Language-Specific Development of Noun Bias Beyond Infancy

Monami Nishio ^1* Ayuha Koyanagi ² Hiromu Yakura ³ Takaya Hanawa ⁴ Shoi Shi ⁵

• ¹ National Center for Child Health and Development (NCCHD), Tokyo, Japan
• ² Fvital inc., Tokyo, Japan
• ³ Max Planck Institute for Human Development, Berlin, Berlin, Germany
• ⁴ The University of Tokyo Hospital, Tokyo, Japan
• ⁵ University of Tsukuba, Tsukuba, Ibaraki, Japan

Speech and language delays can pose significant developmental challenges, often impacting learning, literacy, and social skills. Early detection of these delays, particularly through monitoring vocabulary development, is crucial. A well-known phenomenon in early language acquisition is the "noun bias," where infants acquire nouns faster than verbs. While studies have shown that this bias exists across languages, its trajectory beyond infancy is unclear, especially in terms of how the noun-to-verb ratio evolves over time. In this study, we employed an AI-driven voice analysis pipeline to examine vocabulary development in Japanese- and English-speaking children across a broad age range. Our findings revealed that noun growth plateaued earlier in English than in Japanese, resulting in a more pronounced noun bias beyond infancy in Japanese. These results suggest that the noun bias seen during infancy may gradually align with adult noun-to-verb ratios, which differ substantially between languages (e.g., 23,800:7,921 in English1 vs. 71,460:7,886 in Japanese2). This study highlights the value of AI-based analysis tools in advancing our understanding of language development and underscores their potential for both research and clinical applications, particularly in identifying and assessing language delays or disorders.