日英中間言語における主語に関する統語分析上の誤り

It has been claimed that SLA of syntax, at its early stage at least, is carried out by the full transference of Ll parametric values (Schwartz, 1998a & b; Schwartz & Sprouse, 1994, 1996; Tsimpli & Smith, 1991; Tsimpli & Roussou, 1991). The mechanism of L1 transfer is explained by Tsimpli & Smith in more concrete terms : L1 transfer is an association process where learners map L1 features onto L2 morphophonological forms, i.e. learners “misanalyse” the input. The present paper investigates this possibility through a study of the acquisition of obligatory subjects (referential DPs, and expletives 'it' and 'there') by Japanese learners of English, and gives further support to the “misanalysis” theory. It has been claimed, within the framework proposed by Chomsky (1995), that structural differences between English and Japanese subjects are explicable in terms of the feature specifications associated with them. Japanese allows a null subject in tensed clauses whereas English does not; this is due to the D feature in Tense (T) being weak in Japanese and strong in English (Wakabayashi, 1997, 2002; cf. Yatsushiro, 1999). However, this parametric difference is claimed to be easily conquered by Japanese learners of English (Hirakawa, 2003; Wakabayashi, 1997, 2002; Wakabayashi & Negishi, 2003; Zobl, 1990). The present study examines the validity of the above, by investigating whether noun phrases appearing immediately before the verb in Japanese-English Interlanguage are really subjects in the sense of English. We argue, by presenting experimental data, that at least elementary to intermediate level Japanese speakers misanalyse English Subject-Verb structure as Topic-Null Subject-Verb (Kuribara, 2000, 2003), whilst they know that there must be an overt phrase before a predicate. A grammaticality judgement test has been administered to three groups of learners whose English proficiency ranged from elementary to post-intermediate. The test contains two types of constructions : null subjects placed in various structural contexts, and DP topic-Subject structures. The data show that learners tend to tolerate null subject clauses where the verbs are preceded by phrasal categories (such as DPs) significantly more than clauses where the verbs are preceded by heads or no lexical items at all. As to DP Topic-Subject structures, the majority of elementary and intermediate learners have accepted all the constructions, regardless of whether the topic is O-marked or not, or whether the subject is overt or null. These results clearly indicate that the elementary and intermediate learners have not acquired the strong D feature. What they have learnt seems to be that English generates the subject and/or the topic before the verb (in the sense of Japanese), and that (at least one of) the phrase (s) always requires morphophonological material. (cf. Hawkins, 2001; Kuribara, 2000, 2003).