Research report

January 2006
Jim Milton, University of Wales Swansea

The full version of this report is available to download in pdf format - see box below. On this page you'll find the report's background section.

Background to the Language Train software trial

English native-speaking learners of Spanish commonly mispronounce Spanish words. Sounds common in Spanish either do not exist phonemically in English or behave differently, and are replaced by sounds which do exist in English. This is commonly called negative transfer or interference. Learners in primary school identify pronunciation problems as a major obstacle to learning. They repeatedly report that poor pronunciation would lead to their stigmatisation in the event of using their foreign language. The Language Train software is designed to target the problems English-speaking learners of Spanish experience, and to provide practice and training in the auditory discrimination of some of the sounds learners commonly confuse.

The principle is not entirely novel. Auditory training is at the root of the audio-lingual approach to language teaching and minimal pair drilling remains a feature of pronunciation practice in foreign language classes. But the Language Train software adds a degree of control, precision and intensity in this kind of training that the average foreign language teacher could not manage.

This software models two versions of a word in Spanish. One model is correct in its Spanish pronunciation and includes a phonemic distinction particular to Spanish but which is not made in English. The second model is of the same Spanish word but with the Spanish phoneme replaced by a substitute frequently used by English native speakers. The learners hear these two models and then hear a third version of the same word. This version is a repeat of one of the models and the learners have to indicate which version they heard; the correctly pronounced Spanish version or the interlanguage version. As the program progresses the learners hear models which, through the use of sound morphing software, become progressively more and more similar. The program is sensitive to the competence of the learners at the task and adjusts the difference between the models so that learners are working at their point of competence where they are only just able to discriminate between the two models. The activity is presented in the form of a game where the learners receive feedback on their progress through the game.

The Language Train software is based on earlier software called Phonomena. Language Train, however, provides modelling and training of discriminations in the context of foreign language words while Phonomena divorces training from meaningful contexts and practises purely at the level of the phoneme.

The trial of the Language Train software is placed entirely in the context of modern foreign language teaching in Britain where the time available for teaching is short, and where language teaching is expected to be meaningful and communicative. Methods and materials, if they are to be considered effective, must involve the target language and must be able to demonstrate a learning effect or learning advantage within a few hours of use. Foreign language classes in British primary schools are small, and the sample is consequently also small.

The Language Train software is a prototype and the trail should be viewed in this light. It is intended as material which will support and enhance foreign language learning and is not a complete language learning method in its own right.

top of page