Lack of word-superiority effect in processing letter features

Abstract

Letter recognition is typically faster in words than in nonwords. In this study, we tested the word-superiority effect obtained when either subletter features or misalignment of letters had to be detected. Subjects were presented with both high-and low-frequency words and with legal and illegal nonwords. Space among the letters varied. In the regular space condition, letters within a string were separated by normal spaces; in the irregular space condition, letters were separated by normal spaces, small spaces, or large spaces. In Experiment 1, subjects were required to detect the presence of a bold segment contained in one of the letters of the string. No wordsuperiority effect was obtained. Furthermore, spacing affected neither latencies nor response accuracy. In Experiment 2, subjects were required to detect the presence of a letter misaligned with respect to the others. Again, no word-superiority effect was obtained. However, spacing affected this task, irregularly spaced strings being responded to more slowly and less accurately than regularly spaced strings. The results indicate that at the first stages of analysis, words and nonwords are similarly coded. The pattern obtained is consistent with a multistage model of word recognition in which parallel feature extraction processes are followed by a level at which spatial relationships are computed and by a third level at which abstract grapheme identity is recovered from letter shapes.Since Cattel's studies of 1886, the problem of determining which visual characteristics of a word are encoded during reading has become a central issue of cognitive psychology. The word-superiority effect (WSE)-the higher detectability of letters when they are contained in words-is a well-studied phenomenon (for a review, see Carr, 1986). Advantages of words over single letters (Johnston & McClelland, 1973;Prinzmetal, 1992) and over orthographically regular (Estes & Brunn, 1987;Solman, 1987) and irregular nonwords (Baron & Thurston, 1973;Marchetti & Mewhort, 1986;Sphoer & Smith, 1975) have been reported. The WSE over nonwords in letter detection is manifested both in latencies of letter search tasks (Krueger, 1992) and in accuracy in forcedchoice procedures (Reicher, 1969). The effect could be due either to an enhanced visual perceptibility of words (i.e., the physical characteristics of words actually provide a more perceptible stimulus) or to verbal and/or lexical factors, and hence dependent on more abstract computations.The processes involved in the analysis of visual information are often regarded as roughly hierarchical. Feature extraction processes such as edge and blob detection, texture, size, and color information are commonly thought to be computed during the initial stage of visual analysis. On the other hand, the high-level processes such as object (or word) recognition are thought of as the end product and they can no longer be considered purely visual, Correspondence should be addressed to F. Peressotti, Dipartimento di Psicologia dello Sviluppo e della Socializza...