adverb and the verb must be semantically compatible. The use of a

given mimetic adverb in a given context narrows the choice of the verb so that it is often

predictable and may be omitted. See However, the combinations of mimetic

adverbs and verbs are not fixed. A wide range of verbs can be used with a mimetic verb,

and vice versa.

The frequency of one verb, however, is singularly higher than other verbs. This

verb is suu 'do.' Sm and its continuative form, site iru appear with a large number of

mimetic adverbs as part of idiomatic expressions. See the following.

(2-2) biQkuri suru 'to get surprised'

gikuQ to suru 'to be startled'

kaQ to suru 'to get mad'

karaQ to suru 'to get dry'

kyotoN to suru 'to look stupefied'

tikuQ to suru 'to feel a pricking pain'

zime-zime suru 'to be damp'

gata-gata suru 'to be shaky'

gura-gura suru 'to be wobbly'

piN-piN site iru 'to be energetic'

poro-poro site iru 'to be crumbly'

kiriri to site iru 'to be sharp/smart'

Although not as frequently as verbs, adjectives are also modified by mimetic

adverbs. See the following.

(2-3) poka-poka (to) atatakai 'to be comfortably warm'

kira-kira (to) mabusii 'to glimmer and be blinding'

tiku-tiku (to) itai 'to be painful with pricking pains'

hiNyari (to) tumetai 'to be comfortably cold' Ouotative Particles

When mimetic words are used as adverbs, a particle customarily used for quotation,

either IQ or i, is obligatory with certain classes of them and optional with other classes. In

general, a particle of quotation is obligatory with more colloquial, more iconic mimetic

adverbs and optional with less colloquial, more conventional mimetic adverbs. See 3.1

for details.

There also is a difference between the distribution of to and it. The mimetic adverbs

that appear with it are only those that are at the most colloquial and iconic end of the

continuum, as below (2-4).

(2-4) a. Musiba deki-tara haisya-saN ni
bad tooth form-if dentist-Hon by

giri-giri-giri-giriO" te yar-are -tya -u zo.
do -Pass-end up-Pres Emph

'I'm telling you. If you get a bad tooth, you'll have to

have it drilled by the dentist.'

b. RaioN-saN to zoo -saN o iQsyo ni
lion -Hon and elephant-Hon Obj together Adv

ire-ru to zoo -saN wa raioN-saN o piiN" te
put-Pres if elephant-Hon Top lion -Hon Obj

tatak-u kamo sirenai. Raion-saN wa zoo -saN
slap -Pres maybe lion -Hon Top elephant-Hon
wa zoo -saN ni gabuO" te kamituk-u
Top elephant-Hon to bite -Pres

ka mo sirenai.

If you put a lion and an elephant together, the elephant

may knock the lion around, and the lion may take a big

bite out of the elephant.'

Some representative classes of mimetic adverbs and their examples are listed below

according to the quotative particles with which they appear. In a table of this kind, glosses

are not given because few concise glosses are available that would do justice to the full

complex meanings of the mimetic forms. The meanings of these mimetic forms are

explained fully in Chapters 4 and 5 with sample sentences.

The list below shows that all mimetic adverbs that appear with =e belong to the

category that obligatorily requires a quotative particle. To on the other hand can appear

with all mimetic adverbs.

(2-5) examples
a. oa or I is obligatory

CV-CV(V)N" to/te pi-piN" to/te
CV-CV(V)Q" to/te pi-piQ" to/te
CV(V)N" to/te piiN" to/te
CV(V)Q" to/te piiQ" to/te
(y-)CVCV-CVCVQ" to/te paka-pakaQ" to/te
(y-)CVCV-CVCVN" to/te paka-pakaN" to/te
onomatopoetic forms koke'koQkooQ" to/te

b. tI is obligatory

C(y)V(V)N to piN to
C(y)V(V)Q to paaQ to
C(y)V'Q-CV" to pa'Q-pa" to
c. to is optional

C(y)VN-C(y)VN" (to) pi'N-piN" (to)
C(y)V'V-C(y)VV" (to) pi'i-pii" (to)
(y-)CV'CV-CVCV" (to) pa'ku-paku" (to)
(y-)CVQCV'ri" (to) piQtari" (to)
(y-)CVNCV'ri" (to) uNza'ri" (to)
A comparison of the above and the list in (3-1) of 3.1 clarifies that (2-5a)

corresponds to Class I, the most iconic accentual class. Verb Ellipsis

As mentioned above, a verb following a mimetic adverb may be omitted. In such

cases, the quotative particle is also left out. See the following.

(2-6) a. Retasu pari-pari, kyuuri pori-pori.
lettuce cucumber

'[We munch away at] gri lettuce and crunchy


b. Namae o kii -te gikgQ.
name Obj hear-and

'[I was] startled when I heard the name.'

c. Tokee ga iti-zi o booN.
clock Sbj one-o'clock Obj

The clock [struck] one with a bong.'

d. Hara no musi ga g.Q.
stomach Gen insect Sbj

The insect in my stomach [made] a noise of guQ. = My

stomach growled.'

e. Asi o hayame-te saO saO s sasaO to iki-sugi-ru
feet Obj hasten-and go -pass-Pres

to kane ga goo-----. kaze ga saa----.
when bell Sbj wind -Sbj

'When he quickened the pace of his walking and passed

by the place, he [heard] the bong of the bell [sound] long

and weary and the wind [start] rustling.'

Although the above use is very productive, it is limited only to use for dramatic

and/or rhythmic effects, and in the more ordinary contexts, constructions with verbs


2,2.2. Morpho-Phonologv of Mimetic Adverbs Monosyllabic and Bisyllabic Mimetic Adverbs

Two major types of mimetic adverbs are those based upon monosyllabic "roots" and

those based upon bisyllabic "roots." For simplicity's sake, I call them "monosyllabic

mimetic adverbs" and "bisyllabic mimetic adverbs." Note that reduplication of

monosyllabic "roots" results in bisyllabic mimetic adverbs, but in order to avoid

confusion, these will be identified as reduplicativee mimetic adverbs." For the present

discussion, see the following examples of the two categories. The quotative particle and

prosodic features are omitted from the following listing. The glosses are omitted for the

reasons given in


Monosyllabic Mimetic Adverbs: piN poN paQ
piiN pooN paaQ
piN-piN poN-poN paQ-pa
pi-piN po-poN pa-paQ

Bisyllabic Mimetic Adverbs: pakaQ pakuQ
pakaN pakuN
paka-paka paku-paku
paka-pakaQ paku-pakuQ

In addition, there are a small number of more complex and/or irregular

onomatopoetic forms. KokekoOkoo (the rooster's cry) and hoohokekyoo (the Japanese

nightingale's singing) are two typical examples of this category. Because these forms are

not part of the actual mimetic system, we will largely ignore this category hereafter,

although we will refer to it from time to time whenever the contrast between this category

and the more systematic core of the system reveals significant facts.

In the above, the term "root" was introduced with quotation marks; the term was

intended to refer to /pi-/, /po-/, /pa-/, /paka-/, /paku-/, etc.

The use of the term "root" is not too felicitous in the context of sound symbolism.

In as much as we are dealing with a system of sound symbolism, all forms are ultimately

broken down into semantic units corresponding to phonemes or phonological features. To

say that final /N/ is a morpheme while initial /p/ is merely a sound-symbolic unit thereby

positing a construction such as /pi-N/ and treating /p/ and /N/ as qualitatively different

would be misleading sound-symbolically. It would obscure the fact that /paN/ and /piN/ or

/paN/ and /kaN/ are as systematically related to each other as /paN/ and /paQ/ are to each

other. When we identify the meanings of constitutent elements, then, we will not separate

"morphemes" from "phono-semantic" units.

Admitting that this is a borderline area where classifications based upon traditional

concepts of morphology do not always coincide with dissections achieved by other criteria,
it nevertheless seems meaningful or at least useful, if only for practical purposes, to make a

distinction between "monosyllabic" and "bisyllabic" mimetic adverbs because of the
following morphological reasons.

First of all, we observe that either /CV/ or /C1V1C2V2/ (where C1 # C2) is

obligatory in sound-symbolic forms. All other elements such as vowel length,

reduplication, final /Q/ and /N/ are optional. That is to say, we not only have the basis for

treating /CV/ and /C1V1C2V2/ as units parallel to the traditional concept of "root" but also

have a support for setting up two morphological classes.

Secondly, optional variations such as reduplication and vowel length operate equally

on /CV/ or /C1V1C2V2/ as a single unit. For example, reduplication of /C1V1C2V2/

creates /C1V1C2V2-C1V1C2V2/. In this sense, /C1V1C2V2/ forms a coherent unit above

phonemes. The sequence /CV/ likewise functions as a unit This means that, although

sound-symbolically /C1V1C2V2/ may turn out to be /C1V1 + C2V2/ or even /C1 + V1 +

C2 + V2/, morphologically speaking, /C1V1C2V2/ behaves as a unit distinct from /CV/.

Third, there are certain operations that are not shared by the two categories. For

example, /-ri/ with or without intervening /-Q-/ or /-N-/ is applied only to bisyllabic forms.

An example is /pita-ri/ or /pi-Q-ta-ri/ meaning 'exactly.' A related form /piaQ to/ 'exactly'

clarifies that /pita-/ is the root in these forms.

As it turns out, the classification of mimetic adverbs into monosyllabic and bisyllabic

adverbs proves to be meaningful in the discussion of phonotactic constraints and iconicity.

See and 3.2.

Accepting the basic view above, one might wonder why /pi-piN, po-poN, pa-paQ/,

etc., are so identified as reduplicative monosyllabic mimetic adverbs instead of

nonreduplicative bisyllabic mimetic adverbs in the way /pakaN/ or /pokaN/ are.

Three facts about these forms support the analysis of these as reduplicated forms.

First, as we will see in, the meaning of /pi-piN/ is hurriednesss' which is not

present in bisyllabic mimetic adverbs of the similar shape. Second, /pi-pi-piN/ and longer

expansions with odd numbers of syllables exist with a meaning similar to the above.

These cannot be interpreted as based upon bisyllabic roots. Third, in all these forms, not

only the consonants but also the vowels are identical. In forms unambiguously based on

bisyllabic roots such as those with /-ri/, consonants must not be identical except in a few

k-initial forms, while the first and the second vowels can be either identical or different

This favors the monosyllabic root analysis. General Phonotactic Constraints Monosyllabic mimetic adverbs. A large number of monosyllabic

mimetic adverbs are onomatopoetic; they imitate sounds. This feature allows what appear

to be rather aberrant forms to be incorporated relatively easily into monosyllabic mimetic

adverbs. See the following, for example.

(2-8) [weeN to] (wailing)

[wiQ to] (a drunk's vocalization expressing content)
At least, the first of the above cannot be interpreted as */ueeN" to/ because

monosyllabic mimetic adverbs do not permit a diphthong in the form of */ue/. Nor is it

possible to interpret this as a bisyllabic mimetic adverb, because *[ueeN to] that would

correspond to the characteristic pattern of bisyllabic mimetic adverbs does not occur.

In the rest of Japanese grammar, /we, wi/ appear only in recent Western loanwords.

While monosyllabic mimetic adverbs allow such irregular strings, bisyllabic mimetic

adverbs do not allow them.

Likewise, initial /r/, which is not allowed in bisyllabic mimetic adverbs, appears in

/riiN riiN" to/, an onomatopoetic imitation of the sound of a small bell or an insect

chirping. In addition, /ru'N-ruN/ is used profusely in popular magazines for young people

and in their speech; this means something like 'fashionably and excitedly.'

(2-9) Moo ruN-ruN si-te dekake-masi-ta.
already do-and go out -Pol -Past

'She went out very excitedly.'

However, since /ruN-ruN/ does not appear with the accent pattern of */ruN ruN"

to/, it is not an unambiguously mimetic adverb. As we will see in 3.1, this particular

accent pattern is the earmark of the most iconic mimetic adverbs. In this sense, /ru'N-run/

may be considered not a true mimetic adverb.

The flap /r/ occurs in initial position in monosyllabic roots limitedly as above but

never in bisyllabic roots. The glide /y/, in contrast, is absent from the initial position of

monosyllabic mimetic adverbs, while appearing in a small number of bisyllabic mimetic

words. The glide /y/ is used in vocal segregates that are monosyllabic such as /yaQ/ (a

call) and /yoo/ Hi' but not in mimetic adverbs.

A summary of the frequencies of various initial consonants in monosyllabic mimetic

roots is given below. For the data used in this and other computations, see Appendix A.

The frequencies in (2-10) are based upon N and Q ending mimetic adverbs. N ending

variations on a single root such as CVN, CVVN, CV-CVN, CVN-CVN, etc. are treated as

constituting a single case. Likewise, Q ending variations on a single root such as CVQ,

CVVQ, CV-CVQ, CVQ-CV, etc. are treated as one occurrence. Forms with palatalized

consonants are omitted from consideration. For the consideration of palatalization, see


(2-10) /p/: 9 /b/: 9 /h/: 7 /m/: 3
/t/: 7 /d/: 5 Is/: 6 /z/: 6 /n/: 2
/k/: 9 /g/: 8
/w/: 7 /0/: 8
/r/: 1

The above shows that all obstruents, voiced or voiceless, are more frequently used

than nasals.

Using the same data, we can show the frequencies of vowels in monosyllabic

mimetic adverbs. See below. This shows that /e/ is considerably less frequent than the

other vowels.

(2-11) /i/: 19
/e/: 11
/a/: 18
/0/: 19
/u/: 20 Bisyllabic mimetic adverbs. The phonotactic constraints on the

consonants of bisyllabic mimetic adverbs differ slightly depending upon the types of

bisyllabic adverbs. Reduplicative forms are different from adverbs of the form CVQCVri

or CVNCVri. Figure 2-1 lists the combinations of consonants in reduplicative bisyllabic

mimetic adverbs of the form C1V1C2V2-C1V1C2V2. See Appendix B for the actual


This shows that no form of this type has the same consonant for C1 and C2.

