A comprehensive study of vowel production and perception in two Arabic dialects: Joufy and Qatifi

The thesis investigates the vowel system in two Saudi Arabic dialects: Qatifi and Joufy. It comprises three studies: two production studies and one perception study. The production studies report on the acoustic features of each phonological vowel system and they also investigate a putative merger in Qatifi short high vowels [i~u] and a putative split in Joufy low vowels [a(ː)/ɑ(ː)]. Meanwhile, the perception study examines how vowel spectral and temporal cues affect Qatifi and Joufy listeners’ perception of the duration contrast of vowel pairs.
Analysis of scripted and unscripted texts reveals that short and long vowels differ in their temporal and spectral aspects, forming eight distinctive vowel categories in both dialects. The Qatifi vowel merger between /u/ and /i/ shows no absolute merger when the acoustic categories’ overlap was quantified using the Pillai score. With regards to Joufy low vowels [a(ː)/ɑ(ː)], there appears to be a split in progress that seems to be conditioned in the synchronic data by multiple triggering consonantal contexts, which include uvulars and the co-occurrence of [l] and [ɡ]. However, these observed environments are inconsistent, and many examples show no patterns, so /ɑ(:)/ might be better described as phonemic, especially because they form minimal and near-minimal pairs. The results also suggest that morphological structure and lexical item type may play a role in the spread of this putative sound change.
The perception study examines how vowel spectral and temporal cues affect Qatifi and Joufy listeners’ perception of the duration contrast of vowel pairs. The findings provide additional evidence that cue weighting in vowel perception is vowel-dependent. Individual vowel position and acoustic features play a role in how listeners use spectral and temporal cues. In the overall performance, temporal cues were more attended to than spectral cues by listeners of both dialects. High vowels exhibit greater sensitivity to spectral cues than low vowels, with variations observed between front-back high vowels and also between the two dialects.