Speech errors in fast speech of alveolar and alveo-palatal consonants: A preliminary study using ultrasound

In this study, we observed speech errors that occurred when alternating between alveolar and alveo-palatal consonants during fast speech, through acoustic analysis and ultrasound tongue imaging. We selected phonologically parallel contrasting pairs in Japanese, /sa ɕa/ and /ta tɕa/. Three native Japanese speakers participated in the experiment. They completed a preliminary questionnaire and reported that despite having no history of speech or hearing disorders, they were aware of experiencing speaking difficulties. Participants were asked to pronounce bisyllabic phrases repeatedly in time with the metronome (BPM = 480). Their pronunciation was recorded using a microphone and ultrasound. The experimenter then auditorily judged each token as correct or errant pronunciation. We measured each consonant’s spectral center of gravity (CoG) using Praat and analyzed ultrasound images of each consonant using AAA. We calculated the error rates for each token and found that Speaker 1 had a 44% error rate for both /sa ɕa/ and /ta tɕa/, Speaker 2 had a 25% error rate for /sa ɕa/ and 53% for /ta tɕa/, and Speaker 3 had a 27% error rate for /sa ɕa/ and 53% for /ta tɕa/. Additionally, the CoG and tongue contours during errors often differed from those during correct pronunciations.