Another apparent constraint is the absence of /r/ in initial poistion. Also note the

underrepresentation of /y/. These two consonants are also limited in monosyllabics. The

pattern of occurrence of all other consonants in the first position of bisyllabic mimetic

adverbs is not different from that in monosyllabic mimetic adverbs.

The second consonants appear in the following order: r/ > /k/ > /t/ > /s/ > /b/ > /y/ >

/z/ > /w/ > /g/ > /m, n/ > /d/ > /p/ > /h/. That is, /r/ is most frequent; voiceless obstruents

excluding /p/ follow it; next follows /b/; glides, voiced obstruents excluding /b/, and

nasals follow it; and finally /p/ and /h/ show up.

C\ r k t s b y z v g m n d p h total

r 0

k 11 12 5 2 1 1 2 34

t. 8 10 2 1 1 1 1 1 25

s 4 6 2 2 3 1 1 1 3 23

b 12 8 11 5 2 1 39

y 2 2 1 1 6

z 7 6 1 4 2 1 21

v 4 3 3 3 2 5 2 1 1 24

g 11 5 13 7 6 1 2 1 1 47

m 4 5 2 3 1 5 3 23

n 2 4 4 1 2 2 2 17

d 3 6 1 2 6 1 19

p 11 13 13 5 1 1 44

h 6 4 5 4 3 1 1 1 25

total 85 70 69 36 25 18 13 7 6 5 5 4 3 1 347

Figure 2-1: Consonant Combinations in CV 1C2V2-C1V1C2V2

C1 r t k s p y m v n b z g d h total

r 0

t 4 1 2 1 8

k 8 3 5 2 1 19

3 4 1 3 3 1 12

p 7 11 10 2 30

y 1 1 1 3

m 2 2 1 1 6

v 1 1 2

n 2 3 1 1 1 1 1 10

b 1 6 4 3 1 15

z 1 1 2 1 1 7

g 4 6 2 4 1 1 18

d 2 1 2 2 2 1 10

h 3 1 2 2 1 1 10

total 36 36 35 19 11 4 2 2 2 2 1 0 0 0 149

Figure 2-2: Consonant Combinations in CVQCVri or CVNCVri

Figure 2-2, on the other hand, lists the combinations of consonants in bisyllabic

mimetic adverbs with the form of CVQCVri or CVNCVri. See Appendix C.

The absence of/r/ from both the first and the second position is conspicuous. The

absence of /r/ from the initial position is the feature that permeates the whole mimetic

system. The absence of /r/ from the inventory of the second consonants is limited to this

particular type of mimetic adverbs. This is because of the constraint on consonant clusters

in mimetic adverbs: the clusters /Qp, Qt, Qk, Qs, Nb, Nm, Nz, Nn, Ny, Ng, Nw/ exist;

other combinations do not That is, only voiceless obstruents participate in geminate

clusters, and only voiced obstruents, nasals, and glides follow /N/. The consonant /h/,

which is almost non-existent in the second syllable of bisyllabic mimetic roots in general,

is absent from the second syllable of this type, too. The absence of /Nd/ seems to be

accidental; the frequency of voiced obstruents is not high in this position, and historically

/d/ merged with /z/ before high vowels /i, u/ (Mabuchi 1971).

As for the frequencies of the consonants in the first syllable, there is not much

difference between this type of mimetic adverb and the reduplicative bisyllabic or the

monosyllabic mimetic adverbs. A major difference turns up in the second syllable, where

the order of the consonants is /r/ > /t, k/ > /s/ > /p/ > /y/ > /m, w, n, b/ > /z/ > /g, d, h/.

Note the reverse order of/p/ and /b/ compared to that in the reduplicative type. For the

implications of this, see 6.2.

Another difference concerns /k/. In this type of bisyllabic mimetic adverb, the

constraint on the combination of the first and the second consonant is violated by those

forms possessing /k/ in both positions. These are koOkiri 'meagerly and exactly,' kaOkiri

'exactly,' kiOkari 'exactly'; kuOkiri 'vividly,' and koOkuri 'tasting rich.' These forms are

aberrant in that no other consonant can appear this way in either type of bisyllabic mimetic


Finally, it will be noticed that in both types, voiced obstruents are less frequent than

voiceless obstruents in the second syllable. In the first syllable, voiced obstruents are just

as frequent as voiceless obstruents. This is part of one general characteristic trait of

bisyllabic mimetic adverbs; there are more contrasts in the first syllable than in the second

syllable. Another case in point is palatalization in bisyllabic mimetic adverbs. As we will

see in 6.1, palatalization applies to the first syllable if there is no alveolar consonant in the

second syllable.

The frequencies of the vowels in the first and the second syllable of the reduplicative

bisyllabic mimetic adverbs are as follows.

(2-12) a. in the first syllable

/i/: 61
/e/: 41
/a/: 64
/o/: 94
/u/: 87

b. in the second syllable

/i/: 72
/e/: 15
/a/: 122
/lo: 72
/u/: 66

The above shows that e/ is the least used in both positions. Also, /e/ in the second

syllable is considerably less frequent than other vowels, while /a/ in the second syllable is

considerably more frequent than other vowels. Compare this situation with the first

syllable where the vowels are more or less evenly spread out, and the high frequency of /al

suggests another case of neutralized contrast in the second syllable. Semi-reduplicative mimetic adverbs. A considerable number of mimetic

adverbs consist of two different roots. These roots are habitually put together, and the

accent of the adverbs conforms to the least iconic accent pattern of the reduplicative

mimetic adverbs; that is, the pitch falls after the first mora.

In one type, the vowels in the first and second root differ, but the consonants are the

same. See the following list with sample objects that the mimetic words describe.

(2-13) ka'ra-koro the sound of wooden clogs

ka'ta-koto the sound of small hard objects
hitting against a box

ka'sa-koso the sound of dry leaves

ga'ta-goto the sound of trains

All of the above have /k/ or /g/ initially, and /-a-a/ of the first item is replaced by

/-o-o/ in the second. There are no other examples of this sort, except that expanded forms

of the above exist as fixed expressions such as /kara'N koro'N/ (e.g., the sound of

wooden clogs).

In the second type, the first moras of the two items are different; the second moras

are identical. In the following, glosses are provided where possible; otherwise, sample

objects that the mimetic words describe are given.

(2-14) pe'tya-kutya 'chitter-chatter'

bo'ka-suka sudden and heavy snow fall

he'do-modo 'being flustered'

mu'sya-kusya 'being sullen'

ti'ku-taku (< Tick Tock) 'tick tock'

ti'ra-hora light snowfall

ti'ya-hoya 'flattering'

te'ki-paki 'efficiently'

tu'be-kobe 'complaining about this and that'

de'ko-boko 'with bumps'

do'ta-bata 'stamping noisily'

do'ka-suka 'in a large quantity'

do'gi-magi 'being flustered'


'moving around in vain to free

zya'ka-suka 'noisily'

no'ra-kura 'lazily'

gi'ku-syaku 'in an uncordinated movement'

a'ta-huta 'in a hurry'

u'ro-tyoro 'moving around restlessly'

ya'ki-moki 'worriedly'

In the third type, entirely dissimilar roots are juxtaposed, as in the following.

(2-15) pi'i-tiku shrill peeping of a bird

pa'ti-kuri 'blinking eyes'

bu'tu-kusa 'complaining sullenly'

bu'u-suka the loud sound of a trombone

tyo'ko-maka 'moving around in small steps'

so'so-kusa 'busily'

ga'ta-pisi the noise from an old cart

gu'u-suka 'sleeping deeply'

All of the above have only one accented syllable; they are single words. A small

number of mimetic adverbs consist of two words which are customarily used together.

(2-16) pi'i-tiku pa'a-tiku shrill peeping

dota'N bataN 'stamping noisily'

dosiN bata'N 'stamping noisily'

nora'ri kura'ri lazily'

e'Qtira o'Qtira 'toiling'

2.3. The Grammar of Mimetics as Nominal Adjectives

2.3.1. Syntax

In Japanese grammar, there is a category of words called keevoo-doosi 'adjective

verbs.' Following Kuno (1973), I will call them "nominal adjectives." These words are

similar to adjectives in semantic and grammatical roles; they are also similar to nouns in the

way they appear with the copula. They represent qualities rather than actions and, unlike

nouns, are not the head of a noun phrase Like nouns, however, they do not inflect and

appear unchanged in front of the copula, desu/da. RiQpa 'splendid' is one example. See

the following.

(2-17) riQpa da/desu 'it is splendid'

riQpa na tatemono 'a fine building'

riQpa ni tateru 'to build splendidly'

riQpa de 'it is splendid, and...'

A large number of mimetic words are used in the same way as nominal adjectives,

and their use is commonly identified as "use of mimetic words as adjective-verbs."

Compare the following with above (2-19).

(2-18) bara-bara da/desu 'it is disorganized'

bara-bara no/na keekaku 'uncoordinated plan'

bara-bara ni tateru 'to build uncoordinatedly'

bara-bara de 'it is disorganized and ...'

The only difference between the non-mimetic and the mimetic nominal adjectives is

that the mimetics employ no as well as na before a noun; impressionistically, no seems to

be used more frequently for mimetic nominal adjectives than na.

2.3.2. Morpho-Phonology of Mimetic Nominal Adiectives

The mimetic nominal adjectives are classified into the following categories.

(2-19) a. accentless reduplicative and semi-reduplicatives


(y-)CVCVN-CVCVN betyoN-betyoN 'soaked'

(y-)CVCV-CVCV gata-gata 'shaky'

C(y)VN-C(y)VN puN-puN 'angry'

C(y)VV-C(y)VV gyuu-gyuu 'tight'

b. accented forms with /-ri/

(y-)CVQCV'ri piQta'ri 'just right'

In addition to these, there are a few exceptional forms such as /kiN-kira-ki'N/


Compare the above with (3-1) in 3.1. The similarity between the above classes and

those in (3-1) that belong to the less iconic classes is apparent. The only difference is that

the reduplicative mimetic nominal adjectives are accentless. Also apparent is the paucity of

varieties in mimetic nominal adjectives as compared to the varieties of mimetic adverbs.

For instance, partially reduplicative mimetic nominal adjectives which would correspond to

adverbial forms such as /pa'Q-pa" to/ 'quickly' do not exist. Nor is expansion allowed for

mimetic nominal adjectives. Also, mimetic nominal adjectives of the form /CVCVri/ do not

exist This means that mimetic nominal adjectives comprise a more limited and fixed

inventory of conventionalized forms than mimetic adverbs.

Kindaichi (1978) says that the mimetic adverbs and mimetic nominal adjectives are

inflectional variations of the same "adjectives" differing only in accent if at all. However,

as we shall see in 2.4, when mimetic nominal adjectives and mimetic adverbs are used as

modifiers of nouns, the semantics of mimetic nominal adjectives is not the same as that of

mimetic adverbs. Mimetic nominal adjectives tend to indicate abstract qualities rather than

on-going actions and sounds. In particular, they do not imitate sounds. See the


(2-20) a. paN-paN ni naru 'to stretch out'
(nominal adjective)
b. paN paN to tataku 'to hit with the sound of paN
(adverb) pgN'
(2-21) a. kaN-kaN ni naru 'to get angry'
(nominal adjective)

b. kaN kaN to tataku 'to hit with the sound of kaN
(adverb) kaN'

Since there are more mimetic adverbs than mimetic nominal adjectives and since

there is an iconic-conventional continuum between the mimetic adverbs and the mimetic

nominal adjectives, it seems best to interpret mimetic nominal adjectives as

conventionalized derivations of mimetic adverbs.

2.4. Mimetic Words in Noun Phrases

2.4.1. General Characteristics of Japanese Noun Phrases

Generally speaking, verb phrases that include mimetic adverbs are transformed into

noun phrases just like most other verb phrases in Japanese. Provided that the noun is

thematizable in the sense of Kuno (1973), any noun in a sentence with a verb phrase,

whether the verb phrase include a mimetic adverb or not, can be taken out and relativized

by simply dropping its particle and proposing the verb phrase to the noun. See below.

(2-22) underlying

Satoo-saN ga Yamada-saN ni riNgo no kawa o
Sato -Hon Sbj Yamada-Hon to apple Gen skin Obj

kuru-kuru mui-te age-ta.
peel-Ger give-Past

'Mrs. Sato peeled the skin of an apple into a spiral for

Mrs. Yamada.'

(2-23) a. relativized with ringo as head noun

[Satoo-saN ga Yamada-saN ni kawa o
Sato -Hon Sbj Yamada-Hon to skin Obj

kuru-kuru mui-te age -ta] riNgo
peel-Ger give-Past apple
'the apple whose skin Mrs. Sato spirally peeled for Mrs.


b. relativized with Yamada-san as head noun

[Satoo-saN ga riNgo no kawa o kuru-kuru
Sato -Hon Sbj apple Gen skin Obj

mui-te age-ta] Yamada-san
peel-Ger give-Past Yamada-Hon
'Mrs. Yamada for whom Mrs. Sato peeled the skin of the

apple spirally'

c. relativized with Satoo-saN as head noun

[Yamada-saN ni riNgo no kawa o kuru-kuru
Yamada-Hon to apple Gen skin Obj

mui-te age-ta] Satoo-saN
peel-Ger give-Past Sato-Hon

'Mrs. Sato who peeled the skin of an apple spirally for

Mrs. Yamada'

2.4.2. Mimetic Adverbs with the Verbs lu To Say' and Sita 'Did'

Two types of noun phrases with mimetic adverbs call for our attention. These noun

phrases have either iu 'to say' or sita 'did' as part of the modifier. See below.

