In phonetics, the Shortlist model is a hybrid model of auditory word recognition. The term hybrid means that the initial stage of word recognition is assumed not to be influenced by non-sensory sources of information, and that the later stages of selection and integration are assumed to be affected by these non-sensory sources of information.
The Shortlist model takes a phoneme string as its input. On the basis of this input a shortlist of word candidates is selected, which enter into competition. The shortlist is constantly updated while more phonemic information comes in, and while candidates are eliminated as a result of the competition process. Sentence context influences lexical processing during the stage of selection. Apart from placing the effect of sentence context relatively late during selection (thus decreasing the importance of the sentence context), Shortlist also incorporates a mechanism of bottom-up inhibition: in case a contextually appropriate candidate no longer fits the acoustic information, its activation is decreased.
- Norris, D. 1994. Shortlist: A connectionist model of continuous speech recognition, Cognition, 52, 189-234