本文以臺語的ViViN結構為研究對象，此一構式在動詞重疊後加上一個名詞成分，且說話者對該動後名詞成分必然帶有負面評價。根據動詞重疊的末音節不變調，以及動詞重疊成分與模態詞及情狀副詞的相對位置，本文主張前述動詞雖然移位，但並未離開輕動詞組。此外，由於此一結構中的動後名詞成分不能省略，也不能移到動前位置，本文主張該成分帶有無法詮釋的評價性質，透過兩階段的特徵移位，分別到達DP語段(phase)及vP語段的邊緣，以進行性質分享並達成與評價性詞組主要語的對協（參Pesetsky & Torrego 2007）。此一結構的極長距依存關係，表面上對語段理論(Chomsky 2001, 2008)帶來挑戰，但透過特徵的階段性移位，我們仍可以在語段理論的架構下為實際的語言現象提出解釋。

This study aims to provide an analysis of the ViViN construction in Taiwanese, which is obligatorily accompanied by a negative evaluative reading on the post-verbal noun. Based on observations including the citation tone on the final syllable of the reduplicative verb form and the form’s relative positions with modals and manner adverbs, we suggest that the verb has moved to a higher position without leaving the vP. Also based on the indispensability of the post-verbal noun and the fact that the post-verbal noun cannot be preposed, we propose that an uninterpretable evaluative feature moves from the NP to the edge of the DP phase and then to that of the vP phase, in order to value the head of the Eval(uative)P (based on Pesetsky & Torrego (2007)). Ostensibly, this construction with a very long-distance dependency poses a problem for phase theory (Chomsky 2001, 2008), but the problem can be solved by Move-F in a relaying-style scheme to accomplish the Agree.