Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Effects of syllable frequency in speech production: Replication and extension
Poster Session C, Saturday, September 13, 11:00 am - 12:30 pm, Field House
Julia Chauvet1, Sophie Slaats2, David Poeppel3, Antje S. Meyer1,4; 1Max Planck Institute for Psycholinguistics, 2University of Geneva, 3New York University, 4Radboud University
Speaking requires translating an intended message into a sequence of sounds. There is a noticeable difference in the ease with which frequently produced sequences of speech sounds are articulated compared to those rarely produced (see for review, Laganaro, 2019). One explanation is that well-learned sequences form stable, automatized articulatory routines—syllable-sized chunks in a Mental Syllabary (Crompton, 1982)—while motor plans for infrequent sequences are assembled online, segment by segment. Empirical support for the Mental Syllabary hypothesis includes, for instance, shorter naming latencies for frequent syllables and diverging ERP waveforms for high and low frequency syllables before articulation onset. We aimed to replicate these findings and extend the data to articulatory duration which, so far, had remained untested. If motor plans for frequent syllables are automatized, this should yield shorter and less variable articulatory durations for frequent than infrequent syllables. In two naming experiments, adult native speakers of Dutch produced high- or low-frequency non-word syllables (e.g., ‘kem’ or ‘kes’). Experiment 1 (n = 26) used immediate naming and visual presentation of the syllables. Materials were taken from Cholin, Levelt and Schiller (2006). Using linear mixed modeling and a backward step-wise model comparison approach, we found a main effect of frequency on naming latency: as expected, naming latency decreased with syllable frequency. We also observed an interaction between syllable frequency and the condition it was assigned to. Within the high-frequency condition, naming latencies decreased with frequency as expected. In contrast, within the low-frequency condition (0 to 3.1 occurrences per million), naming latencies slightly increased with frequency. This effect could reflect the effortful retrieval of infrequent, but putatively learned, syllables. Contrary to our prediction, articulatory duration increased with syllable frequency. Experiment 2 (n = 20) used delayed naming and auditory presentation of the syllables. Participants heard a target syllable, then an auditory distractor, and finally produced the syllable upon presentation of a visual cue. We recorded the EEG as in Bürki, Cheneval and Laganaro (2015). Preliminary results replicate their basic pattern of diverging ERP waveforms in the 200 ms time window before speech onset, suggesting that frequent and infrequent syllables differ in processing cost. As expected, there was no frequency effect for the delayed naming latencies. However, articulatory duration again increased with frequency. Experiment 1 is the first replication of Cholin et al.’s behavioral findings with identical materials and Experiment 2 is a conceptual replication of the study by Bürki, Cheneval and Laganaro (2015). The study adds novel results concerning articulatory durations. We speculate that high-frequency syllables will have been produced in more diverse contexts and so be associated with a wider set of permissible instantiations, from which speakers make a random choice in our task. This raises questions about the internal structure of motor plans—e.g. whether they involve gestures or phonetic features or differ in their articulatory targets. Future research should examine how motor plans interact with the prosodic demands of connected speech, including how frequency shapes articulatory duration in fluent, continuous production.
Topic Areas: Language Production,