Paths to Phonological Complexity: Onset clusters in speech production, perception, and disorders
Final Report Abstract
The goal of the present project was to shed new light on the nature of phonetic biases in word‐initial consonant clusters in speech production, speech perception as well as speech disorders. The languages investigated were chiefly Georgian, German, Russian and Korean. Georgian has almost free consonant phonotactics for CC onsets, German allows for an intermediate number of clusters, and Korean allows for none. Russian allows for more clusters than German. All experiments focused on onset clusters, since these have been considered to be a primary locus of markedness effects. Also for overlap patterns between successive consonants, previous research has posited phonetic bias effects in that low overlap (Georgian, Russian) was postulated as less marked than high overlap patterns (German). The overarching goal of our experiments was to test for phonetic biases in overlap patterns and the potentially differential plasticity of native overlap patterns in the languages under investigation. Imitation experiments conducted with German, Georgian, and Korean speakers suggest that low overlap between consonant sequences may emerge as part of a faithful imitation of a given audio stimulus, independently of phonetic biases. Existing native overlap patterns thereby seem to exert a constraining force on the degree of imitation for the Germans. While both German and Korean participants showed overall a high degree of flexibility in imitating various consonant overlap patterns, Georgian speakers failed to imitate at all, possibly due to language‐specific cue weighting effects. This is to be followed up in future research. A perception experiment tested for language‐specific effects in the perceptual sensitivity to overlap differences. First results tentatively suggest that phonetic sensitivity cannot be predicted by a given language's cluster inventory, since Korean listeners show a higher sensitivity compared to Georgian listeners. This, however, needs to be corroborated by further analyses. An extensive cross‐linguistic comparison of CCV overlap patterns in 7 languages further suggested that languages differ not only in the coarticulatory timing patterns as such, but rather they differ in the range of patterns they use. All languages investigated seem to share a common core of high overlap patterns for parts of their inventory. A rate‐scaling experiment conducted on Russian investigated whether cluster emergence which can be observed in forced rate repetition varies as a function of the cluster inventory of a given language. The experiment underscored on the one hand the role of universal forces in these experiments in that the frequency of cluster emergence was independent of whether the emergent cluster is legal in Russian or not. However, subtle effects in the temporal‐spatial properties of the produced gestures witness an interaction of these universal forces with language specific properties: There are possibly less state‐feedback induced suppression effects in the context of clusters which are part of the phonotactic inventory. Lastly, the work conducted on speech disorders compared complexity effects in consonant cluster productions in two patient groups (apraxia of speech, phonological impairment). Results provide evidence for consonant clusters as higher‐order production units at the processing levels involved in both PI and AOS. Their vulnerability to at least the apraxic error mechanism depends on the particular phonetic constituents of a cluster. Notably, both patient groups made substantial numbers of errors that led to more complex phonological forms. Overall the results forcefully underscore the role of particular phonetic factors in the production of clusters for production and perception in both normal and disordered speech. Presumably universal markedness effects cannot be evaluated unless fine‐grained, language‐specific details of articulation, acoustics, and perception are fully taken into account.
Publications
- (2016) Investigating the interaction of dynamic stability with grammar: evidence from Russian. 15th Laboratory Phonology Conference, Cornell University, Ithaca, USA
Pouplier, M., Marin, S., Cunha, C., Kochetov, A.
- (2017) Cross‐linguistic differences in the perception of articulatory timing lag in onset clusters. Annual meeting of the Linguistic Society of America, Austin, Texas, USA
Kwon, Harim, Chitoran, I., Pouplier, M. Lentz, T., Hoole, Ph.
(See online at https://doi.org/10.1121/1.4970140) - (2017). A "phonological mind" in our brains? Clinical evidence. 7th International Conference on Speech Motor Control, Groningen, NL
Ziegler, W., Lehner, K., Jakob. H. & Aichert, I.
- (2017). Consonant clusters in patients with phonological impairment vs. apraxia of speech. 7th International Conference on Speech Motor Control. Groningen, NL
Jakob, H. & Ziegler, W.
- (2018) Machine learning shows that clusters have a temporal component. Satellite Workshop Representing Phonotactics at the 16th Laboratory Phonology Conference, Lisbon, Portugal
Lentz, T., Pouplier, M., Hoole, Ph.
- (2019). Produktion von Konsonantenclustern bei Patienten mit aphasisch‐phonologischer Störung und Sprechapraxie (Production of consonant clusters in patients with aphasic‐phonological impairment and apraxia of speech). Dissertation. Ludwig‐Maximilians‐Universität München
H. Jakob
