Vocal fold vibrations are more difficult to achieve in obstruents than sonorants due to the aerodynamic voicing constraint (AVC), i.e., the fact that a buildup of air pressure in the supraglottal cavity during oral closures reduces the transglottal airflow. The AVC can be circumvented by various voicing adjustment gestures, such as larynx lowering, tongue root advancement and tongue body lowering. The current study employed laryngeal and lingual ultrasound to investigate the use of these strategies in Canadian French. The vertical movement of the larynx was measured using optical flow analysis, while lingual movement was analyzed by tracking X and Y coordinates at distinct fanlines across consecutive images.

Results revealed that there was more pronounced larynx lowering in voiced obstruents and that it tended to be greater in voiced stops than in voiced fricatives. Tongue-related maneuvers displayed more interspeaker variation but tendencies showed that the tongue root was more advanced in voiced stops than in voiced fricatives and slightly more for /d/ than /b/. Significant tongue body lowering was observed for both voiced stops and voiced fricatives only preceding the vowel /a/. Finally, larynx lowering was strongly correlated with voicing duration in voiced obstruents. A similar but weaker correlation was found for tongue root advancement.

Overall, this study suggests that larynx lowering is an efficient strategy to circumvent the AVC in Canadian French but that some speakers may also resort to lingual adjustments. Additional strategies that are known to play a role, such as nasal or oral leakage, were not considered.

Words in Tagalog/Filipino can be either penult-prominent or ultima-prominent. Scholars have been divided on whether the language has stress, or only phonemic vowel length in penults and default phrase-final prominence. Using a corpus of Original Pilipino Music, we find that both prominent penults and prominent ultimas are set to longer notes and stronger beats, even in phrase-medial position. We further find that among pre-tonic syllables, those that would plausibly attract secondary stress are mostly set to longer notes and stronger beats. Text-setting does not faithfully reflect differences in phonetic cues between the two types of prominence, nor is it sensitive to presumed phonetic differences between high and low vowels. We conclude that songwriters’ text-setting decisions reflect phonological stress in Filipino, and that both penult-prominent and ultima-prominent words bear stress.

Accounts of prosody in understudied languages are often impressionistic, potentially leading to conflicting accounts due to different researchers being drawn to different acoustic cues. The debate surrounding the location of primary stress in Plains Cree is such a case. One widely adopted claim states that stress is realized on the antepenult, whereas others argue for a penultimate accent. The present study investigates the phonetic properties of stress (duration, F0, intensity, vowel quality) in multisyllabic words and in phrases to understand the patterns that have led to the current debate. We find that there are cues supporting both previous claims: a high F0 on the antepenultimate syllable compatible with “antepenultimate stress” and a falling F0 on the penultimate syllable compatible with “penultimate accent.” Based on the acoustic evidence, we suggest that Plains Cree is a pitch-accent system, with a predictable penultimate HL word-level pitch-accent. Tonal patterns in other syllables are the result of prosodic boundaries, phonetic interpolation, or tonal spreading.

This study investigates phonological and phonetic details of disjunctive declaratives (ddcls) and alternative questions (altqs) in Arabic. The aim of the phonological and phonetic analyses of these syntactically identical utterances is to find out the cues that are responsible for the disambiguation. Consequently, a production study eliciting ddcls and altqs was run with 20 participants producing 160 utterances (80 ddcls and 80 altqs). Findings reveal that ddcls and altqs are similar in having a global rise-fall contour, but differ in the phonetic implementation of the fall, since minimum F0 values are significantly higher in altqs than in ddcls, suggesting that there is a fall to mid in the former (proposing !H%) and a fall to low in the latter (L%). There are also significant phonological differences in the accentual features between both sentence types, i.e., the conjuncts are always accented in altqs, but they are deaccented in ddcls. The findings are a contribution to the prosody-meaning literature, showing the importance of prosody for syntactic disambiguation. The findings are used to propose a theory for the disambiguation of disjunctive sentences.