Send message to Authors

Paper short abstract

Japanese vocabulary consists of three strata with distinct phonotactics. This study tested whether phonotactics guide kana choice (hiragana vs. katakana) across age groups using low-familiarity and nonce words. Results show strong phonotactic effects with developmental shifts to lexical knowledge.

Paper long abstract

Japanese vocabulary is traditionally divided into three strata—native (wago), Sino-Japanese (kango), and loanwords (gairaigo)—each characterized by distinct phonotactic patterns.Previous research has shown that phonotactic characteristics influence lexical stratum inference, and recent large-scale studies have demonstrated that clusters corresponding to lexical strata can be learned from phonological information alone. Appropriate orthographic choice according to lexical stratum relies on accumulated phonotactic knowledge specific to each stratum from auditory experience. However, the mechanisms by which phonotactic cues are utilized to infer lexical stratum and guide orthographic selection remain unclear. Moreover, little is known about how this mapping process changes across development.

This study investigated whether phonotactics guide kana orthographic choice (hiragana vs. katakana) and how this process differs across age groups. To minimize lexical and orthographic knowledge effects, we employed low-familiarity real words and nonce words. A total of 21 stimuli were used: 10 low-familiarity real words, 2 nonce words, 5 onomatopoeic words, and 4 words with variable orthographic usage. All stimuli were presented auditorily, and participants judged whether each word should be written in hiragana or katakana, with a 15-second response limit. Participants included 21 adults and 96 primary school children.

Results revealed systematic effects of phonotactic patterns on orthographic choice. Stimuli with native-like phonotactics elicited higher hiragana selection rates across all age groups. Conversely, stimuli with loanword-like features (voiced consonants, geminated consonants, long vowels) showed higher katakana selection rates, with particularly strong effects observed for nonce words. Onomatopoeic words and words compatible with multiple lexical strata exhibited more balanced selection patterns. Developmental differences were also observed. Younger children relied more strongly on phonotactic cues, whereas older children showed increased influence of lexical experience and orthographic conventions. For example, hiragana selection for the low-frequency native word /kamachi/ declined substantially from younger to older children.

These findings suggest that phonotactic features serve as robust cues for lexical stratum inference and orthographic selection, even at early developmental stages. Furthermore, the results support a dynamic inferential account in which phonotactic cues initially dominate orthographic choice but are gradually integrated with lexical knowledge and orthographic conventions as linguistic experience accumulates.