A Constructionist Analysis of Emphatic Mimetics in Japanese
* Kimi AkitaJapan Society for the Promotion of Science / the University of Tokyo
Abstract
This paper discusses the semantic properties of so-called emphatic mimetics in Japanese (e.g., huNwa^ri ‘fluffy’, zaQku^ri ‘coarse’). They instantiate a morphophonological construction (i.e., CVCCV^ri) that has both compositional and noncompositional aspects. This construction ac- cepts relatively arbitrary form-meaning relationships, but, like other mimetic constructions, has as its semantic side a holistic event schema that contains both manner and result. The latter prop- erty allows even purely resultative emphatic mimetics (e.g., koNga^ri ‘baked lightly brown’) to take the quotative particle -to, which is assumed to mark manner.
1 Introduction
This paper presents a basic description and a constructionist analysis of the semantics of what I call “emphatic mimetics” in Japanese.1 Japanese has a huge sound-symbolic lexicon with rich semantic variety, whose items are called mimetics (or ideophones, expressives). As is often the case across languages, Japanese mimetics have some typical forms, such as reduplicative (e.g., su^tasuta ‘walking briskly’) and suffixal ones (e.g., hoQ ‘relieved’), and emphatic mimetics can also be defined by their morphophonological shape CVCCV^ri.2 The coda of their first syllable, which is called an “emphatic consonant/mora” (“{C}” in Hamano 1998, “prosodic infix” in Nasu 2004), is realized as the first half of a geminate cluster when followed by a voiceless consonant, as in (1a), and as a moraic nasal when followed by a voiced consonant, as in (1b) (Kuroda 1967, 1979; McCawley 1968: 125).3
(1) a. biQku^ri ‘astonished’, geQso^ri ‘dejected’, koQte^ri ‘heavily fatty’, zaQku^ri ‘coarse’ b. huNwa^ri ‘fluffy’, koNga^ri ‘baked lightly brown’, ziNwa^ri ‘slowly permeated’
Emphatic mimetics are the third most productive class of mimetics in Japanese, and have received some pho- nological considerations, especially in the context of mora augmentation (Davis & Ueda 2002; Akashi 2007). (2) Productivity of mimetic templates: # entries in Kakehi et al. (1996) i. CV(^)CV-CVCV (e.g., pa^tipati ‘snapping repeatedly’) 484
ii. CVCVQ^ (e.g., patiQ^ ‘snapping’) 213 iii. CVCCV^ri (e.g., paQti^ri ‘eye wide open’) 134 iv. CVCV^ri (e.g., pati^ri ‘snapping’) 131
v. CVCV(^)N(^) (e.g., patiN^ ‘snapping’) 101 (adapted from Akita 2009: 110) For at least three reasons, however, little is known about their semantics. First, emphatic mimetics are often re- garded as a secondary type that is derived from CVCV^ri (e.g., biku^ri for biQku^ri, huwa^ri for huNwa^ri). As
Near-final draft, 01/21/11. To appear (June 2011) in KLS 31: Proceedings of the 35th Annual Meeting of the Kansai Linguistic Society.
Kuroda (1979: 205–206) and Moriyama (2002) correctly note, however, about half of them (61 (45.52%) in my database) do not have their “non-emphatic” counterparts (e.g., koQte^ri vs. *kote^ri; koNga^ri vs. *koga^ri), and, as shown in (2), they are more productive than CVCV^ri mimetics. Second, emphatic mimetics are unlikely to be a pure sound imitation, and often have nonmimetic tones (Moriyama 2002; Akita 2009: Chapter 4). Third, as we will see in Section 4, it is by far harder to abstract a common semantic feature from emphatic mimetics than from reduplicative and suffixal ones, whose majority are systematically linked with durativity and punctuality, respec- tively (Akita 2009: Chapter 5). The present paper is thus intended to provide some basic and advanced discussion on the meaning of emphatic mimetics in support of the growing theory of Construction Morphology (Booij 2010, etc.). The discussion will lead us to conclude that, despite their relatively low mimeticity, emphatic mimetics do possess a key semantic property of mimetics, which is termed “holisticity.”
The paper is organized as follows. Section 2 will frame a research question focusing on the particle selection of mimetics. Section 3 will introduce the notion of morpho(phono)logical constructions. Section 4 will investigate the semantics of emphatic mimetics with respect to their constructional properties, both compositional and noncom- positional. Section 5 will extend the discussion in terms of construction-specific and construction-general features of emphatic mimetics. Section 6 will conclude the paper.
2 Research question
In this section, I outline the research question to be tackled in this paper, which is concerned with the apparent ir- regularity of the particle selection of emphatic mimetics. Japanese mimetics are mainly manner adverbs in a broad sense. Under certain phonological conditions (see Nasu 2002 for their details), they are optionally marked by the quotative particle -to, as illustrated in (3) (Tamori 1984; Toratani 2006; Kageyama 2007).
