The Production and Perception of Prosodic Prominence in Urban Najdi Arabic

Hussain Almalki

Major Professor: steven h. weinberger

Committee Members: Harim Kwon, Douglas Wulf

Online Location, #443
April 20, 2020, 10:00 AM to 01:00 PM


This dissertation addresses prosodic prominence in Urban Najdi Arabic (UNA) in both production and perception. Although the production and perception of prosodic prominence is relatively understudied in UNA, prior research from a number of spoken varieties of Arabic has revealed systematic differences among these varieties in terms of prominence marking in production. Recent cross-linguistic research, further, suggests that acoustic and non-acoustic factors may influence the perception of prominence. Accordingly, this dissertation examines how and to what extent native UNA speakers prosodically mark prominence in different information structures. It also investigates the influence of acoustic cues and contextual factors on UNA listeners’ perception of prominence.

The production experiment examines how UNA speakers utilize acoustic cues to mark information structure, namely focus, and whether they actively disambiguate lexically and propositionally identical utterances according to discourse requirements. The results show that UNA speakers acoustically distinguish different aspects of information structure through prosody. The acoustic correlates of prominence are duration, maximum intensity, F0 maximum and F0 range. These acoustic cues are typically associated with stressed syllables. In particular, UNA Speakers use these acoustic cues to disambiguate focus location, focus status, focus size and focus type. However, speakers in this experiment do not overwhelmingly distinguish between ambiguous sentences between contrastive and non-contrastive conditions.

The first perception experiment examines how well can listeners perceive acoustic prominence from the speech signal alone. Using data from the production experiment, listeners are asked to rate prominence of sentences by providing prominence ratings for each word on a 5-point rating scale. In this bottom-up design, listeners only have access to information in the speech signal. The results show that UNA listeners are highly successfully in perceiving prominence based on the speech signal alone. The pattern of the perceived prominence in this experiment mirrors the pattern found in the production experiment.

In the second perception experiment, a top-down design is employed to examine whether the perception of prosodic prominence is affected by contextual cues. Listeners are asked to read a context question and then listen and rate prominence for each word in the answer on a 5-point rating scale. In this highly controlled experiment, the set-up questions and corresponding answers are varied in terms of the question-answer congruence. In the congruent question-answer pairs, the answer is prosodically appropriate to the question. In the incongruent question-answer pairs, the answer is prosodically inappropriate to the question. Further, in the incongruent pairs, listeners hear the same answer after different set-up questions. This is done to specifically test the independent effect of context. The findings from the congruent pairs replicate the findings from bottom-up experiment in that listeners are able to successfully perceive prominence when both acoustic and contextual cues complement each other. The results from the incongruent pairs show that contextual factors could partially affect the perception of prominence. Specifically, in some incongruent pairs, listeners seem to be responding to contextual cues rather than acoustic cues, as reflected by their prominence ratings. However, such effect of the context seems to be phonologically conditioned by the accent distribution on the answer utterance.

Findings from these experiments indicate that the production and perception of prosodic prominence in UNA is a multifaceted process, which is seems to be influenced by several competing factors.