(2-24) a. gooN toiu oto

'sound ringing like gooN = eg. the bong of the bell'

b. pikaO to iu hikari-kata

'a way of shining that says pikaO = flashing'

c. pota-pota to iu oti-kata

'a way of falling accompanied by the sound of pota-pota

= dripping'

(2-25) a. piti-piti sita ugoki-kata

livly movement'

b. taOpuri (to) sita huku

'loose outfit'

Each of the above two types exhibits particular characteristics which separate them

from ordinary noun phrases. For example, the sentences in (2-24) above cannot be tied to

any underlying sentences that have all of the components of the phrases as their

constituents. The sentences in (2-26) below are ungrammatical. The sentences in (2-24)

above are cases of "non-relativizing nominalization" which Makino (1968) describes.

(2-26) a. *Oto ga gooN to iu.
sound Sbj say

'?The sound says bong.'

b. *Hikari-kata ga pikaO to iu.
shine-manner Sbj say

'?The mode of shining is glaring.'

The second type differs from the ordinary noun phrase in that itaL which literally

means 'did,' is semantically tenseless in these constructions. The underlying sentences of

(2-25) are the following.

(2-27) a. Ugoki-kata ga Viti-piti si -te i -ru.
move-manner Sbj do-Ger Cont-Pres

The movement is lijvly.'

b. Huku ga taQDri si-te i -ru.
outfit Sbj do-Ger Cont-Pres

'The outfit is loose.'

Mere inversions of (2-25) result in ungrammatical constructions in (2-28).

(2-28) a. *Ugoki-kata ga piti-piti si-ta.

b. *Huku ga taQpuri si-ta.

In the following section, we will closely look at the characteristics of each.

2.4.3. lu

Iu in conjunction with mimetic words and the obligatory quotative particle to or X is

used to expressively describe "physical actions" by way of "quoting" them iconically.

Most characteristically, iu is used to indicate or imply certain sounds by "quoting"

them with the use of words known as gioNgo "sound-mimicking words." This is

understandable since iu means 'to say.' See the following.

(2-29) a. kokekoOkoo to iu niwatori no nakigoe
coakadoodle-doo Quot say rooster Gen cry

'a rooster's cry that sounds like cockadoodle-doo'

b. moo to iu nakigoe
moo Quot say cry

'the crying sound that sounds like moo = mooing'

c. waaO to iu kaNsee

'clamor that sounds like waaO = a sudden loud clamor'

d. zaa-zaa to iu mizu no oto
water Gen sound

'the sound of water that sounds like zaa-zaa = the

sound of torrents'

e. sakuO sakuO to iu oto

'the sound that sounds like sakuO sakuO = e. g. wooshing

sound of footsteps in the snow'

The mimetic adverbs in the following also may be considered to indicate sounds.

(2-30) a. gata-gata-gataO to iu taore-kata
Quot say fall -manner

'the manner of falling with the sound of

gata-gata-gataO = the way heavy objects fall in a row'

b. zaaO to iu kobore-kata
Quot say spill -manner

'the manner of spilling with the sound of zaaO = e. g. the

way a large quantity of liquid gushes out when spilled'

In (2-30a) and (2-30b) above, only iu may be used to tie the noun indicative of

sound and the sound-mimicking words. Sita is not allowed in this context. On the other

hand, even when the noun denotes sound, if the modifier is not onomatopoetic and

qualifies the noun in some other way, sita must be used. See the following.

(2-31) a. kiriO to sita koe

'cris voice'

b. *kiriO to iu koe

In 3.1, accentual classes of mimetic adverbs will be discussed. Of these classes,

some of the least iconic classes are virtually never used with iu in the above fashion. For

example, accentless monosyllabic adverbs (e.g., paaO to 'rapidly') and bisyllabic adverbs

with /-ri/ (e.g., pisya'ri 'abruptly and firmly') rarely appear this way. The reason

evidently is because these conventionalized forms do not imitate sound. On the other

hand, the more irregular and unsystematic onomatopoetic forms such as those imitating

animal vocalizations (e.g., kokekoOkoo 'dockadoodle-doo') are used only with iu when

modifying nouns. Also the expanded and more expressive forms of mimetic adverbs

appear with in, as we saw in (2-29e) and (2-30a) above; these forms also do not occur

with ia when modifying nouns.

As stated, the above cases involve forms ordinarily known as gioNgo

"sound-mimicking words." A straightforward statement could be made, if only gioNgo

were used with iu. As it turns out, however, there are other cases which are difficult to

interpret as cases of sound-mimicking. See the following.

(2-32) a. tara-tara to iu nagare-kata
flow -mode

'a mode of flowing that says tara-tara = thick


b. hyoiO hvoiO to iu tobi-kata

'a manner of jumping that says hvoiO hvoiO =

consecutive hopping'

c. tikuO to iu itami

'pain that says tikuO = a pricking pain'

d. ziiN to iu mahikaN
numbing sensation

'a numbing sensation that says ziiN = tingling feeling of


e. syariO to iu hazawari
feeling on the teeth

'the feeling on the teeth that says syariO = crisp touch'

f. betaO to iu kaNzi

'the feeling that says betaO = sticky feeling'

Dictionaries do not classify these mimetic forms as gioNgo 'sound-mimicking

words.' For example, Asano (1978) designates these as gitaigo 'mode-mimicking words.'

The common feature of the above uses is that the nouns are either nominalized verbs of

action or nouns of physical sensation. And with this type, also, the more creative and

expresive mimetic forms are permitted. The verb iuj then, is considered to serve to mark

off and enhance the iconic and expressive representation of physical actions.

It was stated that, in the onomatopoetic qualification of nouns, only iu is used to

form mimetic modifiers. When nominalized forms of verbs and nouns of physical

sensation are used with mimetic modifiers, on the other hand, ia often is replaceable by

sita, as we see in (2-33) and (2-34) below. In such cases, the meanings of the first and the

second of each pair are very similar, but the first is more expressive.

(2-33) a. betaO to iu kaNzi

'sticky touch'

b. betaO to sita kaNzi

'sticky touch'

(2-34) a. piti-piti to iu ugoki-kata

'limvel movement'

b. piti-piti to sita ugoki-kata

'lively movement'

Another category of nouns that are modified by mimetic words with in relates to

'facial expressions' and 'bodily gestures.'

(2-35) a. gvoO to iu kao

'the face that sas gyoQ = the facial expression that is

indicative of astonishment = a look of surprise'

b. pokaN to iu hyoozyoo
facial expression

'the facial expression that says pokaN = an

absent-minded look'

c. gaOkari to iu yoosu

'the appearance which tells that one is disappointed = a

disappointed look'

The verb iu in this type of noun phrases is used with mimetic forms which represent

psychological properties and means 'expressive of (a feeling).' The use ofi in this sense,

however, is not limited to mimetic words. The examples in (2-36) below demonstrate

non-mimetic use of the construction.

(2-36) a. dame to iu kaQkoo
no good gesture

'the gesture indicating that it did not go well'

b. aNsiN si-ta to iu hyoozyoo
relief do-Past expression

'the facial expression that indicates that he/she was


2.4.4. Sita

The use of sita as we saw in (2-25) has some noteworthy characteristics. First, it is

semantically almost vacuous. Its tenselessness is one aspect of this. It is semantically

vacuous also in the sense that it does not have a positive meaning in the way other verbs do

or in the way the same verb does in other contexts. In other contexts, sita means

'conducted' or 'accomplished' or stands for other verbs in ways similar to 'did' in

English. See (2-37) below.

(2-37) a. keNkyuu (o) si-ta
research Obj do-Past

'(I) did research'

b. piN piN to si-ta

'(I) did the action of piN piN = e.g., I struck the string


The verb sita in (2-37) can be replaced by a synonym yaOta 'did' as shown in


(2-38) a. keNkyuu (o) yaQ-ta
research Obj do-Past

'(I) did research'

b. piN piN to yaQ-ta

'(I) did the action of piN piN = e.g., I struck the string


On the other hand, the semantically vacuous sita cannot be replaced by yaOta. In the

following, (2-39b) is ungrammatical.

(2-39) a. pika-pika sita iwasi

'shiny herrings'

b. *pika-pika yaOta iwasi

Another characteristic of the noun phrases in question is that their underlying

sentences always have the relativized nouns of the noun phrases as their subjects, and only

as their subjects. In other words, these noun phrases resemble the type of noun phrases

consisting of adjectives and nouns or of nominal adjectives and nouns.

Thus, the role of sita in question is largely to form a grammatical construction of

adjective-like mimetic modifiers. This seems to be why Miyazi (1978) calls such cases of

sita zvootaisei-keesiki-doosi 'stative formal verbs.' It should be noted that these syntactic

characteristics of mimetic words plus sita are the same as those exhibited by certain verbs

classified simply as the fourth class of Japanese verbs by Kindaichi (1950) and as

'D-verbs' by Nagashima (1976). Here, we will use Nagashima's term and refer to them

as 'D-verbs.' Some examples of D-verbs are as follows.

(2-40) a. bakage -ta situmoN
be foolish-Past question

'a foolish question'

b. SitumoN ga bakage -te i -ru.
question Sbj be ridiculous-Ger Cont-Pres

The question is ridiculous.'

(2-41) a. odoke -ta ugoki-kata
be comical-Past move-manner

'a comical movement'

b. Ugoki-kata ga odoke -te i -ru.
move-manner Sbj be comical-Ger Cont-Pres

The movement is comical.'

(2-42) a. ni -ta katati
be similar-Past shape

'a similar shape'

b. Katati ga ni -te i -ru.
shape Sbj be similar-Ger Cont-Pres

The shape is similar.'

Hence, we will call these particular constructions of mimetic adverbs and sita or site

iru as mimetic D-verbs.

2.4.5. Mimetic D-verbs and Mimetic Nominal Adjectives

We have seen that not all mimetic words can be used with iu as modifiers in noun

phrases. We have also seen that restrictions exist as to the semantic categories of nouns

that can be modified by such constructions. Likewise, restrictions exist as to the mimetic

adverbs used in mimetic D-verbs. To start with, contrary to the case ofiu, genuinely

onomatopoetic mimetic words do not appear in mimetic D-verbs. This kind of restriction

is understandable given the nature of D-verbs in general which indicate qualities rather than

actions. In this section, we will look at the contexts of mimetic D-verbs a little more


In accomplishing this task, we need to contrast mimetic D-verbs with mimetic

nominal adjectives. This is because the latter, being similar to adjectives, also indicate

qualities rather than actions and appear in the same grammatical slots as mimetic D-verbs.

The question is if there is any semantic or otherwise systematic difference between

mimetic D-verbs and mimetic nominal adjectives. Indeed, we see that in certain cases, the

two seem to be interchangeable without apparent differences in meaning. See the


boso-boso sita gohaN

boso-boso no gohaN

gori-gori sita imo

gori-gori no imo

beto-beto sita te

beto-beto no te

zara-zara sita kami

zara-zara no kami

piOtari sita namae

piOtari no namae

'dry (improperly cooked) rice'

'dI (improperly cooked) rice'

'hard potato'

'hard potato'

tickyy hand'

'stick hand'

'coarse paper'

'coarse paper'

'approriate name'

'appropriate name'

(2-43) a.


(2-44) a.


(2-45) a.


(2-46) a.


(2-47) a.


(2-48) a. gyookai to beOtari sita seehu 'government which has
business with government inappropriately close ties
with business'

b. gyookai to beOtari no seehu 'government which has
business with government inappropriately close ties
with business'

In the above, both categories appear before the same nouns, and they are

interchangeable; there are many more of such cases. Nevertheless, a closer inspection of

the two categories proves that, with certain nouns, the two categories are either exclusive

or contrastive. Certain nouns appear only with one or the other; or before certain nouns,

the two categories mean distinctively different things. And in such cases, mimetic D-verbs

relate to animate objects, movements, or concrete physical properties, while mimetic

nominal adjectives relate to inanimate objects, stative conditions, or abstract ideas. The

details follow.

Generally speaking, nominalized forms of verbs indicating 'actions' are modified by

mimetic D-verbs rather than mimetic nominal adjectives. See the following.

(2-49) a. pika-pika sita hikari-kata
shine -manner

'glaring shining = glaring'

b. *pika-pika no hikari-kata

(2-50) a. piti-piti sita ugoki-kata

lively movement'

b. *piti-piti no ugoki-kata

(2-51) a. gata-gata sita yure -kata


b. gata-gata no yure -kata

(2-52) a. pota-pota sita oti -kata

b. pota-pota no oti -kata

Of the above mimetic nominal adjectives, only (2-52b) is non-existent; the other

nominal adjectives exist, as in pika-pika no tukue 'shiny desk,' piti-piti no zuboN 'very

tightly fit pants,' and gata-gata no kuruma 'run-down automobile.' 'Nevertheless, only

the mimetic D-verbs are allowed above. Certainly, there are more mimetic D-verbs than

mimetic nominal adjectives, and most of the mimetic D-verbs, like (2-52), do not have

counterpart mimetic nominal adjectives. However, the important point is that we find no

case where the relationship is reversed, i.e., where the mimetic nominal adjective alone is

used with a nominalized form of a verb of action. Hence, this is a structural/semantic

constraint rather than a statistical coincidence.

Likewise, nouns indicating voice, movements, feelings, sensations, personal

characteristics, attitudes, and facial expressions are modified by mimetic D-verbs rather

than mimetic nominal adjectives.