(3) Ai-wa {ke^rakera (-to) warat/ te^kuteku (-to) arui/ do^kidoki (-to) kanzi} -te i-ta. -TOP MIM -QUOT laugh MIM -QUOT walk MIM -QUOT feel -CONJ be-PST
‘Ai was {laughing cackle-cackle/walking lightly/feeling excited}.’
The quotative also follows mimetic adverbs whose bare realization is phonologically ruled out (i.e., those with less than four moras and their derivatives). Importantly, this is true for both manner and result mimetics, as shown in (4). In this regard, -to-marking is a default marking in mimetics.
(4) a. Neko-ga {koroQ^/ koroN^/ koro^ri} *(-to) korogat-ta. (manner) cat-NOM MIM MIM MIM -QUOT roll-PST
‘A cat rolled over.’
b. Ken-ga ebi-o {karaQ^/ kara^ri} *(-to) age-ta. (result) -NOM shrimp-ACC MIM MIM -QUOT deep.fry-PST
‘Ken deep-fried a shrimp crisp.’
Meanwhile, unaccented resultative reduplicatives are obligatorily followed by the copula -ni, as in (5).
(5) Ai-ga beekon-o {karikari/ katikati/ paripari} -{ni/*to} yai-ta. (result) -NOM bacon-ACC MIM MIM MIM -COP/QUOT fry-PST
‘Ai fried bacon {crisp/hard/crunchy}.’
Note that the copula can never be attached to result mimetics of the (4b) type. (4) b´. *Ken-ga ebi-o kara^ri-ni age-ta. (result)
-NOM shrimp-ACC MIM-COP deep.fry-PST
‘Ken deep-fried a shrimp crisp.’
Emphatic mimetics show an apparently unexpected pattern with respect to particle selection. As will be dis- cussed in Section 4, some of them represent manner, whereas others represent result. In parallel with the accented reduplicatives in (3), however, they optionally take -to regardless of their meanings (Tamori 1984).
(6) a. Ai-ga {yuQku^ri/ noNbi^ri} -{(to)/*ni} arui-ta. (manner) -NOM MIM MIM -QUOT/COP walk-PST
‘Ai walked {slowly/leisurely}.’
b. Ai-ga pan-o {puQku^ri/ koNga^ri} -{(to)/*ni} yai-ta. (result) -NOM bread-ACC MIM MIM -QUOT/COP bake-PST
‘Ai baked bread {swollen/lightly brown}.’
Although the copula is not available in most cases, we can find some counterexample resultatives on the Internet, like (7) (last accessed 9 October 2010), and a few emphatic mimetics require or allow -to-marking, as in (8). (7) a. kutibiru-o puQku^ri-ni siage-te kure-ru rippu (http://www.buyma.com/item/4660273/) lip-ACC MIM-COP finish-CONJ give.me-NPST lipstick
‘a lipstick that successfully makes [our] lips plump’
b. Saibasi-de tutui-te-mo karikari-ga wakar-u kurai ryoomen-o long.chopstick-with poke-and-even crisp-NOM know-NPST extent both.side-ACC
koNga^ri-ni yaki-mas-u. (http://cookpad.com/recipe/1227377) MIM-COP fry-POL-NPST
‘Fry both sides (of the tofu) to the extent that [you] feel [its] crispness even when you poke [it] with long chopsticks.’
(8) a. Pan’ya-ga honmono-to soQku^ri -{*to/ni} Anpanman-o yai-ta. baker-NOM real.thing-with MIM -QUOT/COP -ACC bake-PST
‘A baker baked bread just alike with real Anpanman (An-breadman).’ b. Ken-wa huku-o baQti^ri -{to/ni} kime-ta.
-TOP clothes-ACC MIM -QUOT/COP decide-PST
‘Ken made [himself] look striking in [his] clothes.’
It is thus unlikely that the CVCCV^ri form is destined (e.g., phonologically) to be incompatible with the copula-
tive resultative. Then, the question is why most emphatic mimetics only allow -to. This paper will ascribe this mystery to the construction they instantiate, whose semantic representation contains both manner and result.
3 Construction Morpho(phono)logy
This section briefly introduces Construction Morphology or Morphophonology as a theoretical framework. Broadly speaking, constructions are pairings of form and meaning. This notion has been mainly developed in the analysis of sentences and idioms (Fillmore & Kay 1995; Goldberg 1995; inter alia). (9) is an oft-cited example of what is called the caused-motion construction in English, in which the syntactic schema [NP [V NP PPobl]] is paired with the meaning ‘X CAUSES Y to MOVE Z’.
(9) Fred sneezed the napkin off the table. (Goldberg 1995: 156)
The point is that the intransitive verb sneeze is likely to license neither the direct object the napkin nor the direc- tional PP off the table, which are instead attributed to the whole construction. Therefore, at least Goldberg’s ver- sion of Construction Grammar values the notion of noncompositionality.
