More Words in the Neighborhood: Interference in Lexical Decision Due to Deletion Neighbors

Article excerpt

This article reports two lexical decision experiments that provide evidence for the automatic activation of deletion neighbors-that is, words that overlap with the presented word save for the deletion of one letter. Experiment 1 showed slower and less accurate no decisions for nonwords with deletion neighbors (e.g., come in scome), relative to control nonwords. Experiment 2 showed slower and less accurate yes decisions for words with higher frequency deletion neighbors, relative to control words. An important methodological implication of these results is that stimuli should be equated using a different definition of orthographic neighborhood from that which is currently the norm. The results also have significant theoretical implications for input coding schemes and the mechanisms underlying recognition of familiar words.

The effects of orthographic similarity on visual word identification have attracted considerable theoretical and empirical attention (for a review, see Andrews, 1997). The research suggests that processing of a written word results in the automatic activation of orthographically similar words (the word's orthographic neighbors) and that this can affect the speed of lexical access. Establishing which words make up the similarity neighborhood of a letter string can offer valuable insights into the structure of lexical representations and the nature of input coding. Theoretical input coding schemes make predictions about the similarity of pairs of letter strings and, hence, which words are likely to be activated following presentation of a target letter string. For example, some coding schemes predict that the words train and rain are coded in such a way that there is little or no overlap in their respective input codes; such schemes can be falsified if evidence is found that the representation of rain is automatically activated following presentation of train. Another key question concerns whether the activation of orthographic neighbors facilitates or inhibits word recognition; the answer to this question is critical for understanding the mechanisms underlying word identification. The present article reports two experiments that bear upon these issues.

The first lexical decision experiment in which orthographic neighborhood effects were investigated was reported by Coltheart, Davelaar, Jonasson, and Besner ( 1977). They examined the effect of a metric that they called N, which is a count of the number of substitution neighbors (SNs) of a letter string-the number of words sharing the same letter in all but one position. They found that N had no effect on the latency of yes responses, but that no responses to large-TV nonwords were slower than those to small-iV nonwords. A wealth of subsequent research has examined the effect on stimulus recognition of both the size of the similarity neighborhood and the frequency of the words contained in the neighborhood. Virtually all of this research has adopted the definition of orthographic neighbors employed by Coltheart et al (1977).

However, the TV metric is probably only an approximate measure of the size of a word's (or a nonword's) similarity neighborhood. The first indication of this was provided by evidence of transposed letter similarity effects (e.g., Andrews, 1996; Chambers, 1979). A transposition neighbor (TN) of a letter string is a word that is identical save for the transposition of two adjacent letters. The confusability of TN pairs such as clam and calm suggests that the definition of a letter string's neighborhood should be broadened to include not just words that can be formed by letter substitution, but also words formed by letter transpositions.

In this article, we present evidence for a further broadening of the similarity neighborhood to include words formed by letter deletions. We define a deletion neighbor (DN) of a word to be a letter string that differs from that word by the deletion of a single letter; for example, the DNs of plane are plan, pane, and lane. …


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.