Back to list

Detail of contribution

Auteur: Saeko URUSHIBARA

A Morphosyntactic reflex of root and epistemic modals: Evidence from a Japanese negative future element

Abstract/Résumé: Due to its strict head finality Japanese does not exhibit robust evidence based on displacement for an articulated clausal structure. Yet there is a striking morphological division of labor for the interpretation of a negative future element -mai. Of the two regular verbal conjugations vowel and consonant verbs, the former allows two distinct inflections tabe- (1a) and tabe-ru (1b), each corresponding to the root and epistemic intepretations. This is obvious from co-occurence with subject- and speaker-oriented adverbials. (1)a. Taroo-wa naniganandemo gohan-o tabe-mai. Taroo-top at all his costs rice-acc eat-mai 'Taroo will not eat rice at all his costs.' b. Taroo-wa kitto gohan-o tabe-ru-mai. 'Probably Taroo will not eat rice.' Taroo-top probably rice-acc eat-[-past]-mai Assuming that tabe- is a V head and tabe-ru is a V-T complex, I aruge that (1) is explained by positing two functional heads Mod(root) and Mod(epistemic) below and above TP. This conforms to the fact from many languages that root and epistemic modals are interpreted at lower and higher positions than TP, e.g. the scope relation with tense and adverbs. A question arises as to why consonant verbs do not pattern with vowel verbs. Namely, (2a), where a consonant verb precedes -mai, is ungrammatical, and the forms for root and epistemic interpretations are neutralized to (2b). (2)a. *Taroo-wa naniganandemo hon-o yom-mai. Taroo-top at all his costs books-acc read-mai 'Taroo will not read books at all his costs.' b. Taroo-wa naniganandemo/kitto hon-o yom-u-mai. 'Taroo will not read books at all his costs.'' Taroo-top probably books-acc read-[-past]-mai probably I argue that (2a) is ruled out by the Japanese Coda Condition. One might then wonder why an epenthesis strategy active in Japanese verbal conjugation ([a] in yom-a-nai 'read-[a]-neg' and [i] as in yom-i-masu 'read-[i]-polite' and yom-i-nagara 'read-[i]-while') does not save (2a). I argue that this is due to differences in phases and a mismatch in vocabulary insertion at Morphological Structure (MS). Assuming that CP and vP are phases, Mod(root)P containing V and the features of -mai are sent to MS (and subsequently PF), but this entity is 'smaller' than CP, the phase that tiggers epenthesis. (1b), (2b) and the above examples of epenthesis contain functional categories higher than Mod(root), so these are the cases where CP phases are sent to MS and epenthesis takes place where necessary. No epenthesis is necessary in the case of (1a) since the stem of a vowel ver does not violate the Coda Condition. Thus, (2b) can be seen as a morphologically-driven derivation.