詞頭在尾關係子句之語句處理基礎

The Processing Foundation of Head-Final Relative Clauses

由於詞頭在尾之關係子句表面上看似其他名詞前的修飾語，許多分析只將此類關係子句當作如同形容詞或狀語等名詞前修飾語，而將之認定為不同於詞頭在首的關係子句。本文旨在檢視詞頭在尾關係子句之語句處理特性，欲證明詞頭在尾的關係子句仍具有內在句法結構及空詞類，而應被視為獨立於其他名詞前修飾語。主要證據來自三方面：首先，在語句理解中，漢語所有格關係子句受所有者在關係句中句法位置所影響，不同於副詞性關係子句；其次，韓文關係子句的理解研究中亦發現涉及位移的關係子句不同於含有零代名詞的副詞子句；第三，漢語中主語提出的關係子句較賓語提出者容易理解，顯示關係子句含有內在句法結構，且空詞類在結構上所佔的位置影響其理解之難易。因此，本文認為詞頭在尾之關係子句具有內在句法結構及空詞類，在語句理解上顯示他們不同於其他的名前修飾語。

This article examines the special processing characteristics of head-final relative clauses. As head-final relative clauses superficially resemble other prenominal modifiers such as adjectivals, stative verbs, and adjunctive/complement clauses, their status as a distinct kind has constantly been challenged. We present existing processing evidence to show that head-final relative clauses should be distinguished from other prenominal modifiers as they do involve structure-based filler-gap integrations (like those of head-initial relative clauses) and observe universal extraction effects. The evidence includes: (a) processing differences between possessive relative clauses and adverbial relative clauses in Mandarin; (b) processing differences between gapped relative clauses and adjunct clauses with null pronouns in Korean; and (c) subject-extracted relative clauses being easier to process than object-extracted relative clauses in head-final relativization.