本文嘗試從內部時間結構、MARVS與構式語法等三種認知角度來辨析臺灣華語視覺感知構式「看」、「看見」、「看到」與「看完」的語意差異。研究結果發現，相較於內部時間結構與MARVS的分析存在著若干弱點，結合語法化觀點的構式語法理論更能區別及解釋「看」、「看見」、「看到」與「看完」的語意差異，顯示認知語言學內部的各種理論之間也具有解釋力的強弱區別。本文除了發現構式的象徵性、圖式性與組成成份的語法化程度是造成「看」、「看見」、「看到」與「看完」語意差異的關鍵因素外，還發現相對於「看見」、「看到」與「看完」的語意只侷限於視覺感知域，「看」的語意卻可藉由範疇化及概念轉喻而同時涵蓋視覺感知域與認知域。綜合言之，本文的研究使我們更具體的了解到構式語意、概念結構與語法行為間的密切關聯。

According to the view of temporal structure, MARVS and construction grammar, this research attempted to differentiate the semantic structures of the four constructions of visual perception, viz., kan看, kanjian看見, kandao看到 and kanwan看完, in Taiwanese Mandarin. The result showed that the combining of construction grammar and the theory of grammaticalization can differentiate and explain the semantic structures of kan看, kanjian看見, kandao看到 and kanwan看完 more perfectly than the perspectives of temporal structure and MARVS. Based on the construction grammar and the theory of grammaticaliza-tion, the finding displayed that the semantic differences among kan看, kanjian看見, kandao看到 and kanwan看完 were resulted from the symbolicity and schematicity of constructions and the degree of grammaticalization of their units. In addition, it can be shown that the meaning of kan看encompasses both visual perceptional domain and cognitive domain because of the mechanisms of catego-rization and metonymy, while the meaning of kanjian看見, kandao看到 and kanwan看完is limited within the visual perceptional domain. In conclusion, this study found not only the close relationship among constructional semantics, con-ceptual structures and syntactic behaviors, but also the importance of construc-tional semantics in probing the interaction between cognition and linguistic structures.