Phonological and Physiological Constraints on Assimilatory Pharyngealization in Arabic: Ultrasound Study

Abstract

An essential characteristic of continuous speech is the great articulatory and acoustic variability of sounds, owing to the influence of surrounding segments. This segmental variability, referred to as coarticulation or assimilation, has been a central issue in coarticulation models and is subject to the influence of phonological and physiological constraints. This dissertation examines these constraints using emphasis (known as either pharyngealization or uvularization) as a case study and presents a novel investigation of the C-to-C effect of emphasis, i.e. the effect induced by an emphatic consonant on the preceding consonants. It addresses three core questions. The first is whether the contextual effect exerted by pharyngealized consonants is constrained by phonological factors such as phoneme contrast – i.e. if consonants that are contrastive in pharyngealization (e.g. /t s/) and those that are not (e.g. /n l/) demonstrate similar assimilatory effects. Another central question is whether and to what extent assimilation is influenced by the gestural conflict between the target and trigger consonants that arises from the physiological lingual constraints. This is addressed by comparing the assimilation in consonants varying in the degree of lingual constraints: highly constrained [ɡ ʃ j], moderately constrained [χ] and unconstrained [b f]. The third question is whether the assimilatory effect of pharyngealization is a phonetic or phonological process. Ultrasound and audio data were obtained from 15 adult native speakers of Eastern Peninsular Arabic spoken in certain regions of the Arabian Gulf states as they read target words in a carrier sentence. Tongue contours from different time points of the target and trigger consonants were traced manually. Articulatory data confirmed that emphatics /tˤ sˤ/ are pharyngealized (not uvularized) by retracting the tongue root towards the back pharyngeal wall and, consequently, depressing and flattening the tongue body and dorsum. The results of polar SS-ANOVA and root retraction ratios revealed that phoneme contrast does not inhibit assimilation: coronals exhibit categorical effects whether they are contrastive in pharyngealization (e.g. /t s/) or non-contrastive (e.g. /n l/). Assimilation leads to a neutralization of their pharyngealization contrast in /s/ for many speakers and in /t/ for some speakers as the assimilated /t/ and /s/ become articulatorily indistinct from pharyngealized /tˤ/ and /sˤ/, respectively. For other speakers, distinctiveness is maintained by completely resisting assimilation in /t/. A gestural conflict between pharyngealization and another articulation can result in two possible outcomes depending on its severity. An extreme conflict between pharyngealization and tongue root advancement in [ɡ ʃ j] results in a complete resistance to coarticulation. A moderate conflict between pharyngealization and dorsum raising and retraction in [χ] leads to resistance to pharyngealization in some cases and to a reduction of dorsal raising to allow some degree of coarticulatory root retraction in other cases. Finally, the lack of lingual conflict in [b f] allows the consonants to coarticulate unrestrictedly with the pharyngealized consonant. The realization of the effect as phonetic or phonological is contingent on the presence of either phonological and/or physiological constraints. It is categorical (assimilation or resistance) in the case of phonemic contrast as in /t s/ and gestural antagonism as in [χ ɡ ʃ j], and gradient in the absence of both constraints as in [b f l]. The results support the predictions of Recasens’ (1997) degree of articulatory constraints model, suggesting a correlation between the degree of articulatory constraints and the magnitude of coarticulatory pharyngealization. They are, however, only partially consistent with Cohn’s (1990) target-interpolation model. One issue with this model is that it does not account for the speaker variability in assimilating some consonants such as [b f l], which are underlyingly unspecified for pharyngealization and unconstrained articulatorily. The model, however, accounts for the role of phonetic constraints in restricting phonetic interpolation, as in [ɡ ʃ j].