Abstract
字彙觸接是閱讀時達到文字辨識的關鍵性步驟，心理語言學相關的研究一 致假設兩個階段︰ 字型辨認的產生，進而觸接語意訊息。但對心理詞彙 庫的基本組成單位則意見分岐。早期的研究報告在「認知精簡性」的前提 假設下，認為文字物理刺激物的書寫表徵及語音屬性，各自根據不同的感 官特質組織成心 理詞彙庫專有的兩個子系統，語意訊息則與其它抽象概 念 儲存於一般認知系統中；詞彙庫只需記錄語言的外在形式表徵。近年 來，隨著研究腦傷失語病患報告數量的增多和病因類別的區分，以及相關 理論的逐年進展，一切跡象皆 顯示：語意成分亦包含於心理詞彙庫中。 即，不同的語言 各有一套獨特的心理詞彙庫，其中，分別記載著語言形 式 和相關的語意訊息。本論文的目的即從雙語使用者的角度 出發，探討 詞彙單位形式和意義之間的關聯：語義概念及其表達形式之間的關聯究竟 是一對多還是一對一？ The purpose of the thesis is to examine the nature of representations in the mental lexicon, i.e., whether the units of the mental lexicon contained forms for access only, or with their semantic specifications as well? Sternberg's memory serial-scanning paradigm with an item-recognition task was adopted. In the experimental paradigm, the reaction-time (RT) is measured for deciding whether a target stimulus has been presented in a previous set of to-be-memorized items. In general, RT has been found to be a linear increasing function of the set size (i.e., equal increase of each item in the to-be- memorized set). This linear function allows an investigator to generate two parameters; an intercept represents encoding, decison making, and response execution, and a slope stands for the time required by each comparison between a probe item and one member from a memory list. By examining the RT-function of one particular factor, i.e., the symbolic type or language for a memorized set and for a test stimulus, one could make inference about the token specifications in the underlying comparison stage. The hypothesis was that knowledge was stored in some separate, language- dependent, representation, Accordingly, the slope of linear function was predicted to vary depending on the factor that whether category consistency was reached between a positive set and a following stimulus. Four experiments were performed. The results of all experiments here supported an unitary representation containing both form and content in the mental lexicon; the separate storage hypothesiswas argued for.