MIAPARLE: Online training for the discrimination of stress contrasts

MIAPARLE: Online training to improve discrimination of stress contrasts Jean-Philippe Goldman, Sandra Schwab University of Geneva jeanphilippegoldman@gmail.com, sandra.schwab@unige.ch Abstract MIAPARLE is a public web application that is designed to offer a range of CAPT (computer-aided pronunciation teaching) tools for L2 learners. Besides helping language learners to reduce their foreign-accentedness, the goal of the platform is to test these tools, gather feedback and improve them according to their educational impact. In this article, we describe one particular training tool that focuses on stress perception. This tool is particularly useful for speakers whose L1 is a fixed-stress language, such as French. These speakers have difficulties perceiving and discriminating stress contrasts. To help them with this so-called stress 'deafness', the methodology used in the training is based on successive questions in which a visual pattern is associated with the sound of a lexical item. After successively completing their pre-tests, training and post-tests, the participants are given their improvement score. The performance of the training is evaluated by comparing the learner’s results at the pre- and post-test stages. Various methodological parameters, such as the number of training items or the number of visual patterns are tested in parallel in order to quantify their teaching efficiency, and to optimise the overall teaching impact. Keywords:​ CAPT, pronunciation training, prosody, stress deafness 1. Introduction Second language learners tend to imprint the prosody of their mother language onto the second language (L2) (e.g., Barquero et al., 2014; Muñoz, 2010). This prosodic cross-language transfer is often combined with segmental transfers, which can lead to the presence of a foreign accent. A foreign accent can not only hamper communication between learners and natives (like in ​Anderson, Munro & Derwing 1988), but it can also affect the credibility of learners and how they are evaluated by others; this can sometimes lead to social discrimination (e.g., Dailey et al., 2005, Lev-Ari & Keysar, 2010; , Purnell et al. 1999). Moreover, despite the crucial role of prosody in speech flow segmentation (e.g., Cutler 1984; Cutler & Butterfield 1992 for English; ​Bagou 2002 for French), it is rarely taught in language courses, even in foreign-language pronunciation courses. In the framework of ​computer-assisted pronunciation teaching​, the goal of this project is to develop a methodology which improves the prosodic skills of L2 learners in order to reduce cues of foreign accent. After many years of research (Neri et al. 2002, Busà 2008, Chun 2012), there has recently been a growing interest in CAPT applications (e.g. for French , Spanish , English ,and 1 2 3 Norwegian ). However, there have been few applications 4 that focus specifically on prosodic features (among others,, Vakil & Trouvain, 2015 for German stress; Sztahó et al, 2014, for Hungarian intonation, stress and speech 1 ​http://www.coelang.tufs.ac.jp/mt/fr-swiss/dmod/index_en.html 2 http://www.enterate.unam.mx/Articulos/2007/abril/sarahi.html 3 ​http://www.carnegiespeech.com/products/nativeaccent.php 4 ​https://orgesuniversitetet.no/prosjekt/computer-assisted-listening-and-speaking rhythm; Niebuhr et al. 2017, for visualization methods of prosody). This paper focuses on the tool used to assist in discrimination of stress contrasts. Although this method is designed for learners with any first language (L1), training on …

Content truncated to 500 words