1.Introduction

Extensive work has been carried out for grammatical phenomena in dialectal and diachronic varieties of Japanese, but there is another type of variety which has received little attention in theoretical work: Wakamono Kotoba ‘youth language.’ This nonstandard variety of Japanese is mainly used by the teens and twenties (Horio 2015Horio, Kei 2015  Wakamono Kotoba-ni-mirareru Gengo Henka-ni-kansuru Kenkyū [A study on language variations and changes in Japanese Wakamono Kotoba ]. Ph.D. diss., Kyushu University., 14), and it has been hotly debated in sociolinguistics (Coulmas 2013Coulmas, Florian 2013 Sociolinguistics, 2nd edn. Cambridge: Cambridge University Press. , 68–70). Still, its grammatical and functional properties have been largely neglected in theoretical studies. An exception is Seraku and Akiha (2019)Seraku, Tohru, and Takako Akiha 2019 “ Poi in Japanese Wakamono Kotoba ‘Youth Language’.” Lingua 224: 1–15. , who describe formal and functional properties of the modality-related, clitic-like item poi, and articulate an account within Evaluative Morphology (Grandi and Körtvélyessy 2015Grandi, Nicola and Livia Körtvélyessy (eds) 2015 Edinburgh Handbook of Evaluative Morphology. Edinburgh: Edinburgh University Press.).

The present inquiry provides another theoretically-informed case study of Japanese youth language. We reveal morphosyntactic and functional properties of the nominalizing suffix -mi attested in the social media Twitter,11.It is widely held that -mi is an item of Wakamono Kotoba. The age group of twitterers who use -mi is not specified in previous studies, but our data suggest that it largely coincides with the age range for Wakamono Kotoba speakers (i.e. the teens and twenties). Twitter, however, sets a unique environment for the users; it allows only 140 Japanese characters per post, and the posts are open to the public (Uno 2015, 106). We thus assume that the Twitter register constitutes a subvariety of Wakamono Kotoba. and account for them within the theory of nominalization recently proposed in Shibatani (2017Shibatani, Masayoshi 2017 “Nominalizations.” In The Handbook of Japanese Syntax, ed. by Masayoshi Shibatani, Shigeru Miyagawa, and Hisashi Noda, 271–331. Berlin: De Gruyter Mouton. , 2018aShibatani, Masayoshi 2018a “Nominalization.” In The Cambridge Handbook of Japanese Linguistics, ed. by Yoko Hasegawa, 431–462. Cambridge: Cambridge University Press., 2018bShibatani, Masayoshi 2018b “Nominalization in Crosslinguistic Perspective.” In The Handbook of Japanese Contrastive Linguistics, ed. by Prashant Pardeshi and Taro Kageyama, 345–410. Berlin: Walter de Gruyter. , 2019Shibatani, Masayoshi 2019 “What is Nominalization?” In Nominalization in Languages of the Americas, ed. by Roberto Zariquiey, Masayoshi Shibatani, and David Fleck, 15–167. Amsterdam/Philadelphia: John Benjamins. ).

The nominalizer -mi is originally used in the standard variety of Japanese, as in ama-mi ‘sweetness’ (cf. ama-i ‘sweet’). For the sake of brevity, -mi in the standard variety is notated as -mi _SV, with the unmarked notation -mi being saved for -mi in Wakamono Kotoba. As will be argued, -mi shows distinct grammatical and functional properties from those of -mi _SV. For instance, -mi is attached to an adjective stem which resists -mi _SV. Consider (1).

The form tsura-mi is illegitimate in the standard variety of Japanese. According to Uno (2015Uno, Nagomi 2015 “Twitter-niokeru “Atarashī Mi-kei” [The new mi-form attested in Twitter ].” Kokubun 123: 94–106., 99), who targets only adjective-based mi-forms, their first attestation year in Twitter is 2007. At the time, only two forms, tsura-mi and nemu-mi ‘sleepy-nmlz,’ were used, but the number of mi-forms has expanded considerably in terms of type and token. Uno (2015Uno, Nagomi 2015 “Twitter-niokeru “Atarashī Mi-kei” [The new mi-form attested in Twitter ].” Kokubun 123: 94–106., 99) states that as of 2014, there were 225 types of adjective-based mi-forms, and nine of them (e.g. tsura-mi, nemu-mi, kawai-mi ‘cute-nmlz,’ kowa-mi ‘scary-nmlz’) have more than 1,000 tokens, all of which are still used in Twitter. There are several motives for using -mi in Wakamono Kotoba, including the writer’s desire to render his/her message look cuter, the writer’s intention to express his/her feeling euphemistically, and the writer’s attempt to soften his/her wording to mitigate its possible negative impact on the reader. In (1), hontōni tsura-mi-ga fuka-i conveys the writer’s depressed feeling in a more roundabout way than the mi-absent sequence hontōni tsura-i ‘really depressing.’ Moreover, unlike -mi _SV, -mi attaches not only to an adjective stem but also to a verb, a noun, or even a phrase. There are further unique properties of -mi, and our claim is that these properties are naturally captured in Shibatani’s theory of nominalization, with a number of empirical and theoretical consequences.

We first review previous studies in Section 2, setting out the stage for our analysis to be developed in Sections 3–4. Shibatani’s view of nominalization is introduced in Section 3, and it is applied to spontaneous data of -mi in Section 4. Empirical and theoretical implications of the proposed account will be pointed out in Section 5.

2.Previous studies

While there is a wealth of studies on -mi _SV (Itoh and Sugioka 2002Itoh, Takane, and Yoko Sugioka 2002  Go-no Shikumi-to Gokēsei [The mechanism of word and word formation]. Tokyo: Kenkyusha.; Kageyama 1993Kageyama, Taro 1993  Bunpō-to Gokēsei [Grammar and word formation]. Tokyo: Hituzi Syobo Publishing.; Shibatani 2018aShibatani, Masayoshi 2018a “Nominalization.” In The Cambridge Handbook of Japanese Linguistics, ed. by Yoko Hasegawa, 431–462. Cambridge: Cambridge University Press.; Sugioka 2005Sugioka, Yoko 2005 “Meishika Setsuji-no Kinō-to Imi [The functions and meanings of nominalizing suffixes].” In Gendai Keitairon-no Chōryū, ed. by Tsuyoshi Ohishi, Yoji Toyoshima, and Tetsuo Nishihara, 75–94. Tokyo: Kuroshio Publishers., inter alia), little work has been conducted for -mi (Kato et al. 2018Kato, Eri, Sanae Yamashita, and Yasushi Kami 2018 “Twitter-de Tsukawareru Fukai-no Imi [The meanings of fukai used in Twitter ].” Proceedings of the Workshop on Gengo Shigen Katsuyō 3: 274–279.; Mizuno 2017Mizuno, Minori 2017 “Netto Shūdango-niokeru Setsubiji -Mi-no Goki Kakuchō [Extension of the base for the suffix -mi in the Internet language].” Shigen 13: 167–174.; Uno 2015Uno, Nagomi 2015 “Twitter-niokeru “Atarashī Mi-kei” [The new mi-form attested in Twitter ].” Kokubun 123: 94–106., 2017Uno, Nagomi 2017 “Twitter-de mirareru Meishi-ni Kōsetsusuru Setsubiji Mi [The suffix -mi attached to a noun attested in Twitter ].” Journal of the Graduate School of Humanities and Sciences 20: 29–37., 2018Uno, Nagomi 2018 “Twitter-ni miru Onomatope-ni Kōsetsusuru Setsubiji Mi-no Kinō [Functions of the suffix -mi attached to an onomatopoeia attested in Twitter ].” Comparative Japanese Studies Annual Bulletin 14: 183–189.; Yoda 2016Yoda, Ayano 2016 “Twitter-ni Mochiirareru -Mi-no Yōhō [The usage of -mi used in Twitter ].” Shindai Kokugo Kyōiku 26: 1–12.). Presumably, this is partly because -mi is a recent innovation in Wakamono Kotoba and partly because grammatical analyses are usually not attempted for expressions/constructions in Wakamono Kotoba.