(2-53) a. kiNkiN sita koe

'high-pitched voice'

b. saku-saku sita hazawari
touch on the teeth

'crisp light texture (of food)

c. sowa-sowa sita taido


d. tvara-tvara sita taido

'irresponsible attitude'

e. saOpari sita kaQkoo
'simple and clean appearance'

f. saOpari sita azi

'refreshing taste'

Another category of nouns that are modified by mimetic D-verbs is that of animate

objects, as people and live aninals. See the following.

(2-54) a. sina-sina sita karada 'flexible body'

b. kvoro-kvoro sita kodomo 'child full of curiosity

c. haki-haki sita hito 'si ied person'

d. kiOtiri sita hito 'punctual person'

In the above example, only D-verbs may be used. Some nouns may be modified by

both mimetic D-verbs and mimetic nominal adjectives, as we have seen in (2-44) through

(2-49). Also, even nouns derived from verbs may be modified by mimetic nominal

adjectives, if they can refer to the stative result of an action. See below.

(2-55) a. gotva-eotva sita narabe -kata
lay out-manner (=nominalizer)
'messy manner of arrangement'

b. ota-gotva no narabe -kata
lay out-manner (=nominalizer)

'messy arrangement (as a result)'

A large number of nouns which indicate stative objects, on the other hand, may be

modified only by mimetic nominal adjectives, as in (2-56). For these nouns, corresponding

D-verbs produce ungrammatical phrases.

(2-56) a. ore-vore no kooto 'shabby coat'

b. iri-iri no zikaN 'close timing'

c. oti-goti no imo 'hard potato'

d. eura-gura no yu 'boiling hot water'

e. gusvo-gusvo no yoohuku 'drenched dress'

f. gusva-gusva no omuretu 'slopy omlet'

g. gati-gati no yuki 'icy snow'

h. kusva-kusva no kami 'wrinkled paper'

i. hova-hova no yaki-imo 'steaming hot baked yam'

Abstract or non-tangible nouns also may be modified only by mimetic nominal

adjectives. See the following.

(2-57) a. bara-bara no ikeN 'different opinions'

b. pera-pera no eego 'fluent English'

c. pari-pari no nihoNgo 'unadulterated Japanese'

d. tyobo-tvobo no seeseki 'mediocre grades'

So far, the semantic categories of nouns have been correlated with the categories of

modifiers. As it turns out, however, the semantic categorization of nouns is not always

clear-cut Ordinarily "stative" objects may be mobile under certain circumstances. In such

cases, if "movement" is the intended meaning, a mimetic D-verb may be used. See the


(2-58) ura-ura sita isu 'wobbly chair'

Conversely, even when the noun indicates 'people,' if the expression concerns

'immobility' or 'dead bodies,' D-verbs are not used; instead, mimetic nominal adjectives

are used in such cases. See the following.

(2-59) a. bero-bero no yoQparai 'dead-drunk drunkard'

b. *bero-bero sita yoQparai

We have largely focused on cases where only one or the other type of mimetic

modifer is acceptable. In certain cases, however, we find that both categories are allowed

before the same noun with clearly contrastive meanings. See the following for instance.

(2-60) a. bara-bara sita ugoki 'a diffused sequence of

b. bara-bara no ugoki 'uncoordinated different

The phrase in (2-60a) above indicates an uncoordinated, untimed sequence of

movements by an individual or individuals. The phrase in (2-60b) on the other hand

indicates different uncoordinated movements which are going on all at the same time. In

other words, the mimetic D-verb in (2-60a) concerns 'temporal change,' while the mimetic

nominal adjective in (2-60b) concerns synchronicc diversity.' The contrast is basically that

of 'action' and 'non-action.'

The contrast between 'movement' and 'stativeness' occasionally appears in the form

of the contrast between 'physical dimensions' and 'non-dimensional properties.' See the


(2-61) a. karada ni piOtari sita huku
body to outfit
'an outfit that fits the body just right'

b. natu ni piOtari no huku
summer to outfit

'an outfit that is appropriate in summer'

In (2-61a), the outfit fits the physical dimensions of the body; in (2-61b), the outfit

meets the requirements of summer in terms of non-dimensional properties such as color,

thickness, appearance, style, material, etc..

Semantic classifications are always a nebulous business, because one can play with

them and attach unusual meanings to grammatically correct expressions. Nevertheless, the

oddity of expressions that transgress semantic boundaries is striking to the native speaker.

The ridiculousness of (2-62b) as compared to the naturalness of (2-62a) stems from this.

(2-62) a. pika-pika sita iwasi no mure
sardine Gen school

'a school of shiny sardines'

b. *pika-pika no iwasi no mure
sardine Gen school

'a school of shiny sardines (?)'

Because a nominal adjective is used instead of a mimetic D-verb, (2-62b) sounds as

if the sardines were either made of metal and polished or stuffed and greased. One native

speaker, when he heard this expression, reacted to it by bursting out into laughter as he

said, "There is no such thing!"

To summarize, although there are cases where the contrast is not clear as in (2-43)

through (2-48), the mimetic D-verbs generally indicate 'movements,' 'physical

dimensions' and so on, while the mimetic nominal adjectives indicate inanimatenesss' and

more abstract ideas. Taking into account the characteristics of mimetic modifers with ji,

we can identify a continuum of iconicity in mimetic modifiers: forms with ij are most

iconic, and mimetic nominal adjectives are least iconic. This is also reflected in the

morpho-phonological characteristics of respective categories. We have already seen that

forms with ij contain more dramatic, mimetic, and creative or irregular forms than mimetic

D-verbs. Mimetic D-verbs in turn contain more dramatic mimetic forms than mimetic

nominal adjectives. Recall that only reduplicated or semi-reduplicated mimetic forms and

those with /-ri/ are used with mimetic nominal adjectives; unreduplicative forms are not

used. Besides, reduplicative mimetic forms in nominal adjectives carry no accent, which is

one of the least mimetic of all accentual types. Also the majority of mimetic nominal

adjectives are based on bisyllabic roots, which are less mimetic than monosyllabic roots.

This problem of iconic continuum will be taken up again in Chapter 3.

It was suggested in 2.4.4 that sita in mimetic D-verbs is almost vacuous

semantically, but the present discussion leads us to the revised conclusion that the

semantics of sit is partially reflected in the sense of animatenesss.'

2.5. Derivations

2.5.1. Nouns

A number of nouns clearly derive from mimetic words; most of them are used in

slang, colloquial speech, casual writing or baby talk (=adult speech to young children).

Those unambiguously demonstrable as of mimetic origin belong to either the phonological

type in (2-63) or the phonological type in (2-64) below.

The words shown in (2-63) are phonologically indistinguishable from the most

conventional mimetic adverbs; their accent falls on the first syllable and the majority of

them are reduplicative forms of monosyllabic roots. Many of them are used in baby talk,

and, deriving from onomatopoetic mimetic words, they indicate objects or animals that

produce such sounds.

(2-63) a. waN-waN

b. koN-koN

c. po'Q-po

d. po'N-poN

e. tyu'u-tyuu

f. tyuTN-tyuN

g. pi'ka-doN

h. to'N-kati

i. boi'N

'dog' (from the barking of dogs)

'fox' or 'coughing' (from the sound of
foxes or coughing)

'locomotive' (from its sound)

'tummy' (from the sound one would
produce when tapping the belly)

'mouse' (from its sound)

'small bird' (from the sound of chicks)

'A-bomb' (from its sound and flashes)

'hammer' (from the sound of hitting with
a hammer)

'female breast' (from the sound of
hitting a voluminous object)

The words in (2-64) below are phonologically indistinguishable from mimetic

nominal adjectives; they are accentless reduplicative forms. They are different from

nominal adjectives in that they can appear as subject or object of a sentence. Interestingly

enough, only reduplicative forms of bisyllabic mimetic roots are used this way as nouns,

although there are both monosyllabic and bisyllabic mimetic nominal adjectives.

(2-64) a. syabu-syabu




e. nuru-nuru

f. hira-hira

g. kasa-kasa

h. gara-gara

Clearly the above use of accentless

'a Japanese dish with thinly sliced meat'
(from the sounds one makes when moving
meat in the soup)

'small dots or protrusions'

'small dots'

'protrusions on coarse or thick objects'

'sticky substance'


'dry rough place on skin'

'baby's rattle'

bisyllabic mimetic reduplicative forms as nouns

derives from the syntactic similarity of nominal adjectives to nouns. The fact that no

accentless reduplicative forms of monosyllabic roots are used as nouns of the type shown

in (2-64) above seems to be a correlate of the fact that there are few monosyllabic mimetic

nominal adjectives to start with. Therefore, we may consider the absence of accentless

monosyllabic mimetic nouns as a statistical coincidence. Of course, the very fact that there

are few monosyllabic nominal adjectives is not a coincidence. It is the consequence of the

strong iconicity of monosyllabic mimetic roots which defy conventionalization. In this

sense, the absence of accentless monosyllabic mimetic nouns is also the consequence of

this strong iconicity of monosyllabic mimetic roots.

The absence of accented reduplicative bisyllabic mimetic nouns in (2-63), on the

other hand, is not a statistical coincidence, because there are a large number of accented

reduplicative bisyllabic mimetic adverbs. The reason is not clear, but it may be due to the

strong onomatopoetic requirement of baby talk and similar borderline phenomena.

2.5.2. Compound Nouns

A type of compound noun in which a bi-moric mimetic form is proposed to a noun is

shown below.

(2-65) a.










The compounding

is shown below.

(2-66) a.


















b or-mo(')oke



of a reduplicative or









'thick and sticky praise = excessive praise'

tatteredd cloth = rag'

'junk ball = badly pitched ball'

largeg lunch box = large lunch eater'

'torrential rain'

'abrupt and forceful drinking = gulping'

'sneak thief .

'small mustache'

'rugh slicing'

'coarse paper'

'rag profit = easy money'

'dilapidated house'

'soft wind = breeze'

semi-reduplicative mimetic word and a noun

'tight packing'

'torrential rain'

'smiling face'

'lazy person'

lzy disposition'

'boisterous merrymaking'

'ngisy comedy = slapstick comedy'

'scattering murder = mutilation murder'

i. giku-syaku-u'NteN 'clumsy driving'

j. peN-pe'N-gusa 'clinking grass = shepherd's
purse (plant)'

The third type is the compounding of a mimetic word ending with /-ri/ and a noun.

(2-67) a. zuNguri-gata 'gstoc type'

b. tyaOkari-ya 'calculating person'

c. poOkuri-byoo 'suddenly (dying) disease'

d. huNwari-o'muretu 'soft omelet'

e. biOkuri'-bako 'surprise box = jack-in-the-box'

The accent pattern of these compound nouns agrees with the description of

compound nouns by McCawley (1977). If the final element is three or more moras long,

the accent is placed on the first mora of the final element If the final element is less than

three moras long, there are three types of accent according to McCawley. In the above we

see only two. In most of the above compounds in which the final element is less than three

moras long, the compounds are accentless. In (2-651), (2-65m), (2-66j) and (2-67e), on

the other hand, the accent is on the last syllable of the first element

2.5.3. Other Compounds

There are a small number of compound verbs, compound adjectives, and compound

nominal adjectives that contain mimetic words.

Compound verbs are shown in (2-68) below. Compound adjectives are shown in

(2-69). Compound nominal adjectives are shown in (2-70).

(2-68) a. tyoro-makasu 'defeat cheaply = cheat'

b. .tvo-giru 'cut in a quick. short stroke =

c. bura-sagaru 'dangle loosely'

(2-69) a.

b. muzu-gayui

c. hyQro-nagai

(2-70) a. hara-.eko da

b. hana-yet.a da

c. maru-poyg da

d. gaa-aki da

'slightly bitter'

'greepy-itchy = creepy'

'skinny and long = lanky'

'the stomach is flat = (she) is
starving hungry'

'(his) nose is flat'

'(she) is round and plumD = (she)
is plump'

'(the store is) completely

2.5.4. Verbs Derived from Mimetic Words /-tuk-/ and /-mek-/

A large number of verbs seem to be related to mimetic words. A considerable

number of them are clearly traceable to mimetic words. These are formed by addition of

derivational suffixes to mimetic words. The most productive of them is the derivational

suffix /-tuk-/. This suffix is used exclusively with mimetic roots, and its meaning is 'to

overtly/excessively exhibit the feature of (the mimetically expressed condition).' Note that

only bisyllabic mimetic words are used with /-tuk-/. Some examples are listed below










'to be sprinkled'

'to be sticky'

'to be disorganized, to be diffuse'

'to be excessively dry'

'to be shaky'

'to behave frivorously'

'to flicker'

'to be flashy'


dabu-tuk-u 'to be overabundant'

kira-tuk-u 'to glitter'

gira-tuk-u 'to glare'

Note that initial /p/ which is characteristic of mimetic words appears above. The

suffix /-tuk-/ is the only derivational suffix that is affixed to p-initial mimetic roots. With

other suffixes, we see /h/, a reflex of/p/, in initial position.

Another suffix that clearly derives verbs from mimetic words is /-mek-/. In contrast

to /-tuk-/, /-mek-/ means 'to fashionably/appropriately/elegantly exhibit the mild feature of

(the condition expressed by the mimetic words),' as in the following.

(2-72) hisi-mek-u 'to be jammed with people (because of
the popularity)'

toki-mek-u '(the heart) throbs with joy'

doyo-mek-u 'to stir (with applause, etc.)'

zawa-mek-u 'to stir (with applause, admiration)'

kira-mek-u 'to glimmer'

yoro-mek-u 'to reel elegantly'

This suffix is used with nouns, too, as in the following.

(2-73) aki-mek-u 'to look like fall'

har-mek-u 'to look like sring'

tki-mek-u 'to be timly, to be in power'

The contrast between /-tuk-/ and /-mek-/ in terms of

inappropriateness/appropriateness is clear in the following examples.

