Speaking together in time: An investigation of prosodic timing structure and unison production in people with aphasia

Abstract

Aphasia is an acquired language disorder, typically resulting from stroke, that can affect the ability to speak fluently and accurately. Many effective treatments to improve spoken language production in persons with aphasia (PWAs) involve unison production, in which the PWA speaks along with a clinician or recording. However, the mechanism underlying this technique is not well understood. To maximize therapeutic efficiency of unison production, it is critical to understand how it works, and which types of stimuli are most effective.

One key variable that differs across unison production approaches is the prosodic timing structure—the patterning of stressed versus unstressed syllables within the stimulus phrases. Most approaches use a “conversational,” syntax-influenced timing structure, while others, particularly music-based treatments, use a beat-based, “metrical” timing structure. Recent research has shown that beat-based rhythm can particularly benefit spoken output in PWAs, suggesting that it may promote entrainment to the stimulus, i.e., using temporal prediction to guide planning and subsequent production of speech in unison with the model. This dissertation investigates this idea by comparing conversational versus metrical prosodic timing structure during unison production on PWAs’ (a) syllable production accuracy, and (b) timing alignment with the unison stimulus—an objective measure of temporal prediction and stability.

In Study 1, group-level comparisons reveal that speaking along with metrically timed stimuli yields (a) a higher percentage of accurate syllables, particularly for individuals with the nonfluent aphasia subtype, and (b) earlier and less variable timing alignment, suggesting enhanced temporal prediction. Study 2 employs item-level regression analyses to explore whether stressed syllables, which are typically more acoustically and linguistically salient than unstressed syllables, are more likely to be produced accurately and/or show evidence of enhanced entrainment. Findings indicate that (a) stressed syllables are more likely to be produced at least partially correctly, and (b) timing alignment of stressed syllables differs based on prosodic timing structure, suggesting different modes of planning and production in conversational versus metrical contexts.

Overall, results suggest that prosodic timing structure should be considered when designing unison production stimuli, as metrical timing facilitates entrainment, leading to more accurate spoken output among PWAs.