Intention in articulation: Articulatory timing in alternating consonant sequences and its implications for models of speech production

Abstract

Several studies have reported that during the production of phrases with alternating consonants (e.g., top cop), the constriction gestures for these consonants can come to be produced in the same prevocalic position. Since these coproductions occur in contexts that also elicit segmental substitution errors, the question arises whether they may result from monitoring and repair, or whether they arise from the architecture of the phonological and phonetic planning process. This paper examines the articulatory timing of the coproduced gestures in order to shed light on the underlying process that gives rise to them. Results show that overall at movement onset the gestures are mostly synchronous, but it is the intended consonant that is released last. Overall the data support the view that the activation of two gestures is inherent to the speech production process itself rather than being due to a monitoring process. We argue that the interactions between planning and articulatory dynamics apparent in our data require a more comprehensive approach to speech production than is provided by current models.

Keywords:

• Speech production
• Articulatory timing
• Monitoring
• Speech errors
• Cascading activation

Acknowledgements

Work supported by Deutsche Forschungsgemeinschaft (PO 1269/1-1) and NIH (R01 DC008780-01). Thank you to Susanne Waltl for help with data labelling and to our colleagues in Munich and at Haskins for valuable feedback.

Notes

¹Whether these coproductions should be considered speech errors and under which conditions any given coproduction can be deemed errorful has been subject to vigorous debate. The current paper focuses on the temporal variability of articulator kinematics which is observed in situations in which multiple candidates compete strongly during utterance encoding; the question how to negotiate the relationship of ‘variability’ and ‘error’ is not the focus of the paper.

²The current data were collapsed across the experimental variables stress (iamb, trochee), position (initial, final) and vowel (/ı/, /α/) which were included in the original dataset. The full set of utterances collected for each vowel and both speaking rates was (capital letters indicate stress): top COP, TOP cop, COP top, cop TOP; TOP top, top TOP, COP cop, cop COP.

³The differences in metronome rates between subjects cannot be the main factor conditioning the between-subject differences, since the direction of the effect (whether there are more positive or negative GONS lag values) for a given subject is always the same in the two rate conditions. Also, the metronome differences between subjects are smaller than the differences in the number of positive vs. negative GONS values for the two rates. See Table A1 in the Appendix.

⁴The large between-subject variability in the amplitude values is conditioned by the different vocal tract anatomy of the speakers and of no theoretical interest. The numeric amplitude value depends on the speakers’ vocal tract size, the palatal shape, and the occlusal angle. Notice that the magnitude of the difference between intended and intruding is comparable across subjects, about 3–4 mm for TDy and 4–5 mm for TTy. Also the magnitude of the standard deviation difference between intended and intruding is comparable across subjects and articulators.

⁵This study included tokens across all conditions and speakers recorded in Pouplier (2003), that is, for both alternating and nonalternating conditions all rate, stress, position, and vowel variables for seven speakers. Coproductions were identified as in Pouplier (2003), that is, any token for which the articulator height of the intruding gesture deviated more than two standard deviations from the control mean was defined as a coproduction.

⁶The perception experiment was also based on the data of speaker JX, the same speaker whose data were used for the classification test in the Marin et al. study.