(2-74) a. Hosi ga kira-mek-u.
star Sbj

The stars twinkle.'

b. Yasude no soosyokuhiN ga kira-tuk-u.
cheap Gen accessory Sbj

'Cheap accessories glitter.'

(2-75) a. Omoigakenai eNsyutu ni zyoonai wa
unexpected stage effect to theater Top


The unexpected stage effect caused a stir among the


b. Maiku no kosyoo de zyoonai wa
microphone Gen trouble with audience Top

itizi zawa-tui-ta.

'For a short time, there was a general stir in the

audience at the trouble of the microphone.'

In the above, (2-74a) and (2-75a) concern something esthetic, and /-mek-/ is used

there. In (2-74b) and (2-75b), inappropriate events are dealt with, and /-tuk-/ is used. /-ke-/ and /-kas-/

Derivational suffixes /-ke-/ and /-kas-/ also appear quite often with mimetic words.

The meaning of/-ke-/ is something like 'to turn to (the state expressed by the mimetic


(2-76) boya-ke-ru 'to become blurred'

huya-ke-ru 'to swell up, to become sodden'

toro-ke-ru 'to melt'

niya-ke-ru 'to get foppish'

yoro-ke-ru 'to reel'

izi-ke-ru 'to become crooked'

dara-ke-ru 'to become lazy'

The suffix /-kas-/ means 'to turn X into (the condition expressed by the mimetic

word).' It is a causative suffix.

(2-77) boya-kas-u 'to blur'

huya-kas-u 'to steep' /-el and /-as-/

The contrast between /-ke-/ and /-kas-/ is parallel to the contrast between /-e-I and

/-as-/. The meaning of /-e-/ is 'to do (the action/movement indicated by the mimetic

word),' while the meaning of /-as-/ is 'to cause X to do (the action indicated by the mimetic


(2-78) a. yur-e-ru (cf. yura-yura) 'to sway'

b. yur-as-u (cf. yura-yura) 'to rock'

(2-79) a. tar-e-ru (cf. tara-tara) 'to droop'

b. tar-as-u (cf. tara-tara) 'to hang' /-ge-/ and /-gas-/

The voiced variants of/-ke-/ and /-kas-/ are also used with mimetic words, as in

(2-80a, b) below. Another suffix /-gar-/ is similar to /-ge-/ and is used as in (2-81c).

(2-80) a. koro-ge-ru 'to tumble (intr.)'

b. koro-gas-u 'to tumble (tr.)'

c. koro-gar-u 'to tumble (intr.)' /-k-/ and /-g-/

Derivational suffixes /-k-/ and /-g-/ appear in the following forms. Their meaning is

'to do the action/movement indicated by the mimetic word.'

(2-81) hata-k-u (cf. pata-pata) 'to dust, to beat'

hazi-k-u (cf. pati-pati) 'to fillip, to snap'

yura-g-u (cf. yura-yura) 'to sway'

soyo-g-u (cf. soyo-soyo) 'to rustle in the wind'

sawa-g-u (cf. sawa-sawa) 'to clamor' /-mar-/. /-m-/. and /-me-/

The suffix /-mar-/ derives intransitive verbs; its meaning is 'to become (the condition

indicated by the mimetic word).'

(2-82) yuru-mar-u (cf. yuru-yuru) 'to become loose'

kuru-mar-u (cf. kuru-kuru) 'to be rolled up'

The suffix /-me-/ derives transitive verbs; its meaning is 'to turn X to (the condition

indicated by the mimetic word).'

(2-83) yuru-me-ru (cf. yuru-yuru) 'to loosen'

kuru-me-ru (cf. kuru-kuru) 'to wrap up'

heQko-me-ru (cf. peko-peko) 'to flatten'

The suffix /-m-/ derives both intransitive and transitive verbs.

(2-84) yuru-m-u (cf. yuru-yuru) 'to become loose (intr.)'

muku-m-u (cf. muku-muku) 'to swell (intr.)'

kura-m-u (cf. kura-kura) 'to become dizzy (intr.)'

kuru-m-u (cf. kuru-kuru) 'to roll up (tr.)' Miscellaneous Derivations of Verbs

The suffix /-b-/ derives one mimetic verb.

(2-85) koro-b-u (cf. koro-koro) 'to tumble (intr.)'

The following verbs also may be considered to include sound-symbolic


(2-86) mus-u (cf. musi-musi) 'to be sultry'
kosur-u (cf. kosi-kosi) 'to rub'

sur-u (cf. sura-sura) 'to rub'

The resemblances may be coincidental in these cases, however. In any case the

abundance of clearly mimetic verbs is striking, though not surprising, considering the close

relationship between the verbs and the mimetic words in Japanese.

By contrast, very few adjectives can be considered to be of direct mimetic origin.

And those that are tied to mimetic roots are neologisms or slang words, as below.

(2-87) noro-i (cf. noro-noro) 'slow'

tyoro-i (cf. tyoro-tyoro) 'easy'

toro-i (cf. toro-toro) 'slow, stupid'

boro-i (cf. boro-boro) 'easy'


3.1. Accent and Iconicity in Mimetic Adverbs

It is best to treat one phono-semantic correlation in mimetic adverbs separately

before we embark on the question of "sound symbolism" as we generally understand it.

This is the role of accent in mimetic adverbs. The following discussion will demonstrate

that the accentual classification of mimetic adverbs is correlated with the continuum of


For the purpose of this discussion, we will look at p-initial monosyllabic mimetic

adverbs. The actual p-initial forms which are used for this analysis are listed in (3-1)

below. Three prosodic and three non-prosodic criteria classify the forms into four classes

as we will see below. Phonemic pitch fall, or word accent, is marked by //; it is seen in

the last class. Intonational pitch fall is marked by /"/; it is seen in the first two and the last

class. This list, for one thing, is provided to show the possible range of variation for each

group of monosyllabic mimetic adverbs which share a common initial consonant. Since

p-initial mimetic adverbs are the most productive category, comparisons with other

monosyllabic mimetic adverbs prove that the list almost exhausts all the possibilities for

monosyllabic mimetic roots.


Class I: [ to]
piN" to piiN" to piN piN" to piiN piiN" to
paN" to paaN" to paN paN" to paaN paaN" to
poN" to pooN" to poN poN" to pooN pooN" to
puN" to puuN" to ?puN puN" to ?puuN puuN" to
pyoN" to pyooN" to pyoN pyoN" to pyonN pyooN" to
piQ" to piiQ" to piQ piQ" to piiQ piiQ" to

peQ" to p
paQ" to p
poQ" to p
puQ" to p
pyuQ" to p

Class II: [ to]
pi-piN" to
pa-paN" to
po-poN" to
pyo-pyoN" to
pi-piQ" to
pe-peQ" to
pa-paQ" to
po-poQ" to
pu-puQ" to
pyu-pyuQ" to

Class III: [ to or
piN to
paN to
poN to
puN to
pyoN to
piQ to
paQ to
poQ to
puQ to
pyuQ to

Class IV: [ to]
pi'N" to
paN" to
po'N" to
pu'N" to
pyo'N" to

pi'i" to
pa'a" to
pyu'u" to
po'i" to

eeQ" to
aaQ" to
ooQ" to
uuQ" to
yuuQ" to
oiQ" to
uiQ" to
yoiQ" to
ii" to
oi" to
ui" to
yoi" to

peQ peQ" to
paQ paQ" to
poQ poQ" to
puQ puQ" to
pyuQ pyuQ" to

peeQ peeQ" to
paaQ paaQ" to
pooQ pooQ" to
puuQ puuQ" to
pyuuQ pyuuQ" to
poiQ poiQ" to
puiQ puiQ" to
pyoiQ pyoiQ" to
pii pii" to
poi poi" to
pui pui" to
pyoi pyoi" to

pi-pi-piN" to
pa-pa-paN" to
po-po-poN" to
pyo-pyo-pyoN" to
pi-pi-piQ" to
pe-pe-peQ" to
pa-pa-paQ" to
po-po-poQ" to
pu-pu-puQ" to
pyu-pyu-pyuQ" to

piiN to
paaN to
pooN to
puuN to
pyooN to
piiQ to
paaQ to
pooQ to
puuQ to
pyuuQ to

pi'N-piN(" to)
paN-paNC( to)
po'N-poN(" to)
pu'N-puN(" to)
pyo'N-pyoN(" to)
pi'Q-pi" to
pa'Q-pa" to
po'Q-po" to
pu'Q-pu" to
pyu'Q-pyu" to
pi'i-pii(" to)
pa'a-paaC' to)
pyu'u-pyuu(" to)
po'i-poi(" to)

pu'i" to ?pu'i-pui(" to)
pyo'i" to pyo'i-pyoi(' to)

The criteria for classifying the above forms into four classes are (1) the presence or

absence of phonemic accent, (2) the presence or absence of prosodic pitch fall before the

quotative particle Lt (3) the presence or absense of initial pitch rise, (4) the possibility of

lengthening the vowel, (5) the optionality of the particle tQ, and (6) the possibility of freely

expanding forms. The table below summarizes the characteristics of each class using both

accentual and non-accentual criteria.

Table 3-1: Summary of Accentual Classes

phonemic pitch fall initial expandability vowel It
accent before to pitch rise length

I no yes no yes long/short obligatory

II no yes yes yes long/short obligatory

III no no yes no long/short obligatory

IV yes yes yes no short only obligatory/

Class I and Class II forms exhibit a marked pitch fall between the last long syllable

of the mimetic forms and the particle of quotation I or e. Note that this type of accent is

the one used for imitative quotation of animal and other sounds which do not fit in the

general description of the mimetic system, as in the following.

(3-2) doQbooN" to [doQbooN to] (sound of plunging

weeN" to

[weeN to]

koke'koQkoo" to [kokekoQkoo to]

into water)



The above pattern of pitch fall after a long syllable is peculiar because, in Japanese
phonology in general, this pattern is allowed only at the end of an emphatic quotation, as

in (3-3) below.

(3-3) KaidaN" to iQ -ta.
stairs Quot say-Past

'I said, "Stairs."

In the rest of the grammar, a syllable of the shape /(C)VN/, for instance, can be

either unaccented or accented between V and /N/; that is, the pitch can fall only between V

and /N/, not after /N/. See the following.

(3-4) a. daN ga (accented) 'step (Sbj)'

b. kai.daN ga accentlesss) 'stairs (Sbj)'

Notice that the example in (3-2) above is an accentless word, as (3-3b) indicates. In

other words, this type of accentuation overrides the phonemic accent; it is part of the

intonational prosody of a grammatical unit larger than a word. The prosodic pitch fall of

Class I and Class II mimetic words therefore is not phonemic in the ordinary sense; these

mimetic forms, in other words, are accentless. The initial pitch fall in Class IV, on the

other hand, is phonemic; the mimetic forms in this class are accented. The notations /'/ and

/"' are used to mark this contrast Note that there is a slight pitch fall before to in Class IV,

too. This also is interpreted to be /"/, but being at the end of an already descending phrase,

the pitch fall is not as marked as in the case of Class I or Class II.

The basis for distinguishing Class I from Class II is the presence or absence of

initial pitch rise. Class II forms are characterized by a pitch rise after the initial mora; Class

I forms uniformly lack such a rise. See the following contrast.

(3-5) a. [piN piN piN piN to] used, for example, to
(Class I) describe the action of
striking a string

b. [pi-pi-pi-piN to] used, for example, to
(Class II) describe the action of
striking a string
hurriedly several times

The absence of pitch rise in an unaccented initial syllable itself is not unique to the

Class I mimetic adverbs. In the other strata of Japanese, also, we see the same process.

That is, if an unaccented initial syllable contains the syllable-final nasal /N/ or a long

vowel, it is often realized with a flat high pitch without a rise. On the other hand, if the

initial syllable contains only one mora /(C)V/ or contains syllable-final /Q/ instead of /N/, a

rise follows that syllable. See the following.

(3-6) a. [doN.kaN] [doN.kaN] 'numb'

[] [] 'peak, pass'

b. [o.riN.piQ.ku] 'Olympic games'

[] 'pressure'

There is a difference between (3-5a) and (3-6a) above, however. In (3-5a), no rise

appears, however carefully one might articulate the form. In (3-6a), on the other hand, the

initial rise appears when it is carefully articulated.

The above difference is considered to arise from the difference in the unit we are

dealing with. In (3-5b), (3-6a), and (3-6b), each form is accentually a single unit, a word,

each having maximally one pitch fall. In (3-5a), on the other hand, the mimetic portion is

considered to consist of a number of monosyllabic words, no one of which is subordinated

to any other. If an initial pitch rise were permitted for each unit of Class I forms, such

inadmissible strings as *[piN to] or *[piN piN to], which have a rise and a fall in a single

syllalbe, would be produced.

The supporting evidence for this is the parallellism between monosyllabic and

bisyllabic forms, as in (3-7) below.

-- 'L,--
(3-7) a. piN" to [piN to]

b. piN piN" to [piN piN to]

c. pataN" to [pataN to]

d. pataN" pataN" to [pataN pataN to]

The forms in (3-7a) and (3-7b) above are considered to be underlyingly pararell to

(3-7c) and (3-7d); *[pataN pataN to] does not exist, and each is the most iconic form in its

class as we shall see later.

The treatment of Class I forms as consisting of multiple words explains the

irregularity observed in (3-8) below, too. In Class I, a voiced geminate cluster [bb] (/Qb/)

is allowed, violating the predominant constraint on Japanese geminate clusters; it is not

allowed in Class IV.

Class I paQ paQ" to baQ baQ" to
paN paN" to baN baN" to

Class IV pa'Q-pa" to *ba'Q-ba" to
pa'N-paN(" to) ba'N-baN(" to)

Non-prosodic criteria also set apart the Class I forms from the other classes. Class I

forms are expandable; the variations are virtually infinite. The vowels can be lengthened.

