CONTEXT EFFECTS
Ways in which contextual information may (a) influence the interpretation of a word or an utterance or (b) speed up lexical access.
The use of contextual cues demands attentional resources. It should be distinguished from the phenomenon of spreading activation, which is highly automatic and not normally subject to our control. However, the two are sometimes difficult to distinguish. Suppose recognition of the word SPOON is speeded up when it occurs in the utterance: He laid the table with a knife, fork and spoon. To what extent is this effect due to the table-laying context and to what extent is it due to the spread of activation from the forms KNIFE and FORK? There is considerable disagreement as to how contextual information influences the construction of meaning. Possible views include:
Bottom-up driven. Perceptual information is primary; contextual information is used to check and enrich it.
Top-down driven. Context biases interpretation ahead of perception.
Interactive. Contextual information interacts with perceptual information at all stages of processing.
‘Bottom-up priority’. A minimal amount of perceptual data is processed before contextual influences can apply.
Ambiguity resolution. Contextual data is only used in cases of ambiguity.
The timing of context effects is especially a matter of debate in relation to lexical access. Most accounts assume that, when a listener hears part of a word, he/she forms a list of potential matches. Does context operate to limit the number of word candidates that are chosen? Does it (by increasing or reducing activation) influence the choice of the successful candidate? Or is it simply used to check the appropriacy of the candidate that has been chosen?
A connected issue is whether and how context plays a role in resolving ambiguity. Does it simplify the process by enabling the listener or reader to select a single appropriate sense for a polysemous or homonymous word? Or does the listener/reader automatically access all possible senses before using context to determine which is the correct one? A much-cited finding (Swinney, 1979) suggested that the latter is the case. Subjects were presented with sentences containing ambiguous words. (Example: The man was not surprised when he found several bugs in the corner of his room.) Even where there was a highly specific context (several spiders, roaches and other bugs), they showed signs of having retrieved both possible senses (BUG ¼ spy gadget, BUG ¼ insect). The same finding has been reported with homonyms which come from different word classes: i.e. where the context provides syntactic as well as semantic cues. Thus, the word [wi:k] speeds the recognition of both MONTH and STRONG, even when it occurs in a sentence where it is clearly a noun.
Further evidence is based on the eye movements of readers. The presence of an ambiguous word in a text increases gaze duration. Two factors contribute:
a. whether the two senses of the word are balanced in terms of frequency or whether one is dominant (more frequent);
b. whether disambiguating contextual information occurs in advance of the word.
Recent findings indicate that:
Only the dominant sense is retrieved when a context indicates the dominant one.
Both senses of a word are retrieved when a context indicates the subordinate one.
However, the possibility remains that all senses are retrieved– but foregrounded to different degrees depending upon their frequency and contextual appropriacy.
See also: Interactive activation, Lexical access, Listening: higher-level processes, Reading: bottom-up vs top-down, Reading: higher-level processes, Speech perception: autonomous vs interactive, Top-down processing
Further reading: Brown and Yule (1983); Reeves et al. (1998: 202–8); Simpson (1994)
