Phonetic recoding of phonologically ambiguous printed words

Abstract

Speech detection and matching simultaneously presented printed and spoken words were used to examine phonologic and phonetic processing of Hebrew heterophonic homographs. Subjects detected a correspondence between an ambiguous letter string and the amplitude envelopes of both dominant and subordinate phonological alternatives. Similar effects were obtained when the homographs were phonologically disambiguated by adding vowel marks. The matching of the unpointed printed forms of heterophonic homographs to the dominant and subordinate spoken alternatives presented auditorily was as fast as matching the pointed unambiguous forms to the respective spoken words. This outcome was not obtained when print and speech were not presented simultaneously. These results suggest that printed heterophonic homographs activate the two spoken alternatives they represent and provide further confirmation for fast phonetic recoding in reading. Most studies of lexical ambiguity have examined the processing of printed homophonic homographs embedded in text or presented in isolation (e.g., Onifer & Swinney, 1981; Seidenberg, Tanenhaus, Leiman, & Bienkowski, 1982; Simpson & Burgess, 1985; Swinney, 1979). Homophonic homographs (e.g., bug) are characterized by an orthographic structure that has one pronunciation but two different meanings in semantic memory. Research with homophonic homographs has focused on whether the two meanings related to the orthographic structure are activated in parallel or whether one meaning acquires dominance at some stage after the presentation of the ambiguous letter string. Several studies have suggested that, even in a biasing context, all the meanings of a homograph may be automatically activated and retrieved (e.g., Onifer & Swinney, 1981; Seidenberg et al., 1982; Swinney, 1979; Tanenhaus, Leiman, & Seidenberg, 1979). In contrast, it has been shown that biasing contextual information affects lexical processing of homographs at an early stage, selecting only contextually appropriate meanings (e.g., Glucksberg, Kreuz, & Rho, 1986; Schvaneveldt, Meyer, & Becker, 1976). A third approach poses that exhaustive access that does not occur in parallel but is determined by the relative frequency of the two meanings related to the ambiguous word (e.g., Duffy, Morris, & Rayner, 1988; Forster & Bendall, 1976; Hogaboam & Perfetti, 1975; Neill, Hilliard, & Cooper, 1988; Simpson, 1981; and see Simpson, 1984, for a review).