The Digital Literacy Instructor: Developing Automatic Speech Recognition and Selecting Learning Material for Opaque and Transparent Orthographies

While learning a new language can now be a lot of fun because attractive interactive games and multimedia materials have become widely available, many of these products generally do not cater for non-literates and low-literates. In addition, their limited reading capabilities make it difficult for these learners to access language learning materials that are nowadays available for free on the web. More advanced course materials that can make learning to read and spell in a second language (L2) more enjoyable would therefore be very welcome. This article reports on such an initiative, the DigLin project, which aims at developing and testing online basic course material for non-literate L2 adult learners who learn to read and spell either in Finnish, Dutch, German or English, while interacting with the computer, which continuously provides feedback like the most determined instructor. The most innovative feature of DigLin is that in production exercises learners can read aloud and get feedback on their speech production. This is made possible through the use of Automatic Speech Recognition (ASR). In this article we focus on what ASR is and what is needed to employ ASR to develop learning materials for non-literates and low-literates L2 learners. Central issue is the selection of the content for the four languages, which differ in orthographic transparency and present their own specific problems in combination with the mother tongue of the learners.