Socio-indexical cues and paralinguistic information are often beneficial to speech processing as this information assists listeners in parsing the speech stream. Associations that particular populations speak in a certain speech style can, however, make it such that socio-indexical cues have a cost. In this study, native speakers of Canadian English who identify as Chinese Canadian and White Canadian read sentences that were presented to listeners in noise. Half of the sentences were presented with a visual-prime in the form of a photo of the speaker and half were presented in control trials with fixation crosses. Sentences produced by Chinese Canadians showed an intelligibility cost in the face-prime condition, whereas sentences produced by White Canadians did not. In an accentedness rating task, listeners rated White Canadians as less accented in the face-prime trials, but Chinese Canadians showed no such change in perceived accentedness. These results suggest a misalignment between an expected and an observed speech signal for the face-prime trials, which indicates that social information about a speaker can trigger linguistic associations that come with processing benefits and costs.

A perceptual dependence of stop consonants on preceding liquids was previously demonstrated by excising the CV portions of natural tokens of /al da/ and /ar da/ and replacing them with synthetic stimuli from along a /da/-/ga/ continuum. [V. Mann, Percept. Psychophys. 28, 407–412 (1980)] Using these “hybrid disyllables” a “duplex percept” has been created by presenting the F3 formant transition of the CV portion to one ear and the remaining stimulus portion to the other ear. Thus the F3 transition is simultaneously perceived as supporting the /da/-/ga/ distinction, and as an isolated “chirp,” Listeners performed a discrimination task on these materials, attending to either the speech percepts or the chirps. Speech discrimination was a clear function of liquid context: Consistent with previous results, the category boundary along the /da/-/ga/ continuum shifted in favor of more “g” responses when stops followed /al/ than when they followed /ar/. Chirp discrimination, however, was not related to liquid context. Since the effect of the preceding liquid appears specific to “perceiving in the speech mode” an explanation cannot be couched in terms of some low-level auditory effect on perception of the F3 transition. Rather, a phonetic explanation of the effect of liquid context remains plausible. [Research supported by NICHD and BRS.]

We conducted an auditory naming task (n=20) using eight model talker voices previously rated for attractiveness and prototypicality such that the Most Attractive, Least Attractive, Most Typical, and Least Typical voice for each gender served as a model talker. Female shadowers accommodated more than males, particularly to the Most Attractive Female model. This finding led us to question if in the course of accommodation to an attractive female voice, female shadowers themselves become more vocally attractive. We then conducted an AX task where listeners judged whether shadowers' baseline or shadowed productions were more attractive. We found that female voices were generally judged to be more attractive in shadowing conditions, regardless of the qualities of the model voice. In a follow-up experiment examining typicality, the shadowers' voices were used in a reaction time experiment where listeners categorized each voice as male or female. Following Strand (1999) more typical voices are those that elicit faster reaction times. Results of this experiment suggest that shadowers' voices, especially female ones, become more prototypical when shadowing typical voices and less so when shadowing atypical ones.

Studies of phonetic convergence using single-word auditory naming tasks offer insight into how variability in stimuli affect the translation from speech perception to speech production. In this paper, we report on an experiment which compares phonetic convergence in single-word production between high variability (mixed talker condition) or low variability (blocked talker condition) using five female model talkers’ voices for the task. Twenty female participants participated in a production task where they produced baseline tokens and shadowed model talker productions in either the high or low variability condition. Phonetic imitation was quantified using listener judgments in an AXB similarity rating task where a model token was compared to a shadower’s baseline and shadowed token. The results indicate a trend towards more convergence in the low variability condition, but this was highly affected by model voice; one model voice was spontaneously imitated more in the high variability condition than the low variability condition. Several socio-cognitive tests were administered to shadowers, and continued analyses of the data will explore whether these individual socio-cognitive measures predict shadowers’ predispositions toward phonetic convergence.

Race, gender, and age have traditionally been considered the three big social categories perceivers attend to. Recent research has highlighted how perceivers, listeners in this case, use the acoustic-phonetic information from spoken language to evaluate and categorize individuals. For example, studies have shown that different dialects elicit varying judgments of intelligence, friendliness and attractiveness. Such biases have implications regarding our assessment of individuals, be it in the clinic or the courtroom. While we know that our perception of speech can trigger biases, it is important to explore the opposite—whether our biases affect our perception of speech. This study aims to contribute to the knowledge base of the latter by examining the effects of listener biases on speech intelligibility. Listeners will be presented with sentences, independently controlled for clarity, produced by model talkers of different dialects and backgrounds. This auditory stimuli will be accompanied by an image or description of the talker as a method of eliciting or constructing a preconceived linguistic bias. After each sentence, participants will be asked to type out what they heard. Analyses will examine the proportion of correct words, confidence ratings, response time, and other measures exploring participants’ ratings of vocal qualities.