學前華語發展性語言障礙兒童之語句複誦表現

Abstract

本研究旨在探討華語發展性語言障礙（Developmental Language Disorder, DLD）兒童與典型發展（Typical Language Development, TLD）兒童在語句複誦（Sentence Repetition, SR）表現的差異，並分析語句複誦能力與語言能力之間的關聯性。語句複誦涉及聽覺處理、工作記憶、語法解碼等複雜的心理語言歷程，與個體的記憶及語言能力密切相關。研究顯示，語句複誦作業是能夠區分診斷DLD兒童的有效工具，但目前針對華語語句複誦作業編制的研究仍相對較少，因此，本研究發展一套自編華語語句複誦作業，透過七種評分指標（完整複誦得分、音節錯誤數量、詞語複誦得分、詞語錯誤數量、語法得分、核心結構得分、構音錯誤數量）進行兒童語句複誦表現的評分，檢視語句複誦的七項評分指標對於區分DLD與TLD兒童的敏感性和特異性，以驗證此作業臨床診斷DLD兒童的有效性，同時也分析兒童在九種句型（直述句、疑問句、把字句、被字句、兼語句、複雜句、關係子句、條件句、被把條件句）的語句複誦表現。 本研究招募25名五至六歲的華語DLD兒童及21名華語TLD兒童，除了施測修訂學前兒童語言障礙量表和托尼非語文智力測驗（第四版中文版）之外，兩組兒童均施測研究者自編之語句複誦作業，此作業共包含華語的9種主要句型，每種句型共有4個句子，每一個句子視為一題，總計36題。研究者依七項評分指標，分別計算兩組兒童在不同句型的語句複誦得分或錯誤數量。 研究結果顯示，DLD兒童在語句複誦作業的七項評分指標表現均顯著較TLD兒童低落，且兩組兒童的語句複誦能力與其語言理解及表達能力之間皆存在顯著的相關性。兩組兒童在語句複誦作業中九種句型的語句複誦表現，以直述句和疑問句為最佳，兼語句為最差，關係子句和複雜句為次差。以自編語句複誦作業的表現區辨DLD與TLD兒童的分類具良好的敏感性和特異性，亦具備良好的信度，包括評分者間信度、再度信度及內部一致信度。而所有評分指標中又以語法得分指標對於兩組兒童的區辨能力最強，構音錯誤數量指標區辨能力最弱。整體而言，本研究之自編華語語句複誦作業能有效區分DLD與TLD兒童，並具有穩定的敏感性、特異性、信度及效度，適合作為臨床篩檢華語DLD兒童的工具。This study explores the differences in sentence repetition (SR) performance between Mandarin-speaking children with Developmental Language Disorder (DLD) and those with Typical Language Development (TLD), as well as the relationship between SR ability and language skills. Sentence repetition involves complex psycholinguistic processes, including auditory processing, working memory, and grammatical decoding, and is closely associated with memory and language abilities. Previous research has shown that SR tasks are effective tools for diagnosing DLD, but studies on the development of Mandarin SR tasks remain limited. To address this gap, this study developed a Mandarin SR task and evaluated children’s performance using seven scoring indicators: binary score, number of syllable errors, word repetition score, number of word errors, grammatical score, core structure score, and number of articulation errors. The sensitivity and specificity of these indicators in distinguishing DLD from TLD were examined to validate the task's clinical utility. Additionally, children’s SR performance was analyzed across nine sentence types, including declarative, interrogative, ba-construction, passive, pivotal, complex, relative clause, conditional, and ba-construction passive conditional sentences. Participants were 25 Mandarin-speaking children with DLD aged 5 to 6 years and 21 children with TLD of the same age range. In addition to completing the self-designed SR task, individual children were assessed with the Revised Preschool Language Scale (PLS) and the Test of Nonverbal Intelligence-Fourth Edition (TONI-4). The SR task included 36 sentences spanning nine major sentence types, with four sentences per type and one question per sentence. The scores or errors for each sentence type were calculated based on seven scoring indicators. Results indicated that DLD children scored significantly lower than TLD children across all seven SR indicators. Among sentence types, declarative and interrogative sentences were the easiest, while pivotal sentences were the most challenging, followed by relative clauses and complex sentences. Significant correlations were observed between SR performance and both language comprehension and expression in both groups. The self-developed SR task demonstrated strong reliability (inter-rater, test-retest, and internal consistency), validity, sensitivity, and specificity for distinguishing DLD from TLD. Among all indicators, the grammatical score exhibited the highest discriminative power, while the number of articulation errors showed the lowest. Overall, the Mandarin SR task effectively distinguishes DLD from TLD children and could serve as a reliable and valid tool for clinical screening.