Masked phonological priming effects in English - Center for Reading ...

102 K. Rastle, M. Brysbaert / Cognitive Psychology 53 (2006) 97–145investigation both because it has been studied extensively against benchmark effects ofvisual word recognition and because it has been highly influential in this area for a numberof years. Furthermore, the DRC model has been made available for public evaluation (seehttp://www.maccs.mq.edu.au/~max/DRC), thus allowing us to examine its performanceexplicitly on the actual stimuli used in our experiments. Finally, the DRC model is controlledby a number of parameters, each of which can be altered independently. The factthat such parametric alterations are possible in the implemented model provides scope foran evaluation of a number of potential theories of visual word recognition, ranging fromtheories that postulate a very ‘‘weak’’ phonological contribution to those that postulate a‘‘stronger’’ phonological contribution to this process.2. The masked phonological priming effect in English: What are the data?Five experimental paradigms have been used to examine the influence of tachistoscopicallypresented homophonic stimuli on the processing of English target words:forward-masked perceptual identification, backward-masked perceptual identification,forward-masked reading aloud, forward-masked lexical decision, and text reading (eyemovements). Though our own experimental work will use the forward-masked visual lexicaldecision task, we review evidence from all five of the paradigms both for completenessand to identify whether the appearance of the masked phonological priming effect mightbe related to particular characteristics of the performance tasks themselves. Further, weinclude all studies that fall into one of the five paradigms, even though some of these studiesclaimed to look for the boundary conditions of the masked phonological priming effectby using, for example, very short stimulus-onset-asynchronies, homophones in the stimuluslist, or parafoveal nonword stimuli. This decision was made because researchers acrossparadigms have not been consistent in their views concerning the conditions under whichthe masked phonological priming effect occurs. Our discussion of each experimental paradigmis accompanied by tables presenting various details about each experiment consideredin the meta-analysis. To enable readers to ask specific questions about the nature ofthe masked priming effect, we present more information concerning each experiment in thetables than we discuss in text (e.g., prime exposure duration, target duration). Our ownpurpose is to establish the empirical basis for a masked phonological priming effect in English,however, and so we do not entertain these more specific questions here.2.1. Forward-masked perceptual identificationIn the forward-masked perceptual identification procedure (Humphreys et al., 1982), aforward pattern mask (e.g., #########) is followed immediately by a briefly presentedword or nonword prime stimulus, which is then replaced by a briefly presented target stimulusand a backward pattern mask. Prime and target are presented in different cases, andparticipants are asked to identify the target word. The dependent variable is the percentageof target words correctly identified. Three papers (18 experiments) using this paradigm aresummarized in Table 1: Humphreys et al. (1982), Perfetti and Bell (1991, Experiment 3),and Booth, Perfetti, and MacWhinney (1999). 11 Booth et al. (1999) reported phonological priming effects in different age groups of children. We describe hereonly data collected from the older children (6th grade).