This conception was inherited by construction morphologists, especially Geert Booij (2010, etc.), who argues that quite a few morphological phenomena that have traditionally received derivational accounts can be better generalized in terms of constructions. Tsujimura & Davis (2008) propose a constructionist account of Japanese innovative verbs, such as abe^r- ‘shirk one’s responsibility halfway’ (< Abe (ex-premier) [proper noun]), bagu^r-
‘get a bug’ (< bagu ‘bug’ [English loanword]), boko^r- ‘beat’ (< bokoboko ‘beaten to a pulp’ [MIM]), and koku^r-
‘declare one’s love’ (< kokuhaku ‘confession’ [Sino-Japanese]). These verbs, which have diverse origin, are con- structionally schematized as a pairing of the morphophonological template [V …(C)V(C)V^-r-] and the regis- ter-related information like “playfulness” (see Akita 2010b for another case study in Japanese).
In what follows, I will argue that emphatic mimetics instantiate a morphophonological construction that have not only holistic but also compositional properties. The present investigation will thus contribute to the growing theory another case that illustrates the significance of constructions in the morphophonology of Japanese, whose syntax is in general conceived “less constructional” than languages like English.
4 The constructional semantics of emphatic mimetics
In this section, I explore the constructional semantics of emphatic mimetics by focusing on three semantic proper- ties they exhibit: namely, emphasis, unpredictability, and referential diversity.
4.1 Emphasis
I will start with a closer look at the emphatic nature of emphatic mimetics. The emphatic meaning of emphatic mimetics is usually understood to iconically link with their additional mora, compared with CVCV^ri mimetics.4 However, the significance of the segmental realization of {C} also deserves a subsection. It seems that /Q/, which strengthens the air release of the following obstruent, is more straightforwardly associated with the concept of em- phasis than /N/, whose resonance might instead evoke the image of sluggishness or looseness (see Akita 2009:
192).
In fact, /Q/ is also a default realization for other types of mimetic “affixes” that are more or less related to the notion of emphasis (cf. Akita 2010b). First, post-lexical emphatic mora augmentation in unaccented redupilcatives, illustrated in (10), basically involves /Q/ rather than /N/ even if it results in the occurrence of a voiced geminate, as in (10b–c), which is generally avoided (Nasu 2002; Akashi 2007).
(10) a. pikapika ‘glittery’ piQkapika/*piNkapika b. subesube ‘dry and smooth’ suQbesube/*suNbesube
c. gizagiza ‘notched’ giQzagiza/*giNzagiza (adapted from Nasu 2002: 103–104) Second, as cited in (11), 90% of all bimoraic roots that can form suffixal mimetics (i.e., CVCVX) can form a -Q-suffixed mimetic, whereas -N- and -ri-suffixed mimetics are of much more limited productivity.
(11) CVCVQ^ (e.g., koroQ^ ‘rolling’) 207/230 (90.00%) CVCV(^)N(^) (e.g., koroN^ ‘rolling’) 98/230 (42.61%)
CVCV^ri (e.g., koro^ri ‘rolling’) 132/230 (57.39%) (adapted from Nasu 2007: 49) A similar distributional tendency is observable in emphatic mimetics. As shown in Table 1, which is based on the entries of Kakehi et al. (1996), C2 of emphatic mimetics, but not of other bimoraic root-based mimetics, is self-evidently skewed toward voiceless obstruents (χ2 (1) = 42.02, p < .001). (Such an inclination is absent in their C1 (χ2 (1) = .06, p = .81 (n.s.)).) This means that emphatic {C} is predominantly realized as /Q/.5
Table 1. The consonant distribution of emphatic and other bimoraic root-based mimetics
Emphatic Other
Voiced 66 (51.97%) 244 (53.16%) C1
Voiceless 61 (48.03%) 215 (46.84%) Voiced 25 (18.66%) 242 (50.00%) C2
Voiceless 109 (81.34%) 242 (50.00%)
The present data thus serves as indirect reinforcement of the idea that emphasis is a characteristic, if not com- mon, semantic component of emphatic mimetics.6 From the constructionist point of view, the observed phonose- mantic importance of {C} can be considered as a compositional aspect of this morphophonological construction.
4.2 Unpredictability
This subsection and the next put stronger emphasis on the noncompositionality of the emphatic mimetic construc- tion. Here I will demonstrate the formal and semantic unpredictability of emphatic mimetics.
First, it is quite common that one root occurs in more than one mimetic template. For example, the mimetic root doki is realized in do^kidoki, dokiQ^, dokiN^, doki^ri, and doQki^ri ‘startled, nervous, excited’. However, 26 (19.40%) emphatic mimetics in my database have no single root-based relative (e.g., *genaQ^, *genaN^, *gena^ri,
*ge^nagena for geNna^ri ‘dispirited’). According to Akita (2009: 147), lexical isolation of this sort is most striking in emphatic mimetics. In this sense, they are formally less predictable than other major types of mimetics.
