中華心理學刊 民 68, 21卷, 1期, 41-48
Chinese Journal of Psychology 1979, Vol. 21, No.1, 41-48
黃榮村（國立臺灣大學）
摘要
高出現次數英文字往往比低出現次数英文有較高的聯想值，而且聯想反應往往是針對整個刺激字。但在包括雙字以上的中文詞的聯想中，由於中文詞的組成字也各有其獨立的意義在，這些組成字可能因此對聯想過程有所影響。本文旨在提出一與中文詞聯想有關的 「依時性可分離假設」，並提供對該假設的實驗驗證。該假設認為若對中文詞進行聯想的時間夠長，則受試者不僅會從該中文詞中尊找聯想的素材，也會對該中文詞的組成字做聯想反應。依據該假設，則高出現次數中文詞不一定比低出現次數中文詞有較高的聯想值，而且聯想反應的對象也不一定祇是該中文詞本身，可能也會從它的組成字尋找聯想的材料。本文提出兩個實驗嘗試驗證該假設的可行性。實驗一指出了在驗證中文雙字詞聯想的 「依時性可分離假設」中，所可能產生的實驗干擾與解釋上的困難。實驗二進一步分別操弄了中文雙字詞(W)及其組成宇(W1：與 W2)的出現次數，共得五組實驗組，質驗結果指出受試者在剛開始時，主要是利用雙字詞中的第一個組成字（W1） 做聯想，當聯想時間增長後，聯梖速度减慢，受試者閉始利用第二個組成字(W2) 當爲聯想對象。道些實驗結果積極支持了中文雙字詞聯想的「依時性可分離假設」。
Time-Dependent Separability Hypothesis of Chinese Words Association
Jong-Tsun Huang(National Taiwan University)
Abstract
A high correlation of word frequency with word associations has been well documented. It was hypothesized in the present study that such a correlation may not occur in associations of Chinese two-character words. It should be noted that each constituent character of a two-character word has its own unique meaning, independent of the meaning of the word itself. Thus, when the subject is required to associate as many words as possible in response to a presented word without a stringent time restriction, he may first generate words as response to the stimulus word as a whole, and then, if he runs out of his associations, he may generate words by considering each constituent as a stimulus. According to this time-dependent separability hypothesis, there would exist no correlation between word frequency and word association of the two-character words. Two experiments were thus designed to test this hypothesis. Experiment 1 mixed and presented the single- and two-character Chinese words, with their varying frequency counts, to the subject. The subject was required to respond as many associates as possible to each presented word. The results showed that the differential responding rate between the low- and the high-frequency single-character Chinese words was not significantly different from that between two-character ones. Possible experimental confoundings were then discussed and these defects were excluded in Experiment 2. Frequency counts of two-character Chinese words and of their constituent characters were independently varied in Experiment 2, resulting in a total of five experimental groups. The results showed that the subject produced associates by first responding to the first constituent of two-character Chinese word. As the time passed, he shifted to respond to the second constituent. These results strongly supported the time-dependent separability hypothesis.
