Pseudoword context effects on letter perception: The role of word misperception

Abstract

Three experiments using the Reicher–Wheeler task tested for letter report accuracy in pseudoword stimuli as a function of the pseudoword's similarity to real words. When pseudowords had a single word neighbour, performance was enhanced (relative to a letter‐in‐x condition) when the neighbour contained the target letter in the correct position, and diminished when it contained the wrong alternative. Increasing the number of compatible word neighbours of pseudoword stimuli had no additional influence on performance over and above the benefits of a single neighbour. This suggests that read‐out from whole‐word orthographic representations following word misperception is one component of pseudoword context effects observed with impoverished stimulus presentation.

Notes

Correspondence should be addressed to Jonathan Grainger, Laboratoire de Psychologie Cognitive, Université de Provence, 3 pl. Victor Hugo, 13331 Marseille, France. Email: [email protected]‐mrs.fr

The authors thank Ken Paap and two anonymous reviewers for their insightful comments on earlier versions of this work.

As noted in our previous work (Grainger & Jacobs, 1994), we cannot rule out the possibility that the observed correlation between word and letter report accuracy is due to the common influence of some external factor(s) on both processes.

Following the convention adopted by Paap, Chun, and Vonnahme (1999) the first stimulus corresponds to the target, followed by the forced‐choice alternatives and the position tested (D and T presented as alternatives for the fourth letter in the given example).

The interaction also indicates that level of word misperception is not simply correlated with some general fluency in perceptual processing that would affect word and letter perception alike, since the identification of letters in strings of Xs is actually lower in the high misperception group. This could reflect the role of attentional factors related to word identification scores, with “high‐identification” participants allocating more attention to the word level, and “low‐identification” participants allocating more attention to the letter level.

We draw readers' attention to the fact that Paap et al. (1999) did find a pseudoword superiority over irregular anagrams in a neutral baseword condition analogous to that tested in Experiment 1. This discrepancy will be examined in more detail in the general discussion.

Concerning this particular result, the IAM and AVM could be saved by arguing that the negative baseword bias condition violates the rules for stimulus selection in the Reicher–Wheeler task that were designed to prohibit the use of postperceptual guessing strategies. We would argue that inferential processes such as postperceptual guessing determine response selection in the Reicher–Wheeler task even in conditions designed to eliminate such processes.

This raises the issue as to what might be an ideal baseline condition against which performance to pseudoword stimuli should be measured. Should future experiments use letter‐in‐x strings or irregular anagrams or consonant strings? We would argue that there is no ideal baseline condition. Future research should help understand how letter identification performance differs in letter‐in‐x strings compared to irregular anagrams, thereby clarifying the interpretation of prior results involving such conditions.