The nominalizer -mi _SV is often compared with the nominalizer -sa in the literature. It has been widely held that (i) -sa is much more productive than -mi _SV and (ii) the sa-nominalized element tends to denote an objective entity, and the mi _SV-nominalized element tends to denote a subjective entity. As for point (i), -sa is attached to all types of native adjective stem (e.g. fuka-sa ‘depth’ < fuka-i ‘deep’) and many nominal-adjective stems (e.g. jūdai-sa ‘importance’ < jūdai-da ‘important’), but -mi _SV is only attachable to a limited number of native adjective stems (e.g. yowa-mi ‘weak point’ < yowa-i ‘weak’). Point (ii) is illustrated with (2)–(3).

In both cases, the nominalizer is attached to the stem of the same adjective fuka-i ‘deep.’ When -sa is applied to fuka, the resulting form fuka-sa means ‘depth.’ This sharply contrasts with the mi _SV-nominalized counterpart fuka-mi, which means ‘profoundness.’

The morphosyntactic and functional properties of -mi _SV crucially differ from those of -mi in Wakamono Kotoba. Firstly, -mi may be attached to a much wider range of items than -mi _SV, such as the verb neru ‘sleep’ in (4).

Secondly, as has been stated in Section 1, -mi is equipped with various discourse functions (e.g. rendering the writer’s message look cuter, expressing his/her feeling euphemistically). In this paper, we focus on the unique morphosyntactic properties of -mi, but we also consider functional properties of -mi when they are related to the morphosyntactic discussion of -mi.

As for point (i), Mizuno (2017)Mizuno, Minori 2017 “Netto Shūdango-niokeru Setsubiji -Mi-no Goki Kakuchō [Extension of the base for the suffix -mi in the Internet language].” Shigen 13: 167–174. is noteworthy among the previous studies, which offers a wide array of examples as well as some remarks on them. These remarks, however, are rather sporadic and are not systematically presented in theoretical terms. To mention a few, Mizuno recognizes what we call the sentential use of a mi-nominalized element, but it is not discussed in line with other functions of nouns such as an NP-heading use and an NP-modification use (see Section 4). Moreover, some mi-nominalizations are developing a grammatical process, but this issue is not even touched upon.

Lack of a theoretical analysis finds itself in the other extant studies on -mi. Firstly, Yoda (2016)Yoda, Ayano 2016 “Twitter-ni Mochiirareru -Mi-no Yōhō [The usage of -mi used in Twitter ].” Shindai Kokugo Kyōiku 26: 1–12. and Uno (2017)Uno, Nagomi 2017 “Twitter-de mirareru Meishi-ni Kōsetsusuru Setsubiji Mi [The suffix -mi attached to a noun attested in Twitter ].” Journal of the Graduate School of Humanities and Sciences 20: 29–37. deal with only a single form involving -mi: kodomo-mi ‘child-nmlz’ and nemu-mi ‘sleepy-nmlz,’ respectively. They provide many examples, but it remains to be seen to what extent their claims are generalizable. Secondly, Uno (2015Uno, Nagomi 2015 “Twitter-niokeru “Atarashī Mi-kei” [The new mi-form attested in Twitter ].” Kokubun 123: 94–106., 2018Uno, Nagomi 2018 “Twitter-ni miru Onomatope-ni Kōsetsusuru Setsubiji Mi-no Kinō [Functions of the suffix -mi attached to an onomatopoeia attested in Twitter ].” Comparative Japanese Studies Annual Bulletin 14: 183–189.) and Kato et al. (2018)Kato, Eri, Sanae Yamashita, and Yasushi Kami 2018 “Twitter-de Tsukawareru Fukai-no Imi [The meanings of fukai used in Twitter ].” Proceedings of the Workshop on Gengo Shigen Katsuyō 3: 274–279. present several interesting data (e.g. example where -mi applies to an onomatopoeia), but their theoretical significances are not appreciated. In a nutshell, it is not obvious how the findings of these previous studies are construed in general linguistic terms and how they relate to the study of nominalization in Japanese and other languages.

In addressing these theoretical issues, we uncover several new data. For instance, based on Mizuno’s (2017Mizuno, Minori 2017 “Netto Shūdango-niokeru Setsubiji -Mi-no Goki Kakuchō [Extension of the base for the suffix -mi in the Internet language].” Shigen 13: 167–174., 170) finding that -mi may be attached to the finite form of a predicate, we observe variations in modal forms, such as yar-anakya-mi ‘do-must-nmlz.’ Further, based on the previous finding that -mi may be attached to not only an adjective stem but also a verb and a noun, we add that -mi may be attached to a pronoun, a demonstrative, and even an already-nominalized form.

Against these theoretical and empirical issues, we propose an account of -mi in the light of Shibatani’s view on nominalization. To this end, the next section (Section 3) introduces his theory, which will be applied to mi-nominalization examples in Section 4.

3.Framework

In a series of recent studies (2017, 2018aShibatani, Masayoshi 2018a “Nominalization.” In The Cambridge Handbook of Japanese Linguistics, ed. by Yoko Hasegawa, 431–462. Cambridge: Cambridge University Press., 2018bShibatani, Masayoshi 2018b “Nominalization in Crosslinguistic Perspective.” In The Handbook of Japanese Contrastive Linguistics, ed. by Prashant Pardeshi and Taro Kageyama, 345–410. Berlin: Walter de Gruyter. , 2019), Shibatani proposes a novel approach to nominalizations (see also Shibatani 2009Shibatani, Masayoshi 2009 “Elements of Complex Structures, where Recursion Isn’t.” In Syntactic Complexity, ed. by Thomas Givon and Masayoshi Shibatani, 163–198. Amsterdam/Philadelphia: John Benjamins. , 2013Shibatani, Masayoshi 2013 “What Can Japanese Dialects Tell Us about the Function and Development of the Nominalization Particle No?” In Japanese/Korean Linguistics 20, ed. by Bjarke Frellesvig and Peter Sells, 429–451. Stanford, CA: CSLI Publications.; Shibatani and bin Makhashen 2009Shibatani, Masayoshi, and Khaled Awadh bin Makhashen 2009 “Nominalization in Soqotri, a South Arabian Language of Yemen.” In The Linguistics of Endangered Languages, ed. by Leo Wetzels, 9–31. Leiden: Brill.; Shibatani and Shigeno 2013Shibatani, Masayoshi, and Hiromi Shigeno 2013 “Amami Nominalizations.” International Journal of Okinawan Studies 4 (1): 107–138.), radically departing from the standard notion of nominalization (Comrie and Thompson 2007Comrie, Bernard, and Sandra Thompson 2007 “Lexical Nominalization.” In Language Typology and Syntactic Description, vol. 3, 2nd edn., ed. by Timothy Shopen, 334–381. Cambridge: Cambridge University Press. ; Yap et al. 2011Yap, Foong Ha, Karen Grunow-Hårsta, and Janick Wrona (eds) 2011 Nominalization in Asian Languages. Amsterdam/Philadelphia: John Benjamins. ). Shibatani views nominalizations as metonymic phenomena; the nominalized structure evokes an entity-type denotation metonymically tied to the denotation of the base. Firstly, “entity” is broadly construed to include an individual, an event, a proposition, and so on. An entity-type denotation is a basis for reference. For instance, the nominalized form singer denotes ‘a person who sings,’ which is a basis for referring to a certain individual in context. This contrasts with the verb sing, which denotes a relational concept and serves as a basis for predication. Secondly, “metonymy” yields various form-concept pairs through contiguity schemas such as ‘the event for the protagonists’; employer, for instance, denotes the agentive protagonist of the employing event.

