Is the Masked Priming Same-Different Task a Pure Measure of Prelexical Processing?

Abstract

To study prelexical processes involved in visual word recognition a task is needed that only operates at the level of abstract letter identities. The masked priming same-different task has been purported to do this, as the same pattern of priming is shown for words and nonwords. However, studies using this task have consistently found a processing advantage for words over nonwords, indicating a lexicality effect. We investigated the locus of this word advantage. Experiment 1 used conventional visually-presented reference stimuli to test previous accounts of the lexicality effect. Results rule out the use of different strategies, or strength of representations, for words and nonwords. No interaction was shown between prime type and word type, but a consistent word advantage was found. Experiment 2 used novel auditorally-presented reference stimuli to restrict nonword matching to the sublexical level. This abolished scrambled priming for nonwords, but not words. Overall this suggests the processing advantage for words over nonwords results from activation of whole-word, lexical representations. Furthermore, the number of shared open-bigrams between primes and targets could account for scrambled priming effects. These results have important implications for models of orthographic processing and studies that have used this task to investigate prelexical processes.

Non-cognate translation priming effects in the same–different task: evidence for the impact of “higher level” information

“…Nonetheless, the results of Experiments 4 and 5 clearly do indicate that pre-lexical orthography is not the entire story in the masked prime same-different task, at least with different-script (e.g., Japanese-English) bilinguals. That is, consistent with the claims made by Kelly et al (2013), it does appear that, at least for word targets, the story also requires that one assume that information beyond the orthographic level is activated and, more importantly, used in the same-different task.…”

“…This analysis, provided by Norris and colleagues' account, is quite consistent with the results of Experiments 4 and 5. Therefore, the only aspect of the story that the results of Experiments 4 and 5 alter is that they require that one acknowledge that evidence from the prime supporting the reference stimulus can come from either the lexical (Kelly et al, 2013) or semantic levels (as those are the levels at which non-cognate translation equivalents share representations). In other words, in contrast to the claims of Norris and colleagues (Kinoshita & Norris, 2009, 2010Norris & Kinoshita, 2008;Norris et al, 2010), performance in this task is not based solely on prelexical orthographic information.…”

“…Likewise, Kelly, van Heuven, Pitchford, and Ledgeway (2013) have recently noted that priming for non-word targets does not always work the same way as for word targets, suggesting that the sources of the priming in the two cases might be different. In particular, Kelly et al have argued that the extant data do not rule out the possibility that, in the case of non-word referents and targets, participants might be using pre-lexical representations, whereas, in the case of word referents and targets, they could be using lexical representations (i.e., that representations beyond the orthographic level are playing a role in this task, at least for the word targets).…”

Action and perception in literacy: A common-code for spelling and reading.

“…The modelling of form priming of the lexical decision task is best done by models which can simulate the latter; in current models, this is a multiply-constrained, non-linear, process and provides a continuing source of debate regarding the interpretation of many such results (Kelly et al, 2013;Kinoshita & Norris, 2009;Lupker & Davis, 2009). The remaining studies in this section extend the empirical reach of the SE-PC model by investigating effects not considered to involve lexical processes, and not, to my knowledge, currently simulated by other letter-coding models for reading.…”