Cognitive processes in speech perception

The recognition of spoken language involves the extraction of acoustic-phonetic information from the speech signal, and the mapping of this information onto cognitive representations. To develop accurate psycholinguistic models of this process, we need to know what information is extracted from the signal, and when and how it is integrated with stored knowledge. The central knowledge store for speech perception is the mental lexicon, that is, a stored representation of words. The utterances we hear may be new to us, but they are made up of known words; by recognising the words and parsing their sequence we are able to understand what has been said. Word recognition, we argue, is at the heart of speech perception.