Criterion bias and search sequence bias in word recognition
2017-01-06T05:23:42Z (GMT) by
This thesis was concerned to test the adequacy of criterion bias explanations of the word frequency effect and the semantic priming effect under conditions where the identity of word letters is easily determined.<br><br> Evidence to support the action of criterion bias was provided by the first four experiments which employed the lexical decision task. There it was found that nonwords which were misspelled versions of high frequency words (e.g. MOHTER) were more likely to be misidentified as words than were nonwords that were misspelled versions of low frequency words (e.g. BOHTER). It was also found that more errors occurred for misspelled words preceded by a semantically related word (e.g. LETTER ENVLEOPE) than for those preceded by an unrelated word (e.g. POLICE ENVLEOPE). These results would be expected if high frequency and semantic priming cause words to be identified before all stimulus information is processed.<br><br> The next two experiments involved a further test of the criterion bias approach by determining whether frequency and priming effects survive a task which demands an unusually detailed examination of the stimuli. This task was the misspelling decision task, where subjects were required to determine whether letter sequences were misspelled words (for example the item MOTHER should receive a no response, MOHTER a yes response, and MIRPREAT a no response). It was found that frequency and priming effects were largely unaffected by this process, and it was suggested that a modified search model (employing search sequence bias) would provide a better account of the results of the misspelling decision experiments.<br><br> The final three experiments examined whether the misspelling decision results were changed if items were presented in randomly mixed case (e.g. mOhTer). It was found that case mixing did not alter the magnitude of frequency or priming effects, and it was argued that this opposed conventional criterion bias models (e.g. Morton, 1970) and ruled out the possibility that the detailed checking required by misspelling decision could have been carried out using a physical representation of the raw stimulus (Becker, 1976).<br><br> In the final chapter it was argued that criterion bias has little value in explaining the results outlined and that word recognition models should appeal to alternative principles of explanation.