Next, even emphatic mimetics that have root-based relatives are often isolated in meaning. In my database, 51
(47.22%) non-orphan emphatic mimetics are referentially different from their relatives. Specifically, we need to posit root-level semantic extension of some kind in order to capture the relationship between those emphatic mi- metics and their relatives. For example, the ‘exact’ meaning of the emphatic mimetic tyoQki^ri can be derived via metonymy and metaphor from the ‘clipping’ meaning of its relatives tyokiN^ and tyo^kityoki, for a clear cut made by clipping can be abstracted into the concept of exactness without difficulty. On the other hand, the pleasant chill depicted by the emphatic mimetic hiNya^ri ‘pleasantly cool’ is a kind of the low temperature represented by its cognate nonmimetic verb hiyas- ‘make cool’. This synecdochical relation is independent of the metonymical rela- tion between this verb and the non-emphatic mimetics hiyaQ^, hiya^ri, and hi^yahiya ‘thrilled’ (see Akita 2010c). The semantic relationships between emphatic mimetics and their root-based relatives are summarized in (12). It tells us how hard it is to predict the meanings of emphatic mimetics from their relatives.
(12) a. Emphatic mimetics with relatives:
Metaphorically related (e.g., yaNwa^ri ‘mild’ vs. yawaraka- ‘soft’) 5 (3.73%) Metonymically related (e.g., doQka^ri ‘stately’ vs. do^kadoka ‘tramping’) 40 (29.85%) Synecdochically related (e.g., siQto^ri ‘pleasantly moist’ vs. sitoQ^ ‘damp’) 2 (1.49%) Homonymically “related” (e.g., guQsu^ri ‘sound asleep’ vs. gu^sugusu ‘snuffling’) 4 (2.99%) Identical (e.g., gaQsi^ri ‘solid’ vs. gasiQ^ ‘solid’) 57 (42.54%) b. Emphatic mimetics without relatives (e.g., kiQpa^ri ‘decisive’) 26 (19.40%)
4.3 Referential diversity
Finally, this subsection is focused on the diversity of referential meanings of emphatic mimetics—namely, the diversity of what they specifically emphasize. Emphatic mimetics can be semantically classified into four, as shown in (13), although there are some boundary cases (see the author’s website for the exhaustive lists of words). (13) a. Manner (e.g., yuQku^ri ‘slow’, niNma^ri ‘smiling complacently’) 39 (29.10%) b. Result:
i. Ambiguous (e.g., beQta^ri ‘plastering, sticky’, huNwa^ri ‘gentle, fluffy’) 5 (3.73%) ii. Vague (e.g., boQki^ri ‘breaking completely’, suQpo^ri ‘slipping in’) 15 (11.19%) iii. Genuine (e.g., deQpu^ri ‘portly’, siNna^ri ‘pliant’) 45 (33.58%) c. Emotion (e.g., gaQka^ri ‘disappointed’, siNmi^ri ‘pensive’) 15 (11.19%) d. Degree/quantity (e.g., hoNno^ri ‘faint’, kiQka^ri ‘exact’) 15 (11.19%) Manner emphatic mimetics are used to modify predicates, as in (14a), and emotional ones are typically com- bined with the dummy verb su- ‘do’ to form mimetic verbs, as in (14b) (Akita 2010c). The degree/quantity type is illustrated in (14c). Since mimetics of this type are more likely to instantiate “extremity” than emphasis, we can posit a construction-level semantic extension based on endpoint metonymy (Brugman 1988): extremity is the high end of the scale of emphasis.
(14) a. Hutari-wa paathii-o koQso^ri nukedasi-ta. (manner) 2.people-TOP party-ACCMIM slip.out-PST
‘The two slipped stealthily out of the party.’
b. Pan-ni kabi-ga hae-te gaQka^ri si-ta. (emotion) bread-DAT mold-NOM grow-and MIM do-PST
‘[I] was disappointed that the bread got moldy.’ c. Basu-wa go-zi tyoQki^ri-ni ki-ta. (degree) bus-TOP 5-o’clock MIM-DAT come-PST ‘The bus came exactly at five o’clock.’
The most complicated but most crucial type is the result type, which can be further divided into three. First, some emphatic mimetics are ambiguous (i.e., polysemous) between manner and result meanings. The ambiguity can be confirmed by their (in)compatibility with certain nonmimetic manner/result adverbials. For example, in (15a), huNwa^ri has a result meaning compatible with ‘soft’ but not with ‘hard’, whereas, in (15b), it has a manner meaning compatible with ‘gently’ but not with ‘violently’. This seems to be parallel with what is said for some nonmimetic adverbials like kirei-ni ‘carefully/clean’ and zatu-ni ‘roughly/messy’ (see Kuroda & Lee 2005). (15) a. Ken-ga pan-o huNwa^ri-to {yawarakaku/ *kataku} yai-ta. (result)
-NOM bread-ACCMIM-QUOT soft hard bake-PST ‘Ken baked bread fluffily {soft/*hard}.’
b. Ken-ga aiken-ni moohu-o {yasasiku/ *ranboo-ni} huNwa^ri-to kake-ta. (manner) -NOM pet.dog-DAT blanket-ACC gently violent-COPMIM-QUOT spread-PST
‘Ken spread a blanket {gently/*violently} over [his] pet dog.’