The particle of quotation, IQ or e is obligatory.

As explained above, Class II forms also have the pitch fall before the particle of

quotation, but they consist of only one accentual unit followed by 1t. Class II forms

exhibit initial pitch rise, and they are partially reduplicated forms. The vowel of the last

unit can be either short or long. The particle of quotation again is obligatory. Class I and

Class II are very similar.

Class III and Class IV monosyllabic forms differ from the above and from each

other as follows.

Class III forms are accentless; there is no phonemic pitch fall. Moreover there is no

pitch fall before 1t; the mimetic form and the quotative particle 12 form an inseparable word

unit. In this respect, they are different from Class I and Class II forms. Like Class II

forms, however, the accentless long syllable of these mimetic forms may exhibit a pitch

rise. When articulated carefully, the unaccented initial syllable of a Class Im form is

accompanied by a pitch rise between the first and the second mora.

Class III forms do not contain reduplicated, triplicated, or otherwise expanded

forms. Unlike Class I and Class II forms, their expandability is limited. The vowels of

Class III forms, however, can be short or long like Class I and Class II forms. Also, like

Class I and II, oI is obligatory.

Class IV forms, on the other hand, are accented on the first syllable. This class

alone exhibits phonemic pitch fall.

Class IV forms are fixed in shape: their vowels are always short, and there are only

two options of expansion, unreduplicated or reduplicated. Forms such as /*pi'iN-piN" to/

and /*pi'N-piN-piN" to/ do not exist. As for the optionality of I, there are two

subclasses. With fully reduplicated forms such as /pi'N-piN/ and /pi'i-pii/, Io is optional.

In other forms, It is obligatory.

We now clearly see a dichotomy between the more dynamic and flexible mimetic

adverbs of Class I and Class II and the less flexible and fixed mimetic adverbs of Class III

and Class IV. In Class I and Class II, various mechanisms mark off the mimetic forms

from the rest of the sentence. At the other extreme, the mimetic elements merge with the

rest of the sentence because of the omission of I or because of the accentual integration of

IQ with the mimetic forms.

The above formal characterization of the contrast makes sense in view of the

functional and semantic differences of the classes. Class I and Class II appear where a

more dramatic and sound-symbolic effect is expected. Class IV, on the other hand,

typically appears in idiomatic expressions, as in the following.

(3-9) do'N-doN 'rapidly'

ho'i-hoi 'without thinking

pu'N-puN suru 'to get angry'

mu'N-muN suru 'to be sultry'

tu`N-tuN suru 'to be touchy/cross'

piN-piN suru 'to be energetic'

pi'i-pii suru 'to be distressed without

All of the above expressions retain a substantial trace of sound-symbolic meanings

that allows one to tie them with their Class I counterparts. Nevertheless, their meaning are

considerably narrower than or shifted from those of their Class I counterparts so that they

cannot be used in the same context See the following, for example.

(3-10) a. DoN-doN de -te ku -ru.
exit-Ger come-Pres

They come out rapidly in a large quantity.'

b. DoN doN"to de -te ku -ru.
exit-Ger come-Pres

They come out with drum-like sounds.'

(3-11) a. Pi'N-piN si-te i -ru.
do-Ger Cont-Pres

'She is healthy [for her age].'

b. ?PiN piN" to si-te i -ru.
do-Ger Cont-Pres

'She is plucking the string twice (?).'

The foregoing statement is not intended to mean that all Class I and Class IV

counterparts are as different from each other as above (3-10) and (3-11). On the contrary,

most of Class I and Class IV counterparts are interchangeable with only a slight difference

in their nuances. For example, see the following.

(3-12) a. paN paN" to tataku 'to beat flatly and vigorously

b. DaN-paN tataku 'to beat flatly and vigorously

What should be emphasized here is rather that, if there is a substantial difference, it

is the Class IV forms that carry idiomatic and less sound-symbolic meanings.

The preceding discussion is based upon a contrast between reduplicative adverbs of

Class I and Class IV. The same can be said of the contrast between Class I and Class III

non-reduplicative adverbs to a large extent, although the situation is a little more complex

in this case as we shall see later in 3.3. See the following for the moment.

(3-13) a. ZaOto mitumoQ-ta kagiri de wa ni -toN no
roughly estimate-Past extent Adv Top two-ton Gen
suna ga hituyoo des -u.
sand Sbj need CopPol-Pres

'Roughly estimated, we need two tons of sand.'

b. Sono syuNkaN zaO"to suna ga haiQ-te ki -ta.
that moment sand Sbj enter-Ger come-Past

That moment, sand rushed in with a gushing noise.'

(3-14) a. Kono heya no hoo ga guato akarui
this room Gen side Sbj considerably bright

des -u.

This room is considerably better-lighted.'

b. Sono syuNkaN watasi wa haNdoru o
that instant I Top steering wheel Obj

gu" to nigirisime-masi-ta.
grasp -Pol -Past

That instant, I tightly grasled on to the steering


(3-15) a. Syakai -too wa koNkai no seNkyo de mo
socialist-party Top this time Gen election in also

paO to si -mas-eN desi -ta.
be conspicuous-Pol -Neg CopPol-Past

The Socialist Party didn't do well in this election,


b. TotuzeN biQkuri-bako no huta ga paO" to hirai-ta.
suddenly surprise-box Gen lid Sbj open-Past

'Suddenly the lid of the Jack-in-the-box sprang widely


The above demonstrates that the more idiomatic, less sound-symbolic uses of

non-reduplicative mimetic forms call for the Class III accent pattern. Note that it is far

easier to define the meanings of the mimetic forms in (3-13a), (3-14a) and (3-15a) than

those in (3-13b), (3-14b) and (3-15b); the former are idiomatic and semantically more

restricted than the latter.

Even when we cannot perceive such a clear-cut difference as above, semantic

differences do exist Hearing a request Yuka o do'N-doN to humi-narasi-te kudasai

Please stamp on the floor like do'N-doN.' a speaker of a Kantoo dialect stamped on the

floor several times (more than twice) continuously. When she heard a request Yuka o doN

doN" to humi-narasi-te kudasai "Please stamp on the floor like doN doN,' she responded

by stamping on the floor twice. Thus, the meaning of reduplication in Class IV is

'plurality or continuity at a relatively rapid speed,' while the meaning of reduplication in

Class I is 'twice.'

We may say that we see a continuum of 'iconicity' in the continuum of Class I to

Class IV. Using specific forms, we will see how a word gets conventionalized and

becomes less iconic in the transition from Class I to Class III or Class IV. See the


(3-16) a. taiko o doN doN" to tataku (Class I)
drum Obj hit

'to hit a drum twice vigorously'

b. taiko o do'N-doN to tataku (Class IV with t)
drum Obj hit

'to hit a drum several times vigorously'

c. taiko o do'N-doN tataku (Class IV without W)
drum Obj hit

'to hit a drum several times vigorously'

d. ame ga do'N-doN huQ-te kuru (Class IV without 1g)
rain Sbj fall-Ger come

'the rain comes down vigorously in a large quantity'

e. *ame ga doN doN" to huQ-te kuru (Class I)
rain Sbj fall-Ger

In the above, (3-16a) alone has the meaning of 'twice,' which is directly related to

the reduplication of the mimetic element. Also, as we shall see later, the meaning of the

mimetic adverbs in (3-16a) above is tied to the physical aspect of the drum which here is

large and sturdy in construction. Note that, as we go down the above examples, the

meaning of 'twice' is first lost, followed by the meaning of 'hitting a flat thick non-metalic

surface.' One may suspect that (3-16d) contains the latter meaning, but the fact that

(3-16e) is ungrammatical suggests that this is not the case. Likewise, in the following

example, the use of the mimetic adverb is clearly idiomatic and restricted sound

symbolically. Here again, the Class I counterpart cannot be used.

(3-17) a. tenki ga doN-doN kawaru (Class IV without I1)
weather Sbj change

'the weather changes rapidly (from one weather to

b. *tenki ga doNdoN"to kawaru (Class Io)
weather Sbj change

There is a cooccurence restriction on one segmental feature. Diphthongs do not

appear in Class Ill in (3-1). This is apparently due to their strong iconicity.

So far, we have dealt with the accent of monosyllabic mimetic adverbs. We can

identify parallel phenomena in bisyllabic mimetic adverbs. The following are the major

types of bisyllabic mimetic adverbs. The first three clasees are parallel. Class I is

missing. Class V is an additional class.

Class I: (y-)CVCV(V)Q" to pataQ" to
patyaQ" to
petaaQ" to
petyaaQ" to
(y-)CV(Q)CV(V)N" to pataN" to
petyaN" to
poQtoN" to
poQtyaN" to
pataaN" to
pasyaaN" to
paQtaaN" to
poQtyaaN" to
expansion of above pataQ pataQ" to
pataN pataN" to

Class II: (y-)CVCV-CVCVQ" to pata-pataQ" to
pasya-pasyaQ" to
(y-)CVCV-CVCVN" to poto-potoN" to
poro-poroN" to
expansion of above pota-pota-potaQ" to

Class IV: (y-)CV'CV" to pa'ti" to
pi'ka" to
(y-)CV'CVN" to pataN" to
patiN" to
(y-)CV'CV-CVCV(" to) pa'ta-pataC' to)
pa'tya-patya(" to)
(y-)CV'CVN-CVCVN(" to) pataN-pataN(" to)
pokaN-pokaNC' to)
Class V: (y-)CV(Q)CV'ri(" to) paku'ri(" to)
pisya'ri(" to)
piQta'ri(" to)

The parallelism of the above classes to the classes of monosyllabic mimetic adverbs

is clear. Class I of bisyllabic mimetic adverbs is comparable to Class I of monosyllabic

mimetic adverbs in terms of the position of the accent nucleus, the restriction on the vowel

length, the obligatoriness of IQ, and the expandability of the forms. A class comparable to

Class In of monosyllabic mimetic adverbs is missing: the accentless pattern does not exist

for bisyllabic mimetic adverbs. However, Class V is unique to bisyllabic mimetic adverbs.

The contexts where each of the bisyllabic classes is used are also parallel to those

where a comparable class in monosyllabic mimetic adverbs is used. Class I of bisyllabic

mimetic adverbs is used in the most dramatic contexts, while Classes IV and V are used in

the least dramatic contexts.

3.2. The Iconic Continuum

Mimetic words in general form a continuum between those categories that are more

typically mimetic and those that are less so.

For example, mimetic nominal adjectives are less mimetic than mimetic adverbs.

Recall that reduplicative nominal adjectives, which constitute the majority of mimetic

nominal adjectives, lack any pitch fall, intonational or mimetic, just like Class Ill

monosyllabic mimetic adverbs. Besides, their morphological variations are limited, as we

saw in 2.3.2. Moreover, in 2.4.5, we saw that mimetic nominal adjectives are

semantically tied to more abstract qualities than mimetic D-verbs which are based on

mimetic adverbs.

In turn, bisyllabic mimetic roots are less mimetic than monosyllabic mimetic roots,

and the majority of mimetic nominal adjectives are based on bisyllabic mimetic roots.

Also, in, we saw that the derivational suffixes /-tuk-/ and /-mek-/ are suffixed

only to bisyllabic mimetic roots.

Also, p-initial mimetic words are more typically mimetic than other forms, as we

shall discuss in the last part of this section. The phenomenon of the accentual continuum

observed so far is part of this larger problem.

While the connection between the iconic continuum and accent has been overlooked

in the past, the idea of an iconic continuum is not entirely new. The traditional dichotomy

between giongo 'sound-mimicking words' and gitaio 'mode-mimicking words,' for

instance, reflects the two ends of this mimetic continuum. And such classifications have

been used to describe or account for certain aspects of the mimetic system. For instance,

Inaba (1972) and Yamaguchi (1973) use these concepts to explain some historical facts

about mimetic words.

According to Inaba, from Old Japanese through Moder Japanese, the

second-syllable vowels in bisyllabic iseego 'voice-mimicking words' are always /i, u/,

while the second-syllable vowels in bisyllabic gitaig 'mode-mimicking words' are

predominantly /a, o/. Also, those having /i, u/ as the second-syllable vowel, whether

'voice-mimicking' or 'mode-mimicking,' are rarely suffixable with derivational suffixes.

Within 'voice-mimicking words,' those having the same vowels in the first and the second

syllable are more likely to be suffixed with derivational suffixes than those having different

vowels in the first and the second syllable. Verbal suffixes -meku and -tuku are suffixed

to giseego as well as gitaigo, while nominal-adjectival suffixes -ka. -yak., and -raka are

suffixed only to gitaigo.

Yamaguchi, on the other hand, utilizes the dichotomy between gingo

'sound-mimicking words' (the more iconic) and gitaigo 'mode-mimicking words' (the less

iconic) to identify the pronunciation of some mimetic words in Middle-Old Japanese.

According to Yamaguchi, bisyllabic mimetic roots in Middle-Old Japanese in which

both Cland C2 are obstruents are classified into the following four types.

(3-19) a. C1 = [+voice]; C2 = [-voice]

b. C1 = [+voice]; C2 = [+voice]

c. C1 = [-voice]; C2 = [-voice]

d. C1 = [-voice]; C2 = [+voice]

Of these, Yamaguchi states that the first category largely consisted of giong and the

last largely of gitaigo; the second and the third category consisted of both giong and

gitaigo. In other words, at the stage of Middle-Old Japanese, the first category was most

iconic, while the last was least iconic. In agreement with these facts, the first category was

absent from the ordinary stratum of Old Japanese.

