「顏色」並非無聲、「淺」也不是深：驗證無聲成分的兩個個案研究

Color isn’t Silent, Shallow isn’t Deep: Two Case Studies of Evaluating Silent Elements

由於 Kayne 一系列作品的影響，近年來句法學文獻中「無聲成分」(silent element) 的數量激增。然而無聲成分在文獻中卻缺乏明確的定義，因此也少有針對特定無聲成分的實證或測試研究。本文採用 Her & Tsai (forthcoming) 對詞彙項目 (lexical items) 的分類，將無聲成分清楚定位，並且將文獻中的無聲成分區分出不增加語意的「本質 (intrinsic)無聲成分」與增加語意的「非本質 (extrinsic) 無聲成分」，後者是句法所不允許的。我們接著對 Kayne 提出的兩個無聲成分的分析進行嚴格的語意與句法驗證。一是 The lake is LITTLE BIT shallow 中的 LITTLE BIT（大寫表示無聲）。我們證明其屬性為「非本質無聲成分」，因其造成來源形式 (source form) 與表層形式 (surface form) 的語意不同。驗證的另一無聲成分是 John’s car is a bright green COLOR 中的 COLOR。其雖為「本質無聲成分」，但我們以句法測試證明 COLOR/color 並非自由變化 (free variation)，而且顏色形容詞很清楚地具有名詞的兼類。因此，顏色形容詞在句法上都伴隨著 COLOR 的分析難以成立。我們進一步從語言習得的觀點論證，以上兩個無聲成分也都難以成立。本文同意 Her & Tsai (forthcoming) 的看法：雖然無聲成分是語言中不可或缺的，但研究者所提出的每個無聲成分都必須在語意與形式上有充分的動機與證據。

A notable recent trend in syntax is the proliferation of silent elements (SEs), due largely to work by Kayne. Yet the exact grammatical status of such SEs is not always made explicit. Consequently, few have been explicitly and empirically tested. In this paper we adopt the taxonomy of lexical items in Her & Tsai (forthcoming) and distinguish two kinds of SEs in the literature: an intrinsic SE is one that does not add any meaning, while an extrinsic SE is one that does and is thus by definition illicit. We then put two of Kayne’s SE analyses to rigorous semantic and syntactic tests. One concerns the adjective shallow, as in ‘the lake is LITTLE BIT shallow’ (Kayne 2006), where capital letters indicate silence. We demonstrate that LITTLE BIT are extrinsic SEs, as the source form and the surface form are not semantically equivalent. The other SE examined is COLOR, an intrinsic SE in ‘John’s car is a bright green COLOR’ (Kayne 2005b:Chapter 10). We demonstrate with syntactic tests that it is not free variation between COLOR/color and that color adjectives also function unmistakably as nouns. Thus, the assumption of an SE head noun COLOR for all color adjectives cannot be justified. We further demonstrate that from the perspective of language acquisition the two SE accounts are likewise problematic. This study therefore concurs with Her & Tsai (forthcoming) that, while SEs are indispensable in language, for each SE proposed there must be sufficient semantic justification and formal motivation.