A dual read-out model of word context effects in letter perception: Further investigations of the word superiority effect.

Abstract

A dual read-out model of context effects in letter perception is described that predicts forced-choice accuracy in the Reicher paradigm and its relation to word reportability. It is hypothesized that a correct choice to a letter in a word context is made when either the correct letter representation or a word representation containing the correct letter in the correct position reaches a response threshold (a criterion level of activation). This hypothesis was implemented using the basic architecture of the interactive activation model (J. L. McClelland & D. E. Rumelhart, 1981) in its semistochastic variant (A. M. Jacobs & J. Grainger, 1992). The model successfully captures the data of J. C. Johnston (1978), otherwise thought to be critically damaging for this type of model, and accurately predicts performance in a series of new experiments using the Reicher paradigm.

The relative benefit of word context is a constant proportion of letter identification time

“…Using this criterion and the data presented in Figure 4 of Massaro and Cohen (1994), the estimated SD threshold is 93.3 msec for letters in nonwords and 65.1 msec for letters in words, a ratio of 1.4ϫ. A similar analysis of the data from Figure 4 of Grainger and Jacobs (1994) shows thresholds of 32.3 msec for letters in nonwords and 24.7 msec for letters in words, a ratio of 1.3ϫ. Both of these studies used the more traditional two-alternative forced choice paradigm, but varied the SD.…”

“…The two experiments reported here, as well as the data in Massaro and Cohen (1994) and Grainger and Jacobs (1994), all suggest that observers require about 1.4ϫ as much inspection time to identify letters within nonword letter strings as they do letters within words. The data reported here suggest that this is true regardless of retinal location (fovea or periphery) or the degree of visual crowding.…”

The Split Fovea Theory and the Leicester critique: What do the data say?

“…They are a crude measure when it comes to study the fine-grained, online processes involved in normal word recognition. High accuracy in the Reicher-Wheeler task can be reached on the basis of correct perceptual identification instead of recognition at word level (Grainger & Jacobs, 1994) and can thus reflect other processes than we intend to measure. Jordan et al (2008) (Nazir, Ben-Boutayab, Decoppet, Deutsch, & Frost, 2004).…”

Lexical organization of language-ambiguous and language-specific words in bilinguals

“…This effect has been termed as the word superiority effect (e.g., McClelland, 1976;Prinzmetal, 1992; see also Grainger, Bouttevin, Truc, Bastien, & Ziegler, 2003). Similarly, previous studies using the Reicher-Wheeler paradigm have also shown that readers struggle less when identifying letters embedded in legal non-words than those embedded in illegal non-words (i.e., the pseudo-word superiority effect; see Grainger & Jacobs, 1994;Grainger et al, 2003), demonstrating that orthographic coding is highly influenced by the degree of familiarity with the orthographic structure of the target language. Interestingly, Grossi, Murphy and Boggan (2008) extended these effects to bilingual readers, showing that non-balanced Italian-English bilinguals display sizeable (pseudo-)word superiority effects when tested in their two languages.…”

“…Following the assumption that orthographic coding is modulated by the degree of familiarity of the reader with the orthographic structure of language (see Grainger et al, 2003), we hypothesized that bilingual participants would display significantly better performance in letter identification for Basque words and non-words that are orthographically legal in Spanish (i.e., orthographically unmarked items) than for Basque words and non-words that feature orthographic cues implausible in Spanish (i.e., orthographically marked items). Under the assumption that orthographically unmarked words lead to language non-selective activation, the predicted enhanced recognition of letters embedded in orthographically unmarked strings stems from the increased feedback received by these letters from higher-level lexical representations in the two languages (as predicted by interactive-activation models of bilingual and monolingual word recognition; e.g., Grainger & Jacobs, 1994;Grainger, Midgley, & Holcomb, 2010;9 McClelland & Rumelhart, 1981;van Heuven & Dijkstra, 2010). According to the hypothesis that orthographically marked words would lead to language-selective lexical-orthographic activation, letters embedded in language-specific Basque strings would receive a lesser degree of feedback activation from higher levels of processing (they would not co-activate Spanish words), thus resulting in impoverished recognition of sub-lexical units.…”