The timing of Accentual Phrases in read and spontaneous speech: data from Acadian French

Abstract

While comparisons of read and spontaneous speech materials note prosodic differences such as the number and position of pauses and the number and position of tone unit boundaries, only a few studies have examined the rhythm of these two speech styles. In these studies, rhythm is measured with metrics based on the durations of segmental – that is, vocalic and consonantal – intervals. In this paper we consider the timing of larger units known as Accentual Phrases (APs), which are groups of syllables that are demarcated by a primary stress. Our study uses AP-based rhythm metrics to examine differences between read and spontaneous speech materials.

Data are from sociolinguistic interviews conducted with 12 speakers of a variety of Acadian French spoken in northeastern New Brunswick. Each speaker produced both reading and spontaneous styles. Approximately 34 minutes of speech – about 2,400 APs – were analyzed. APs were identified by two native speakers. Calculations of AP-based rhythm metrics were made with the same equations that are used for segmental interval measures (delta-AP, Varco-AP) and for segmental pairwise variability measures (nPVI-AP, rPVI-AP).

APs in spontaneous speech are shorter in duration than those in read speech, even though the average number of syllables per AP is similar in both styles. APs in spontaneous speech also have greater durational variability, and they show greater inter-speaker variation. Discrimination analysis suggests that the normalized metrics –Varco-AP, nPVI-AP – contribute most to distinguishing between the two styles. One implication of this study is that AP-based rhythm metrics can contribute to a framework for the comparison of read and spontaneous speech materials. More generally, the study confirms that there are interesting patterns of speech timing that are located at a level above vocalic and consonantal segments.