The Word Status of Japanese Adjectives

1 Problem

It is widely assumed in the grammar of Japanese that pure adjectives, which characteristically have the (nonpast) i-ending, conjugate "in a manner similar to verbs" (Kuno 1973:28). They are called "canonical adjectives or CAs" in the terminology of Nishiyama 1999. Indeed, given a paradigm like (1) (slightly modified from Nishiyama 1999: 190), we may be led to believe that CAs do conjugate by themselves. However, Nishiyama has recently offered an interesting analysis of CAs, noting that "except for the present forms, all other forms contain [/k/] after the adjectival root" and therefore "it is the present form that is exceptional in that it does not contain /k/ overtly" (1999:190). From this observation, he goes on to argue that underlyingly, adjectival predicates in Japanese have the form of an adjectival root plus k(u), where he takes /u/ as an epenthetic vowel, followed by the copula ar; thus, he views the adjectival predicate taka-katta 'high-PAST' in (1), for example, as deriving from taka-k(u) + at-ta 'high-k(u) + be-PAST'. This analysis is reasonable, since the copula is certainly discernible in the various forms of the CAs in (1). Furthermore, Nishiyama extends this analysis to present forms such as taka-i 'high-PRES' by proposing that the actual generation of this form from taka-k(u) + ar-u 'high-k(u) + be-PRES' takes place in Morphological Structure, a grammatical module placed after syntax in Distributed Morphology (Halle and Marantz 1993), by way of Merger and Fusion. Roughly put, the k(u) ending of a CA and the present tense copula ar-u are fused and realized as i in Nishiyama's theory of adjectives. Hence, the paradigm in (1) is actually along the lines of (2), as far as syntax is concerned. Notice here that what seem to be conjugating adjectives are in fact a combination of CA-k(u) and the regular copula.

(1)		'high'	'wide'	Copula
	Present	taka-i	hiro-i	ar-u
	Past	taka-katta	hiro-katta	at-ta
	Presumptive	taka-karoo	hiro-karoo	ar-oo
	Conditional	taka-kereba	hiro-kereba	ar-eba
	Gerundive	taka-ku-te	hiro-ku-te	at-te

(2)		'high'	'wide'	Copula
	Present	taka- k(u) ar-u	hiro- k(u) ar-u	ar-u
	Past	taka- k(u) at-ta	hiro- k(u) at-ta	at-ta
	Presumptive	taka- k(u) ar-oo	hiro- k(u) ar-oo	ar-oo
	Conditional	taka- k(u) ar-eba	hiro- k(u) ar-eba	ar-eba
	Gerundive	taka- k(u) at-te	hiro- k(u) at-te	at-te

This analysis has remarkable explanatory power, and I fully agree with this way of deriving the surface forms of Japanese adjectival predicates. Nevertheless, I must disagree with Nishiyama's analysis that k(u) is a "predicative copula" (1999:189), the conclusion he draws from observations based on the following two sentences. (3a) contains a CA, taka 'high', whereas (3b) contains a "nominal adjective or NA" (e.g., Kuno 1973), sizuka 'quiet'.

(3)

1. a. Yama-ga         taka-k(u)-mo ar-u.¹

   mountain-NOM high-k(u)-even be-PRES

   'The mountain is even high.'

2. b. Yoru-ga   sizuka-de-mo ar-u.

   night-NOM quiet-de-even be-PRES

   'The night is even quiet.'

(Nishiyama 1999:185; slightly modified)

Nishiyama pays special attention to the seeming structural parallelism between these sentences. Assume for the moment that this parallelism is genuine; notice, then, that k(u) in (3a) corresponds to de in (3b), which is often identified as one conjugational form of the copula da (e.g., Masuoka and Takubo 1992). Following Bloch (1946a,b), who suggests that the copula is indispensable for predication, Nishiyama proposes to call de the "predicative copula" (1999:188). As for the tense-carrying ar that follows it, Nishiyama labels it the "dummy copula" (1999:188). By analogy, therefore, k(u) must also be a predicative copula that is followed by ar. Thus, k(u) in (3a) is considered to be an independent morpheme on a par with de in (3b). In fact, in Nishiyama's theory de...