Past Event: Friday Lunch Talk: Deriving sibilant-vowel phonotactics from a soft bias in perception (Ayla Karakaş)

Talk Abstract: Past research has demonstrated that listeners discretize the continuous speech signal as sequences of sounds. In auditory “sequences” of fricatives and vowels, e.g., /sV/ and /ʃV/, listeners can anticipate the identity of the vowel during aperiodic energy of a preceding fricative, even before hearing the periodic energy of the vowel (Yeni-Komshian & Soli, 1981; Galle et al., 2019).  An obvious source of this apparent anticipation is coarticulation: for some period of time, both the fricative and the vowel are being articulated simultaneously. This coarticulation can be observed in the acoustic signal, concurrently cueing listeners to both the fricative and the vowel. 

However, coarticulation may not be the whole explanation for anticipatory behavior in speech perception. Anticipation is a domain-general cognitive ability. While the perceptual informativity of coarticulation is well established by past empirical results (e.g., Whalen, 1989; Nittrouer & Whalen, 1989), it is not yet clear whether it accounts for the totality of the anticipation effects that have been observed. 

We propose an additional possibility: listeners may be sensitive to a relationship in the acoustics of the aperiodic energy of a fricative and the periodic energy of a subsequent vowel, even beyond the well-known effects of coarticulation. In this talk, we will present a model of this relationship in the framework of Dynamic Field Theory. We then share results of our empirical study, which extends past perceptual studies to include conditions with and without coarticulatory cues to the vowel, and show how the model informed the predictions of our study.