Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Universal vs language-specific processing of acoustic cues: an EEG study of English and Korean
Poster Session D, Thursday, October 1, 4:30 - 6:30 pm, Wangari Maathai
This poster is part of the Sandbox Series.
Chiara Repetti-Ludlow1, Jinhee Kim2, Barbara Shinn-Cunningham2, Christina Bjorndahl2; 1Georgia Institute of Technology, 2Carnegie Mellon University
Listeners make use of acoustic cues to distinguish the sounds of their language. For example, the English laryngeal contrast (e.g. voiced /bitʃ/ beach vs. voiceless /pitʃ/ peach) is cued primarily by voice onset time (VOT) and secondarily by fundamental frequency (f0). Neurolinguistic research on English suggests that N1 indexes the continuous nature of cues such as VOT, while P3 indexes more categorical information, with larger amplitudes for more prototypical cues. However, languages differ in which contrasts they employ and in the relative weighting of acoustic cues. It therefore remains unclear which aspects of these English findings reflect language-specific processing rather than universal perceptual processes. To address this gap, we turn to Korean, which has a 3-way laryngeal contrast (e.g. lenis /pul/ fire vs. fortis /p*ul/ horn vs. aspirated /pʰul/ grass), with each pair having a different relative cue weighting. In comparing English and Korean data, we ask two primary questions: (1) Is continuous acoustic information as indexed by N1 encoded perceptually in the same way across languages, and (2) Does P3 index information about cue weights within a given contrast, or general acoustic informativeness? We designed two oddball tasks, one with English speakers and a second with Korean speakers. The English study had stimuli varying along two laryngeal continua ([b~p] and [d~t]), which were manipulated in Praat for VOT (5 steps) and post-stop f0 (3 steps), with a total of 1200 trials. The Korean study also had stimuli varying along two laryngeal continua ([p~p*~pʰ] and [t~t*~tʰ]), which were manipulated in Praat for VOT (3 steps) and post-stop f0 (3 steps), with a total of 1080 trials. In each experiment, one of the phonemes (out of four options in English and six options in Korean) was the target, and participants had to press one button if the word they heard was the target, and another button if it was not the target. ERP analyses are carried out with statistical validation via linear mixed-effects models. Data collection is complete for English (n=30) and ongoing for Korean (target=30). For the first question, results for English show that N1 is sensitive to VOT but not to f0. We predict that this will be similar in Korean, lending credence to the idea that N1 reflects a more universal perceptual effect which is sensitive to an abrupt stimulus like a VOT burst, but not a less salient cue, like f0. Turning to the second question, in English, a stimulus with a prototypical VOT and ambiguous f0 elicits a greater P3 amplitude than a stimulus with a prototypical f0 and ambiguous VOT. We hypothesize that P3 indexes information about cue weights within a given contrast, and we therefore predict to see variation in the Korean data depending on whether the target is lenis, fortis, or aspirated. By comparing how two typologically distinct laryngeal systems are processed, this work explores which neural indices of speech perception are shaped by linguistic experience and which reflect more general auditory processing.
Topic Areas: Speech Perception, Phonology