Information structure, prosody and the origins of Formosan relativization

Project: Research

Layman's description

Some Formosan languages (aboriginal languages of Taiwan) display a word order pattern which language typology would lead us not to expect to find anywhere: verb-initial word order combined with relative clauses preceding the noun. The purpose of this project is to examine reasons for this by means of various linguistic tests, to see what this can reveal about the origins of relativization.

The Formosan languages (the aboriginal languages of Taiwan), and some close Philippine relatives, are the only known verb-initial languages which allow head-final relativization patterns. In fact, Rel-N is the unmarked option in (at least) Amis, Bunun and Tsou.

The purpose of this project is to examine possible reasons for this typologically unexpected situation. There are at least three possible explanations for Rel-N order in Formosan: a) that the head noun and the relative clause historically represent two independent NPs in a flat (exocentric) structure, making traditional word order typology irrelevant; b) that the head noun is historically the subject of an internally headed relativization construction with underlying Austronesian VOS order and; c) that the relative clause is extracted leftwards by a process which places new information preceding given information.

Each of these hypotheses is to some extent supported empirically by various facts in the languages involved and each has interesting theoretical implications (e.g. touching on the possibility that relativization may be a quite recent development in Formosan languages). Based on intonational, functional and distributional clues, we will evaluate the three hypotheses against each other. For this purpose, data on a wide range of constructions will be collected from three Formosan languages: Seediq (predominantly N-Rel), Bunun (Rel-N) and Puyuma (where both orders occur in free variation).
Effective start/end date2010/01/012013/12/31