The situation has not changed much to the present. The last category above is still

the least conventional. A change has occurred at the other end, however, according to

Yamaguchi; the first category in contemporary Japanese contains a considerable number of


In addition, Yamaguchi notes that, while the weakening of intervocalic /h/ to /w/

occurred in gitaigo, as in sawa-sawa (
ordinary stratum, as in kowai (
process in such 'sound-mimicking words' as gohoN (sound of coughing).

At various places, I have repeated that p-initial mimetic adverbs are most typically

mimetic. The historical status of p-initial forms, however, is problematic. Yamaguchi

(1973) and Suzuki (1965) take the view that p-initial mimetic forms did not exist in

Japanese until the stage of Recent Japanese (1603-1867). Their claim is based upon the

survey of written documents including Jesuit grammars and writings in which Japanese

was transliterated using the Portugese writing conventions. The possibility remains,

however, that the absence of /p/ in written documents was caused by its extreme

expressiveness, unfit for the erudite nature of literature up to this time. Perhaps the

ensuing popularization of literature in the Edo period (1603-1867) allowed the inclusion of

the more expressive forms previously looked down upon for literary purposes.

It is a standard hypothesis in Japanese linguistics, first claimed by Kazutoshi Ueda,

that the ordinary stratum of Proto-Japanese probably had /p/ in initial as well as medial

position (Hashimoto 1950). The hypothesized disappearance of/p/ is explained as a case

of weakening. In view of this hypothesis and the concept of iconic continuum, it seems

more natural to hypothesize that /p/ in initial position existed in Proto-Japanese mimetic

words, too, as part of a symmetrical structure and that it was later replaced only in the

non-iconic part of the language as a result of a general phonetic change of weakening (p >


In any case, the synchronic status of p-initial mimetic forms is clear-cut

Qualitatively and quantitativly, they are central to the mimetic system in contemporary

Standard Japanese. Features that are sporadic in the mimetic system in general are absent

in p-mimetic forms. Voiced obstruents and nasals, which are rare as the second-syllable

consonant, are entirely absent from this position in p-initial mimetic words. Consequently,

the sequence /NC/, where /C/ is a voiced obstruent or a nasal, is lacking in p-initial mimetic

forms. The prosodic features of p-initial forms, as we have seen in 3.1, are also typically

mimetic. Irregular patterns, such as the one seen in the marginally mimetic form

/hono'-bono/ 'faintly light,' are never used with p-initial forms. Also, sequential voicing

of the initial obstruent of the second element, as in /hono'-hono/, is non-existent in p-initial

forms. The derivational suffix /-tuku/, which is typically used with mimetic words, is

suffixed to p-initial mimetic forms, but another derivational suffix /-meku/, which is

suffixable to non-mimetic nouns such as har 'spring' and aki 'autumn,' is not used with

p-initial mimetic forms.

3.3. Accent of Lone Vowels

One problem has not been settled for the unreduplicative forms of Class m

monosyllabic words. The uses of Class If (unaccented) unreduplicative adverbs with

long vowels cannot be accounted for simply as a case of conventionalization.

Unreduplicative mimetic adverbs with long vowels carry the Class IIl accent pattern

considerably more frequently than those with short vowels in otherwise identical contexts.

Apparent exceptions are those with /ee/ such as peeO" to (spitting) and purely

onomatopoetic forms such as mo(o" to (cow), miiN" to (locust), riiN" to (chime), and

nya')a" to (cat); these never occur with the Class Ill accent pattern.

The result of a pilot study showed that 56% of the long vowels in such forms were

unaccented (i.e., Class II), while 24% of the short vowels were unaccented. A follow-up

experiment using controlled pairs of sentences consistently agreed with this result.

The experiment was carried out with two speakers of Kantoo dialects. Neither of

them was a subject in the pilot study and neither of them had any linguistic background.

Twenty-two pairs of otherwise identical sentences using unreduplicative mimetic adverbs

with long and short vowels were prepared and presented to them in a randomized list.

They then were requested to read the list aloud twice at normal speed, and their

pronunciations were tape-recorded and transcribed later. Appendix D lists all the sentence

pairs used in this experiment. A sample pair is given below. Because the Japanese writing

system does not mark accent, the subjects could accentuate the test items in any way they


(3-20) a. Huta o toQ -tara arukooru no nioi ga puN to si-ta.
lid Obj take-Past alcohol Gen smell Sbj do-Past

'When I took off the lid, I smelled a strong smell of


b. Huta o toQ -tara arukooru no nioi ga puuN to si-ta.
lid Obj take-Past alcohol Gen smell Sbj do-Past

When I took off the lid, I smelled a very strong smell

of alcohol.'

The following results were obtained. (Incorrectly read mimetic forms were omitted

from the calculation.)

Table 3-2: Percentage of Accentless (= Class III) Forms

short vowels long vowels

Speaker 1 1st time 6/22=27.3% 11/22=50%
2nd time 7/22=31.8% 12/22=54.5%

Speaker 2 1st time 4/21=19% 8/19=42%
2nd time 4/21=19% 10/20=50%

The results show that, for both speakers, twice as many unreduplicative mimetic

adverbs with long vowels carry the Class m accent (i.e. are accentless) as those with short

vowels. In other words, piiN to for instance is more likely to be pronounced as [piiN to]

(as opposed to [piiN to]) than piNt is as [piN to]. It is not clear whether it is

phonologically caused or whether it is related to the semantic role of vowel lenghthening.

Interestingly, the accent patterns were not fixed lexically Nor did the two speakers show

any consistent agreement.

In any case, given the general correlation between the iconicity scale and accentual

patterns, when two forms are contrasted in this thesis, they will be drawn from the same

accent classes wherever possible.


4.1. Introduction

In this chapter, I will describe the sound symbolism of monosyllabic mimetic

adverbs. This will clarify most of the meanings that are associated with various

modifications of the roots in the sound-symbolic system in general. It will also clarify the

meanings of the individual consonants and vowels as they appear in monosyllabic roots.

Most of the examples for this analysis are p-initial mimetic adverbs. This is

because, as we have seen in 3.2, p-initial mimetic adverbs are most characteristically

mimetic. Forms starting with other consonants will supplement the analysis.

The types of mono-syllabic mimetic forms that are dealt with here are mimetic

adverbs of the shapes, C(y)VV, C(y)V(V)y)V(V)N, )V(V)Q and their reduplications. C

stands for /p, h, b, t, d, k ,g, s, z, n, m, y, w, (0)/. V stands for /i, e, a, o, u/. VV

stands for either long vowels /ii, ee, aa, oo, uu/ or diphthongs /ai, oi, ui/. /N/ stands for

the syllable-final nasal that constitutes a full mora. /Q/ stands for the first half of geminate

clusters,and it also constitutes a full mora.

The semantic domains and features involved in each constituent part of the forms

are briefly summarized below. This is provided as an outline for the more detailed

description that follows. For a fuller summary see 4.11. The numbers indicate the

sections where the features are dealt with. Note that the features in (4-1) are mostly listed

from the first element in the sequence to the last. In the analysis following, however, the

order of presentation is reversed in order to deal with the features in ascending order of


(4-1) Initial consonantal element: Type of movement. (4.7,
4.8, 4.9)

Point of ariticulation: Surface structure on
contact. (4.9)

Manner: Manner. (4.9)

Voicing: Quality of surface. (4.8)

Palatalization: Energy and childishness. (4.7)

Vowels: Size and shape of movement. (4.6)

Diphthongization: Circular movement. (4.5)

Vowel length: Length. (4.4)

Final consonants: Final aspect. (4.3)

Repetition: Phase. (4.2)

4.2. Repetition

4.2.1. Reduplicatives vs. Non-Reduplicatives

First, we will look at the contrast between a single base form, a root plus

modifications, and a complete reduplicative: e.g., /piN'/ vs. /piN-piN; /piiN'/ vs.


Generally speaking, the single form indicates that the action takes place once, while

the reduplicative form indicates repeated or continuous actions. See the following


(4-2) a. Teeburu-kurosu o piNto hiQpaQ-te hosi-ta.
table -cloth Obj pull -and dry-Past

I carefully pulled the table cloth into shape and hung it

on a line.'

b. Teeburu-kurosu o piN piN to hiQpaQ-te hosi-ta.
table -cloth Obj pull -and dry -Past

I carefully pulled the table cloth into shape several

times and hung it on a line.'

(4-3) a. Tukue no ue o paNto hoN de tatai-te kyoo wa
table Gen top Obj book by hit -and today Top

kore de osimai to iQ -ta.
this with end Quot say-Past

'Slapping the book down on the desk, she said, "So much

for today."'

b. Tukue no ue o paN paN to hoN de tatai-te
table Gen top Obj book with hit -and

kyoo wa kore de osimai to iQ -ta.
today Top this with end Quot say-Past

'Slapping the book down on the desk several times, she

said, "So much for today."'

(4-4) a. Okyaku ga de -ta to omoQ-tara Yasue-saN
customer Sbj exit-Past Quot think-when Yasue-Hon

tara sio o pDato doa no hoo e
Top salt Obj door Gen direction to
nage -tuke -ta no.
throw-attach-Past Atten

'As soon as she saw the [nasty] customers leave,

Yasue threw a dash of salt toward the door [to purify

the place].'

b. Okyaku ga de -ta to omoQ-tara Yasue-saN
customer Sbj exit-Past Quot think-when Yasue-Hon

tara sio o paO aO to doa no hoo e
Top salt Obj door Gen direction to

nage -tuke -ta no.
throw-attach-Past Atten

'As soon as she saw the [nasty] customers leave,

Yasue threw dashes of salt toward the door [to purify

the place].'

In the examples (4-2a), (4-3a) and (4-4a) above, the use of a single mimetic base

indicates that the action took place once. In (4-2b), (4-3b) and (4-4b), on the other hand,

the use of a reduplicative form indicates that the action took place more than once.

In the above examples, the simple base form and the reduplicative form can appear

contrastively in an identical context. In some other sentences, however, they cannot be

used in this fashion; only one or the other is allowed in a certain context. See the


(4-5) a. PiN to har -are -ta harigane ni zaimoku
stretch-Pass-Past wire to log

o sibari-tuke -te sita ni orosu no
Obj tie -attach-and below to lower Comp
des -u.

They tie the logs to the tightly stretched wire and

lower them down.'

b. *PiN piN to/*piN-piN to har-are-ta harigane ni

zaimoku o sibari-tuke-te sita ni orosu no des-u.

(4-6) a. Ano ko tui saikiN inaka kara p. to
that child merely recently country from

de -te ki -ta N des -u.
come out-and come-Past Comp CopPol-Pres

That girl arrived from the countryside just recently
looking unprepared. = She came right out of the sticks.'

b. *Ano ko tui saikiN inaka kara QOQoO to de-te

ki-ta N des-u.

In the above (4-5) and (4-6), the reduplicative forms are unacceptable, because the

context makes it clear that there is a single event. For (4-5), it is irrelevant whether the

wire was pulled once or twice; rather the focus is on the entire event of stretching the wire

tightly over a span. Likewise, in (4-6), the event of coming out of the country is the

focus. In the following example on the other hand, only the reduplicative forms are

permitted because the context specifies that the event takes place continuously or


(4-7) a.Tugi kara tugi e to yatugibaya ni po'N-poN to
next from next to Quot fast like arrows

kotoba ga toN-de ku -ru.
word Sbj fly-Ger come-Pres

'Words come out of his mouth one after another like

arrows. = He speaks a mile a minute.'

b. *Tugi kara tugi e to yatugibaya ni Nto kotoba

ga toN-de ku-ru.

The contrast of reduplicative forms vs. single forms is also attested in mimetic

adverbs with initial consonants other than /p/. See the following.

(4-8) a. Okaa-saN no te o guito hiQpaQ-ta.
mother-Hon Gen hand Obj pull -Past.

'[The child] forcefully pulled the mother's hand ne.'

b. Mizu -iro no atarasii musi -tori -ami o
water-color Gen new insect-catch-net Obj

kakae-ta otoko no ko ga epuron o tuke-ta
carry-Past male Gen child Sbj apron Obj wear-Past

mama no okaa -saN no te o i-g
as it is Gen mother-Hon Gen hand Obj

hiQpaQ-te ki -mas-u.
pull -and come-Pol-Pres

'A boy with a new blue butterfly net comes this way as

he (continuously and forcefully) drags his mother,who

is still with an apron on, by the hand.'

(4-9) a. Zyuusu o sutoroo de tyuOto suQ -ta.
juice Obj straw with suck-Past
'[The child] slued up the juice with the straw.'

b. Zyuusu o sutoroo de tyuO tyuO to suQ-ta.
juice Obj straw with suck-Past

'[The child] drank up the juice with the straw with

two slurps.'

(4-10) a. ToNto kata o tatai-ta.
shoulder Obj tap-Past

'I taped her on the shoulder onc.'

b. ToN toN to kata o tatai-ta.
shoulder Obj tap -Past

'I tapped her on the shoulder twice.'

The contrast is usually clear for Class I forms. For Class Im and Class IV,

however, the contrast is not always clear. Since they are generally less iconic and more

conventionalized, Class In adverbs and Class IV adverbs often fail to produce clear

semantic correspondences. That is, either the non-reduplicative form and the reduplicative

form are semantically quite different from each other, or one of the pair is non-existent

And even when the semantic resemblance is apparent, the contrast may not be immediately

tied to the number of occurrences. However, even in such cases, it is often possible to

trace the contrast to the contrast of repetition. For instance, in (4-11) below, the contrast

between the non-reduplicative adverb and the reduplicative adverb appears in the form of

the stative condition of an object and the behavioral characteristic of a person. We can say

that 'stativeness' is a kind of one-time occurrence and that 'behavior' involves repetition.

(4-11) a. Ano oziityaN wa sesuzi ga piNto si-te
that old man Top spine Sbj do-Ger

i -ru.

