Tones in Hakka Infant-Directed Speech: An Acoustic Perspective

Abstract

This study explores lexical tones in Hakka infant-directed speech (IDS) in comparison with adult-directed speech (ADS), investigating whether lexical tones are hyperarticulated and distorted in Hakka IDS. Special attention is directed at the tonal contrast between unchecked and checked tones, an issue never investigated previously. Sixteen mother-infant dyads participated in this study, with infants' age ranging from 6 to 26 months. The speech stimuli contained 18 disyllabic phrases in the form of C_1V_1C_2V(C), where C_1 and C_2 are voiceless consonants and V_1 (a corner vowel [i, a, u]) carried the target tone. Mothers interacted with their infants and the researcher naturally at their homes. Interviews were recorded as IDS and ADS. For each recording, the first two clear tokens of each target tone were segmented out for acoustic analysis of fundamental frequency (F_0) by PRAAT. Results show that lexical tones in IDS are phonetically enhanced by exaggerated F_0 contour, elevated F_0 mean, widened F_0 range, steepened F_0 slope, lengthened F_0 duration, and expanded tonal distance. Yet they are not distorted because they can still be distinguished by one or more F_0 cues. All hyperarticulated cues contribute to perceptually salient linguistic signals, and help infants with tonal identification and categorical learning. More significantly, checked and unchecked tones have different tonal behaviors in ADS and IDS. Both tone types overlap in ADS, but separate in IDS. This tonal separability results from glottalization of the [p, t, k] codas in the production of checked-tone syllables in spontaneous and continuous speech.