Akita 2008 SS subsystems

全文

(1)Akita, Kimi. 2008. Two cognitive subsystems of sound symbolism in Japanese. KLS 28: Proceedings of the Thirty-Second Annual Meeting of Kansai Linguistic Society, 23-33.. Two Cognitive Subsystems of Sound Symbolism in Japanese* Kimi Akita Kobe University and JSPS. 1. Introduction This paper proposes a dichotomy of sound-symbolic elements in Japanese, which is very suggestive of Talmy’s (1988, 2000) two subsystems of language as a cognitive system and highly useful for identification of grammatically relevant aspects of sound-symbolic meaning, definition of the categorial status of sound-symbolic words (or mimetics), and consideration of sound symbolism in a crosslinguistic context. This proposal, with its attention to some general issues like aspectuality and categorial properties, goes beyond the traditional mimetic studies, which are characterized with their excessive concentration on dictionary-oriented descriptions, and establishes a foundation of sound-symbolic grammar.. 2. Dictionary-Oriented Descriptions in Previous Studies Previous studies on Japanese mimetics show an extreme trend toward dictionary-oriented descriptions of their phonology, morphology, and phonosemantics. This characteristic inclination is readily noticeable in a stock of articles and monographs (to name a few, Izumi 1976; Kindaichi 1978; Miyaji 1978; Tamori 1983; Nishio 1988; Hamano 1998) as well as numerous dictionaries focusing on this word class (e.g., Amanuma, ed. 1973; Asano, ed. 1978; Chang 1990; Kakehi et al. 1996; Tobita and Asada, eds. 2002; Yamaguchi 2003). As Tsujimura (2004, 2005), Tsujimura and Deguchi (to appear), and Kageyama (2006, to appear) state, we can ascribe this tendency to preceding researchers’ strong interest in the “peculiar ” properties of the form and meaning of mimetics and their enthusiasm for helping learners of Japanese as a second language acquire this “very Japanese” vocabulary. One of the major achievements the previous investigations have made is a list of various “ sound-symbolic elements ” of mimetics, which can be defined as morphological/phonological partials or processes that iconically or motivatedly convey some meaning. Here I summarize representatives of such elements with some examples, mainly based on Hamano (1998): (1) Sound-symbolic elements in Japanese: a. Reduplication of a CVCV-root: pokipoki (repetitive crunches), iraira (continuous irritation), uyouyo (crawling) b. Stem-final elements:1 saQ(-to) (a quick movement), paN(-to) (one shot), kirari(-to) (one twinkle) c. Vowel length: baN(-to) (one short bang) vs. baaN(-to) (a prolonged bang) d. Intensifier {C}: gusari(-to) (thuck) vs. guQsari (deep thuck); huwari(-to) (softness) vs. huNwari (fluffiness) e. Accent: nu^runuru (dynamic: adverb/verb) vs. nurunuru (static: adjective/noun) f. Palatalization: nurunuru (sliminess) vs. nyurunyuru (disgusting sliminess) g. Voicing of obstruents: koroQ(-to) (roll of a light object) vs. goroQ(-to) (roll of a heavy object) h. Symbolism of C1 of CVCV-based mimetics (Hamano 1998: 172): p taut surface light; small; fine b taut surface heavy; large; coarse t lack of surface tension; subduedness light; small; fine d lack of surface tension; subduedness heavy; large; coarse.

(2) k hard surface light; small; fine g hard surface heavy; large; coarse s non-viscous body; quickness light; small; fine z non-viscous body; quickness heavy; large; coarse h weakness; softness; unreliability; indeterminateness m murkiness n viscosity; stickiness; sliminess; sluggishness y leisurely motion; swinging motion; unreliable motion w human noise; emotional upheaval i. Vowel symbolism (Hamano 1998: 100, 172-173): protrusion line/tenseness small large /i/ + /u/ + + /o/ /a/ + /e/ Vulgarity. To give a brief description of each element, reduplicative forms of two-mora roots have a repetitive, continuous, or plural meaning, as in (1a) (see Hurch, ed. 2005 for crosslinguistic explorations in reduplicative semantics). (1b) illustrates stem-final elements, /-Q/, /-N/, and /-ri/, all of which symbolize some kind of ending of a telic event. As (1c) shows, the contrast of a short and a long vowel imitates the temporal contrast, short and long, of the referent event. Intensified forms illustrated in (1d) are often called “intensified adverbs, ” especially in the context of “mora augmentation” (see Davis and Ueda 2002). As Kuroda (1979: 205-206) notes, despite its name, an intensified adverb does not necessarily have its “nonintensified” counterpart (e.g., *ukari vs. uQkari). As (1e) exemplifies, regarding reduplicatives of a two-mora root, accentuation counts as a distinctive feature in Tokyo dialect. As Kageyama (to appear) suggests, the pitch fall (i.e., dynamic change in frequency) contained in mimetic adverbs and verbs can be analyzed as mimicking their dynamic semantics. Palatalization, which is one of the most frequently discussed sound-symbolic processes (Mester and Itô 1989; Schourup and Tamori 1992; Hamano 1994, 1998), adds an unfavorable connotation, as in (1f), although, as in the intensification in (1d), a minimal semantic contrast is not always detected between a palatalized-nonpalatalized pair (e.g., horohoro (dropping) vs. hyorohyoro (faintness)). The contrast in (1g), as well as those in (1h), shows that voicing of obstruents gives rise to semantic shifts in weight, size, or granularity. (1h) lists consonant symbolisms at the onset of the first syllable of two-mora roots. Finally, (1i) summarizes vowel symbolisms, which Hamano claims, unlike consonant symbolisms, do not employ positional distinction—being coherent at any position. Indeed these previous enthusiasms may have been successful and sufficient for their goal (i.e., dictionary composition). However, having a broader perspective to the mimetic study as a whole, there are some problems. First, the extreme inclination toward descriptive studies has caused the absence of theoretical investigation into the mimetic grammar or that of analyses of the internal organization of the sound-symbolic system from a grammatical perspective, beyond a simple enumeration of phonosemantic correspondences. This seems to have set mimetics apart from general linguistic areas. Second, the absorption in the study of their own language has given Japanese linguists a closed perspective that stays away from crosslinguistic or typological explorations. This paper supplements both lacunae by positing a two-way distinction of the sound-symbolic elements listed in (1).. 3. Prosodic vs. Segmental Sound Symbolism In this section, I propose and define two kinds of sound symbolism by means of the segment/prosody distinction.2 As I mentioned in the previous section, Japanese linguists have described a variety of sound-symbolic elements in a simple, disordered manner. A closer look, however, allows us to find that there are two distinct types in those elements. The first one is what I call “prosodic (sound) symbolism. ” (1a-e) are prosodic-symbolic elements, which constitute the prosodic structure or contour of a word. Reduplication, stem-final element attachment (or suffixation), vowel lengthening, and intensification all augment one or.

(3) two (or more) morae and determine the prosodic structure of a word; accentuation also fits the label “prosodic. ” Prosodic-symbolic elements except stem-final elements spread out of the mimetic stratum and can function as an interface with other strata (see note 9). The other kind is “segmental (sound) symbolism,” which is concerned with phonemes or phonetic features of mimetic roots (i.e., CV or CVCV): (1f-i). The sound-symbolic processes of palatalization and voicing as well as both consonant and vowel symbolism add neither morae nor a pitch fall, but rather constitute the basic content of a mimetic to fill in the skeletal structures that prosodic-symbolic elements form.3 Take the mimetic kyorokyoro ‘looking around restlessly’ for example. In terms of the present dichotomy, this mimetic can be analyzed into CVCV-reduplication as a prosodic-symbolic element and the following segmental-symbolic elements of the mimetic root kyoro: C1 /k/ (or [+velar, +stop, +oral, -voiced]), C2 /r/, /o/ at two Vs, and palatalization at C1. This distinction seems compatible with the cognitive subsystems of language Talmy (1988, 2000) discusses. According to him, “closed-class” elements, which constitute what he calls “the grammatical subsystem,” depict the structure of “cognitive representation” whereas “open-class ” elements, which constitute what he calls “the lexical subsystem,” depict its content. With their depiction properties in mind, we can reasonably hypothesize that prosodic and segmental symbolisms are a member of the grammatical and the lexical subsystems, respectively. 4 In the rest of this paper, I will demonstrate how much efficiency the present dichotomy gives to aspects of mimetic research, which simultaneously suggests its compatibility with Talmy’s cognitive-subsystem model. In Section 4, the dichotomy will make it possible to discuss what part of mimetic semantics is grammatically relevant. In Section 5, the distinction will enable us to determine how mimetics should be located in the Japanese lexicon. Finally in Section 6, I will extend the idea developed in Section 5 to a crosslinguistic investigation into sound symbolism.. 4. Grammatical Relevance of Sound-Symbolic Meaning In this section, I will consider grammatically relevant aspects of meaning of mimetics in terms of the dichotomy of sound symbolism proposed in the previous section. First, syntactic relevance will be examined for each of the two kinds of sound symbolism with special attention to the aspectuality of sentences containing a mimetic (see Pinker 1989). Second, morphological relevance will be examined through observation of semantic constraints on mimetic verb formations. 4.1. Syntactic relevance: The case of aspectualit y Aspectual properties of some sound-symbolic elements of Japanese mimetics have been pointed out frequently but partially in the literature (among others, Izumi 1976; Kita 1997; Lu 2006: 96; Kageyama, to appear; for similar phenomena in other languages see Alpher 1994: 163; Bybee et al. 1994: 166-174; Nuckolls 1996: 58-59). It has hardly been scrutinized exactly what kinds of aspectual properties each element has. Reduplication, which is an instance of prosodic symbolism, seems to have attracted the strongest attention. Tsujimura and Deguchi (to appear) present an exceptional consideration which has succeeded in testing aspectual reduplication focusing on the telicity of sentences that contain a reduplicative mimetic. They point out that reduplicative mimetics of a two-mora root disambiguate a sentence as atelic, as in (2): (2) a. Mizu-o go-hun-{kan/de} non-da. (ambiguous in telicity) water-ACC 5-min.-{for/in} drink-PST ‘[I] drank (the) water {for/in} five minutes.’ b. Mizu-o go-hun-{kan/*?de} gokugoku non-da. (CVCV-red  atelic) water-ACC 5-min.-{for/in} MIM drink-PST ‘[I] drank (the) water vigorously {for/*?in} five minutes.’ (Tsujimura and Deguchi, to appear) Parallel tests to sentences containing mimetics of various kinds of prosodic symbolism allow us to notice that what Tsujimura and Deguchi observe is the tip of the iceberg. In (3), in contrast.

(4) with (2), mimetics containing a stem-final element, /-Q/, /-N/, or /-ri/, specify the sentence as telic. (I suppose that specific kinds of ending are segmental-symbolically determined.) (3) a. Mizu-o iti-byoo-{kan/de} non-da. (ambiguous in telicity) water-ACC 1-sec.-{for/in} drink-PST ‘[I] drank (the) water {for/in} one minute.’ b. Mizu-o iti-byoo-{*kan/de} goku{Q/N/ri}-to non-da. (stem-final elements  telic) water-ACC 1-sec.-{for/in} MIM -QUOT drink-PST ‘[I] drank (the) water in one gulp {*for/in} one minute.’ Moreover, different temporal adverbials can test the durativity of mimetics. In (4), a reduplicative mimetic and a mimetic with /-ri/ ending bring a durative and a punctual meaning, respectively, to their sentences. (4) a. Ken-ga Mai-o {issyun/sibaraku-no aida} mi-ta. (ambiguous in durativity) Ken-NOM Mai-ACC {for.an.instant/for.a.while} look-PST ‘Ken looked at Mai for {an instant/a while}.’ b. Ken-ga Mai-o {*issyun/sibaraku-no aida} ziroziro mi-ta. (CVCV-red  durative) Ken-NOM Mai-ACC {for.an.instant/for.a.while} MIM look-PST ‘Ken stared at Mai for {*an instant/a while}.’ (Akita 2007c) c. Ken-ga Mai-o {issyun/*sibaraku-no aida} zirori-to mi-ta. (/-ri/  punctual) Ken-NOM Mai-ACC {for.an.instant/for.a.while} MIM-QUOT look-PST ‘Ken glared at Mai for {an instant/*a while}.’ Vowel length, which is another example of prosodic symbolism, exhibits a contrast in both telicity and durativity, although the contrast is incomplete in telicity.5 (5) a. Mizu-o iti-byoo-{*kan/de} guQ-to non-da. (short V  telic; cf. (2a)) water-ACC 1-sec.-{for/in} MIM-QUOT drink-PST ‘[I] drank (the) water in one gulp {*for/in} one second.’ b. Mizu-o san-byoo-{kan/de} guuQ-to non-da. (long V  ambiguous in telicity; cf. (2a)) water-ACC 3-sec.-{for/in} MIM-QUOT drink-PST ‘[I] drank (the) water in a long gulp {for/in} three seconds.’ c. Mai-ga {issyun/*sibaraku-no aida} tuna-o guQ-to hii-ta. (short V  punctual) Mai-NOM {for.an.instant/for.a.while} rope-ACC MIM-QUOT draw-PST ‘Mai jerked vigorously on the rope for {an instant/*a while}.’ d. Mai-ga {??issyun/sibaraku-no aida} tuna-o guuQ-to hii-ta. (long V  durative) Mai-NOM {for.an.instant/for.a.while} rope-ACC MIM-QUOT draw-PST ‘Mai jerked vigorously and steadily on the rope for {??an instant/a while}.’ The table below is a summary of the observation here with the information about the eventuality type(s) with which a mimetic with the property is compatible added: Table 1. Prosodic symbolism and aspectuality Telicity Durativity Possible eventuality types CVCV-reduplication (1a) atelic (2b) durative (4b) activity, state Stem-final elements (1b) telic (3b) punctual (4c) achievement(, semelfactivity)6 Vowel lengthening (1c) ambiguous (5b) durative (5d) activity, accomplishment, state There are some apparent counterexamples to these phonosemantic correspondences. For example, the mimetic ziQ(-to) contains the stem-final element /-Q/, and the above generalization expects it to denote a telic, punctual eventuality. However, this is not the case..

(5) (6) a. Ken-ga Mai-o go-byoo-{kan/*de} ziQ-to mi-ta. (/-Q/  atelic) Ken-NOM Mai-ACC 5-sec.-{for/in} MIM-QUOT look-PST ‘Ken stared at Mai {for/*in} five seconds.’ b. Ken-ga Mai-o {?issyun/sibaraku-no aida} ziQ-to mi-ta. (/-Q/  durative; cf. (4a)) Ken-NOM Mai-ACC {for.an.instant/for.a.while} MIM-QUOT look-PST ‘Ken stared at Mai for {?an instant/a while}.’ This seeming contradiction can be settled by assuming two levels of meaning for one mimetic: namely, sound-symbolic meaning and lexical meaning (or word meaning) (see Tamori and Schourup 1999: 8). Following the above generalization, the sound-symbolic, precisely prosodic-symbolic, meaning of ziQ(-to) is supposed to be telic and punctual. However, its lexical meaning ‘staying patiently’ can betray it and “coerce” the aspectuality as atelic and durative. In this case, we can solve the contradiction problem by analyzing the stem-final element /-Q/ as conveying an extended meaning like ‘intensely’, which does not pertain to a particular aspectual feature (see Akita 2007b for further discussion). In this respect, we could not reject the opinion that it is the segmental symbolism of /z/ and /i/ that at least partially determines the eventuality type ziQ(-to) depicts as activity. Nevertheless, ziN(-to), another mimetic that is based on the same root, does conform to Table 1, although only in durativity, as in (7).7 (7) a. Atama-ga {issyun/sibaraku-no aida} itan-da. (ambiguous in durativity) head-NOM {for.an.instant/for.a.while} hurt-PST ‘[My] head ached for {an instant/a while}.’ b. Atama-ga {issyun/*sibaraku-no aida} ziN-to itan-da. (/-N/  punctual) head-NOM {for.an.instant/for.a.while} MIM-QUOT hurt-PST ‘[My] head stung for {an instant/*a while}.’ Therefore, we must conclude that the aspectual contribution of segmental symbolism, if any, is not as systematic as that of prosodic symbolism. This conclusion gets further support from the uncontroversial fact that it is common that one mimetic root can appear in several prosodic templates (e.g., do^kidoki, dokiQ, dokiN, dokiri, doQkiri, etc. for the root doki). Thus, at the level of sound-symbolic meaning, aspectual properties of mimetics crucially depend on or link with prosodic, not segmental, symbolism. Aspectual contributions are detected with the rest of prosodic-symbolic elements as well. First, as Tamori (1983), Yamanashi (2000: 244-246), and Lu (2006: 95-96) remark, the CVCCVri form (i.e., the form of “intensified adverbs”) has a resultative state meaning. Accordingly, it is durative but ambiguous in telicity, and thus compatible with accomplishment (e.g., Pan-o huNwari yaita ‘[I] baked bread fluffy’) and state predicates (e.g., Pan-ga huNwari-siteiru ‘[This] bread is fluffy’). Second, since, as exemplified in (1e), the accentual differentiation has the two-mora reduplicative form as its prerequisite, its aspectual properties follow those of two-mora reduplicatives given in Table 1. This prosodic feature is in charge of a further specification of eventuality types: when accented, the reduplicatives are ambiguous between activity (e.g., Kumo-ga hu^wahuwa ukandeiru ‘A cloud is floating in a fluffy manner’) and state (e.g., Kono wata-wa hu^wahuwa-suru ‘This cotton feels fluffy’), whereas, when unaccented, they are used as a part of a state predicate (denoting attributes of their referents; e.g., Kono wata-wa huwahuwa-da ‘This cotton is fluffy’) or of a resultative phrase (e.g., Pan-o huwahuwa-ni yaita ‘[I] baked bread fluffy’) (Toda 1942; Kita 1997; Kageyama, to appear). Then, why is only prosodic symbolism relevant to aspectuality? The most likely answer is that it is because prosodic-symbolic elements define the temporal contours of their referent eventualities through mimicking them by means of their prosodic contours.8 This is not possi-.

(6) ble for segmental-symbolic elements, for they only determine the essential semantic types (e.g., magnitude, shape, color) of subtle nuances of mimetic roots. The structure- and content-representing natures of prosodic and segmental symbolisms, respectively, here imply their memberships in the grammatical and lexical subsystems, respectively. 4.2. Morphological relevance: The case of mimetic verb formations In this subsection, I further discuss the aspectual properties of prosodic symbolism observed in the previous subsection in terms of two instantiations of verb formation of mimetics. It is widely known that, in Japanese, a number of mimetics can form a verb in combination with the verb suru ‘do’ (see Akita, to appear, a). What is further interesting, subsets of such mimetics can participate in some other different verb formation processes as well. First, some mimetics can form a (semelfactive) verb by the attachment of the verb kuru ‘come’. What is significant here is that candidates for the attachment are in principle limited in form: all such mimetics have a stem-final element. Note that two mimetics having the root ira (i.e., iraQ(-to) and iraira) show different verb formation possibilities, which suggest that segmental symbolism is not the key in this morphological process. (8) guQ-to-kuru ‘be moved’, kaQ-to-kuru ‘get upset’, muQ-to-kuru ‘get disgruntled’, piN-to-kuru ‘be inspired’, ziN-to-kuru ‘be deeply moved’, iraQ-to-kuru ‘be irritated’, katiN-to-kuru ‘be offended’, siQkuri-kuru ‘have a nice fit’, ??tikutiku-kuru ‘feel prickled’, *iraira-kuru ‘be irritated’, *moyamoya-kuru ‘feel gloomy’ This formal restriction can be attributed to the aspectual meaning of the mimetic verb construction (i.e., punctual emotional experience), which is only compatible with stem-final elements among prosodic-symbolic elements. Likewise, verbs in the [mimetic root + suffix (-tuku/-meku)] form are morphophonologically or phonosemantically restricted: as Tamori (1993), Hamano (1998: 56-57), and Tsujimura and Deguchi (to appear) describe, they are exclusively made from the roots of two-mora reduplicatives. (9) a. gira-tuku (< giragira) ‘glisten’, kasa-tuku (< kasakasa) ‘(or skin) become dry and rough’, muka-tuku (< mukamuka) ‘get disgruntled’, zawa-tuku (< zawazawa) ‘hum’, *geso-tuku (< geQsori) ‘become skinny and faint’, *mu-tuku (< muQ(-to)) ‘get disgruntled’ b. kira-meku (< kirakira) ‘glitter’, tuya-meku (< tuyatuya) ‘become fascinatingly beauty ful’, yura-meku (< yurayura) ‘swing’, zawa-meku (< zawazawa) ‘hum’, *mata-meku (< maQtari) ‘sweeten’, *syu-meku (< syuQ(-to)) ‘become slim’ This restriction is again explainable from the perspective of the semantics of the verb forms. Since both mimetic verb forms denote an activity or temporary change of state, two-mora reduplicatives, which link with atelic, durative aspectuality, fit them best (see Tamori 1993; Hamano 1998: 56-57 for further semantic conditions for these verb forms). These two cases of morphological relevance of sound-symbolic meaning reinforce the idea suggested above that prosodic symbolism has basic aspectual motivation while segmental symbolism does not.. 5. The Categorization of Mimetics In this section, I focus on another property of the grammatical subsystem of language—closedness—to discuss that prosodic symbolism also plays a significant role in defining the mimetic category in the Japanese lexicon. “ The categorization problem ” of mimetics is investigated in some detail by Tamori and Schourup (1999), who have reached twelve mimetic-specific morphophonological characteristics, such as “freedom from rendaku (sequential voicing),” “freedom from the nasalization of intervocalic /g/,” “abundance of /p/-initial words,” and “suffixation of /-Q/, /-N/, and /-ri/” (pp. 210-211). Importantly, however, words like dokidoki and burabura, which show none of the mimetic-specific properties evidently (not sure about the “freedom from rendaku” as the third consonants of these words are natively.