That old man's back is straight.'

b. Ano obaatyaN wa piN-piN si-te i -ru.
that old woman Top do-Ger Cont-Pres

That old woman is energetic.'

4.2.2. Partial Reduplication Class II

In this section, we consider the partially reduplicated forms such as pi-DiQ. pa-Pa

and their expansions in Class II. These indicate that there is a preparatory phase to the

event corresponding to the length of the initial repetition(s). They also indicate that the

action is carried out forcefully, rushingly, or vigorously. See the following.

(4-12) a. Hue o pi-piO to narasi-te kuruma o
whistle Obj blow -and car Obj


'[The policeman] hurriedly blew a whistle a couple of

times and stopped the cars.'

b. Kona o pa-paO to ire -ta.
flour Obj put in-Past

'I hastily scooped in some flour.'

c. Do-dooNto toozyoo.

'Enter rushingly and forcefully. = Introducing a

revolutionary breakthrough!' Class II

Another type of partial reduplication in the forms of pi'O-pi, paO'-pa and so on

means that the event takes place hurriedly, unhesitantly, and/or carelessly. See the

following for example.

(4-13) a. Kona o paO-pa to ire -ta.
flour Obj put in-Past

'I scooped in some flour unhesitantly.'

b. ToO-to to kie-use -ro.

'Disappear right away. = Be lost.'

c. SaO-sa to si-nasai.

'Do it right away. = Be quick about it.'

This type of reduplication is limited to a handful of forms with /Q/. No form with

/N/ can be reduplicated this way. *Pan-p for instance, does not exist. Another

phonotactic constraint on the forms belonging to this category comes from the general

constraint on gemminate clusters in Japanese which disallows geminate clusters of voiced

obstruents. As a result of this constraint, the initial consonants in these forms are

invariably voiceless obstruents. Thus, *zaQ-za and the like are normally precluded.

4.3. Final Elements /N. 0. 0/

In this section, the semantic roles of the final elements are described. The

phonological contrast between /N, Q, 0/ is correlated with a semantic contrast between the

manners of ending the event or movement

/N/ indicates that the direction of the motion or the quality of the sound changes

toward the end. That is, it indicates that there is a reaction or reverberation to the initial

movement It may indicate that the object is flexible or elastic and is capable of such

reactions or reverberations. /Q/ on the other hand indicates that the object is inflexible and

that the action is carried out vigorously in one direction or without changing its nature.

Note that in the following examples, the elasticity of the object or the lingering effect of

the action is indicated by the context Therefore, only /N/ is permitted.

(4-14) a. Gitaa no ito o piN to hazii-ta.
guitar Gen string Obj pluck-Past

'I plucked the string of the guitar.'

b. *Gitaa no ito o piO to hazii-ta.

(4-15) a. Nuno o DN to haQ -ta.
cloth Obj stretch-Past

'I stretched a cloth across.'

b. *Nuno o piQ tohaQ-ta.

(4-16) a. Gomu -huuseN ga aN to ware -ta.
rubber-balloon Sbj break-Past

The rubber balloon blew up.'

b. *Gomu-huuseN ga paO toware-ta.

(4-17) a. Koosui ga puN to nioQ-ta.
perfume Sbj smell-Past

The perfume smelled strongly with a lingering effect.

= I got a wiff of her perfume.'

b. *Koosui ga uO to nioQ-ta.

Contrast the above with the following examples in which /Q/ rather than /N/ is the


(4-18) a. Go-isi o piO to hazii-ta.
S-stone Obj snap-Past
'I snapped go-stones (=small round and flat stone

pieces used in the chess-like game of g ) with my


b. *Go-isi o piN to hazii-ta.

(4-19) a. Nuno o piQ to hiki-sai-ta.
cloth Obj pull-rip-Past

'He quickly ripped the cloth into narrow strips.'

b. *Nuno o piN t hiki-sai-ta.

(4-20) a. Iro ga paQto kawaQ-ta.
color Sbj change-Past

The color changed all of a sudden.'

b. *Iro ga aN to kawaQ-ta.

(4-21) a. AkaNboo ga puO to onara o si-ta.
baby Sbj wind Obj do-Past

The baby passed wind.'

b. *Akanboo ga puN to onara o si-ta.

In the above, the objects or movements are not elastic or lingering; the events are

started and terminated abruptly. The contrast between (4-17a) and (4-21a) is interesting.

When a lingering smell or fragrance is the main concern as in (4-17a), /N/ appears; in a

context as in (4-21a) where one chooses to pay attention to the abruptness of the event

instead of the lingering effect, /Q/ appears.

Sometimes, forms with /N/ and /Q/ appear contrastively in otherwise identical

contexts. The contrast is an interesting one in the following.

(4-22) a. EndaN o RoN to kotowaQ-ta N
marriage proposal Obj refuse -Past Comp


I hear that she turned down the offer of marriage as

if she was bouncing a ball. = I heard that she flatly

turned down the offer of marriage.'

b. EndaN o pO to kotowaQ-ta N
marriage proposal Obj refuse -Past Comp


'I heard that she suddenly turned down the offer of


In (4-22a) above, the critical issue is the boldness, daredevilness, or audacity of the

person who turns down an offer which is considered to be very attractive. This comes

from the association of /poN/ with the action of throwing a ball and similar objects. In

(4-22b), by contrast, the issue is the abruptness of the event. In other words, /poQ/ in

(4-22b) indicates that there was no indication of the move in advance.

Forms with other initial consonants also support the above claim. In the following

pair, both /kiiN/ in (4-23a) and /kiiQ/ in (4-23b) are translated with the verb 'screech.'

However, /kiiN/ is used to describe a long resonating sound, while /kiiQ/ is used to

describe a sound made when stopping a car abruptly.

(4-23) a. BeiguN no bakugekiki ga kiiN toiu oto
US Army Gen fighter plane Sbj say sound

o hibik-ase -te ik-u naka...
Obj sound-Caus-Ger go-Pres while

'As a fighter plane of the US Army screeched over [my


b. Kuruma wa kiiO to oto o tate-te
car Top sound Obj make-and

stop -Pol -Past

The car screeched to a sudden stop.'

Other forms with /N/ with the meaning of 'lingering effect' or 'elasticity' include the

following. The right-hand column identifies the examples of events or objects that the

adverbs describe.

(4-24) a. koN the resonating sound of a bell

b. kaN the resonating sound of a bell

c. toN hitting a drum; tapping on the

d. tiN sound produced by hitting a small
bell; hitting a triangle (=musical

e. baN the sound of a gun

f. boN striking the key of the piano

g. buN-buN humming of a bee

h. tyuN chirping of a small bird such as a

The meaning of 'the involvement of the nasal cavity' is attested in the following forms.

(4-25) a. kuN sniffing at; whining

b. tuN stimulative odor

c. syuN blowing the nose

d. huN puffing air through the nose

Note that all the above forms and /puN/ in (4-17a) share a common vowel /u/. As

described in 4.6 later in this chapter, the vowel /u/ indicates a small round and/or

protruded opening, and it seems that the combination of this meaning and the meaning of

'resonation' or 'redirection' of /N/ produces the specific meaning of 'nasality' in these

examples. At the same time, it is not just the vowel /u/ that has the meaning of 'nasality.'

In (4-26) to follow, (4-26a) with /N/ indicates the involvement of the nasal cavity, but

(4-26b) with /Q/ indicates the involvement of the oral cavity instead.

(4-26) a. HuN te yaQ-te.
do -Ger

'Blow air through the nose.'

b. HuO te yaQ-te.
do -Ger

'Blow air through the mouth.'

Like (4-26b) above, forms with /Q/ commonly relate to abrupt, short, and forceful

movements; they do not indicate 'elasticity' or 'lingering effect' at all. See below.

(4-27) a. kuQ turning abruptly; bending a pipe

b. syuQ abruptly ejecting chemical from a

c. huQ blowing out a candle

d. tyuQ sucking

e. kaQ flying into a rage

f. baQ rushing out; throwing something out

g. boQ flaring up abruptly
The number of the forms without either /N/ or /Q/ is limited in the class of forms

with initial /p/. However, where there is a contrast, we observe that the forms without /N/

or /Q/ indicate that the movements are carried out in one direction without much vigor or

emphasis. See (4-28) below. It will be noticed that the form without /Q/ in (4-28b)

describes a less vigorous sound.

(4-28) a. Hue o piiO piiO to huk -u.
whistle Obj blow-Pres

'She blows a shrill whistle.'

b. Hiyoko ga ii-pii nak-u.
chick Sbj cry-Pres

The chick eeps.'

The same relationship as above holds true with the following pair.

(4-29) a. PuiO to yoko o mui-ta.
side Obj face-Past

'She turned her head away disgustedly (and


b. Puito yoko o mui-ta.
side Obj face-Past

'She turned her head away crossly (but calmly).'

When we expand our scope beyond p-initial forms, we find the following examples

which support the claim about the meaning of final /0/.

(4-30) a. Biiru o guiO to nodo ni nagasi -kom -u.
beer Obj throat to let flow-push in-Pres

'She gulps down the beer (dashingly).'

b. Biiru o gui to nodo ni nagasi -kom -u.
beer Obj throat to let flow-push in-Pres

'She drinks down the beer forcefuly but calmly).'

(4-31) a. HaaO to tameiki o tui -ta.
sigh Obj attach-Past

'He drew a dee sigh.'

b. Haa to tameiki o tui -ta.
sigh Obj attach-Past

'He drew a sigh.'

4.4. Vowel Length

The short vowels, /i, e, a, o, u/, indicate that the event is completed instantaneously

and/or that the distance involved is short. The long vowels, /ii, ee, aa, oo, uu/, are

extenders; they indicate that the action takes longer spatially and/or temporally and that the

action is more strenuously carried out. See the following.

(4-32) a. PiNto hiQpaQ-te.

'Stretch it tight.'

b. PiiN to hiQpaQ-te.

'Stretch it very tight.'

(4-33) a. Tanuki no onaka ga paN to haretu si-te
badger Gen stomach Sbj explosion do-Ger

simaQ-ta no des -u.
end up-Past Comp CopPol-Pres

The badger's stomach blew open with a (short) bang.'

b. Tanuki no onaka ga paaN to haretu si-te
badger Gen stomach Sbj explosion do-Ger

simaQ-ta no des -u.
end up-Past Comp CopPol-Pres

The badger's stomach blew open with a (long) bang.'

(4-34) a. Reizooko o ake -tara sakana no nioi ga
refrigerator Obj open-when fish Gen smell Sbj

puN to si-ta.

'When I opened the door of the refrigerator, I was

hit with the powerful smell of fish.

b. Reizooko o ake -tara sakana no nioi ga
refrigerator Obj open-when fish Gen smell Sbj

puuN to si-ta.

"When I opened the door of the refrigerator, I was

hit with the very powerful smell of fish, [so powerful

that I could have smelled it from a distancel'

(4-35) a. PiO to momeN no nuno o hiki-sai -ta.
cotton Gen cloth Obj pull-tear-Past

'She tore a short strip of cotton cloth.'

b. PiiO to momeN no nuno o hiki-sai -ta.
cotton Gen cloth Obj pull-tear-Past

'She tore a long strip of cotton cloth.'

(4-36) a. Tukusi wa neQtoo ni sake syoo-syoo
horsetail Top hot water in sake a small amount

o tarasi-ta naka de yude-ru to
Obj drop -Past inside in boil-Pres when

iro ga aOQ to utukusiku nari -mas-u.
color Sbj beautiful become-Pol-Pres

'As for the horsetails, boil them in hot water with a

dash of ske, and they will turn a vivid color all at

once in a second.'

b. Tukusi wa neQtoo ni sake syoo-syoo
horsetail Top hot water in sake a small amount

o tarasi-ta naka de yude-ru to
Obj drop -Past inside in boil-Pres when

iro ga paaO to utukusiku nari -mas-u.
color Sbj beautiful become-Pol-Pres

'As for the horsetails, boil them in hot water with a

dash of sjak, and they will turn a vivid color rapidly

(but not all at once.)

As we have seen above, many instances of monosyllabic mimetic adverbs can

employ both the short and the long forms. However, there are cases where the contexts

disallow the use of one or the other. See the following.

(4-37) a. Watasi no kehai o kaNzi-ru to 2aO to
I Gen indication Obj feel-Pres when

kao o age-ta.
face Obj raise-Past

'When she felt me approaching, she raised her head


b. *Watasi no kehai o kaNzi-ru to paaO to kao o


(4-38) a. Iti-zikaN mo si-nai uti ni paaO to
one-hour even pass-Neg while

naku naQ -tyaQ -ta.
absent become-complete-Past

'Before even an hour passed, it was all gone (fasL).

= In less than an hour, it was all gone.'

b. *Iti-zikaN mo si-nai uti ni DaO to naku


In (4-37) above, (4-37b) is not permitted, at least not in the sense of (4-37a),

because the action of raising one's head normally is not a diffused act. It is only

marginally acceptable in the sense that there were a large number of people who raised their

heads almost but not exactly at once. In (4-38), on the other hand, the short form is not

permitted because the context clarifies that the event took place gradually over a period of

about one hour, not at once.

Forms with initial consonants other than /p/ also exhibit the contrast of

temporal/spatial distribution. See the following.

(4-39) a. Kaa-saN-gitune wa sono te ni haO to
mother -fox Top that hand to

iki o huki-kake -te yari-masi-ta.
breath Obj blow-cover-Ger give-Pol-Past

The mother fox warmed the [baby fox's] paws with
her (short) breath.'

b. Kaa-saN-gitune wa sono te ni haaO to
mother -fox Top that hand to

iki o huki-kake -te yari-masi-ta.
breath Obj blow-cover-Ger give-Pol-Past

The mother fox warmed the [baby fox's] paws with
her (lng) breath.'

Full Text
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd