為瞭解語言依賴因素(language-dependent factors)與語言組織原則如何影響手語詞彙之像似性(iconicity)與透明度(transparency)之感知過程，同時也期與前人研究做跨語言間的對比（美國手語：Bellugi & Klima 1976；義大利手語：Grosso 1993; Pizzuto & Volterra2000），本文以具有正常聽力之受試者為對象進行二組心理語言學實驗來探討其對台灣手語(Taiwan Sign Language, TSL)詞彙之理解過程。二組實驗（實驗一：詞彙意義無提示猜謎作業(free guessing task)；實驗二：詞彙意義選擇題方式猜謎作業(multiple choice task)）所用的詞彙語料以蘇秀芬(2004)對台灣手語像似性所做的劃分標準為基礎，自台灣手語影像辭典(Tai 2001-2005)選用九十個詞彙。二組實驗結果顯示：(1)大多數台灣手語詞彙皆具模糊性（實驗一：83%；實驗二：75%）；(2)實驗一或實驗二中至少有一名受試者猜對的詞彙比例皆比美國手語研究(Bellugi & Klima 1976)的數據高，但至少有一半以上的受試者猜對的詞彙比例比義大利手語研究(Pizzuto & Volterra 2000)的數據低（台灣手語：17%；義大利手語：24%）；(3)從詞彙像似性種類來分析受試者答對比例可發現實驗一中答對的比例從映像(image: 95%)、轉喻象(metonymy: 30%)、擬象(diagram: 30%)至喻象(metaphor18%)呈現下降趨勢，而實驗二中並未呈現此下降趨勢。綜合實驗結果，本研究提出以下二點建議：(1)前人研究所用之語料有其重新檢視的必要性，並建議以統一之劃分標準來選取手語詞彙以便跨語言之比較研究，以避免因採用熟悉度的判斷(familiarity judgment)所造成詞彙間像似性比例的不平均，進而影響研究結論；(2)可比照Pizzuto & Volterra(2000)所執行之歐洲地區跨語言研究，進行大規模亞洲地區之手語像似性感知度之研究，以期更瞭解語言依賴因素與語言組織原則對跨地區及不同文化背景下之手語像似性感知過程的影響為何。

The aim of the current study is to explore the comprehensibility of the lexical items in Taiwan Sign Language (TSL) by hearing subjects as a way of knowing more about the language-dependent factors and language-organizational principles that may influence the perception of lexical signs' iconicity and transparency. Grounded on the previous studies of American Sign Language (ASL) (Bellugi & Klima 1976) and Italian Sign Language (LIS) (Grosso 1993; Pizzuto & Volterra 2000), we pursued the issue by conducting two experiments (Exp. 1: free guessing task, Exp. 2: multiple choice task) based on 90 TSL signs from Digital Graphic Dictionary of TSL (Tai 2001-2005), but with a different criterion in material selection based on Su's (2004) categorization of TSL signs. The results obtained revealed that: First, most TSL signs are opaque (83% in Exp. 1 and 75% in Exp. 2) while only a small number of the signs are transparent. Second, the proportion of the signs that received correctly guessed meanings at least by one subject in either Exp. 1 or 2 was higher than that in ASL (Bellugi & Klima 1976) but the proportion from at least more than 50% of the subjects was smaller than LIS (17% for TSL; 24% for LIS). Third, looking into the proportion of correct answers given for different types of iconic signs, the proportion of correct answers in Exp. 1 patterned a downward tendency from image (95%), metonymy (30%), diagram (30%), to metaphor (18%), but not in Exp. 2. Drawn from the results, we argue for the necessity of re-examination of the materials used in previous studies since the possible imbalanced proportion of the iconic signs may lead to imprecise conclusions drawn from the cross-linguistic comparisons. In addition, Asian cross-linguistic or cross-cultural studies that focus on the role of the culture-related factors in comprehension of sign languages by hearing are more than welcome.