Effects of stress and metrical pattern on duration perception in Spanish

Abstract

This thesis focuses on prosodic expectations, especially on the research done by Zheng & Pierrehumbert (2010). They lengthened syllables of meaningful English sentences and investigated the conditions in which the lengthening was better detected. Target syllables varied in metrical stress, metrical pattern, serial position and speech rate. Detection was better for strong syllables in all metrical patterns, serial positions and speech rates. The trochaic metrical pattern yielded a higher detection rate than the other two patterns. Zheng & Pierrehumbert claim that their findings confirm the Attentional Bounce Hypothesis (Pitt & Samuel 1990). According to this hypothesis, listeners use the sequencing of strong and weak syllables to predict where stress will fall. Interestingly, the influence of meter was contrary to the findings by Quené & Port (2005) for English, in which the effect of metrical expectancy was not confirmed. The present thesis aimed to replicate the study by Zheng & Pierrehumbert for Spanish. This language has a different metrical stress, with stress mostly on the penultimate syllable instead of on the initial syllable (Sebastián-Gallés & Costa 1997), thus making it presumable that (the trochaic) meter would not have an effect on lengthening detection. An experiment was conducted with native speakers of Castilian Spanish. Results indicated that, even though present, the effect of stress is weaker in Spanish than in English. Metrical pattern has a large effect on duration perception: the dactylic pattern facilitates perception in Spanish, and not the trochaic pattern, which had the highest hit rates in English. Reaction times indicated that serial position is the relevant factor for faster responses, with faster reaction times at the end of the stimuli. The influence of the attentional bounce hypothesis is discussed.