Nominalizations are classified on two grounds. Firstly, the product of nominalization is “lexical” if the product is a noun, subject to a part-of-speech categorization, or “grammatical” if the product is larger in size than words. Secondly, the input is “verbal” if the base is a verb or an adjective, or “nominal” if the base is a noun. These classifications give rise to the four types of nominalization in Table 1. (Verbal/adjectival-based and nominal-based are referred to as “V-based” and “N-based,” respectively.)

While nominalization is generally treated as a process of turning a non-nominal element to a nominal element, Shibatani holds that the input to the process may be nominal, as exemplified in London > Londoner. One of the most innovative aspects of his approach is N-based grammatical nominalization. For Ken > Ken’s, the nominalized form Ken’s denotes an entity metonymically related to Ken, such as the book bought by Ken and the car owned by Ken, with an intended interpretation being chosen through pragmatic inference. In this analysis, the so-called “genitive” construction is merely an instance of nominalization.

Shibatani’s account uniformly handles the four types of nominalization in Table 1. This is morphosyntactically reflected in many languages, where the same morpheme appears in more than one type of nominalization. In Japanese, -no is employed for both V-based and N-based grammatical nominalizations, as illustrated in (5)–(6).22.Shibatani argues that -no is a “nominalization marker,” the idea being that a nominalization takes place independently of -no. Shibatani (2017Shibatani, Masayoshi 2017 “Nominalizations.” In The Handbook of Japanese Syntax, ed. by Masayoshi Shibatani, Shigeru Miyagawa, and Hisashi Noda, 271–331. Berlin: De Gruyter Mouton. , 290) supports this claim by mentioning Izumo Japanese, which does not deploy a particle like -no in examples like (5). See Shibatani (2019)Shibatani, Masayoshi 2019 “What is Nominalization?” In Nominalization in Languages of the Americas, ed. by Roberto Zariquiey, Masayoshi Shibatani, and David Fleck, 15–167. Amsterdam/Philadelphia: John Benjamins. for more data from diverse languages.

The distinction between lexical and grammatical nominalizations is in principle clear; the lexically nominalized element is a noun, registered in the lexicon, while the grammatically nominalized element is phrasal, created for the nonce. Still, the lexical/grammatical divide is elusive in practice. Table 1 cites singing cheerfully as a case of grammatical nominalization, but singing is an instance of lexical nominalization in cheerful singing, where the adjective cheerful (rather than the adverb cheerfully) is indicative of its noun status. Shibatani (2019Shibatani, Masayoshi 2019 “What is Nominalization?” In Nominalization in Languages of the Americas, ed. by Roberto Zariquiey, Masayoshi Shibatani, and David Fleck, 15–167. Amsterdam/Philadelphia: John Benjamins. , Section 2.2) also states that the lexical/grammatical divide forms a continuum. In the present article, we distinguish lexical nominalizations from grammatical ones in terms of (i) whether the product is a word or phrasal and (ii) what morphosyntactic properties the product has (e.g. modified by an adjective or an adverb). If points (i)–(ii) cannot be elucidated sufficiently, we will not specify whether the nominalized structure is lexical or grammatical.33.Apart from points (i)–(ii), Shibatani (2019Shibatani, Masayoshi 2019 “What is Nominalization?” In Nominalization in Languages of the Americas, ed. by Roberto Zariquiey, Masayoshi Shibatani, and David Fleck, 15–167. Amsterdam/Philadelphia: John Benjamins. , Section 2.2) sets out other criteria, but they are tangential to our data. For instance, Shibatani points out that lexical nominalizations tend to be overtly marked and that grammatical nominalizations are zero-marked in many cases. This criterion is not pertinent to our examples, where the nominalized structures always contain -mi.

In the remainder of this section, we review further properties of lexical and grammatical nominalizations and then point out some residual theoretical issues.

3.1Lexical nominalization

Shibatani (2018aShibatani, Masayoshi 2018a “Nominalization.” In The Cambridge Handbook of Japanese Linguistics, ed. by Yoko Hasegawa, 431–462. Cambridge: Cambridge University Press., Section 3) provides a number of examples for lexical nominalization. Sino-Japanese suffixes such as -ka and -shu perform N-based nominalizations, as in shōsetsu-ka ‘novel-nmlz’ (= ‘novelist’) and unten-shu ‘drive-nmlz’ (= ‘driver’). What he calls “stem-nominalization,” which attaches -i or Ø (zero) to a verb stem, realizes V-based nominalization, as in suber-i ‘slide-nmlz’ (= ‘sliding’) and nagare-Ø ‘flow-nmlz’ (= ‘flowing’).

In lexical nominalizations, the evoked concepts tend to be stable across contexts. Thus, unten-shu ‘drive-nmlz’ denotes ‘driver,’ and it cannot mean, e.g., the act of driving in any context. Some ambiguity, however, is detected in stem-nominalizations; suri-Ø ‘pickpocket-nmlz’ has the agentive reading ‘pickpocket’ and the activity reading ‘pickpocketing.’ Still, as stem-nominalization is a lexical process, there are irregular gaps in the meaning pattern. Thus, yama-nobor-i ‘mountain-climb-nmlz’ has the activity reading ‘mountain climbing,’ but not the agentive reading ‘mountaineer.’

In Shibatani’s terms, -mi _SV is said to manifest V-based lexical nominalization.

(7)

Kono

this

shōsetsu-wa

novel-top

fuka-mi-ga

deep-nmlz-nom

aru.

exist

‘This novel is profound.’

The nominalized form fuka-mi denotes an entity metonymically related to the denotation of the base fuka-i ‘deep.’ One such an entity, as evoked in the context of (7), is ‘profoundness.’ In another context, such as (8), the mi _SV-nominalized element receives a different reading.

(8)

Kawagishi-o

riverside-acc

aruitei-te

walk.ipfv-cvb

fuka-mi-ni

deep-nmlz-all

ochita.

fell

‘When I was walking along the riverside, I fell into a deep place (of the river).’

In this way, fuka-mi allows an interpretational leeway, as in the cases of stem-nominalization, but the denotations of mi _SV-nominalized forms are relatively stable across contexts, compared with grammatical nominalization, to which we now turn in the next subsection.

3.2Grammatical nominalization

Though the product of grammatical nominalization cannot be equated with a noun, it behaves like a noun in denoting an entity-type concept. More precisely, it may head an NP, modify an NP, or modify a predicate. We will call these noun-like uses the “NP-heading use,” the “NP-modification use,” and the “adverbial use,” respectively.44.In several places of his writings, Shibatani calls them the “NP-use,” the “modification-use,” and the “adverbial-use,” respectively. These are in principle irrelevant to lexical nominalizations since their products are nouns. Given that the lexical/grammatical divide is blurred, however, we hold that such uses are relevant to nominalizations unless they are straightforward cases of lexical nominalizations.

