作者: J.E. Serkhane , J.L. Schwartz , L.J. Boë , B.L. Davis , C.L. Matyear
DOI: 10.1016/J.WOCN.2006.10.002
关键词: Set (psychology) 、 Contrast (statistics) 、 Speech recognition 、 Babbling 、 Coarticulation 、 Formant 、 Psychology 、 Transcription (software) 、 Speech production 、 Vocal tract
摘要: Abstract Articulatory exploration enables the infant to discover abilities of vocal tract and learn relationships between movements percepts. However, neither direct measurements nor transcription methods have access tongue configurations in pre-linguistic vocalizations. A statistical articulatory–acoustic model integrating non-linear growth human was used describe behavior before at beginning canonical babbling. Analyses were developed assess from a set (F1, F2) formant frequencies reported 4 7 months two separate corpora most likely articulatory degrees freedom model. Results indicate that 4-month corpus is centered around neutral configuration. It involves least three parameters, including one for tongue. The jaw seems play minor role this exploration. In contrast, 7-month corpus, range increases: case plays dominant role, leading large exploitation open–close contrast associated F1 diversification space. simulation co-occurrences closants vocants framework Frame-Content theory provides portrait largely consistent with previously experimental data. Locus scatter-plots also simulated compared available data on development coarticulation CV syllables. This kind analysis could be applied infants’ vocalizations various ages understand speech production relation tract.