Catherine L. Harris & Alison L. Morris Boston University
Two experiments failed to replicate Chialant and Caramazza's (1997) finding that identity RB decreases as a function of lag between the two critical words, while for orthographic neighbors, RB increases (up to a point) as a function of lag. We explain their findings as due to guessing strategies, combined with research paticipants being affected by repetition priming at long lags and repetition blindness at short lags, with non-identical words benefiting less from repetition priming.
Our own experiments on both identity and orthographic RB have convinced us that both of these show a decrease in RB as a function of lag. But could orthographic RB still be just a result of neighborhood inhibition? In the current paper, we argue that this is not a parsimonious explanation, since we find equally robust RB for critical words that are neighbors (tower lower) and for non-neighbors (flowerpot lower). We also find RB for words sharing 3 consecutive letters (dangle danger), 3 alternating letters (donkey danger), 3 middle letters (tangle danger), initial and final letters (doctor danger) and even a single initial letter (doodle danger), with amount of RB increasing as a function of number of repeated letters.
In our replication of Chialant and Caramazza's study, we found that at the longest lags, subjects could often (but not always) report both repeated words. Because subjects were aware that many of the trials contained repeated words, at least some subjects appear to have adopted the strategy of reporting a repetition. This strategy will result in a correct score for the identical condition but an incorrect score for the neighbor condition. If subjects are most likely to use this strategy when they can accurately see at least some of the word, this will produce C&C's results, as it will lead to an inflated percent correct for the identical condition at the longest lag, and a inflated error score for the orthographic neighbor condition at the longest lag.
Questioning the findings of Chialant and Caramazza is important because their claims have profound implications for models of word recognition and lexical representation. They interpreted their data as evidence that repetition blindness is sensitive to lexical identity, not merely orthographic similarity. They suggested that RB could be a new tool for probing the nature of the mental lexicon. Our results cast doubt on the assertion that RB can be used in this way. But even if RB is only sensitive to visual similarity, it is still a valuable tool for understanding visual word recognition. We will use our data to describe how orthographic RB can aid in identifying the orthographic codes used in mediating between visual feature and word recognition.