(7) voiced), sound unambiguously mimetic to native speakers. Then, what gives rise to the mimetic tones? My answer is “prosodic symbolism does. ” We should here notice the fact that the Japanese grammar imposes a morphophonological restriction on what are called mimetics. As I point out in Akita (2007a), all Japanese mimetics fundamentally take one of the following forms (see Lu 2006: 81-88 for diachronic observation of mimetic templates): (10) a. For CV-roots: CVQ^, CVN^, CViQ^, CV(^)V(^), CVV^-CVV, CVV-CVV, CVN^-CVN, CVNCVN, CVi^-CVi b. For CVCV-roots: CVCVQ^, CVCVN^, CVCV^ri, CVCCV^ri, CV^CVCVCV, CVCVCVCV Apparent exceptions like bururuN(-to) (CVCVCVN^) and kururiN(-to) (CVCVriN^) can be explained as unconventional mimetics which are derived from one of the fifteen fixed mimetic templates for intensification or some other purpose. In fact, the examples here have their basic forms that conform to the templates: buruN(-to) (CVCVN^) and kururi(-to) (CVCV^ri), respectively. This restriction (or necessary and sufficient condition) can be qualified as an ultimate factor that distinguishes mimetics from nonmimetic words, which are free from this restriction. Regarding segmental symbolism, it is true that, as Tamori and Schourup say, there are some segmental characteristics that can differentiate mimetics from (the) other lexical strata (e.g., abundance of /p/-initial words, freedom from the nasalization of intervocalic /g/). However, they alone never qualify a word as mimetic. In fact, the loanword piro^siki ‘pirozhki’, which has a /p/-initial but violates the above formal restriction, least likely sounds mimetic. Conversely, [ga^baŋaba], which would be produced if it underwent the nasalization of intervocalic /g/ with its CV^CV-CVCV form retained, can still have a mimetic tone. Now, the categorization problem has been settled and we can conclude that not segmental but prosodic symbolism functions as a criterion for a word being mimetic in Japanese. (At the same time, it may be possible to restate that segmental symbolism is active and effective only when put in a template prepared by prosodic symbolism (see Akita, to appear, b for an experimental backup).)9 This property of prosodic symbolism seems to be attributable to its closed-class nature, and the nature in turn gives further support to the belief that prosodic symbolism belongs to the grammatical subsystem of language.. 6. A Crosslinguistic Perspective to Sound Symbolism At last, based on the discussion in the previous section, I briefly give an implication for crosslinguistic explorations in sound symbolism, which are quite rare in the literature compared with language-internal ones (see Hinton et al., eds. 1994; Nuckolls 1996; Voeltz and Kilian-Hatz, eds. 2001 for exceptional investigations). As Herlofsky (1983) remarks, English, together with other Germanic and Romance languages, is generally believed to be scarce of sound-symbolic vocabularies (see also de Saussure 1959: 69), but linguists since Firth (1930) have also been aware of the abundance of “phonaesthemes ” (minimal units of sound symbolism, often between a phoneme and a morpheme in size: e.g., -ash in clash, dash, smash; gl- in glimpse, glisten, glitter; -kle in prickle, tickle, twinkle) in the language. Phonaesthemes can be regarded as sets of segmental-symbolic elements in the current context. Accordingly, it may be possible to say that English has a somewhat rich segmental-symbolic system. One might criticize that English also abounds in prosodic-symbolic templates like the closed syllable CVC exemplified by words with -ash, -kle, etc. Nevertheless, this criticism misses the point. In English, although not in Japanese, CVC is a common syllable structure and not appropriate as a criterion which guarantees a membership in a particular lexical category. Also, English does not adopt reduplication as frequently as Japanese, Korean, Basque, Bantu, and other languages reported as abundant in sound symbolism (it adopts certain phonaesthemes like -er and -le instead (Herlofsky 1983: 20-21); see also Hurch, ed. 2005). We can thus speculate that the absence of special, productive prosodic templates for sound symbolism makes native English speakers like Herlofsky feel that their language is poor in sound-symbolic phenomena..

