連結學齡前兒童自發語言模式與未來語言能力：以典型比例作為發展指標之三年同期及縱貫追蹤隊列研究

// Connecting Preschoolers' Spontaneous Speech Patterns to Future Language Skills: A 3-Year Concurrent and Longitudinal Cohort Study of Canonical Proportion as a Developmental Index

本研究利用深度學習分析155名2至5歲兒童自然語音錄音｜測量語音中正確音節比例，並與標準化語言測試做比較｜結果顯示此比例能有效預測語言能力，特別是子音發音與詞彙｜顯示自動化語音分析可補充傳統實驗室評估，提升語言發展監測效能。

// This study investigated whether automated analysis of naturalistic child-centered audio can track speech-language development from ages 2 to 5. Using deep learning, researchers measured the canonical proportion of well-formed syllables from recordings of 155 preschoolers and compared these to standardized speech-language tests. Results showed that canonical proportion strongly predicted later speech abilities, especially consonant articulation and vocabulary, and moderately related to phonological skills. The findings suggest automated naturalistic speech measures can effectively complement traditional lab assessments of language development.

Published 2026年8月6日