Second, some emphatic mimetics specify a resultant state and imply a certain manner as well. For example, baQsa^ri not only represents a resultant state of the object that has been cut, as shown in (16a), but also implies a drastic manner of cutting, as shown in (16b).
(16) a. Biyoosi-wa Ai-no kami-o {mizikame/ *nagame} -ni baQsa^ri kit-ta. (result) beautician-TOP -GEN hair-ACC shortish longish -COPMIM cut-PST
‘The beautician cut much of Ai’s hair {shortish/*longish}.’
b. Biyoosi-wa {omoikit-te/ ?tameraigati-ni} Ai-no kami-o baQsa^ri kit-ta. (implied manner) beautician-TOP dare-and hesitant-COP -GEN hair-ACC MIM cut-PST
‘The beautician {boldly/?hesitantly} cut much of Ai’s hair.’
This kind of “implication” seems less clear in their corresponding nonmimetic adverbials, such as mizikaku ‘short’ and omoikit-te ‘boldly’ in (17).
(17) a. Biyoosi-wa {omoikit-te/ tameraigati-ni} Ai-no kami-o mizikaku kit-ta. (result) beautician-TOP dare-and hesitant-COP -GEN hair-ACC short cut-PST
‘The beautician {boldly/hesitantly} cut Ai’s hair short.’
b. Biyoosi-wa omoikit-te Ai-no kami-o {mizikame/ nagame} -ni kit-ta. (manner) beautician-TOP dare-and -GEN hair-ACC shortish longish -COP cut-PST
‘The beautician boldly cut Ai’s hair {shortish/longish}.’
Finally, many emphatic mimetics depict result without any detectable manner implication, as shown in (18). Inter- estingly, most of the exceptional copulative resultatives like (7) and (8) above belong to this group.
(18) Ken-wa {sintyoo/ tyuuisanman} -ni pan-o koNga^ri-to yai-ta. (result) -TOP careful careless -COP bread-ACCMIM-QUOT bake-PST
‘Ken {carefully/carelessly} baked bread lightly brown.’
In sum, the emphatic mimetic construction has continuously related multiple meanings, which will be further discussed in a general context of mimetic semantics in the next section.
5 General discussion
This section develops our constructionist account of emphatic mimetics in terms of their unique (Section 5.1) and general properties (Section 5.2).
5.1 Construction-specific features
Emphatic mimetics have some unique features. To begin with, their formal complexity (i.e., CVCV root + {C} +
^-ri) distinguishes them from other types of mimetics both intra- and crosslinguistically. Unlike reduplicative and suffixal forms, they are unlikely to have their true equivalents in other languages. Historically, their emergence is later than other major modern mimetic forms (Yamaguchi 2002: 34–35). Furthermore, assuming that the primary meaning of this construction is “emphasis,” its form-meaning relationship is somewhat hard to regard as iconic, for mora augmentation presupposes the existence of CVCV^ri. This is a possible source of the relatively low mi- meticity of emphatic mimetics, which in turn allows multiple (often abstract) meanings to come in.
The idea of the relative arbitrariness of emphatic mimetics receives partial support from the fact that, in highly colloquial speech, a subpart of this mimetic class—which sound particularly nonmimetic—can undergo a sound change in their “suffix,” as in (19a). This change is common in nonmimetic words with the same morphological contour, as shown in (19b), but never takes place in CVCV^ri mimetics, as shown in (19c). This phenomenon suggests the low sound-symbolic significance (i.e., low iconicity) of the suffix in emphatic mimetics (see (8); cf. Hamano 1998: 107).
(19) a. baQti^ri ‘perfect’ baQti^si; gaQka^ri/gaQku^ri ‘disappointed’ gaQku^si/gaQku^si; piQta^ri ‘fit- ting’ piQta^si; soQku^ri ‘just alike’ soQku^si; tyoQki^ri ‘exact’ tyoQki^si (emphatic)
(cf. koNga^ri ‘baked lightly brown’ *koNga^si; suQpo^ri ‘slipping in’ *suQpo^si)
b. aNmari ‘little’ aNmasi; baQka^ri ‘just’ baQka^si; yaQpa^ri ‘after all’ yaQpa^si (nonmimetic) c. gaku^ri ‘slumping’ *gaku^si; pita^ri ‘clinging’ *pita^si; tyoki^ri ‘clipping’ *tyoki^si (CVCV^ri)
5.2 Construction-general feature
The emphatic mimetic construction shares an important semantic feature with other mimetic constructions. As discussed in Section 4.3, certain emphatic mimetics (i.e., (13bii), (16)) are vague with respect to their reference to manner and result. This semantic property is likely to be what has been generally discussed as the “holisticity” of eventuality depiction in the literature on mimetics (Kita 1997; Akita 2010a). We can in fact point out man- ner-result vagueness in other types of mimetics as well, as exemplified in (20).