When the product of grammatical nominalization displays an NP-heading use, it behaves like a noun in that it is case-marked and bears a grammatical relation (e.g. subject, object).

(9)

The doctor said that singing cheerfully is good for health. <V-based nominalization>

(10)

If you need a car, why don’t you borrow Ken’s? <N-based nominalization>

In (9), singing cheerfully is nominative-marked and functions as the subject of is good for health. In (10), Ken’s is accusative-marked and functions as the object of borrow.

When a noun modifies another noun, it typically restricts the denotation of the modified noun. In dog food, where dog modifies food, the set of the entities referred to by dog food forms a proper subset of the set of the entities referred to by food. This modification function is also seen in the grammatically-nominalized structures in (11)–(12).

(11)

Do you know the men who are singing cheerfully there? <V-based nominalization>

(12)

Ken’s car is expensive. <N-based nominalization>

In (11), who are singing cheerfully there is usually seen as a relative clause, but for Shibatani, it is an NP-modification use of V-based grammatical nominalization. That is, as dog modifies food in dog food, the nominalized element are singing cheerfully there modifies the NP men, restricting its denotation. The so-called relative pronoun who is thought of as a nominalization marker (or a nominalizer). In (12), the nominalized element Ken’s modifies car, so that Ken’s car refers to the car owned (or bought, etc.) by Ken.

The adverbial use is illustrated with the suffix -sa (see Section 2). Shibatani (2018aShibatani, Masayoshi 2018a “Nominalization.” In The Cambridge Handbook of Japanese Linguistics, ed. by Yoko Hasegawa, 431–462. Cambridge: Cambridge University Press., 448) observes that -sa in contemporary Japanese marginally exhibits an adverbial use when it is combined with a grammatical particle such as -ni. Consider (13).

(13)

Takashi-wa

T-top

[hon-o

[book-acc

yomi-ta-sa ]-ni

read-des-nmlz]-prt

toshokan-ni

library-to

itta.

went

‘Takashi went to the library out of his desire of reading books.’ (Shibatani 2018aShibatani, Masayoshi 2018a “Nominalization.” In The Cambridge Handbook of Japanese Linguistics, ed. by Yoko Hasegawa, 431–462. Cambridge: Cambridge University Press., 448)

Apart from the adverbial use, -sa has neither an NP-heading nor an NP-modification use. In this respect, grammatical nominalizations with -sa are atypical; in contemporary Japanese, -sa is typically used to achieve lexical nominalisations. Shibatani then goes on to speculate that grammatical sa-nominalizations may have been productive in the past, putting forward the non-discrete conception of lexical and grammatical nominalizations. In connection with this, Shibatani (2019Shibatani, Masayoshi 2019 “What is Nominalization?” In Nominalization in Languages of the Americas, ed. by Roberto Zariquiey, Masayoshi Shibatani, and David Fleck, 15–167. Amsterdam/Philadelphia: John Benjamins. , 124–125) suggests that grammatical nominalizations are likely to give rise to lexical nominalizations in the course of time, and mentions that the opposite cases need to be documented where a nominalizer (or a nominalization marker) for lexical nominalizations is extended to instantiate grammatical nominalizations.

3.3Residual issues

This section has outlined Shibatani’s novel approach to nominalizations. We presume that his account is further developed if we tackle the following issues:

(14)

Theoretical issues

1. While Shibatani (2019Shibatani, Masayoshi 2019 “What is Nominalization?” In Nominalization in Languages of the Americas, ed. by Roberto Zariquiey, Masayoshi Shibatani, and David Fleck, 15–167. Amsterdam/Philadelphia: John Benjamins. , 124–125) suggests that grammatical nominalizations are likely to engender lexical nominalizations, can we find cases where lexical nominalizations engender grammatical nominalizations?

2. Shibatani argues for V-based and N-based nominalizations. Are there any other types of nominalization where the base is neither V(erbal/adjectival) nor N(ominal)?

3. Shibatani focuses on an NP-heading use, an NP-modification use, and an adverbial use of a nominalized element, but are there any other noun-comparable uses?

These questions are fruitfully answered if we investigate morphosyntactic and functional characteristics of mi-nominalizations, as will be demonstrated in the next section.

4.Analysis

First of all, it should be explicated why we adopt Shibatani’s framework. There are several motivations for this. Firstly, a novel feature of his theory is its integration of V-based and N-based nominalizations, and it is this very feature that allows us to uniformly handle examples where -mi is attached not only to an adjective/verb but also to a noun. Secondly, his account presupposes the non-discrete view of lexical and grammatical nominalizations. This is suited for describing our examples, where -mi realizes lexical and grammatical nominalizations with varying degrees of nouniness. Some mi-nominalized items show a number of noun-properties and are seen as typical cases of lexical nominalizations, and others show few noun-properties and are considered typical cases of grammatical nominalizations.

We examine V-based and N-based nominalizations in Sections 4.1–4.2. Peculiar features of mi-nominalizations are further highlighted by additional data in Section 4.3. Our data come from Twitter, a primary locus of mi-nominalizations (Uno 2015Uno, Nagomi 2015 “Twitter-niokeru “Atarashī Mi-kei” [The new mi-form attested in Twitter ].” Kokubun 123: 94–106.).55.In extracting mi-examples, I entered in Twitter’s “advanced search” box the sequence mi-(case)-(predicate). I then pick out noteworthy examples, rather than collecting examples in a systematic and statistical fashion. This is because our aim is not to present a quantitative analysis but to unearth empirically and/or theoretically significant features of mi-nominalizations. Following previous studies, we only consider examples where -mi is notated in hiragana (i.e. み).66.The Japanese writing system includes hiragana (mainly used for native words), katakana (mainly used for foreign words), and kanji (adapted Chinese characters). Twitter contains examples where -mi is notated in kanji (e.g. 味, 身) or katakana (i.e. ミ), but there are good reasons for assuming that -mi was originally notated in hiragana and came to be notated in kanji and katakana as well. Firstly, nemu-mi with hiragana -mi appeared in 2007 (Uno 2015Uno, Nagomi 2015 “Twitter-niokeru “Atarashī Mi-kei” [The new mi-form attested in Twitter ].” Kokubun 123: 94–106.), while nemu-mi with non-hiragana -mi emerged later: 2009 for 味 (literally meaning ‘taste’), 2011 for 身 (literally meaning ‘body’), and 2010 for ミ. Secondly, there are cases of -mi such as mijika-mi ‘short-nmlz,’ where -mi is notated only in hiragana (searched on 16/May/2020). I thank one of the reviewers for bringing my attention to this notational issue.

4.1V-based nominalization

V-based nominalizations encompass structures derived through a verb or an adjective. In fact, as we shall see below, -mi is attached to both types of predicate.

4.1.1Adjective (stem)

Unlike -mi _SV, -mi is attached to various types of adjective, including not only native adjective stems (see (1)) but also Sino-Japanese nominal-adjective stems such as shiawase- ‘happy’ in (15) and foreign loan nominal-adjective stems such as hansamu- ‘handsome’ in (16).

(15)

[The writer posts the photo where he and his friends enjoy eating Chinese meals in a restaurant, and says that it was a very enjoyable moment.]

Shiawase-mi-shika

happy-nmlz-npi

nakat-ta.

exist.neg-pst

‘I didn’t feel anything but happiness.’ [11/Mar/2017]

(16)

[The writer posts the photo of a male character of a Japanese computer game.]

