An ultrasound study of American English laterals produced by first graders

Longitudinal studies of child speech production development have traditionally relied on perceptual evaluation and acoustics to track acquisition norms and phonological patterns. The present study administers novel techniques in speech imaging to investigate the articulation and acoustics of speech sounds in developing children, focusing on the lateral sound /l/ as produced by first graders. Real-time three-dimensional ultrasound images with synchronous audio and video recordings were collected from L1 English speaking children. The stimuli were sourced from the GFTA-3—a clinical picture-naming instrument—which was administered by a speech-language pathology student-clinician. Three-dimensional tongue surface reconstructions and the accompanying audio signals were then compared across tokens and across speakers to find common and divergent patterns for lateral sounds. The results show highly variable articulatory strategies that nevertheless can be categorized into three groups. The wide articulatory variety is not obviously tied to acoustic variability, which is relatively small; formant positions are relatively consistent across tokens and across subjects, with few outliers. From the acoustics alone, these children consistently produce what is traditionally described as the dark /l/, but articulation suggests greater variety.