(20) a. Rufi-wa Thiiti-o {bokoboko/ *hangorosi} -ni nagut-ta. (result manner) -TOP -ACC MIM half.dead -COP beat-PST
‘Luffy beat Teach {to a pulp/*half dead}.’ (Usuki 2010: 4; see also Toratani, to appear) b. Ai-wa hakusai-o za^kuzaku-to {ara/ ?mizin} -giri-ni si-ta. (manner result) -TOP Chinese.cabbage-ACC MIM-QUOT rough particle -cut-COP do-PST
‘Ai boldly chopped Chinese cabbage into {several/?fine} pieces.’
In (20a), the “result” mimetic bokoboko(-ni) ‘beaten to a pulp’, with its manner implication, enables the verb na- gur- ‘beat’ to form a resultative sentence, which is otherwise unavailable. In (20b), the manner mimetic za^kuzaku
‘chopping roughly’ implies the largeness of the pieces resulting from this particular chopping manner.
The question in Section 2 can now be answered. Even purely resultative emphatic mimetics take -to because their morphophonological shape CVCCV^ri as such is constructionally linked with the event schema containing both manner and result. The degree of manner specification differs idiosyncratically from item to item. However, the overall abundance of manner denoting/implying emphatic mimetics (i.e., (13a) + (13bi) + (13bii) = 44.03%) characterizes the entire construction as a manner-result category. This reasoning is reminiscent of Leonard Talmy’s remark on “Figure verbs” in Atsugewi and “Manner ideophones” in Basque. He suggests that these grammatical categories are so characterized due to the fact that their major members refer to a particular Fig- ure/Manner (Ibarretxe-Antuñano 2005: 20–21).
6 Conclusion
In this paper, I have investigated the semantics of Japanese emphatic mimetics, which can be schematized as a morphophonological construction. This construction turned out to have both compositional (i.e., the presumable meaningfulness of /Q/) and noncompositional aspects (i.e., formal/semantic unpredictability, referential diver- sity/holisticity) and to be an exceptional room in the mimetic lexicon that allows extensive inclusion of arbitrari- ness. The current paper thus supports the general hypothesis that the tremendous mimetic lexicon of Japanese is systematically structured by a limited number of productive word-level constructions, including reduplicative, suf- fixal, and emphatic ones.
Notes
* An earlier version of this paper was read at KLS 35 in June 2010. I thank Akio Nasu, Noriko Saito, Lawrence Schourup, and the two anonymous reviewers for their comments and help. Remaining shortcomings are my own.
This study was partly supported by Grant-in-Aid for JSPS Fellows (21-2238) and Spanish Ministry of Science and Innovation (FEI2010-14903). Email: [email protected]; website: http://sites.google.com/site/akitambo/.
1 They are also known as “intensified adverbs” (Kuroda 1979) or “ri-entyoo-kyoosei-giyoogo” (extended em- phatic -ri-mimetics) (Kuroda 1967).
2 The abbreviations and symbols used in this paper are as follows: N = moraic nasal; Q = first half of a geminate cluster; ^ = accent nucleus (pitch fall) (only for mimetics); ACC = accusative; CONJ = conjunctive; COP = copula;
DAT = dative; GEN = genitive; MIM = mimetic; NEG = negative; NOM = nominative; NPST = nonpast; POL = polite;
PST = past; QUOT = quotative; TOP = topic
3 Kadooka (2007) inappropriately adopts the term “circumfix” to treat -{C}- -ri as a complex affix.
4 It seems possible to attest the emphatic semantics of emphatic mimetics that have “non-emphatic” (i.e., CVCV^ri) counterparts, which are said to be closer to the mimetic prototype (Moriyama 2002; Kadooka 2007: Chapter 7). For instance, at least to some speakers, boQki^ri ‘breaking (a stick) completely’ evokes a particular result (i.e., two parts of a stick being apart from each other) more strongly than its non-emphatic counterpart boki^ri ‘breaking (a stick) forcefully’ (see Section 4.3), resulting in low cancelability, as in (ia), and availability in a result description, as in (ib) (see Mikami 2006 for a related discussion).
(i) a. Eda-o {?boQki^ri/ boki^ri} -to ot-ta-ga kanzen-ni-wa ore-nakat-ta. branch-ACC MIM MIM -QUOT make.break-PST-but complete-COP-TOP break-NEG-PST
‘[I] broke a branch {?completely/forcefully} but [it] did not break completely.’ b. {BoQki^ri/?boki^ri} -to ore-ta eda-ga oti-te i-ta. MIM MIM -QUOT break-PST branch-NOM drop-CONJ be-PST
‘There lied a branch that had been broken {completely/?forcefully}.’
