本研究目的是發展一套適用於華語學前兒童的非詞複誦作業，探討華語學 前一般發展兒童非詞複誦能力的發展，並檢驗此作業應用作為特定型語言障礙 兒童早期臨床篩選指標的有效性。實驗一招募4 至6 歲的一般發展兒童，包含4 歲組31 名、5 歲組29 名以及6 歲組30 名兒童，蒐集其標準化智力和詞彙測驗 以及非詞複誦的表現。作業包含二至六音節的非詞，以音節為單位計算出非詞 複誦作業正確率指標。結果顯示，臺灣學前兒童的非詞複誦能力有隨著年齡成 長而成熟的趨勢，尤其是在音節長度較長的非詞中更為明顯。不同的年齡組內 均可發現明顯的音節長度效應，顯示本項作業可以測量到兒童的語音短期記憶 容量。另外，華語學前兒童非詞複誦能力和詞彙理解能力之間有正向相關性， 亦顯示本項作業具有良好的效度。實驗二比較華語學前特定型語言障礙兒童和 一般發展兒童在本項作業的表現差異，以及檢視本項作業的臨床診斷正確性。 共有26 名SLI 兒童以及29 名一般發展兒童參與。結果發現顯著的組間差異， 華語4 至6 歲SLI 兒童的非詞複誦表現較一般發展兒童差，此差異顯現在不同 音節長度上。利用區別分析所得的敏感性、特異性與概似比作為臨床診斷正確 性的指標時也顯示，本研究所設計的非詞複誦作業的臨床診斷正確性達到研究 建議的良好範圍之內。期望本項非詞複誦作業未來能夠發展為標準化測驗，以 協助SLI 兒童早期篩選的工作。

Purposes: The aim of this study was two-fold. First, we developed a self-designed nonword repetition (NWR) task to evaluate the performance of NWR in 4–6-yearold Mandarin-speaking children with typical development. Second, we examined the performance of NWR in children with specific language impairment (SLI) and agematched typically developing (TD) children in order to explore the possibility of using NWR accuracy as a clinical marker for Mandarin-speaking preschool children with SLI. Methods: In Experiment 1, 90 TD preschool children (with 31 being aged 4 years, 29 being aged 5 years, and 30 being aged 6 years) were recruited. The children’s performance on the standardized nonverbal IQ test, receptive vocabulary tests, and selfdesigned NWR task was assessed individually. The NWR task included 30 nonwords comprising low-frequency Chinese real words ranging from 2 to 6 syllables in length. The total number of correctly repeated syllables for each child was used to calculate the percentage of correctly repeated syllables that was considered as the NWR accuracy index. In Experiment 2, we compared the NWR abilities of 26 children with SLI who were aged between 4 and 6 years and 29 age-matched TD children. Children who scored 1 standard deviation below the means in two of the three standardized language tests were confirmed to have SLI. The sensitivity and specificity of the NWR task for discriminating between the two groups of children were analyzed using discriminant analysis. Additionally, positive and negative likelihood ratios were reported. Results/ Findings: In Experiment 1, Significant effects of syllable length were found across the three age groups: children’ repetition accuracy decreased as the length of nonwords increased. This phenomenon supported the validity of the self-designed NWR task for assessing the capacity of phonological short-term memory in preschool-aged children. Moreover, the NWR accuracy increased with age, and the age-related change was more evident in nonwords with longer syllables. The NWR accuracy was highly correlated with children’s receptive vocabulary ability. Compared with the results from NWR studies conducted using other languages, our data provided supportive evidence for the clinical usage of the proposed Mandarin NWR task. In Experiment 2, the results obtained after partialling out the influence of nonverbal IQ revealed that compared with the TD children, the children with SLI performed more poorly on the NWR task consistently for the different syllable lengths. The overall diagnostic accuracy of the NWR task was 87%, suggesting that the proposed NWR task could adequately distinguish between Mandarin-speaking preschool children with and without SLI. Conclusions/Implications: In summary, our study provided evidence indicating the phonological short-term memory deficits in Mandarin-speaking children with SLI and the possibility of using the proposed NWR task in the early stages of SLI screening procedure. We expect to extend this self-designed NWR task to a standardized test and expand its applicability in future studies to improve the early identification of children with SLI.