Hansamu-mi-o

handsome-nmlz-acc

kanjiru-nda-yo-nā.

feel-mm-fp-fp

‘I am enchanted by his handsomeness.’ [04/Apr/2019]

In these examples, the denotation of the nominalized structure is metonymically related to that of the base. To take (15) as an example, shiawase-mi denotes the concept ‘happiness,’ which relates to the property ‘happy.’

In (15)–(16), -mi is attached to an adjective stem. Interestingly, -mi may be attached even to a finite form of an adjective (see Mizuno 2017Mizuno, Minori 2017 “Netto Shūdango-niokeru Setsubiji -Mi-no Goki Kakuchō [Extension of the base for the suffix -mi in the Internet language].” Shigen 13: 167–174., 170 for related data).

(17)

[The writer posts the photo of the trout which he caught by himself.]

Jimoto-de

hometown-loc

yamame

trout

tsur-e-ta!!

catch-pot-pst

Metcha

very

ureshi-i-mi-desu.

happy-npst-nmlz-cop

‘I caught trout in a river of my hometown! I am really happy.’ [14/Apr/2019]

(18)

[The writer went out with his friends. He posts the photo which he took then, where he and his friends are smiling.]

Tanoshikat-ta-mi-ga

enjoyable-pst-nmlz-nom

fukai.

deep

‘It was so enjoyable.’ [25/Apr/2019]

In (17), -mi is attached to the non-past form ureshi-i ‘happy-npst,’ whereas in (18), -mi is attached to the past form tanoshikat-ta ‘enjoyable-pst.’

The mi-forms in (15)–(16) are viewed as cases of lexical nominalization because they do not have a phrasal structure. On the other hand, the mi-forms in (17)–(18), which are phrasal, may be regarded as instances of grammatical nominalization. In (18), tanoshi-i ‘enjoyable’ is conjugated as the past form, constructing an adjective phrase. A more telling example is (17), where ureshi-i-mi is modified by the degree adverb metcha ‘very.’ This is indicative of its phrasal status. Examples (17)–(18), however, are less frequently attested than those where -mi is attached to an adjective stem.77.A quick Google search shows that ureshi-mi-ga (with the stem ureshi-) has 220,000 hits and ureshi-i-mi-ga (with the finite form ureshi-i) has 282 hits (accessed on 21/Aug/2019). The nominative case particle -ga was included in the search in order to exclude irrelevant sequences such as ureshi-i-mitai ‘happy-npst-seem’ (= ‘seem happy’). Together with the fact that -mi _SV instantiates only lexical nominalizations, it may indicate that -mi started to be used for lexical nominalizations alone but that its grammatical-nominalizing function has currently been emerging.

As mentioned in Section 3.2, nominalized structures exhibit an NP-heading use, an NP-modification use, and an adverbial use. In (18), tanoshikat-ta-mi displays an NP-heading use in that it is nominative-marked and functions as the subject of fuka-i ‘deep.’ In (17), however, metcha ureshi-i-mi cannot be construed in virtue of any of these three uses. Rather, this mi-nominalized structure seems to be comparable to a one-word sentence, where the utterance of a word performs a speech act of various kinds. Shibatani (2017Shibatani, Masayoshi 2017 “Nominalizations.” In The Handbook of Japanese Syntax, ed. by Masayoshi Shibatani, Shigeru Miyagawa, and Hisashi Noda, 271–331. Berlin: De Gruyter Mouton. , footnote 25) briefly notes this function of nouns, stating “nouns also function as a sentence (so-called one-word sentences) when it is used to perform a speech act, as in the act of warning in an utterance like ‘Fire!’.” Following Shibatani, we functionally define ‘sentence’ as a grammatical unit to be utilized to perform a speech act (e.g. asserting, ordering, warning). In (17), metcha ureshi-i-mi, together with the copula -desu, functions as a sentence; the writer makes an expressive assertion, conveying his positive emotion toward his catching of fish. One might object that metcha ureshi-i-mi-desu in (17) is, strictly speaking, not comparable to a one-word sentence since it contains the adverb metcha and the copula -desu. This objection does not go through in (19), where mijika-mi ‘short-nmlz’ (without modifiers or the copula) conveys the writer’s emotion about how fast time flies.

(19)

[The writer looks back on this year on the New Year Eve.]

Kyō-de

today-temp

ichinen

one.year

owari-ka…

finish-fp

Mijika-mi .

short-nmlz

‘This year ends today… How short (the year is)!’ [31/Dec/2018]

Given that a one-word sentence is a sentence (rather than a phrase or a word), which is used to perform a specific speech act, we call this use of a nominalized structure in (17) and (19) the ‘sentential use’. Uno (2015Uno, Nagomi 2015 “Twitter-niokeru “Atarashī Mi-kei” [The new mi-form attested in Twitter ].” Kokubun 123: 94–106., 96) also notes in passing (but does not analyze) examples such as (19), observing that this type of example first appeared in 2010.

The sentential use of nominalized structures is closely linked to ‘insubordination’ (Evans 2007Evans, Nicholas 2007 “Insubordination and its Uses.” In Finiteness, ed. by Irina Nikolaeva, 366–431. Oxford: Oxford University Press.), or what Yap et al. (2011)Yap, Foong Ha, Karen Grunow-Hårsta, and Janick Wrona (eds) 2011 Nominalization in Asian Languages. Amsterdam/Philadelphia: John Benjamins. call the ‘stance construction.’ It has been cross-linguistically widely observed that a nominalized structure comes to stand as a full-fledged sentence over the course of time.

(20)

Ken-ga

K-nom

sore-o

that-acc

katta-no!

bought-nmlz

‘Ken bought that!’88.One of the reviewers stated that -no in (20) could be treated as a final particle. While this view is reasonable in contemporary Japanese, -no in examples like (20) is historically traced to a nominalizer (Wrona 2012Wrona, Janick 2012 “The Early History of No as a Nominalizer.” In Studies in Japanese and Korean Linguistics, ed. by Bjarke Frellesvig, Jieun Kiaer, and Janick Wrona, 201–220. London: Lincom.). Similar examples are found with -mi (see, e.g., (21)). In contrast to -no, this use of -mi has not been fully established, and it is thus less likely that -mi in such examples is viewed as a final particle. I thank the reviewer for helpful discussion regarding the present point.

In (20), the nominalized structure serves as a sentence, with which the speaker performs an assertive speech act. Compared with Ken-ga sore-o katta ‘Ken bought that,’ the degree of assertion in (20) is higher (Horie 2011Horie, Kaoru 2011 “Versatility of Nominalizations.” In Nominalization in Asian Languages, ed. by Foong Ha Yap, Karen Grunow-Hårsta, and Janick Wrona, 473–496. Amsterdam/Philadelphia: John Benjamins. , 403–406). Shibatani (2019Shibatani, Masayoshi 2019 “What is Nominalization?” In Nominalization in Languages of the Americas, ed. by Roberto Zariquiey, Masayoshi Shibatani, and David Fleck, 15–167. Amsterdam/Philadelphia: John Benjamins. , 95) sees this as a case of insubordination, remarking that “[t]here are several clear cases in which nominalizations, which usually do not stand as sentences, develop into sentences.” In this (as well as Evans’) notion of insubordination, the unit undergoing the process has a clausal structure. By contrast, in (19), the nominalized structure is based on an adjective, not a clause. We thus distinguish between the sentential use of a nominalized element and the insubordination of a nominalized element, though it is possible that the former is a motive for the development of the latter. In fact, Shibatani (2017Shibatani, Masayoshi 2017 “Nominalizations.” In The Handbook of Japanese Syntax, ed. by Masayoshi Shibatani, Shigeru Miyagawa, and Hisashi Noda, 271–331. Berlin: De Gruyter Mouton. , footnote 25) also suggests, though briefly, that one-word sentences may be a basis for the development of insubordination structures.

