國小二年級學童雙字詞聯想常模

Second-Graders’Word Association Norms of Two-Character Chinese Words

本研究建構華人中文世界中迄今所知的第一個兒童聯想常模，用以探索中文在世界語系中獨特的語言特性，及所對應的發展變化。研究由327位台灣小學二年級學童及272位台灣小學六年級學童，針對150個中文雙字詞進行自由聯想且建立聯想常模，分析每個刺激詞對應的聯想反應類別數、共通性與個別性，每個反應詞的聯想強度，並且計算22 種刺激詞與反應詞聯想關係分類的佔比，以及不同分類對應的聯想強度。本研究發現，功能性的聯想關係所佔比例高於類別關係，不同聯想關係類別在佔比以及聯想強度上有不同的分布。研究也發現學童語意網絡的發展變化，隨著年齡的增加，功能聯想關係與類別聯想關係的聯想比例均呈現上升的趨勢，但是類別聯想關係的相對增幅比功能聯想關係來得大，顯示類別關係在10 歲之後逐漸躍居於語意網絡的主要角色。本常模資料可增進我們對於兒童語意知識及其發展軌跡的了解，提供未來心理語言學相關研究選擇適當實驗刺激使用，有很高的教育領域應用性，並且凸顯了中文研究的理論重要性。

This is the first longitudinal word association norms of Chinese children from first to sixth grade to our best knowledge. The norms are constructed to explore the unique linguistic characteristics of Chinese. In this study, we demonstrated the norm of 327 second graders (170 boys and 157 girls) in Taiwan answered the Chinese word association task of 150 bi-character words. The participants were recruited from three public elementary schools in Taipei with various socio-economic background. The stimulus words were selected with four steps. First, we chose the 400 highest frequency bi-character words in the corpus built from textbooks, various books and essays of first to fourth graders (Minister of Education [教育部國語推行委員會], 2002). Second, we deleted the functional words, names of persons or places, and words usually appeared in idioms to make a 253 words list. Third, we used the method of Nation and Snowling (1999), asked 7 elementary school teachers and 16 third graders to rate the familiarity of these 253 words. Fourth, we selected the 195 highest familiarity words and randomly picked 150 of them as the stimulus words. The frequency of these 150 words ranged from 3,218 to 123 per million characters. The 150 words were arranged in four different orders, printed on sheets which participants were tested in groups in two 40-minute sessions. In this study, we analyzed the responses of 327 second graders to build the free association norms of these 150 words, along with the corresponding numbers of responses, commonality and idiosyncratic index. The average number of response (by type) is 98.04, ranging from 42 to 144. The average of commonality is 0.38, ranging from 0.10 to 0.75. The average idiosyncratic index is 0.22, ranging from 0.06 to 0.38. About the association relationship between stimulus word and response word, we defined 22 non-exclusive types, including superordinate category, same-level category, subordinate category, thematic-instrument, thematic-script, thematic association, meaning-similar, meaning-opposite, causation, sentence, orthography, phonology, feature, compound word, part of, analogy, orthographical mistake, phonological mistake, adjective, adverb, verb, and other. In order to demonstrate the distribution of different association relationship types, we calculated the percentages of all types among the highest association strength pairs, the highest five association strength pairs, and the highest ten association strength pairs of the 150 stimulus words. The results showed that when adding more stimulus-response association pairs into the pool, the percentages of each type varied in different directions. For example, the percentages of thematic-script type increased from 16% (the highest) to 21% (the highest five) and 21% (the highest ten). On the other hand, the percentages of meaning-similar type decreased from 33% (the highest) to 19% (the highest five) and 16% (the highest ten). The results showed that meaning-similar association type was quite popular among the pairs with highest consensus, but the percentages were decreased when more association pairs were adding into the pool. To investigate the developmental changes of semantic network, we further compared the differences between this second-grader norm and the sixth-grade norms of 272 participants which collected longitudinally 4 years later. In this study, we randomly selected 6 nouns, 6 verbs, and 6 adjectives, to compare the percentages of functional and categorical association types between second and sixth graders. The results showed that there was a greater increase in categorical relations than functional relations, suggesting a prominent role in taxonomic relatedness within semantic network for Taiwanese children older than 10 years. Besides investigating developmental changes, we could also compare this longitudinal association norms of Taiwanese elementary school children with norms of different languages. In this study, we compared the associations of the word“school”of different languages (English, Portuguese, and Chinese) and different ages (American adults, Portuguese third graders, Taiwanese adults, and Taiwanese second and sixth graders). The results showed clear difference cross ages and languages. In sum, the association norms enhance our understanding of children’s semantic knowledge and its developmental course, provide stimuli for psycholinguistic experiments, and could be used in the educational field and build up theoretical framework cross different languages.