Premium
An error analysis of the word‐superiority effect
Author(s) -
Solman Robert T.
Publication year - 1988
Publication title -
british journal of psychology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.536
H-Index - 92
eISSN - 2044-8295
pISSN - 0007-1269
DOI - 10.1111/j.2044-8295.1988.tb02272.x
Subject(s) - stimulus onset asynchrony , word (group theory) , psychology , stimulus (psychology) , perception , context (archaeology) , speech recognition , arithmetic , audiology , communication , cognitive psychology , linguistics , mathematics , computer science , medicine , paleontology , philosophy , neuroscience , biology
In this experiment subjects were asked to report the identity of a position‐cued critical letter in a linear array of four letters. There were four types of arrays used: (1) unpronounceable non‐words; (2) pronounceable non‐words; (3) words in which the critical letter was minimally constrained, by the context letters; and (4) words in which the critical letter was maximally constrained by the context letters. A different group of subjects was assigned to each of these four array conditions. All four‐letter stimuli were presented in two parts. A leading array in which the information from two quadrants of a vertical by horizontal division of each letter was presented, and, after intervals of 0, 20, 40, 80, 160, 320, 640 ms and infinity (i.e. no trailing array) a trailing array of the complementary letter parts. Examination of identification errors showed that performance in the word conditions was consistently superior to performance in the non‐word conditions, and the magnitude of this difference (i.e. the word‐superiority effect) increased with increasing stimulus onset asynchrony up to 160 ms and then gradually declined. It was argued that these systematic changes in the size of the word‐superiority effect were suggestive of the presence during encoding of robust word and fragile letter codes. This was consistent with the Interactive Activation Model of Word Perception. However, this model also assumes that all strings of letters are encoded in the same way independent of their linguistic properties, and comparisons of the relative proportions of the types of errors made were not necessarily consistent with this assumption. The identification errors were classified either as ‘intrusion’ (i.e. a report of a letter not in the array), or as ‘ location’ (i.e. a report of a non‐critical or context letter from the array), and an analysis of the proportions of the latter indicated that mislocations contributed more to the total errors in the unpronounceable non‐words than in the other three types of stimuli. This suggested that the non‐words may have been encoded differently.