A test of the generality of the word-context effect

Abstract

Three experiments tested the generality of the finding of Reicher (19691 and Wheeler (19701 that a word context facilitated the accuracy of recognition of an embedded target letter. In the present experiments, the context was a letter and the target was a letter fragment. The letter context greatly facilitated the recognition of the letter fragment. In Experiment I, this improvement was in many cases from chance level to perfect performance. An interpretation of this effect in terms of greater familiarity of the fragment and its context was disconfirmed in Experiment II. A procedure of preexposure cueing in Experiment III eliminated the superior recognizability of fragments having a letter context.Reicher (1969) and Wheeler (1970) have reported a result which. apparently. is sufficiently counterintuitive and counter current assumptions to have generated considerable interest. Recognition was found to be better for a target letter embedded in a four-letter word than for the same target letter presented alone for the same brief exposure duration. For most experiments and real-life situations. the redundancy of letter combinations in a word is a possible basis for subjects to guess the target letter. But. under the conditions of the Reicher-Wheeler experiments. a guessing interpretation of their effect appears untenable and the superiority of recognition for a target letter in a word appears to reflect a property of the processing of the letter.In a typical trial in one of their experiments, the word READ. for example. was presented briefly and then followed by the response alternatives E and O. The subject's task was to indicate which letter had appeared in the displayed word. Since ROAD is also a common English word. no clue as to the identity of the target letter is given by knowing any of the letters common to the words READ and ROAD.Considering the relative recency of Reicher's (Note I) original report. a considerable amount of research has been produced concerned with identifying the processes underlying the word-context effect (WCE). Little attention. however. has been paid to the question of whether or not the variables responsible for the WCE extend to situations beyond that of a target letter presented in a word or a pseudoword context. Biederman and his associates (Biederman.

Perceptual versus postperceptual mediation of visual context effects: Evidence from the letter-superiority effect

“…The size of the effect, about 4 %, was far smaller than the LeE reported by Schendel and Shaw (1976). However, in contrast to the effect reported by Schendel and Shaw, the present effect cannot be explained by structural redundancy.…”

“…Subjects may use their knowledge of the structural redundancies in letters, that is, the correlation between letters and their fragments, in improving identification performance in the fragmentin-context condition, whereas such information is not available in the fragment-alone condition. In fact, this is very similar to the interpretation of the LCE adopted by Schendel and Shaw (1976). They derived additional support for that explanation from the results of their Experiment 3.…”

“…Third, context effects in the perception of letters have not been extensively researched. To date, we know of only one report documenting such an investigation (Schendel & Shaw, 1976). Schendel and Shaw (1976) attempted to test the generality of the finding that briefly presented letters are identified more accurately when embedded in words than when presented alone (i.e., the WLE; e.g., Reicher, 1969).…”

“…To date, we know of only one report documenting such an investigation (Schendel & Shaw, 1976). Schendel and Shaw (1976) attempted to test the generality of the finding that briefly presented letters are identified more accurately when embedded in words than when presented alone (i.e., the WLE; e.g., Reicher, 1969). The stimuli employed in their experiments were letter pairs such as Hand N, as well as the critical letter fragments that distinguished between the members of each pair (e.g., -and \ for H vs. N).…”

“…The original demonstrations of the OSE (Weisstein & Harris, 1974) and the OLE (Williams & Weisstein, 1978) were aimed at drawing an "analogy between object and word perception, and opening further opportunities for studying the effects of contexts on the perception of constituent elements without being restricted to linguistic material" (Williams & Weisstein, 1978, p. 85). Similarly, the welldocumented context effects in word perception inspired the exploration of a variety of visual-context effects with stimuli such as faces (e.g., Gyoba, Arimura, & Maruyama, 1980;Homa, Haver, & Schwartz, 1976;Purcell & Stewart, 1986van Santen & Jonides, 1978), letters (Schendel & Shaw, 1976), and real-world scenes (Biederman, 1972(Biederman, , 1981Biederman, Glass, & Stacey, 1973;Palmer, 1975). This research was supported by NSERC Grant A2579 to P. J. and NSERC Grant 001'0105451 to E. M. R. We are indebted to Liz Bosman, Phil Merikle, and Derek Besner for many constructive discussions of the issues, to Keith McGowan for writing the software, and to Susan Fraser for assisting in data collection.…”