(8) In sum, we can hypothesize that languages that are believed as having a rich sound-symbolic system indeed have a well-organized prosodic-symbolic system and that languages that are believed as poor in such vocabularies are in fact not necessarily poor in segmental symbolism but only their grammatical, prosodic-symbolic systems are underdeveloped.. 7. Conclusion This paper has proposed and discussed the validity of the dichotomy of sound symbolism in Japanese. Based on the above observations, we can make a general statement that prosodic symbolism is of higher priority than segmental symbolism with respect to syntactic, morphological as well as lexical relevance. I suggested that this priority of prosodic symbolism can be ascribed to its status as a member of the grammatical subsystem of language. The present proposal also showed a possible direction of the crosslinguistic study of sound symbolism. I hope that this fundamental proposal will take the place of the traditional enumerating descriptions of sound symbolism and provide a basis for further serious investigations into mimetics and sound symbolism in general.. Notes *. This paper is a part of the author’s research supported by Grant-in-Aid for JSPS Fellows (#19·536). I thank Professor Yo Matsumoto for his insightful comments. My gratitude also goes to doctors and nurses at Kobe Red Cross Hospital, from whom I have received lots of warm support in various forms. Remaining inadequacies are, of course, mine. 1 Abbreviations and symbols used in this paper are as follows: ACC = accusative; MIM = mimetic; N = moraic nasal; NOM = nominative; PST = past; Q = first half of the geminate cluster; QUOT = quotative; red = reduplication; ^ = accent nucleus, pitch fall (used only in need) 2 Different terms are used for “prosodic” elements in different frameworks. For example, structuralists and generativists call them “suprasegmentals” and “autosegmentals,” respectively, instead (see Yoshiba 2004). 3 I am indebted to Professor Haruo Kubozono for the labeling of the two kinds of sound symbolism. Moreover, he pointed out that there are in fact three kinds of sound-symbolic elements: palatalization and voicing are distinguishable from consonant and vowel symbolisms, for they add a segmental feature. Nevertheless, I ignore this finer-grained distinction because all these sound-symbolic elements can be grouped as constituents of mimetic roots and it is this point that is crucial for the discussions below. For the same reason, I do not adopt a feature-based derivational analysis for the consonants and vowels in (1h, i) (e.g., [+labial, +stop] > /p/ [+labial, +stop, +oral, -voiced], /m/ [+labial, +stop, +nasal]). 4 These memberships are at the level of morphophonology. In fact, Talmy classifies mimetics (or ideophones) as a word class into the lexical subsystem. Furthermore, it is true that segmental as well as prosodic symbolism is closed (i.e., limited in number). Nevertheless, we can safely say that the former is opener than the latter. This criticism would be cast on the open/closed dichotomy as a whole. 5 The ambiguity here stems from the fact that the prolonged mimetic forms are almost always derived from a mimetic with a stem-final element and are subject to the prosodic symbolism of stem-final elements (Hamano 1998: 67-72, 106-107). As a consequence, the prolonged forms are ambiguous between the telic reading of their inherited stem-final elements and the atelic reading of their newly added long vowels. 6 Semelfactivity here refers to a punctual activity like hitting, kicking, and possibly surprise. This eventuality type is compared with multifactivity, which refers to an iterative activity like knocking and beating (see Smith 1991; cf. Vendler 1957). The latter can be also represented by mimetic predicates in the form of reduplicative (e.g., tonton tataku ‘hit iteratively’). 7 Itamu ‘hurt’ is a stative verb and never allows a shift to a telic meaning. The same restriction seems to hold for the eventuality ziN(-to) depicts, which is the lexical meaning of the mimetic. 8 This property of sound symbolism is further interesting in terms of the typology of lexicalization patterns (Talmy 2000). Japanese is known as a verb-framed language, which prefers lexicalizing the core-frames of motion events (i.e., paths) in verbs. In this regard, the phenomenon observed here, in which an adjunct defines the core-frame of an eventuality, goes in an exactly opposite direction (Yo Matsumoto, personal communication). 9 Kita (1997) gives a semantic analysis that attempts to distinguish mimetics from the rest of the lexicon. However, he also admits the existence of certain uses of mimetics with an intermediate status—sharing some semantic properties with nonmimetic parts of a sentence (cf. Tsujimura and Deguchi, to appear). See Nasu (2001), Hamano (2006), Lu (2006), Akita (2006, to appear, a, b), and Kageyama (to appear) for arguments claiming or suggesting the gradient nature of the mimetic category..

