PARALLEL PROCESSING
A model of language production and perception where the user operates at several different levels of representation (sounds, words, semantics, syntax) at the same time.
It is easy to assume that processing is serial, with a listener, for example, first assembling phonemes into a word, then accessing the word’s meaning and then fitting the word into a developing syntactic structure. However, current evidence suggests that the listener begins to process the message about 0.2 seconds (roughly the length of a syllable) after the speaker starts speaking. This would seem to favour a system in which all levels of processing are operative simultaneously (in parallel). Similar evidence for reading suggests that a written word is also processed at several different levels at the same time: letter features (lines and curves), letters, order of letters, whole word.
Many parallel processing models are interactive. Each level of processing communicates with those above and below it to produce evidence which supports or rules out a particular interpretation. However, parallel processing is not necessarily incompatible with a modular system. In such a system, each level has to complete its processing operation before passing on the outcome to the level above; but this does not exclude the possibility of all levels being active simultaneously.
The distinction between serial and parallel processing is relevant to models of lexical access. Serial models of access (e.g. Forster’s search model) assume that we work through lexical entries in turn until we find a match for a word in the stimulus. In parallel models such as Morton’s logogen system, potential word matches are represented as being in competition with each other: evidence builds up for and against each candidate until a correct match is made.
See also: Lexical access, Search model
Further reading: Forster (1990); Jackendoff (1987: Chap. 6)
