本研究主要探討：一、自閉症兒童非語言溝通能力表現；二、非語言溝通能 力及一般認知能力與表達性詞彙間的同時及預測相關性；三、哪些因素最能預測 同時及六個月後的表達性詞彙發展。研究採用「嬰幼兒綜合發展測驗」、「早期社 會溝通量表」及「華語嬰幼兒溝通發展量表」等蒐集資料，共有30 位自閉症兒童 參與，平均生理年齡為45.97 個月，認知發展平均年齡為22.36 個月，表達性詞彙 數在施測第一個時間點的平均為115.67，在六個月後第二個時間點的平均為 218.30。研究發現：一、自閉症兒童最常主動表現的非語言溝通功能是「主動做要 求」，占所有主動溝通行為的64%，其次是低層次的「主動性共享式注意力」，占 24%，再其次是高層次的「主動性共享式注意力」，占12%；二、三項非語言溝通 功能「主動做要求」、「主動性共享式注意力」和「回應性共享式注意力」在控制 生理年齡及認知發展年齡後，彼此沒有顯著相關性；三、最能預測時間點一表達 性詞彙數的同時相關性因素是「回應性共享式注意力」及認知發展年齡，但當二 者一齊考量時，僅「回應性共享式注意力」是時間點一詞彙數的顯著同時預測因 素：四、最能預測六個月後表達性詞彙數的相關因素是「回應性共享式注意力」 及認知發展年齡，且二者皆對六個月後詞彙的預測有顯著貢獻。本研究支持「回 應性共享式注意力」及一般認知能力在自閉症兒童詞彙發展上的獨特影響，並認 為應對不同非語言溝通功能和語言發展的關係進行更多的探討。

This study examined the relationship between prelinguistic communication and language development in children with autism and focused on three main research questions: (1) the prelinguistic nonverbal communication skills of children with autism; (2) the concurrent and predictive relationships among prelinguistic communication skills, cognitive developmental age, and expressive vocabulary at Time 1 and Time 2(6 months later); and (3) the best and significant predictors of expressive vocabulary at Time 1 and Time 2. Methods: Thirty children with autism spectrum disorders participated in this study. The average chronological age(CA) was 45.97 months, and the average cognitive mental age(MA) was 22.36 months. Three assessment tools were used in this study, namely the Early Social Communication Scales (ESCS), Taiwan Mandarin-Chinese version of the CDI, and the cognitive development subtest of a child development test battery for Taiwanese children. A combined expressive vocabulary inventory including all the words in the CDI-toddler form was used to avoid the ceiling effects of using only the infant form and the floor effect of using standardized language tests reported in previous studies. The average number of expressive vocabulary in the CDI-toddler form at Time 1 and Time 2 testing were 115.67 and 218.30, respectively. Results/Findings: (1) Of all spontaneous nonverbal communication behaviors during ESCS testing, 64% were initiating behavioral request (IBR) , and 24% were lower-level initiating joint attention (IJA) behaviors using only eye contact or gaze alternation. Furthermore, 12% were higher-form IJA behaviors involving pointing, pointing combined with eye contact, or showing objects. (2) No concurrent correlations were found among IBR, IJA, and RJA(responding joint attention) after the influences of CA and cognitive MA were partialled out. (3) The correlation coefficients between cognitive MA and the expressive vocabulary at Time 1 and Time 2 were .63 and .80, respectively. After CA and cognitive MA were partialled out, RJA was the only prelinguistic communication skill that had a partial correlation with Time 1 and Time 2 expressive vocabulary. (4) Although both cognitive MA and RJA were significantly correlated with time 1 expressive Vocabulary , when both factors were entered for multiple regression analysis, the only concurrent significant predictor for time 1 Expressive Vocabulary was RJA. (5) When both RJA and cognitive MA at Time 1 were entered for Time 2 expressive vocabulary regression analysis, both cognitive MA and RJA were significant predictors; however, the semipartial correlation coefficients indicated that RJA contributed considerably more to the prediction. Conclusions/Implications: Because the average expressive vocabulary of the subjects at Time 1 and Time 2 was 115.67 and 218.30 respectively, the results confirmed the unique contribution of RJA and general cognitive development to this stage of expressive vocabulary development, expanding the previous results regarding the importance of RJA for children with autism at a younger age (Wu Chiang, & Yu, 2010) or at different language developmental stages (e.g., Siller & Sigman, 2008). In addition, the nonsignificant partial correlations among IJA, RJA, and IBR after CA and cognitive MA were controlled for indicated that the underlying basis for the independence of these three skills should be investigated.