4.1.2Verb

We have seen that -mi is attached to either an adjective stem or the finite form of an adjective. As shown in (21), -mi may be attached to the finite form of a verb, too.99.Our data contain no clear example where -mi attaches to a verb stem. Also, the examples cited in this section become quite degraded if the finite verbal form is converted to the verb stem. It is not clear at present why -mi cannot attach to a verb stem; it might suggest that -mi first expanded its nominalizing functions for adjective stems and adjectives (see Section 4.1.1) and then developed nominalizing functions for verbs (but not verb stems).

(21)

[The writer feels tired because of summer heat and says that he wants to have a sleep after playing a computer game called Minecraft.]

Natsubate-kashira.

summer.fatigue-wonder

Mainkurafuto

M

yat-te

do-cvb

neru-mi .

sleep-nmlz

‘I think I am suffering from summer fatigue. I’ll go to sleep after playing Minecraft (computer game).’ [14/Aug/2019]

In the finite form ne-ru ‘sleep-npst,’ -ru is glossed as npst (non-past), but more strictly, ne-ru is a form for ‘non-past tense,’ ‘perfective aspect,’ and ‘declarative mood.’ As illustrated in (22)–(23), the verb may also present various modal forms.

(22)

[The writer says that he is so busy that he has no time for his hobby (drawing).]

Oekaki

drawing

shi-ta-sa

do-des-nmlz

aru-kedo

exist-but

nemuta-sa-mo

sleepy-nmlz-also

aru-shi

exist-and

kadai

assignment

yar-anakya-mi-mo

do-must-nmlz-also

kanji-te

feel-cvb

shin-deru.

die-ipfv

‘I want to draw a picture, but I am sleepy. I also feel the obligatoriness of doing school assignments and I am so exhausted.’ [03/Jul/2016]

(23)

[The writer asked his friend to go to an event together, but says on his friend’s Twitter post that he probably cannot go.]

Ore-ga

I-nom

ike-na-sou-mi-ga

go-neg-seem-nmlz-nom

fuka-i.

deep-npst

‘It is highly likely that I cannot go.’ [9/Aug/2019] 1010.One of the reviewers pointed out that -mi in (22)–(23) may be replaced with -kan ‘feeling.’ While such a replacement is possible, it would not be likely that -kan is viewed as an element of Wakamono Kotoba since the suffix appears in non-casual discourse. Example (i) is an excerpt from a book (Sato, F. 2018. Ima-yori Ii Kaisha-ni Shūshokusuru Hōhō. [The method of getting a job at a better company] Tokyo: CM Publishing.). (i) “Sōki-taishoku-seido”-toiu “early-retirement-system”-quot kotoba-mo expression-also jojoni gradually yononaka-ni world-to teichakushiteitta-kan-ga has.established-feeling-nom arimasu.exist.hon ‘It is fair to say that the expression early retirement system has also been gradually established in our society.’

Moreover, -mi is attached to a tai-derivative; -tai is an adjectivizer, which turns a verb into an adjective with a ‘desiderative’ interpretation. This is illustrated in (24), where hair-itakat-ta ‘enter-des-pst’ (-itakat < -tai) is also in the finite form.

(24)

[The writer, an undergraduate student, thinks of his academic major.]

Shōrai-no

future-gen

koto

thing

nukinisureba

setting.aside

shigaku-ka

history-department

hair-itakat-ta-mi-wa

enter-des-pst-nmlz-top

aru.

exist

Tokui-de

good.at-and

sukina

like

koto

thing

shi-takat-ta-nā.

do-des-pst-fp

‘If I do not take into account my future career, I wanted to study at the department of history. I wanted to do something which I am good at and also like.’ [15/Jul/2019] 1111.In Shibatani’s framework, the genitive construction is merely an instance of N-based grammatical nominalization (see Section 3). In (24), however, -no in Shōrai-no koto ‘future-gen thing’ is glossed as gen for the sake of simplicity. The same remark applies to -no in subsequent examples.

Given that all of (21)–(24) involve a verb (rather than a verb stem) and that these verbs select an argument (e.g. subject ore ‘I’ (23), object shigaku-ka ‘history department’ (24)), it is fair to say that the nominalized structures in (21)–(24) are phrasal. It then follows that these examples are cases of grammatical nominalizations. In (22)–(24), the nominalized structures present a NP-heading use, and (21) seems to manifest ‘insubordination’ (see Section 4.1.1).

The survey of mi-nominalizations so far indicates that -mi is a productive nominalizer in that (i) it is applied to various types of adjective (stem) and verb and (ii) it achieves V-based lexical and grammatical nominalizations. The survey has also revealed a new noun-like use of a nominalized structure: the ‘sentential use.’ In this use, the nominalized structure stands as a sentence, with which the writer performs a speech act, as if the speaker performs a speech act with a one-word sentence.

4.2N-based nominalization

Unlike -mi _SV, -mi realizes N-based nominalizations. Consider (25).

(25)

[The writer posts the photo of a shrine at night.]

Reikan-toka

spiritual.inspiration-etc.

mattaku

npi

nai-kedo,

exist.neg-but

yoru-no

night-gen

jinja-wa

shrine-top

ikai-mi-o

spiritual.world-nmlz-acc

kanji-te

feel-cvb

wakuwakusuru.

get.exited

‘I am not sensitive to spiritual things at all, but when I see a shrine at night, I feel it is something like a spiritual world and I get so excited.’ [12/Aug/2019]

Again, -mi is a productive nominalizer for N-based nominalization, too. It may be attached to not only the common noun ikai ‘spiritual world’ in (25) but also the proper noun majotaku (film name) in (26) and the foreign loan word mama ‘mom’ in (27).

(26)

[The writer posts the photo of townscape where there are many houses with an orange roof. This townscape reminds him of the film Kiki’s Delivery Service (whose Japanese title is Majo-no Takkyūbin, with Majotaku being an informal abbreviation).]

Majotaku-mi-o

M-nmlz-acc

kanjiru.

feel

‘(When I see this kind of townscape,) I feel it is something like the world depicted in the film Kiki’s Delivery Service.’ [12/Aug/2019]

(27)

[The writer drew a picture of a male character of a Japanese animation.]

[[Mama-mi-ga

[[mom-nmlz-nom

aru]

exist]

Dai-chan]-ni

D-dim]-to

nat-teshimat-ta.

become-end.up-pst

‘In this picture, Dai ended up having an aura of a good mother.’ [29/Sep/2019]

N-based mi-nominalizations seem to take place for a phrasal element, too. Consider sākuru-no hime ‘princess of a club’ in (28).

(28)

[The writer posts the photo where a female enjoys drinking with a number of males, and says that this atmosphere is something like the one where the only female student in a club (e.g. tennis club) at a university is favored by the male students in the club, as if she were the princess of the club.]