5 /r/ never stands in C2 of emphatic mimetics (e.g., *koNro^ri) (Hamano 1998: 46–47), and C2 /b/ is often replaced by /p/ in emphatic mimetics (e.g., ga^bogabo ‘drinking much water’ vs. gaQpo^ri ‘making a pile’) (Nasu 1999). These phenomena could also be understood in favor of the preference of emphatic mimetics for a geminate {C}.
6 Akio Nasu suggested the possibility that /N/ is phonologically disfavored due to its high operational cost. The present discussion does not reject this alternative account. It might be the case that the obtained consonant distribu- tion depends on both phonosemantic and phonological factors.
References
Akashi, Kaori (2007). Variation of emphatic mora augmentation in Japanese mimetics. Phonological Studies, 10, 3–10.
Akita, Kimi (2009). A Grammar of Sound-Symbolic Words in Japanese: Theoretical Approaches to Iconic and Lexical Properties of Mimetics. Ph.D. dissertation, Kobe University.
Akita, Kimi (2010a). A collocational analysis of Japanese mimetics as special frame evokers. Ms., University of California, Berkeley.
Akita, Kimi (2010b). A constructionist analysis of “euphonic” verbs in Japanese. Ms., University of California, Berkeley.
Akita, Kimi (2010c). An embodied semantic analysis of psychological mimetics in Japanese. Linguistics, 48, 1195–1220.
Booij, Geert (2010). Construction Morphology. Oxford University Press, Oxford.
Brugman, Claudia (1988). The Story of Over: Polysemy, Semantics and the Structure of the Lexicon. Garland Press, New York/London.
Davis, Stuart & Ueda, Isao (2002). Mora augmentation processes in Japanese. Journal of Japanese Linguistics, 18, 1–23.
Fillmore, Charles J. & Kay, Paul (1995). Construction Grammar Coursebook. Ms., University of California, Ber- keley.
Goldberg, Adele E. (1995). Constructions: A Construction Grammar Approach to Argument Structure. University of Chicago Press, Chicago.
Hamano, Shoko (1998). The Sound-Symbolic System of Japanese. CSLI Publications, Stanford, Calif.
Ibarretxe-Antuñano, Iraide (2005). Leonard Talmy: A windowing to conceptual structure and language, Part 1: Lexicalization and typology. Annual Review of Cognitive Linguistics, 3, 325–347.
Kadooka, Ken-Ichi (2007). Nihongo onomatope-goi-ni okeru keitaiteki-/on’inteki-taikeisei-ni tuite [On mor- phological and phonological systematicity in Japanese mimetics]. Kurosio Publishers, Tokyo.
Kageyama, Taro (2007). Explorations in the conceptual semantics of mimetic verbs. In Frellesvig, B., Shibatani, M., & Smith, J. C. (Eds.), Current Issues in the History and Structures of Japanese, pp. 27–82. Kurosio Pub- lishers, Tokyo.
Kakehi, Hisao, Tamori, Ikuhiro, & Schourup, Lawrence C. (1996). Dictionary of Iconic Expressions in Japanese. Mouton de Gruyter, Berlin/New York.
Kita, Sotaro (1997). Two-dimensional semantic analysis of Japanese mimetics. Linguistics, 35, 379–415.
Klamer, Marian (1999). Austronesian expressives and the lexicon. Toronto Working Papers in Linguistics, 16, 201–219.
Kuroda, Kow & Lee, Jaeho (2005). Kekka-koobun-ni okeru kekka-no <hyookasya>-no yakuwari-no zyuuyoosei:
<Kekka>-towa <dare-ni totte>-no, <donna> henka-no kekka-nanoka? [The importance of the role of result
“evaluator” in the resultative construction: Result “for who” and of “what” change?]. Ms., National Institute of Information and Communications Technology.
Kuroda, Shige-Yuki (1967). Sokuon oyobi hatuon-ni tuite [On soku-on and hatu-on]. Gengo kenkyuu, 50, 85–99. Kuroda, Shige-Yuki (1979). Generative Grammatical Studies in the Japanese Language. Garland Press, New
York/London.
Lu, Chiarung (2006). Giongo/gitaigo-no hiyuteki-kakutyoo-no syosoo: Ninti-gengogaku-to ruikeiron-no kan- ten-kara [Aspects of figurative extensions of mimetics: From cognitive linguistic and typological perspec- tives]. Ph.D. dissertation, Kyoto University.
McCawley, James D. (1968). The Phonological Component of a Grammar of Japanese. Mouton, The Hague. Mikami, Kyoko (2006). Nihongo-no giongo/gitaigo-ni okeru imi-no kakutyoo: Konsekiteki-ninti/yokiteki-ninti-no
kanten-kara [Semantic extension in Japanese mimetics: From the perspective of vestigial/prospective cogni- tion]. Studies on Japanese Language and Japanese Literature, 57, 199–217.
Moriyama, Keiko (2002). CVQCVri-kei-no onomatope [CVQCVri-type onomatopes in Japanese]. Journal of International Student Center, 10, 53–71. Nagasaki University.
