Research has shown that gestures can have beneficial effects on second language (L2) pronunciation learning. However, we still do not have a coherent picture of these effects, because different studies have investigated different gestures and usually either considered segmental (Hoetjes & van Maastricht, 2020; Iizuka et al., 2020; Li & Somlak, 2019; Xi et al., 2020) or prosodic (Morett & Chang, 2015; Yuan et al., 2019; Zhang et al., 2020; Zheng et al., 2018) features, but not both, and mostly either tested learners’ production or perception. Importantly, most of these studies have been conducted in a lab setting rather than in an authentic classroom setting. 

This project aims to study the effects of embodied training, using gestures, on pronunciation learning in Swedish, in two lab studies and two classroom studies. Two studies (one in each setting) will focus on a segmental feature, namely the Swedish vowel contrast /i/≠/y/, while the other two will focus on a prosodic one, more specifically the Swedish length contrast (vila ≠ villa). 

The studies will assess both learners’ perception and production of these phonological features before and after a training phase, through a pre-/post-/delayed post-test design. The instruction phase will consist of a video training learners on the given contrast in three different conditions (between subjects), plus a control group (no training): one group without gestures (audiovisual speech only condition), and the other two groups will receive training with two different gestures. The gesture conditions are currently being defined based on interviews conducted with teachers of Swedish as a second language.

Learners’ production in the pre- and post-tests will be assessed through native speaker ratings and acoustic analysis. Perception will be measured through identification or discrimination tasks, but also using eye-tracking, providing a continuous measure of learners’ processing abilities.