We present an analysis of alignment from a multimodal perspective in two telecollaborative settings. Our aim is first to establish an annotation scheme for conversational alignment to deal with the multimodality of desktop videoconferencing environments, and second to explore the possible differences between the two populations.
In our paper we present an analysis of conversational alignment (Pickering & Garrod, 2004) from a multimodal perspective (Guichon & Tellier, 2017) in two telecollaborative settings. Our main aim is to analyse alignment during desktop videoconferencing in its multimodality, including lexical and structural alignment (Michel & Cappellini, 2019) as well as alignment in facial expressions. We analyse two datasets coming from two different models of telecollaboration. The first one is based on Le Français en (première) ligne model which puts in contact future teachers of a foreign language with learners of that language. The second one is based on the teletandem model, where students of two different mother tongues interact to help each other use and learn the other's language. Our dataset is composed of two hours, one from each model, coming from a diad and a triad of students completing the same pedagogical activity. Our aim is twofold. First, we want to establish an annotation scheme for conversational alignment to deal with the multimodality of desktop videoconferencing environments. Second, we want to explore the possible differences in conversational alignment between the two populations in order to understand with which kind of interlocutor alignment is more likely to take place. In the first part of our presentation, we will present our contexts, explaining both the characteristics of the telecollaborative settings and of the interlocutors. In the second part, we will show how the initial model of conversational alignment was adapted to study our data. In the third part, we will present and discuss the results of our analysis.
