Articulatory and Acoustic Properties of Saudi Arabic Coronals and Cross-language Perceptual Assimilation by Naïve Listeners: The Case of Saudi Arabic and Seoul Korean

The Perceptual Assimilation Model (PAM) (Best, 1995) claims listeners directly perceive articulatory gestures of the vocal tract rather than the acoustic/auditory signals. Accordingly, the articulatory similarities and discrepancies between native and non-native sounds determine the perceptual assimilation patterns of the non-native sounds. This dissertation provides empirical data to verify PAM's claims by testing two under-studied languages with typologically less common consonants. In Saudi Arabic, coronals can be pharyngealized (emphatics) and consonant duration is contrastive (singletons vs. geminates). On the other hand, Seoul Korean has the three-way laryngeal contrast including aspirated, lax, and tense stops with the lax stop subject to intervocalic voicing, and it has the two-way fricative contrast. This dissertation investigates cross-language perception in which naïve Saudi Arabic listeners hear Seoul Korean sounds (/th t t* sh s*/) and Seoul Korean listeners hear Saudi Arabic sounds (/t tʕ d s sʕ/ and their geminate counterparts). By doing so, I examine how each group of listeners assimilates the non- native sounds from the other language to their native categories, and to what extent articulatory similarities and discrepancies between the sounds in the listeners’ native language and those in the stimuli predict the way listeners perceive unfamiliar sounds.

This dissertation presents one production experiment and two perception experiments. Unlike Seoul Korean, little is known about the articulatory and acoustic properties of Saudi Arabic sounds. While the understanding of Seoul Korean sounds is based on the previous phonetic works on Seoul Korean obstruents (e.g., Kim et al., 2010), the production experiment, in this dissertation, aims to understand the articulatory configurations of previously unstudied Saudi Arabic coronal obstruents, particularly in the Southern dialect. As the first step to explore the relationship between articulation and perception, the production experiment investigates the tongue root configuration and its acoustic properties during Saudi Arabic speakers’ productions of coronal obstruents, using the ultrasound tongue imaging technology. As previous studies focus on either voicing or gemination or on either stops or fricatives in an initial position, I additionally examine the effect of gemination, voicing, and manner (stops vs. fricatives, including emphatics) on the tongue root configurations during intervocalic obstruents, as well as their acoustic properties, such as durations of constrictions, voice onset time (VOT), presence and absence of vocal folds vibration, and flanking vowel durations. The articulatory findings indicate that tongue root position is significantly affected by gemination and voicing such that tongue root advancement is observed among voiced and geminate consonants.The current study also reveals manner does not make a significant difference in the tongue root position such that stops and fricatives behave similarly.

In terms of acoustic properties, the current results confirm that the Southern dialect of Saudi Arabic has contrastive consonant duration in word-medial positions where geminate coronals have a longer constriction duration than singleton ones. The acoustic findings put the Southern dialect of Saudi Arabic in the category of the true-voiced languages with pre-voiced, unaspirated emphatic, and aspirated stops. The duration of V1 is short, and V2 is long in the geminate sound contexts, and the opposite is true in the singleton sound contexts.

Based on PAM predictions, the investigation in the production experiment, and on the previous phonetic works on Seoul Korean obstruents, I offer a set of testable predictions on how

Saudi Arabic and Seoul Korean listeners perceive sounds from the other language. The predictions are tested in cross-language perceptual assimilation experiments. The results appear to be mostly consistent with the predictions except for a few notable patterns. Listeners generally assimilated non-native sounds to the native counterparts involving the same articulator, constriction location, VOT and/or constriction duration, as found in the case of Saudi Arabic /t d s sʕ/ and /tː dː sː sʕː/ and Seoul Korean /th t t* sh s*/. However, Saudi Arabic emphatic stops /tʕ tʕː/ are assimilated more frequently as Seoul Korean labials /p p*/ than coronals. This emphatic-to- labial assimilation pattern is not expected because /tʕ tʕː/ and /p p*/ are dissimilar articulatorily (have different places of articulation). A possible explanation for this assimilation pattern can be found in their acoustic, but not articulatory, similarity since pharyngealized sounds and labial sounds usually lower F2 of the adjacent vowels. This acoustic similarity arguably may cause the Seoul Korean listeners to assimilate the emphatic consonants to labial consonants. Taken together, the findings suggest that non-native perception is not exclusively articulatory but also acoustic.