Abstract

Perceptual similarity of Swedish and German vowels: A free classification task with naive listeners

Articulatory gestures on a speaker’s face can provide segmental information about speech sounds. Therefore, seeing the speaker can support speech perception in both first (L1) and second language (L2) listeners. This so-called audiovisual benefit is modulated by several factors, for example, the quality of the acoustic signal, visual salience, and language experience. My PhD project focuses on the visual salience and crosslinguistic similarity of vowels and investigates the influence of these factors on the audiovisual benefit in L2 learners of German and Swedish.

In a first baseline study, we defined the visual salience of Swedish vowel contrasts by measuring lip configurations in continuous speech. In a second (ongoing) baseline study, which will be the focus of this talk, we aim to operationalize the crosslinguistic similarity of Swedish and German vowels. To this end, Swedish and German long vowels are embedded in monosyllabic pseudowords and extracted from continuous speech. These vowels are presented both audiovisually and auditorily-only. In a free classification task, naïve listeners (i.e. without prior experience in German and Swedish) sort these vowels freely into a grid based on perceptual similarity. I will describe the study’s material, procedure, and hypothesized outcomes, and discuss how vowel classification data from naïve listeners can serve as a multimodal baseline of crosslinguistic similarity.