<yotsuka@hawaii.edu>

Department of Linguistics
University of Hawaii at Manoa

Consequential 'o-clauses in Tongan and licensing of  a null anaphor

Tongan shows an ergative (ERG) pattern with respect to certain syntactic  operation: e.g., only absolutive (ABS) can undergo relativization by the gap strategy and Equi-deletion under pea-coordination. Otsuka 2002, 2003 proposes that syntactic ergativity should be understood as a consequence of morphological ergativity, arguing that the relevant syntactic operations require that the gap be licensed by Case feature matching. Equi-deletion under 'o-coordination also shows an ERG pattern in that ERG is distinguished from ABS: the antecedent must be ABS. At first glance, this condition cannot be explained in terms of Case matching. I argue,  however, that this condition can also be viewed as a by-product of  morphological ergativity. Specifically, I propose that the empty category in 'o-clauses is a null anaphor (proSE) and that the binding condition for  proSE is Case-sensitive. In this talk, we first put forward the claim that syntactic ergativity results from morphological ergativity by examining relativization and pea-coordination in Tongan. Then, we compare 'o-constructions with pea-constructions to conclude a) that unlike pea, 'o should be regarded as a complementizer rather than a conjunction; and b) that the gap in 'o-clauses is not an outcome of deletion, but a null anaphor. We will then discuss a Minimalist approach to binding proposed by Reuland (2001) and see how it accounts for the distribution and behavior of proSE in Tongan. The current analysis not only explains how the null argument in 'o-clauses is interpreted as coreferential with the ABS-marked NP of the matrix clause, but also correctly predicts that coreference between proSE and the ERG-marked NP is impossible. Furthermore, it also accounts for the well-known fact that long distance (SE) anaphors in accusative languages are subject-oriented.