記憶廣度及語境效應對閱讀歧義詞彙的影響：來自眼動的證據   免費試閱

Memory Span and Contextual Effects on Lexical Ambiguity Resolution During Chinese Sentence Reading: Evidence from Eye Movements

本研究整合過去的研究，旨在探究工作記憶容量、語境位置及語境意義偏向三個變項對於中文讀者在閱讀含有多義詞的句子時眼動型態之影響，並深入探討語意關聯程度及語境位置的效果。研究發現，讀者擷取詞義的歷程，在語意整合階段會因工作記憶容量的大小而呈現不同的型態。此外，語境位置與語境意義偏向也相互作用於詞義擷取歷程，支持再排序觸接模型。當讀者閱讀多義詞的次要意義時，若語境在前，所有讀者皆受影響且在詞彙觸接時已進行歧義處理，而高廣度讀者則在前語境下相較於後語境，多凝視多義詞以解歧。當語境在後時，低廣度讀者則受到語境意義偏向，以及語意關聯程度的影響，多凝視解歧區而非多義詞。最後，根據研究結果，提出文本安排上的建議。

This study examined the effects of using context to solve the lexical ambiguity in reading sentences by recording participants' eye movement. Variables such as the location of disambiguating context, the semantic bias of context, the degree of meaning relatedness of the ambiguous word, and the participants' memory span were included in the study. The purpose is to investigate how memory span influences reader's reading while they process the ambiguous word in a context supported sentence and the advantage of the contextual clues to reader's reading. The results of the study showed that in the integration stage the location of disambiguating context plays an important role in ambiguous sentences reading. The factors affected the processing of ambiguous sentences, showed the differences depending on the individual's working memory capacity. The study also revealed that both the location and the semantic bias of disambiguating context affect access procedures. In the context-before condition, all participants were affected by the subordinate meaning and deal with the ambiguous word when lexical access. However, in the context-after condition, low working memory span readers were more affected by the semantic bias of context and the degree of meaning relatedness when compared to high span readers. As to memory span, when the location of disambiguating context was compared, high working memory span readers fixate longer on ambiguous word in the context-before condition, and low working memory span readers fixated longer on disambiguating region but not ambiguous word in the context-after condition. These results supported the reordered access model. Based on these results, implications and suggestions were discussed.