(9) References Akita, Kimi. 2006. Suru and to for Japanese mimetics: A semantic-integration account. Handbook of the Fifth International Conference on Practical Linguistics of Japanese, pp. 26-27. ---. 2007a. A constructionist approach to mimetic morphophonology. Morphology and Lexicon Forum 2007, Kobe University, 30 June 2007. ---. 2007b. Eventuality type determination of mimetic predicates in Japanese. Conference Handbook of the 134th Semiannual Meeting of the Linguistic Society of Japan, pp. 276-281. ---. 2007c. One experience viewed in two ways: A viewpoint approach to the case marking patterns of Japanese psych-verbs. Lexicon Foru m 3: 283-297. ---. To appear, a. The acquisition of the constraints on mimetic verbs in Japanese and Korean. Japanese/Korean Linguistics 16. ---. To appear, b. Defining the mimetic category in the Japanese lexicon: A reply to Tamori and Schourup (1999). Proceedings of the Eighth Annual Meeting of the Japanese Cognitive Linguistics Association. Alpher, Barry. 1994. Yir-Yoront ideophones. In Hinton et al., eds., 1994, pp. 161-177. Amanuma, Yasushi, ed. 1973. Giongo/gitaigo ziten [A dictionary of mimetics]. Tokyo: Tokyodo. Asano, Tsuruko, ed. 1978. Giongo/gitaigo ziten [A dictionary of mimetics]. Tokyo: Kadokawa. Bybee, John L., Revere Perkins, William Pagliuca. 1994. The Evolution of Grammar: Tense, Aspect, and Modality in the Languages of the World. Chicago/London: The University of Chicago Press. Chang, Andrew C. 1990. A Thesaurus of Japanese Mimesis and Onomatopoeia: Usage by Categories. Tokyo: Taishukan. Davis, Stuart, and Isao Ueda. 2002. Mora augmentation processes in Japanese. Journal of Japanese Linguistics 18: 1-23. de Saussure, Ferdinand. 1959 (orig. 1916). Course in General Linguistics. New York: Philosophical Library. Firth, John Rupert. 1930. Speech. London: Benn’s Sixpenny Library. Hamano, Shoko. 1994. Palatalization in Japanese sound symbolism. In Hinton et al. eds., 1994, pp. 148-157. ---. 1998. The Sound-Symbolic System of Japanese. Tokyo: Kurosio Publishers. ---. 2006. Accent and sound symbolism in Japanese. Japanese/Korean Linguistics 14: 3-14. Hinton, Leanne, Johanna Nichols, and John J. Ohala, eds. 1994. Sound Symbolism. Cambridge: Cambridge University Press. Hurch, Bernhard, ed. 2005. Studies on Reduplication. Berlin/New York: Mouton de Gruyter. Izumi, Kunihisa. 1976. Giseigo/gitaigo-no tokusitu [Some characteristics of mimetics]. In Takao Suzuki, ed., Nihongo-no goi-to hyoogen [Japanese vocabulary and expressions], pp. 105-151. To kyo: Taishukan. Kageyama, Taro. 2006. Gitaigo-doosi-no toogo-koozoo [The syntactic structures of mimetic verbs]. Humanities Review 56, 1: 83-101. Kwansei Gakuin University. ---. To appear. Explorations in the conceptual semantics of mimetic verbs. Proceedings of the 2nd Oxford-Kobe Seminar on Linguistics. Kakehi, Hisao, Ikuhiro Tamori, and Lawrence Schourup. 1996. Dictionary of Iconic Expressions in Japanese. Berlin/New York: Mouton de Gruyter. Kindaichi, Haruhiko. 1978. Giongo/gitaigo gaisetu [An outline of mimetics]. In Asano, ed., 1978, pp. 1-25. Kita, Sotaro. 1997. Two-dimensional semantic analysis of Japanese mimetics. Linguistics 35, 2: 379-415. Kuroda, S.-Y. 1979. Generative Gram matical Studies in Japanese. New York: Garland. Lu, Chiarung. 2006. Giongo/gitaigo-no hiyuteki-kakutyoo-no syosoo: Ninti-gengogaku-to ruikeiron-no kanten-kara [Aspects of figurative extensions of mimetics: From Cognitive Linguistic and typological perspectives]. PhD dissertation, Kyoto University. Mester, R. Armin, and Junko Itô. 1989. Feature predictability and underspecification: Palatal prosody in Japanese mimetics. Language 65, 2: 258-293. Miyaji, Yutaka. 1978. Giongo/gitaigo-no keitairon syookoo [Morphology of Japanese onomatopoeia]. Kokugogaku 115: 33-39. Nasu, Akio. 2001. Onomatope-no gokeisei-to akusento [ Word formation and accent of onomatopoeia]. Studies in Japanese Language and Culture 11: 9-24. Osaka University of Foreign.