Sākuru-no

club-gen

hime-mi-ga

princess-nmlz-nom

sugoi.

great

‘She really has an aura of the princess of a club.’ [12/Aug/2019]

The nominalized elements in (25)–(27) would be cases of lexical nominalizations because they are not phrasal, though their stability as nouns would vary. For instance, since majotaku-mi involves the proper name majotaku (film name), the formation of this nominalized element would be ad-hoc, compared with the other two examples. This may suggest that majotaku-mi is a less stable noun than the other two.1212.This expectation seems to be reflected in their frequencies. In Twitter, mama-mi was used 67 times within the last four hours, while ikai-mi was used ten times and majotaku-mi five times within the last one month (searched on 14/Sep/2019). On the other hand, (28) may be regarded as a case of grammatical nominalization since sākuru-no hime-mi is phrasal. Still this nominalized form is also found repeatedly in Twitter.1313. Sākuru-no hime-mi was tweeted five times, by different twitterers, within the last one year (searched on 15/May/2020). It would thus be reasonable to hold that sākuru-no hime-mi is a non-prototypical case of grammatical nominalization. This is compatible with the non-discrete view on lexical and grammatical nominalizations (see Section 3).

A more striking case is (29), where what we may call ‘double nominalization’ occurs.

(29)

[The writer says that she made a tasty Japanese rolled omelet.]

Kyō-wa

today-top

shiroppu-o

syrup-acc

ire-te

put-cvb

tsukuri-mashi-ta!!

make-hon-pst

[…]

 

Shizenna

natural

ama-sa-mi-ga

sweet-nmlz-nmlz-nom

totemo

very

kuchi-ni

mouth-to

najimi-mashi-ta ❣❣

suit-hon-pst

‘Today, I used syrup to make a Japanese rolled omelet! Its natural sweetness suited my taste❣❣’ [21/Jul/2019]

In (29), ama-sa ‘sweet-nmlz’ (= ‘sweetness’) is a noun that has already undergone a lexical-nominalization process, and -mi is attached to this nominalized element. Compared with the mi-absent form ama-sa, the doubly-nominalized form ama-sa-mi renders the Twitter post look cuter. This is consistent with the writer’s use of a heart-shaped exclamation mark placed at the end of the sentence. Another example of double nominalization is given in (30).

(30)

[The writer often met a woman in a nursery room in hospital after childbearing, and she writes about this woman in her post.]

[[Junyūshitsu-de

[[nursery.room-loc

yoku

often

isshoni

together

naru]

become]

kata]-wa

person]-top

kaiinsekkai-dat-ta

episiotomy-cop-pst

-rashī-nda-kedo,

-seem-mm-but

[…]

 

kanari

very

tsura-sō-deat-ta.

painful-seem-cop-pst

Tsura-sa-mi …

painful-nmlz-nmlz

‘I heard that the woman I often met in the nursery room had undergone an episiotomy. It seemed very painful. Painfulness…’ [11/Aug /2019]

In (30), tsura-sa-mi instantiates the sentential use of mi-nominalizations. Note that the sa-nominalized form tsura-sa does not have this use, and without -mi, (30) is highly unnatural. Double nominalizations (29)–(30) are possible because (i) ama-sa and tsura-sa are nouns (i.e. output of lexical nominalization), (ii) -mi enables N-based nominalizations, and (iii) -mi adds a meaning/function not achieved by -sa alone, as shown in (29) (i.e. rendering the post look cuter) and in (30) (i.e. serving as a sentence).1414.There are seven cases of ama-sa-mi and 38 cases of tsura-sa-mi in the last six months (searched on 17/May/2020). Other examples include setsuna-sa-mi ‘painful-nmlz-nmlz’ (27), kowa-sa-mi ‘scary-nmlz-nmlz’ (20), and nemu-sa-mi ‘sleepy-nmlz-nmlz’ (13); the number in parentheses indicates the number of occurrences in the last six months (searched on 17/May/2020). Of note is yo-sa-mi ‘good-nmlz-nmlz,’ which appeared 39 times in the last one hour (searched on 17/May/2020). One of the reviewers pointed out that the order of -sa-mi cannot be reversed, suggesting that this may reflect the order of morphemes at the right-periphery in Japanese (i.e. a ‘subjective’ morpheme comes before an ‘intersubjective’ morpheme); see Onodera (2014Onodera, Noriko 2014 “Setting Up a Mental Space.” In Discourse Functions at the Left and Right Periphery, ed. by Kate Beeching and Ulrich Detges, 92–116. Leiden/Boston: Brill. , 111–113). This suggestion is illuminating. First, -sa is less subjective than -mi _SV (Section 2); given that -mi also expresses a subjective meaning, as shown in many examples in the text, it is safe to assume that -sa is less subjective than -mi. Second, -mi may be used in consideration of hearers/readers (e.g. to mitigate a negative impact of the utterance on the hearer/reader; see Section 2). That is, -mi is more intersubjective than -sa. I am grateful to the reviewer for his/her insightful suggestion.

Finally, our data include examples where -mi attaches to a noun-like element such as the personal pronoun kare ‘he’ in (31) and the demonstrative sore ‘that’ in (32).

(31)

[The writer posts the photo of an outfit in a men’s wear section. This outfit is similar to the outfit of the main character in a Japanese animation who tries to become the king of pirates.]

[[Bō-kaizoku-ō-o

[[certain-pirate-king-acc

mezashi-teiru]

aim-ipfv]

kare]- mi-ga

he]-nmlz-nom

sugoi.

great

‘(This outfit) really reminds us of a certain guy who tries to become the king of pirates.’ [28/Jun/2019]

(32)

[The writer posts photos depicting the cityscape of Kyoto.]

Kyoto-de

K-loc

itoko-no

cousin-gen

kekkonshiki.

wedding

Sore-mi-ga

that-nmlz-nom

fukai.

deep

‘My cousin’s wedding is held in Kyoto. (The scenes depicted in the photos are) full of Kyoto atmospheres.’ [25/Jun/2017] 1515.This post has two photos, one of which depicts a riverside in Kyoto, and the other of which depicts a traditional wooden room (probably, used as a wedding place) with open windows, from which one can see the cityscape of Kyoto. We thus assume that sore in (32) is anaphoric to Kyoto.

In Japanese, pronouns and demonstratives behave like nouns in that they are modifiable. This can be seen in (31), where kare ‘he’ is modified by the preceding part. Another type of noun-like item is ‘emoji.’ Our data include an example where -mi is directly attached to the ‘emoji’ stamp portraying the flag of Turkey, with the interpretation ‘Turkishness.’

It has been demonstrated that -mi is a productive suffix for N-based nominalizations, too. It is attached to various types of noun and noun-like item and realizes lexical nominalizations (and possibly grammatical nominalizations). The ‘sentential use’ is also detected in N-based mi-nominalizations.

4.3Further issues

In addition to N-based and V-based nominalizations, there are examples where one cannot easily pinpoint the category of a base. Most notably, -mi is attached to an onomatopoeia.

(33)

[The writer posts the photo of a female celebrity.]

Utsukusi-sa-to

beautiful-nmlz-and

kawai-sa-no

cute-nmlz-gen

kyōzon.

coexistence

Kirakira-mi-ga

onmp-nmlz-nom

sugoi.

great

‘(She is) beautiful as well as cute. Her sparkling attractiveness is great.’ [29/Aug/2017]

(34)

[The writer has got a fever and stays home.]

39-do-da-ze!

39-degree-cop-fp

Furafura-mi .

onmp-nmlz

‘My temperature is at 39! I feel dizzy.’ [10/Dec/2018]

In (33), kirakira ‘shining, sparkling,’ when it is applied to a woman, describes a sparklingly attractive woman. The nominalized structure kirakira-mi denotes ‘sparkling attractiveness,’ manifesting an NP-heading use in that it is nominative-marked and functions as the subject of sugoi ‘great.’ Next, in (34), furafura describes a situation where a person feels dizzy, and the nominalized structure furafura-mi denotes the concept of ‘dizziness.’ It stands as a sentence on its own, expressing the writer’s woozy feeling (i.e. sentential use).

