Abstract
This experiment was designed to examine the effects of varying the relationship between temporally separated strings of letter parts on the word-superiority effect. The subjects were asked to report the identity of a position-cued critical letter in an array of four letters. One group of sixteen subjects reported critical letters from words and a second group of sixteen subjects reported the same letters from non-words. All the letter strings 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 40, 80,120,160, and 200 msec, a trailing array of letter parts. The relationship between the part-letters in the leading and trailing arrays was systematically varied as follows: (1) both the critical and the context letter parts were complementary (i.e., when combined they formed complete letters); (2) the critical letter parts were complementary but the context letter parts were not; (3) the critical letter parts were not complementary but the context letter parts were; and (4) neither the critical letter parts nor the context letter parts were complementary. The critical letter was reported more accurately in the words than in the non-words, and the magnitude of this word-superiority effect was significantly reduced when no trailing letter-parts were complementary. Also, the accuracy of critical letter report was independently depressed by non-complementary critical letter information and by non-complementary context letter information. Both these results raise interpretation problems for the Interactive Activation Model of word perception, with the former indicating that pattern masking is not necessary for the emergence of a large word-superiority effect, and the latter suggesting that readout from both the letter and word level contributes to this word advantage.