早期療育醫療人員常困擾當幼兒被診斷為「遲語兒」後，要如何預測他們日後的語言成長型態和速度，以及在學齡前期出現「兒童語言障礙」的風險。本研究主要目的在探索遲語兒的早期語言發展型態和詞彙學習歷程，並評估這些早期指標的預測效果。研究參與者為32位兩歲「遲語兒」，量測他們靜態的語言表現（如，詞彙表達量、語句表達複雜度、接收性語言、表達性語言）和動態的詞彙學習歷程（快速連配、音韻工作記憶和聲調知覺）等能力，日後再於他們三歲與四歲時蒐集接收性語言與表達性語言能力。在四歲時根據語言能力，區分成「語言障礙」或「語言晚熟」組。結果顯示，這兩組兒童在生理年齡、非語文認知能力和生理狀態方面沒有顯著差異，但兩組兒童於兩歲時已在接收性語言、詞彙學習歷程的快速連配和四聲聲調指認有顯著差異；階層模式（HLM）分析發現控制母親教育程度、語言療育和幼兒教育等因子，「快速連配」與「四聲聲調指認」仍能正向調控遲語兒兩歲至四歲的語言成長斜率；另羅吉斯迴歸分析發現兩歲的「接收性語言」能預測遲語兒在四歲診斷為「兒童語言障礙」的風險，「快速連配」的預測性在邊緣顯著範圍。總結這些結果，意涵遲語兒於兩歲時，理解口語訊息、聲調知覺及詞彙學習歷程表現是預測日後語言發展變化的早期指標。此外，研究結果有利於發展評估兩歲遲語兒童，日後出現語言障礙風險的評量工具。

Exploring the predictors of late-talking (LT) children's language development trajectories from toddlerhood to the preschool stage and the benefits of diagnosing a language disorder (LD) at the preschool stage is essential in the clinical practice. Purpose: This longitudinal study examined whether early language abilities and word-learning abilities were predictors of language development trajectories of LT toddlers from toddlerhood to the preschool stage. Specifically, the study investigated whether 2-year-old LT toddlers who were later diagnosed with language disorders at the age of 4 exhibited distinct language profiles and word-learning processes at various ages. Methods: The data in this panel study were collected at three time points (age 2, age 3, and age 4). This longitudinal study recruited 32 two-year-old LT toddlers. At age 4, LT children were classified as having LD or late bloomers (LBs). At age 2, data on the vocabulary size, the sentence complexity, fast-mapping ability of word-learning process, phonological working memory, and lexical-tone perception were collected. In addition, data on receptive and expressive language skills and cognitive ability were gathered at the three ages. Results/Findings: In nonverbal cognitive abilities, the LD and LB groups performed similarly at ages 2, 3, and 4. However, at age 2, the receptive language skill, fast-mapping ability, and lexical-tone perception of the LD group were poorer than those of the LB group. Additionally, after controlling for mother's education level, months of language intervention and attending daycare, the results of the hierarchical linear model revealed that 2-year-old LT toddlers with better fast-mapping ability and lexical-tone perception exhibited faster development of receptive and expressive language skills from ages 2 to 4. Finally, the results of the logistic regression analysis demonstrated that 2-year-old LT toddlers with better receptive language skills were less likely to be diagnosed with LD later at age 4. Conclusions/Implications: At age 2, receptive language skill, fast-mapping ability, and lexical-tone perception were potent predictors of language development trajectories of LT children from toddlerhood through the preschool stage and LD diagnosis later at age 4. Receptive language skill, fast-mapping ability, and lexical-tone perception were essential in fostering the language learning process of LT toddlers. The findings of this study did not only reveal the predictors of LT toddlers' language development but also shed light on designing assessment tools for identifying 2-year-old toddlers who are at a high risk of exhibiting LD later.