Speech sample measures in Japanese children and adults who stutter

Abstract

While more information about Japanese stuttering has become available, there is a lack of standardization of speech sample measures. To advance cross-linguistic knowledge about stuttering, we need data on the types of stuttering-like disfluencies (SLDs) in Japanese and more investigation into their developmental patterns. Spontaneous speech from 10 Japanese speakers, aged 3–48 years, was transcribed and coded for morae, syllables, ‘bunsetsu’ or Japanese word units, and disfluencies. A corpus of 122 SLDs was obtained. Most (82.9%) of the SLDs were repetitions. Both C-initial (/t͡s t͡s t͡sɯk:amaeɪɾɯ/) and V-initial (e.g. /u u uɾɑ/) sound repetitions were relatively rare (3.1%). About two-thirds (68.7%) of SLDs were syllable repetitions. Light syllable repetitions (e.g. /hɑ hɑ hɑt͡ʃi/) comprised a significantly larger proportion of SLDs than heavy syllable repetitions (HSRs), in which a C-onset transitioned into a diphthong (e.g. /nɑɪ nɑɪ nɑɪjo/) or into a moraic nasal (e.g. /sɑn sɑn sɑnkai/). HSRs negatively correlated with age. In the younger speakers about a third of SLDs were HSRs, whereas for the adolescents and adults, approximately 10% of the SLDs were HSRs. The present findings suggest a conversion of 2.40 syllables to bunsetsu would be appropriate for speakers of Japanese. Results corroborated findings from larger corpora, and future directions for investigations into consonant and voicing part of speech are recommended.