Retour vers liste

Détail de la contribution

Auteur: Wei-Tien Dylan TSAI

Rethinking Formal Licensing

Abstract/Résumé: Rethinking Formal Licensing Wei-Tien Dylan Tsai Since Rizzi's (1990) seminal work on Relativized Minimality, it has become clear that there are two main factors involved in forming a legitimate syntactic dependency, namely, identification vs. formal licensing. The former, encoded by antecedent government (Chomsky 1982, 1986) and generalized binding (Aoun 1986), has been the focus of inquiry in the GB literature, while continuing to prosper in the form of Agree under the Minimalist Program (Chomsky 2000). The latter notion, by contrast, receives relatively little attention in the form of head/theta-government (Aoun et al. 1987; Rizzi 1990), and has almost lost all its theoretical status in the post-GB era. In face of strong evidence from Chinese extraction and ellipsis, this paper proposes to reinvent the notion of formal licensing in terms of sisterhood, and reinstates it as a built-in requirement of Merge, i.e., an "insurance" on the behalf of the No Tampering Condition (Chomsky 2007, 2008) such that no phonetic content will be lost in an unwarranted manner. A phenomenon of particular interest along our line of inquiry is that there seems to be two types of adjectival markers in Chinese nominal projections: Unlike a classifier, measure "de" cannot license topicalization/ellipsis from its complement. By contrast, modifier "de" is able to license extraction/ellipsis of the same nature, behaving in line with the classifier: The same measure-modifier asymmetry shows up in passivization, object fronting and verb copying, as well as NP-ellipsis. This leads us to Saito & Murasugi's (1990) observation that a complement of a functional category can be deleted if and only if the functional head has a specifier which it agree with. Along this line, modifier "de" may well serve as the head of a modifier phrase (ModP) in Rizzi's (2004) sense, we propose to develop the idea into the formulation of formal licensing. Measure "de", by contrast, behaves like a clitic to the measure phrase (MeaP), and hence lacks the capability of licensing the empty category in question (also cf. Cinque 2011). The rationale behind our analysis has already been established by the classic notion of head-government; namely, the merger between a functional head and its complement is driven by a form of Agree on default selectional features (cf. Saito 2003), in which an adjunct like MeaP would play no part. All in all, we have shown that, contra Saito et al. (2008), Chinese "de" do not form a homogeneous group. Evidence instead argues for a pure syntactic treatment of the locality effects in terms of formal licensing under the minimalist-cartographic approach, i.e., no tempering of any phonetic content unless fully licensed by head-complement selection.