Second Language Learners’ Spoken Discourse: Practice and Corrective Feedback through Automatic Speech Recognition

Background: Automatic Speech Recognition (Asr)

Standard ASR systems are generally employed to recognize words. The ASR system consists of a decoder (the search algorithm) and three 'knowledge sources': the language model, the lexicon, and the acoustic models. The language model (LM) contains probabilities of words and sequences of words. Acoustic models are models of how the sounds of a language are pronounced; in most cases so-called hidden Markov models (HMMs) are used, but it is also possible to use artificial neural networks (ANNs). The lexicon is the connection between the language model and the acoustic models. It contains information on how the words are pronounced, in terms of sequences of speech sounds. Therefore, the lexicon contains two representations for every entry: an orthographic transcription representing how a word is written and a phonological transcription representing how a word is pronounced. Since words can be pronounced in different ways, lexicons often contain more than one entry for some words, i.e. the pronunciation variants, which indicate possible pronunciations of one and the same word.