詞頻效果是語言習得研究中最穩定的現象，並用以支持訊息累積的觀點。亦即經驗值增加，是建立詞彙表徵的關鍵因素。但近期研究指出，詞頻對詞彙習得的助益，主要來自於高頻詞更容易出現在不同的文本中（語境多樣性）。而語境多樣性反映的是單純的經驗累積，還是不同文本語境的語意變異，仍有爭議。另有研究使用潛在語義分析方法來計算語意多樣性，以分離語言經驗值與文本中的語意變化。然而語境多樣性與語意多樣性高度相關，其對詞彙學習的影響本質有待釐清。因此本研究使用事件相關電位探討語境多樣性與語意多樣性如何影響中文詞彙辨識的本質。實驗一操弄雙字詞的語境多樣性與語意多樣性，邀請大學生進行具體性判斷作業，其中具體詞與抽象詞各半，分析語境多樣性、語意多樣性與具體性對N400振幅的影響。結果發現具體性與語意多樣性有交互作用，語意多樣性效果僅在抽象詞中被觀察到，語意多樣性愈高，N400愈往負向，因此支持事件基礎理論框架對詞彙表徵建立的看法，亦即詞彙特性與語境的語意變化度皆很重要。然而實驗一並未發現語境多樣性效果，有可能因為語境多樣性和語意多樣性有顯著相關，或是年輕人累積的語言經驗中，語境多樣性的差異無法彰顯。因此實驗二以健康長者為受試者，並調查其閱讀經驗，結果發現具體性、語境多樣性和語意多樣性三者具交互作用，而且老年人的閱讀時間也有影響。在具體詞上，高語境多樣性或是閱讀時間多時，語意多樣性愈高，N400愈往正向。綜合結果支持語境多樣性反映熟悉度，語意多樣性則反映語境意義的變化程度。而詞彙的具體性，也對這兩項指標有不同的影響程度。;Word frequency has been a well-established effect in word acquisition to support the information accumulation perspective. Recent studies indicated that at least part of the word frequency effect is caused by high-frequency words being experienced in more contexts (contextual diversity, CD). However, it is unclear whether contextual diversity is just another index to reflect the accumulation of linguistic experience or the semantic variation across contexts. Thus, researchers also used latent semantic analysis to measure semantic diversity (SD). This study used the event-related potentials (ERPs) to investigate the nature of CD and SD in Chinese word recognition. Experiment 1 invited college students to perform the concreteness judgment task on 280 Chinese two-character words that varied in CD, SD, and concreteness. The single-trial analysis and linear mixed model for statistical analysis were applied to evaluate how these factors modulate the N400 amplitude. The results showed a significant interaction between concreteness and SD. As SD increased, the amplitude of N400 was more negative in reading abstract words. However, there was no CD effect. These findings support the instance-based theoretical framework that the concreteness of a word and the variations of contexts are also crucial for a word’s lexical quality. However, the absence of the CD effect in young adults may be due to the high correlation between CD and SD and their insufficient linguistic experience to reveal the CD effect. Thus, Experiment 2 conducted the same experiment on healthy old adults and also conducted their reading experience by questionnaire. The results showed significant three-way interactions. The SD effect, namely the increased SD elicited more positive N400, was significant in reading concrete words with HCD and for elders with more reading experience. In conclusion, our findings support that CD mainly reflects the word familiarity based on information accumulation while SD reflects the semantic variation across contexts. The multiple exposures of a word in diverse and meaningful language environments are crucial for word leaning, especially for abstract words.
