MODULARITY2 (also AUTONOMY)
المؤلف:
John Field
المصدر:
Psycholinguistics
الجزء والصفحة:
P181
2025-09-18
473
MODULARITY2 (also AUTONOMY)
A view that each level of language processing operates independently of the others.
In an autonomous model, each level achieves an output, which it then passes on to the level above. This enables the processor to accurately interpret a sequence such as /ka:dn/ at phoneme level without suffering bias from the fact that the sequence closely resembles the word GARDEN. The latter consideration only comes into play when the output of phoneme processing reaches word level. The advantage of having information encapsulated in this way is that the processor can focus on one set of criteria at a time.
An alternative view regards language processing as interactive. At the very least, this implies that information from a particular level can influence processing at the level immediately above or below: thus, in the example just given, there would be an interaction between the phonemic evidence and the knowledge from word level of the existence of GARDEN.
In a highly interactive model, all levels of processing influence all others in both a bottom-up and a top-down way. Thus, our recognition of the word GARDEN might be influenced by the preceding context: My neighbour’s planting flowers in his-. The advantage of this kind of model is that all sources of information can be considered at once in order to resolve ambiguities. The disadvantage is that decisions become much more complex and thus, in principle, make heavier demands on working memory.
Often cited in favour of modularity is evidence (Swinney, 1979) that both meanings of a homophonous word become activated, even when the preceding context indicates which is the appropriate one. This suggests that the lexical stage has to be completed before the listener/reader is able to consider evidence at a higher (contextual) level.
Discussion of modularity is much influenced by Fodor (1983). He views the mind as composed of a set of central systems which handle generalised operations such as attention or memory. These are supplied with information by input systems which process sensory information and language. The input systems are modular and each has specific functions. Fodor characterises the systems as:
Domain specific. Input via the ears is processed as simple auditory input in the case of music or the noise of traffic, but is recoded phonologically by the speech module if it takes the form of speech.
Mandatory. We cannot help hearing an utterance as an example of speech.
Fast. The processes are highly automatic.
Informationally encapsulated. A module receives information from other modules and passes it on, but its immediate operation is not affected by information contained elsewhere. So, while engaged in processing a spoken word we cannot use context to identify the word more quickly (this does not preclude the use of contextual information at a later, post-perceptual stage).
Localised. Input systems are part of the hard-wiring of the brain; there is a fixed neural architecture for each.
See also: Interactive activation
Further reading: Fodor (1983); Wingfield and Titone (1998: 253–7)
الاكثر قراءة في Linguistics fields
اخر الاخبار
اخبار العتبة العباسية المقدسة