The attachment of -mi to an onomatopoeia has been described in Uno (2018)Uno, Nagomi 2018 “Twitter-ni miru Onomatope-ni Kōsetsusuru Setsubiji Mi-no Kinō [Functions of the suffix -mi attached to an onomatopoeia attested in Twitter ].” Comparative Japanese Studies Annual Bulletin 14: 183–189.. Her novel finding is that some mi-attached onomatopoeias are employed for greeting purposes.

(35)

[A person first says that he/she needs to go to sleep, and another person replies to it.]

1. Mā

   well

   īya,

   do.not.care

   ne-neba.

   sleep-must

   ‘Well, then, I must go to sleep.’

2. Suyasuya-mi .

   onmp-nmlz

   ‘Good night.’ (Uno 2018Uno, Nagomi 2018 “Twitter-ni miru Onomatope-ni Kōsetsusuru Setsubiji Mi-no Kinō [Functions of the suffix -mi attached to an onomatopoeia attested in Twitter ].” Comparative Japanese Studies Annual Bulletin 14: 183–189., 187–188) [17/Oct/2017]

In (35b), -mi attaches to suyasuya, an onomatopoeia describing a sound sleep. Uno interprets this string to mean ‘good night.’ In Shibatani’s framework, this type of example would not be surprising because greeting is one of the functions associated with nouns, such as Morning!. Since greeting is regarded as a type of speech act, we maintain that suyasuya-mi in (35b) manifests the sentential use of a nominalized structure.1616.One of the reviewers noted similarities between oyasumi ‘good night’ and suyasuya-mi; both forms may be used as a reply to ‘I must go to sleep,’ and they have close sound structures. Such similarities may be a motive for the use of suyasuya-mi with a greeting function. In connection with this, the reviewer mentioned Du Bois’ (2014)Du Bois, John 2014 “Towards a Dialogic Syntax.” Cognitive Linguistics 25 (3): 359–410. Dialogic Syntax, where a supra-sentential syntactic structure (called diagraph) captures parallelism (at various levels) between dialogically juxtaposed utterances. Given the aforementioned similarities between oyasumi and suyasuya-mi, when these forms appear in the same discourse, evoking the sense of parallelism, such parallelism may be described in Dialogic Syntax terms. This is an intriguing issue, and I leave it for future research.

In Japanese, onomatopoeias behave like adverbs. Thus, suyasuya is used as in suyasuya nemu-ru ‘have a sound sleep,’ where suyasuya modifies the verb nemu-ru ‘sleep.’ One could think of these nominalized structures as A(dverbial)-based nominalizations, different from V-based and N-based nominalizations. Apart from onomatopoeias, there do not seem to be other cases where -mi achieves A-based nominalizations, but if one looks at other languages, there are cases of A-based nominalizations, though they are by no means common. A case in point is Old Romanian.

(36)

începu

start.ps.3sg

de la

from

apoii

afterwards.pl.def

până

until

la

 

întăii

first.pl.def

‘He started from the last ones to the first ones.’ (Pană-Dindelegan 2018Pană-Dindelegan, Gabriela 2018 “Inversion and Nominal Ellipsis.” In Comparative and Diachronic Perspectives on Romance Syntax, ed. by Gabriela Pană-Dindelegan, Rodica Zafiu, Adina Dragomirescu, Irina Nicula, Alexandru Nicolae, and Louise Esher, 5–22. Newcastle upon Tyne: Cambridge Scholars Publishing., 13)

According to Pană-Dindelegan (2018Pană-Dindelegan, Gabriela 2018 “Inversion and Nominal Ellipsis.” In Comparative and Diachronic Perspectives on Romance Syntax, ed. by Gabriela Pană-Dindelegan, Rodica Zafiu, Adina Dragomirescu, Irina Nicula, Alexandru Nicolae, and Louise Esher, 5–22. Newcastle upon Tyne: Cambridge Scholars Publishing., 12), apoia ‘afterwards.def’ is a nominalized adverb, or in our terms, a noun that has undergone an A(dverbial)-based lexical nominalization process. A-based nominalizations, therefore, are a category independently motivated beyond Japanese youth language.

5.Implications

In this article, we have addressed mi-nominalizations in Japanese youth language, shedding light on its unique morphosyntactic and functional properties and demonstrating that these properties are flexibly captured in Shibatani’s theory of nominalization. By way of conclusion, let us point out empirical and theoretical implications of our proposed account.

Starting with empirical outcomes, we offered new sets of data on mi-nominalizations. -Mi is a productive nominalizer, productive in the sense that it instantiates V-based and N-based (as well as A(dverbial)-based) nominalizations. For V-based nominalization, both lexical and grammatical operations are possible; for N-based nominalizations, the process is in general lexical, with some examples which might be taken as cases of grammatical nominalizations. Furthermore, -mi is attachable to an adjective (stem), a verb, a noun, their phrasal counterparts, a sa-nominalized form, and even an onomatopoeia. These sets of data serve as a rich source of inquiries into nominalizations on the one hand and grammatical investigations of nonstandard linguistic varieties on the other hand.

Theoretically, let us recall that Section 3.3 raised the issues (14), repeated here as (37).

We provide our answers to these questions by considering our findings about morphosyntactic and functional properties of mi-nominalizations.

As for (37a), Section 4.1.1 has pointed out that -mi, when it is applied to an adjective, instantiates lexical nominalizations in many cases and grammatical nominalizations in less cases; together with the fact that -mi _SV has only a lexical-nominalizing function, we suggested that -mi seems to be developing a grammatical-nominalization function. This functional extension is most clearly found for adjectives. A complication arises for verbs, however, since -mi is only attached to the finite form of a verb (not a verb stem), and there is no unequivocal case where -mi realizes lexical nominalizations with a verb (see footnote 9).

As for (37b), Section 4.3 suggested that one could set out the category of A(dverbial)-based nominalizations. This is based on the finding that -mi is attached to an onomatopoeia, which, at least in Japanese, behaves like an adverb. While it is true that adverbs are generally not nominalized in Japanese (as well as many other languages), there are languages in which adverbs undergo a nominalization process, as in the case of Old Romanian (see (36)).

As for (37c), Sections 4.1–4.3 claimed that mi-nominalized structures have a “sentential use” whereby the writer employs a nominalized element as a sentential unit and performs an illocutionary speech act. The sentential use of a nominalized structure is found for V-based and N-based (as well as A(dverbial)-based) nominalizations. Moreover, when -mi is attached to an onomatopoeia, it may allow the writer to perform a ‘greeting’ speech act, which is also an instance of the sentential use of a nominalized structure

More generally, the present study lends further support to Shibatani’s account, which has been corroborated by a wide range of examples in standard varieties of diverse languages, by showing that it is fruitfully applicable even to a nonstandard variety of a language: Japanese Wakamono Kotoba ‘youth language.’

In addition to these theory-internal consequences, our study has broader implications. As stated at the outset, there are few studies which analyze grammatical and functional properties of Japanese youth language. The present research provides a theoretically-informed analysis of Japanese youth language, complementing Seraku and Akiha’s (2019)Seraku, Tohru, and Takako Akiha 2019 “ Poi in Japanese Wakamono Kotoba ‘Youth Language’.” Lingua 224: 1–15. attempt to explore Wakamono Kotoba from a theoretical perspective.