The locus of context effects in word processing

Abstract

Understanding the processes involved in decoding written material is of major interest in psychology. This goal has led to the development of information processing models in which the different functions involved in decoding are thought of as separate stages. For example, given a reaction time task in which subjects are required to discriminate between word and nonword items, a model attempting to describe the processing events leading to a response could include three stages: (1) An initial encoding stage where the item is transformed into an appropriate representation; (2) a lexical access stage where it is determined if the item is stored in memory; and (3) a decision response stage in which the word-nonword decision is made and the motor response is executed. Within this framework, a variable which is known to affect item processing is thought of as affecting at least one stage. Employing the additive factors method allows one to make inferences concerning the locus of the effect of a variable. The variable to be studied is manipulated along with another variable that is believed to affect a particular stage. Observing the statistical patterns of 1 additivity and interaction between the two factors provides an outline for the interpretation; additivity indicates that the variables are affecting different stages, while interaction indicates a common stage. The purpose of this study was to determine in which stage or stages of item processing context has its effect. According to the logic of the additive factors method, in order to determine if context effects are perceptual in nature, context should be manipulated along with an encoding stage variable. Stimulus quality is assumed to be such a variable. If the effects of stimulus quality and context are additive, it is interpreted as meaning context is a post-perceptual variable. However, an interaction between these two factors is assumed to be indicative that the factors are influencing a common stage; in the event of such an outcome, the inference would be made that context affects the encoding stage. Regardless of the context X stimulus quality outcome, in order to determine if and where context has effects on item processing beyond the encoding stage, it should be manipulated along with a variable which affects a post-perceptual stage. Previous research appears to have established that context has no effect on the lexical access stage. However, prior studies have not included the decision stage as a possible locus for contextual effects. Varying the percentage of words and nonwords presented (the relative response frequency) is assumed to affect this stage. Therefore, varying both context and relative response frequency should indicate whether context has any effect on the decision stage.