| Summary: | This thesis investigates two topics: (i) the distribution of argument ellipsis (AE) in languages that permit AE, and (ii) the status of strong islands (SIs). Regarding (i): The present thesis develops a novel account of the distribution of AE. According to this account, AE is subject to a locality constraint: in languages with AE, an argument may undergo AE only if it is local to V. Crucially, the definition of locality varies as a function of whether the language in question is VO or OV. In VO languages, an argument is local to V if and only if it is immediately preceded by V. In OV languages, an argument is local to V if and only if it is a sister of V or V’. With respect to OV languages, it is assumed, following Neeleman and Weerman (1999), that some OV languages base-generate external arguments (EAs) in [Spec, VP], whereas others base-generate EAs as adjuncts to VP. [Spec, VP] is local to V; the adjunct-to-VP position is not. Accordingly, the present account correctly predicts that AE-allowing OV languages fall into two classes: those that allow AE of EAs and those that do not. More significantly, it predicts which AE-allowing OV languages fall into which class. Those that generate EAs in [Spec, VP] allow AE of EAs; those that generate EAs as adjuncts to VP do not. The account of AE proposed here is argued for on the basis of two studies: one, a case study of AE in Brazilian Portuguese, a VO language; the other, a study of a variety of AE-allowing OV languages. Regarding (ii): it is generally held that SI-crossing movement is possible, provided some special step is taken (e.g., ellipsis of the island, insertion of a resumptive pronoun). This thesis argues, on the basis of reconstruction facts in Brazilian Portuguese, Hebrew, Norwegian, and Swedish, that SIs are absolute barriers to movement. Under no circumstance can movement take place out of an SI.
|
|---|
