中文字詞在閱讀中的激發時間歷程 多數研究以中文多字詞的辨認歷程是整體的還是每單字是被個別處理來探討中文 的閱讀單位問題，但實徵研究並未有一致的結論。除斷詞的研究外，亦較少論及閱讀理 解時的處理單位。Kintsch 的建構整合模型描述了一個英文的文章閱讀理解的歷程，認 為每一個單字的所有關聯意義在閱讀中都會被激發，然後僅有與文章意義關聯的部分被 留下並整合出整體文章的意義。本研究以建構整合模型為基礎，觀察文章中單字與雙字 詞的關聯意義激發的時間歷程。探討中文在閱讀中的建構單位應該是單字還是雙字詞。 360 位大學生將被隨機分派到4 促發-目標關係（單字關聯、雙字詞關聯、文章關聯與控 制組）× 3 刺激開始異步（SOA，300ms、400ms、750ms）情境中。文章以逐字方式呈現， 並以文章中的特定字詞為促發刺激，受試者必須在閱讀文章的同時對促發字詞後出現的 目標字進行真假字判斷，如果中文閱讀的單位是單字，我們預測單字關聯目標字的反應 時間在短SOA 時會顯著快於控制組，隨SOA 增加這個促發效果會漸漸消失；而雙字詞關 聯目標字在較短的SOA 時沒有效果，隨SOA 增加雙字關聯目標字則可出現抑制效果；文 章關聯目標字同樣的較短的SOA 時沒有效果，但在較長的SOA 時會有顯著的促發效果。 The Activation Time Course of Chinese Words in Reading Comprehension Most researchers deal with the question Chinese are red by multi-characters words or by characters. They tried to identify the recognition process of Chinese multi-characters words is holistic or by characters, but they did not find a consistent conclusion. Except some researched parsing process of reading Chinese, these researches did not consider situation of Chinese text reading. Kintsch’s Construction-Integration Model describe process of reading comprehension of English text: every related meaning of a word in text would be activated while reading English text, no matter it related to context or not. And irrelevant meaning would be inhibited while reader integrates all meaning of words in text. This research try to find which one is the construction unit in Chinese reading, two-character words or character only. 360 college students will be randomly assigned to one of 4 prime-target relationships (character-related, two-character word-related, context-related, and control) × 3 stimulus onset asynchrony (SOA, 300ms, 400ms, 750ms) situations. The Chinese texts would be displayed on monitor character by character, and target character will be displayed after specific prime character. Participants should do reading comprehension task to the Chinese text and do lexical decision task to the target character at the same time. The response time will be recorded by computer. We hypothesize the construction unit of Chinese reading is single character. We predict response time of character-related target will be faster significantly than control target at shorter SOA, and this priming effect will decrease with SAO increasing, and no effect at longest SOA. The response time of word-related target will be no significant difference with control at short SOA, and will have inhibit effect at longest SOA. The response time of context-related target will be no significant difference with control at short SOA, and will have significant priming effect at longest SOA.