Respiratory activity during speech

Abstract

The aim of this project was to examine the extent of variability of speech breathing patterns in two groups of healthy subjects using non-invasive respiratory measurement (respiratory inductance plethysmography and linearised magnetometers). In the first two studies (Chapters 3 and 4), lung volume measurement of 6 young women was combined with observation of the temporal and linguistic aspects of speech breathing, as well as speech intensity measurement, as the subjects returned for up to 10 repeat sessions in 3 weeks.

During oral reading, lung volumes during speech were found to be significantly variable over time, both between and within individuals, whereas the location of inspirations was relatively invariant: most pauses for breath occurred exclusively at grammatically appropriate places in the texts. Some of the intrasubject lung volume variation was associated with changes in speaking intensity, and significant associations were found between anticipatory lung volumes and the length of the upcoming breath group, suggesting that subjects scan ahead during reading and breathe accordingly.

During spontaneous speech too, linguistic factors were important influences in the lung volume variation, with the volume of air inspired linked to the length of the following utterance. Further, the majority of breaths were taken at structural (clause) boundaries. In contrast to the findings during reading, speaking intensity, however, was not strongly associated with lung volume variation in spontaneous speech. Results of a self-reported mood state questionnaire (The Profile of Mood States) suggested that some of the subjects' lung volume variation was associated with day to day changes in mood, and the possibility of variation due to hormonal factors was also suggested, with data collected on the subjects' menstrual cycles indicating an association.

The Lombard effect, that of increased speech intensity in background noise, was used in Study 3 (Chapter 5) to achieve louder speech in a group of 5 healthy young women, while simultaneous respiratory and speech measurements were made. Despite the wide range of speech intensity encompassed, lung volume responses were not predictable, and there were discrepancies between reading‘ and spontaneous speech. It is clear that increased lung volumes are not always the preferred mechanism for achieving the higher subglottic pressures required for louder speech. The previously reported association between inspired lung volume and the length of the upcoming utterance again held true for this study, suggesting that linguistic factors are stronger in influencing lung volume variation than speaking intensity.