Processing of underlying and surface phonological representations in visual word recognition
Committee Members: Robin Morris (Director), Anne Bezuidenhout, Eric Holt, Kathleen Kirasic
The current study investigated the nature of phonological codes used in visual word recognition. The phonological unit selected for the study was the underlying and surface representations of word-final two-consonant codas. The underlying representation consisted of both consonants whereas the surface representation had the final consonant dropped. A set of three phonological lexical decision tasks were conducted to compare the reaction time and acceptability rate of these two levels of representation that were encoded with pseudohomophones (e.g., contakt and contak). Evidence was found that the involvement of underlying and surface representations was conditioned by coda phonotactics. These two levels of representation did not differ in either reaction time or acceptability rate for the two stop-stop codas (/kt, pt/), but the underlying representation produced faster reaction time and higher acceptability rate than the surface representation for the two nasal-stop codas (/nt, nd/). The difference between the two coda categories was attributed to the store of frequency distribution information of surface representation in language speakers' psycholinguistic knowledge that allowed for direct lexical access. However, evidence was also obtained that the underlying representation of both types of codas produced the same reaction time and acceptability rate, which indicated that processing of the underlying representation was independent of coda phonotactics. Given the result that the reaction time of the underlying representation was either the same as or faster than the reaction time of the surface representation, it was suggested that the underlying representation might be the default phonological code used in visual word recognition. The advantage of the underlying representation was associated with orthography effect such that a phonological representation faithfully mapping the orthography-phonology correspondence was used for lexical access in visual word recognition.