(10) Studies. Nishio, Toraya. 1988. Gendai-goi-no kenkyuu [Studies on the modern lexicon]. Tokyo: Meiji Shoin. Nuckolls, Janis B. 1996. Sounds Like Life: Sound-Symbolic Grammar, Performance, and Cognition in Pastaza Quechua. New York: Oxford University Press. Pinker, Steven. 1989. Learnability and Cognition: The Acquisition of Argument Structure. Cambridge, MA: The MIT Press. Schourup, Lawrence, and Ikuhiro Tamori. 1992. Palatalization in Japanese mimetics: Response to Mester and Ito. Language 68, 1: 139-148. Smith, Carlota S. 1991. The Parameter of Aspect. Dordrecht: Kluwer Academic Publishers. Talmy, Leonard. 1988. The relation of grammar to cognition. In Brygida Rudzka-Ostyn, ed., Topics in Cognitive Linguistics, pp. 165-205. Amsterdam/Philadelphia: John Benjamins. ---. 2000. Toward a Cognitive Semantics, Volume II: Typology and Process in Concept Structuring. Cambridge, MA: The MI T Press. Tamori, Ikuhiro. 1983. Onomatope: On’in-keitai-to goisei [Onomatopoeia: Its morphophonology and lexicality]. Journal of Cultural Science 19, 2: 179-201. Kobe University of Commerce. ---. 1993. Nihongo-onomatope-no toogo-hantyuu [Syntactic categories of Japanese mimetics]. In Hisao Kakehi and Ikuhiro Tamori, eds., Onomatopia: Gion-/gitai-go-no rakuen [Onomatopia: The utopia of mimetics], pp. 17-75. Tokyo: Keiso Shobo. ---, and Lawrence Schourup. 1999. Onomatope: Keitai-to imi [Onomatopoeia: Its morphology and semantics]. Tokyo: Kurosio Publishers. Tobita, Yoshifumi, and Hideko Asada, eds. 2002. Gendai giongo/gitaigo yoohoo ziten [A dictionary of the usage of current mimetics]. Tokyo: Tokyodo. Toda, Kichiro. 1942. Giseigo [Mimetics]. Nihongo 2, 1. Tsujimura, Natsuko, and Masanori Deguchi. To appear. Semantic integration of mimetics in Japanese. Papers from the Thirty-Ninth Regional Meeting, Chicago Linguistic Society. Vendler, Zeno. 1957. Verbs and times. Philosophical Review 66: 143-160. Voeltz, Erhard Friedrich Karl, and Christa Kilian-Hatz, eds., 2001. Ideophones. Amsterdam/Philadelphia: John Benjamins. Yamaguchi, Nakami. 2003. Kurasi-no kotoba: Gion-/gitai-go ziten [ Words of life: A dictionary of mimetics]. Tokyo: Kodansha. Yamanashi, Masa-aki. 2000. Ninti-gengogaku genri [The principles of Cognitive Linguistics]. To kyo: Kurosio Publishers. Yoshiba, Hiroshi. 2004. Segment/prosody dichotomy and cognition. In Donald L. Smith, Shuichi Nobe, Peter Robinson, Gregory Strong, Minako Tani, and Hiroshi Yoshiba, eds., Language and Comprehension: Perspectives from Linguistics and Language Education, pp. 1-29. To kyo: Kurosio Publishers.. Department of Linguistics Graduate School of Humanities and Social Sciences Kobe University 1-1 Rokko-dai-cho, Nada-ku, Kobe-shi Hyogo 657-8501 Japan e-mail: akitambo@lit.kobe-u.ac.jp.

(11) 日本語音象徴語における二つの音象徴秋田喜美（神戸大学大学院／日本学術振興会特別研究員）【要旨】. Ö!, {^¿W²ÑÞ-

(12) ÔîÚ|Ôòî=Ôó îÚ|ÌÀò Ä²=ðQ²¨l c0I,îï²=w²ãd=á·ó% DV¢(i) ) +, uÒÔv ¢, Ôv¼°, ßÐÔ¡Ü0ØÖ,ªÖ²h³0½ (i) a. ïzÚ|õîÚ|Ô Ýä0Ú,òåË, ÔrÌÀ, ç¶ î, tÕjî, t[ó b. V»Ú|õîÚ|Ô Ô0,ò î=jî, `È\, i\ó ïzÚ|!s ¦V»Ú|)+'RJxY0¹>, ïzÚ|!(iii) )K¾²14926²ZE ¸,>, Ô.®,DÔ (ii) )¯,¨Ã$E ), V»Ú|!K¾0Í (ii) a. P Æ0{*>´ô ê}じろじろÍ òDåË  durativeó b. P Æ0{>´ô* ê}じろりとÍòÔrÌÀ-+ punctualó (iii) DåË ÔrÌÀ ç î. atelic telic ambiguous. durative punctual durative. ¨!, ´ê²}²Âñ0É[Ôr(-)- ,] w h²ÔrÌÀ0 îÚ |Ôì*-,wÖ²©Ú'ë/,ò* , vs. *** ,ó ¹D, ïzÚ|!îÚ|Ô¢¼°0Îl,Ô îÚ|Ô!h²S 例 CVQ^, CVCV^ri, CVCCV^ri, , ïzÚ|ÌÀ)+w-ìO ïzægò□ CV^CV-CVCVó0¤_, ïzæg L- ¤ îÚ|Ô,~Ì]VH , !, ,àÔ îÚ|Ô* ò¨l V»¨)+'óïzæ 例 %'ö+: mimetic vs. #ö&: g0¤ b Nk,CmÊE*-,ò□ nonmimeticó ¹?, )¨XîÚ|Ôæg §!, ,ÐÔ ,îÚ|Ôò¼°ó Ó × - fa, y-,M", íeÔ(742Ôº, Ùoî Ú|Ôv0,-,ÐÔ!, åËwFÉ-,«¬²îÚ|Ôïzæg Ï p-, , ÇÔ(8;<4Ôº, îÚ|ÔvB-,ÐÔ!, )æg ±â ) G@ Ïp!, Talmy (1988, 2000) ,ÐÔ0,D AJÓµ345: ªÖ²ÁT,Åc£' ,_, ïzÚ|! qÚCÚ ê²Ýä0U,¨ *14926¨ë/+, èð Û *îÚ|Ô¼° Îl h¥,¦, ¢345:òÓµÉ Ú à0 èðó ¨0¸/,>, V»Ú|!ïzÚ|ÌÀ ,æg0¤Ô îÛ=²cTn0¦, Ôv345: òÓµÉ Ú Tn0 (¡Ü²)éðó ¨0¸/,.

(13)