功能延伸與語法化：台灣南島語名詞組修飾標誌的類型研究

Functional Extension vs. Grammaticalization: A Typological Study of Modification Markers in Formosan Nominals

本文探討原始台灣南島語的名詞組結構有可能歷經由並列結構到非並列結構的歷史演變，而其中又可分為不帶有格位標誌以及帶有格位標誌兩種不同語言類型的並列結構。第一種並列結構適用於所有台灣南島語，且經由功能延伸的方式，產生了借用已有的從屬結構標誌做為名詞組修飾標誌的新句法現象。第二種並列結構則只適用於格位固有內化的台灣南島語，且經由語法化的方式，造成了格位標誌也能做名詞組修飾標誌的新句法功用。除此之外，還有其他五種相關的歷史演變也似乎見存於台灣南島語：（一）格位和對等標誌先變成句子的從屬標誌再變成名詞組的修飾標誌，（二）名詞組修飾標誌先出現於關係子句再出現於非關係子句，（三）曾發生名詞修飾成份的固有格位內化，（四）曾發生SVO的語序變化，以及（五）曾出現詮釋條件的規範。

It is proposed in this paper that the internal structure of proto-Formosan nominals may have undergone a historical development from juxtaposed to nonjuxtaposed type of syntactic structure, in which two distinct ways of juxtaposition DP-DP*, without juxtaposition of the functional projection of case, and KP-KP*, with juxtaposition of the functional projection of case, may have been involved. In the historical processes of using subordination and case markers as modification markers for Formosan non-juxtaposed nominal structures, the recursive DP-DP* pattern may have been accessible to all kinds of Formosan languages and, via functional extension, may have given rise to the appearance of subordinator-marked modifiers, in which no adverbial/complement or coordination relation is still obtained in synchronic syntax. With the recursive KP-KP* pattern, which may have been accessible only to intrinsically case-inflected Formosan languages, case-marked modifiers may also have appeared via grammaticalization, in which case markers no longer express grammatical function and referential specification. In addition, five other relevant historical processes may also be observed in Formosan languages. (a) Case and coordination markers may have been developed first into subordination markers in clauses and then into modification markers in nominals. (b) Modification markers may have been associated first with relativization type of modifiers and then with non-relativization type of modifiers. (c) An intrinsic case inflection of nominal modification expressions may have taken place. (d) An SVO word order change may have taken place. And (e) an interpretational processing requirement may have taken place.