Nasu, Akio (1999). Onomatope-ni okeru yuuseika-to [p]-no yuuhyoosei [Mimetic voicing and the markedness of [p]]. Journal of the Phonetic Society of Japan, 3, 52–66.
Nasu, Akio (2002). Nihongo onomatope-no gokeisei-to inritu-koozoo [Word formation and prosodic structure of Japanese mimetics]. Ph.D. dissertation, University of Tsukuba.
Nasu, Akio (2004). Inritu-settyuuzi-to sasetusei: Nihongo onomatope-no kyootyoogo-keisei [Prosodic infixes and left-adjacency: Emphatic word formation of Japanese mimetics]. Nihongo-to nihon-bungaku, 38, 1–14. University of Tsukuba.
Nasu, Akio (2007). Onomatope-no gomatu-sokuon [The word-final moraic obstruent in Japanese mimetics]. Journal of the Phonetic Society of Japan, 11, 47–57.
Tamori, Ikuhiro (1984). Japanese onomatopoeias: Manner adverbials vs. resultative adverbials. Journal of Cul- tural Science, 20, 163–178. Kobe University of Commerce.
Toratani, Kiyoko (2006). On the optionality of to-marking on reduplicated mimetics in Japanese. In Vance, T. J. & Jones, K. (Eds.), Japanese/Korean Linguistics, 14, pp. 415–422. CSLI Publications, Stanford, Calif.
Toratani, Kiyoko (to appear). Hukusi-onomatope-no tokusyusei-ni tuite [On peculiarities of adverbial mimetics]. Tsujimura, Natsuko & Davis, Stuart (2008). A construction approach to innovative verbs in Japanese. Paper pre-
sented at the Fifth International Conference on Construction Grammar, the University of Texas at Austin, 26 September 2008.
Usuki, Takeshi (2010). Onomatope-kara miru gengo-ruikeiron: “Yootai/keiro-hennyuu”-ni kansuru koosatu [Language typology viewed from mimetics: Notes on “manner/path conflation”]. Paper presented at the Eighth Fukuoka University Linguistics Colloquium, Fukuoka University, 13 September 2010.
Yamaguchi, Nakami (2002). Inu-wa “biyo”-to naite ita: Nihongo-wa giongo/gitaigo-ga omosiroi [Dogs used to cry “biyo”: It is mimetics that are interesting in Japanese]. Kobunsha, Tokyo.
パンをこんがりと
．
焼く理由
―いわゆる強調副詞的擬態語の構文形態音韻論的分析― 秋田喜美
日本学術振興会／東京大学 要旨
本稿は、日本語における「ざっくり」や「ふんわり」などの強調副詞的擬態語の意
味特性を構文文法の観点から考察するものである。これらのオノマトペは、CVCCV^ri
という抽象的な形態音韻レベルで、ある種の構文として特定の意味と対を成している と考えられる。この考えに立つことにより、いくつかの指摘が統一的に可能となる。
まず、この形態音韻構文には、構文文法でしばしば重要視される非合成性のみなら
ず、合成性ないし構成要素の重要性が認められる。つまり、C2の有声性により相補分
布を見せる第二拍（強調子音）は、その 81%が、撥音よりも強調性と直接的に結び付
くと考えられる促音として実現する（例：ぐっさり）。このような有声性分布はそれ以
外のオノマトペのC2には見られないため、強調子音という部品の分節的実現が担う音
象徴的役割を示す現象と言える。一方で、語根を共有する語の頻繁な欠如（例：げん
なりvs. *げなげな、*げなり）や、語根を共有する語との意味的隔たり（例：ちょっき
り（正確性）vs. ちょきちょき、ちょきん（切断））といった形式的・意味的予測不可
能性は、当該構文の非合成的側面を例示するものである。
また、同構文には固有な面と一般的な面の両方が見られる。例えば、「あんまし」等
の一般語の如く、口語において語尾の音変化を許すもの（例：がっくし；cf. *がくし）
が存在する点からは、この種の語のオノマトペ性・類像性の低さが窺われる。こうし た独自性は、本構文がオノマトペという類像的語彙に恣意性が体系的に入り込む余地 として機能していることを示唆する。この恣意性は、強調副詞的擬態語の意味的一般
化を阻んできた主要因と言えよう。一方、「ばっさり」が切断量の多さという結果に加
えて、切断の思い切りのよさという様態を暗示するように、強調副詞的擬態語には包 括的事象描写というオノマトペ一般の意味特性も観察される。構文文法的見地からす ると、純粋に結果状態を表すと考えられる「こんがり」のような語までもが、一般に
様態を標示すると想定される「と」を取るのは（例：パンをこんがり｛と／*に｝焼い
た）、それが具現化する形態音韻構文が、このように様態と結果を包括する全体的事象
スキーマと結び付いているためと説明される。
本構文の存在は、「日本語の巨大なオノマトペ語彙は、有限個の生産的形態音韻構文
（例：重複構文、接尾辞構文）により体系的に構築されている」という、より一般的 な仮説を支持するものである。
