FUNCTION WORD PROCESSING
There is evidence that function words are processed differently from those bearing lexical meaning, which leads some commentators to conclude that the two sets of items may be stored separately.
There are many examples of Slips of the Tongue where content words exchange places but function words are correctly positioned (rules of word formation ! words of rule formation). This suggests that semantic and syntactic assembly are two distinct processes. Further evidence comes from studies of patients who have suffered damage to Broca’s area in the brain. Their lexicon remains relatively intact, but access to grammar (word order, inflection and function words) is often impaired. This may be because the ability to assemble syntactic structure has been affected; or it may be that the patient’s attentional capacity has diminished so that they restrict themselves to the words which are most semantically and phonologically salient, namely the lexical items. Some patients manage to retrieve function words if they are given time to do so.
A third possibility is that these aphasics have lost access to the set of ‘closed class’ items, but not to the general lexicon. This would suggest that the two categories are stored separately and accessed differently. Early support for the ‘double storage’ view came from a study which suggested that, while the recognition of lexical words is affected by their frequency, that of function words is not. The conclusion was that there was a separate ‘frequency free’ processing route for functors. The finding has since been challenged, and more recent results suggest that, even if it holds for reading, it does not apply in listening.
However, recent evidence from brain imaging appears to confirm that function words are stored and processed separately. They seem to be more localised than lexical words and thus more available for rapid processing. It is also suggestive that language-acquiring infants produce function words relatively late, despite their high frequency. This may indicate that a different retrieval process needs to be established.
Processing a lexical word involves two operations: matching a string of sounds against a mental store of spoken words and accessing meaning. By contrast, processing a function word need only involve the first. Some models of speech recognition assume that strong syllables in the speech stream trigger full lexical access (form and meaning) while weak syllables are simply pattern-matched against a separate ‘closed class’ store. In this way, the listener exploits an association in English between weak stress and function words.
Further findings in this area have been produced by studies of verbatim recall which show that content words are remembered more accurately than functors. This may indicate that it is functors which feed first into the construction of a mental representation, and are thus more quickly lost to recall. Or it may suggest that function words are processed at a lower level of attention: they are less informative because they are more frequent. A recent suggestion has been that we do not store language input in verbatim form at all. If we have to report a speaker’s exact words, we do so by identifying words in our lexicon which have been recently activated. This is possible with lexical words but not with functors, which have not been subject to the same activation process.
See also: Aphasia, Lexical access, Lexical segmentation, Lexical storage, Verbatim recall
Further reading: Caplan (1992: 329–50); Cutler (1989); Grosjean and Gee (1987); Potter and Lombardi (1990); Shillcock and Bard (1993)
