How bilinguals' two languages interact with each other has stimulated considerable research. However, little of this research has focused on objective measures of speech production. This study aims to investigate bilinguals' production of Guoyu and Taiwanese voiceless sibilant fricatives to see how language contact and language dominance influence their production patterns. Guoyu is the Mandarin spoken in Taiwan and Taiwanese is one of the Southern Min dialects. Guoyu has a three-way contrast among three voiceless sibilants, namely, a dental /s/, an apical post alveolar or retroflex /¿¿¿¿¿/, and an alveolopalatal. However, Taiwanese only has one voiceless sibilant, a coronal /s/, which has a “palatalized” quality in the context of a following /i/ or /j/. The mismatch between the two sibilant systems provides an opportunity to compare how bilinguals with different language acquisition orders and language dominance align the two different systems.
The two specific questions that this study aims to address are (1) how bilingual adults of three generations' sibilants productions change over time due to language contact and (2) how bilingual children from 2 to 6 years old in the same context acquire Taiwanese and Guoyu sibilants. Participants were 64 bilingual adults, ranging in age from 20 to 80 years with approximately 20 in each of three adult generations (elderly, middle-aged and young), and 60 bilingual children, ranging in age from 2 to 6. Their productions of target sibilants were elicited in an audio- and picture-prompted repetition task. There were two separate recording sections for Guoyu and Taiwanese. In each section, all the target sibilants were in word-initial position with comparable vowel contexts. Adults' productions were analyzed by measuring (1) the centroid frequency in a spectrum estimated over a window taken at the central 20 ms of each sibilants and (2) the second formant frequency at the consonant-vowel boundary. The children's productions were first analyzed by transcription and then using the same two acoustic measures as for the adults.
When the two acoustic measures are plotted against each other, the plots for the elderly and middle-aged adults show two main clusters for all sibilants. One cluster is for Guoyu alveopalatal and the palatalized allophone of Taiwanese /s/, and the other cluster is a merged category of Taiwanese /s/, Guoyu /s/ and Guoyu /¿¿¿¿¿/. For the elderly and middle-aged males, these two clusters are differentiated primarily in the second formant frequency dimension, and both clusters show centroid values that are intermediate between the high values expected for a dental /s/ and the low values expected for post-alveolar fricatives. However, middle-aged females show signs of fronting their /s/ (i.e., producing a higher centroid value) in order to accommodate to the more complex Guoyu sibilant inventory. Moreover, unlike the older two generations, young adults have a more distinctive Guoyu three-way contrast of sibilants, a higher centroid values for a dental /s/, lower centroid value for a retroflex /¿¿¿¿¿/ and intermediate centroids for an alveolopalatal .
Analysis of the transcriptions of the children’s productions showed that these bilingual children acquire Guoyu dental and alveopalatal at the same age, which is much earlier than their acquisition of the retroflex. The errors for the dental and alveopalatal sibilants are mostly substitutions of stops and affricates at the same place of articulation as the target sibilants. Furthermore, /s/ is often used for Guoyu /¿¿¿¿¿/. In addition, their acoustic data indicates that children make a contrast more like young adults but even at 6 years old, they still do not make clear contrast like adults. Based on the data from the four generations of bilinguals, it seems that in this language contact context, there may be an ongoing language shift from Taiwanese to Guoyu.