講演会のご案内 SACL (Sendai Area Circle of Linguistics) では、下記の要領で、Uli Sauerland 氏 (ZAS)と高橋大厚氏（東北大学）の講演会を開催することになりました。奮って ご参加下さい。 日時： ７月１８日（月） 午後３時より 場所： 東北学院大学土樋キャンパス ８号館３Ｆ第４会議室 （正面玄関から入って、一階上がったところが３Ｆ） Ｉ 午後３時∼４時１５分 Uli Sauerland (ZAS) タイトル：Complex sentences within biolinguistics 要旨：The human ability to produce and comprehend complex sentential structures is of central interest to the biolinguistic enterprise (Chomsky 1957, Hauser et al. 2002). Specifically, recursive application of a generative rule constitutes such complexity. For concreteness, we distinguish between three levels of generative rules: phoneme-level ‘Concatenate’, morpheme-level ‘Merge’ and, at the highest level of granularity, sentence-level embedding. Of these three, we focus on the highest level: sentence embedding. In this talk, we present novel evidence on the variability, but also universality of sentence embedding from Teiwa, Pirah a, Matses, and child language and its tight link to false belief. In contrast to our view, a major strand of work in historical linguistics assumes that syntactic complexity has arisen recently in the history of humanity: Deutscher (2000, 2005) claims that Old- Babylonian (1750 b.c.) did not have complex syntax and suggests a more general ‘me-Tarzanyou-Jane’-stage of language. More recently, Dahl (2004), Giv on (2009), Heine and Kuteva (2007), Sampson et al. (2009) speak of a genesis of complex syntax. Finally, at least some researchers (Everett 2005, Klamer 2010) claim that complex syntax is also not a universal property of all living languages. These views are integrated by Wray and
Grace (2007) into a general claim that complex syntax only arises in speech communities where talk with strangers happens, i.e. communities greater than single hunter-gatherer tribes. Embedded sentences are one of the primary uses of recursion in syntactic analysis. But in many languages, embedded sentences cannot be purely syntactically characterized: In formal grammar theory, only central self- embedding requires recursive context-free rules (Chomsky 1956). Hence in languages without centre embedding of clauses, embedding must be justified by appeal to semantic universals. In particular, the assumption that independent sentences (1) can alternate in order without a change in meaning, while a sentence-embedding sequence (2) cannot. (1) John predicted it. Mary won the lottery. == Mary won the lottery. John predicted it. (2) John dreamed Mary won the lottery. =/= Mary won the lottery. John dreamed. We first discuss the distinction between direct and indirect speech. Specifically, we show that in English child language (Hollebrandse 2007) and in languages that have only direct speech (Matses), direct speech allows extraction and shares other properties of indirect speech. Therefore, we see license put aside the distinction between direct and indirect speech for the time being. We then discuss two living languages that have been claimed not to allow embedded sentences: Teiwa and Pirah. For both, we present novel evidence from sentence production in a false-belief scenario and from sentence comprehension. The production evidence shows that in sentence production a specific pattern with a fixed order of the verbs for say/think and a second verb characterizing the belief is used by the speakers to describe the false belief. The comprehension evidence on the most frequent construction further substantiates that they are understood as embedded sentences. This establishes the presence of sentence-embedding in two living languages where it has been disputed. The Teiwa evidence furthermore is important to the interpretation of historical data. Namely, Teiwa uses the sequence pronoun+say as complementizer: (3)
Natan a na walas-man [ a wa a xaf karim ol ] N.he me tell-not
[ he say he boat small buy]
‘Natan didn’t tell me that he bought a small boat.’ In many languages, the complementizer is derived from the verb say or similar verbs (cf. Guldemann 2008). This has been taken as evidence of a recent origin of embedding from paratactic sequences involving say (e.g. Deutscher 2000). However, Teiwa would represent the origin of such a historical development, but we have shown that it has full
embedding. This entails that embedded sentences may be historically older than previously assumed. Overall our evidence supports the claim that complex syntax is an evolved trait closely tied to the expression of belief states. II ４時３０分∼５時４５分 高橋大厚（東北大学） タイトル：Identity for Argument Ellipsis 要旨：Assuming with the recent literature that nominal arguments can be elided in Japanese, this study shows that the mere presence of antecedent nominals is not sufficient for elision of arguments, and argues that antecedents need to occur in domains of the same type as elliptic arguments. The data used to establish the point involve nominal arguments accompanied by the negative polarity item (NPI) sika ‘anything/anyone but’ and we have examined where antecedent arguments accompanied by the NPI need to occur to license elision of corresponding arguments. The results indicate that ellipsis of objects (namely, DPs in the complement position of VP) necessitates that antecedents appear in VP and that ellipsis of subjects (or DPs in the specifier position of TP) demands that antecedents be in the specifier position of TP. It is argued that this “domain parallelism” between elided arguments and their antecedents naturally follows from the derivational model of syntax assuming that structure building takes place unit by unit (the phase theory). To the extent that the present work is on the right track, it should contribute to elucidation of the properties of argument ellipsis in particular and help deepen our understanding of ellipsis in general. 以上 連絡先：阿部 潤 （東北学院大学文学部英文学科）
[email protected]
