PROC: The online processing of homophonous words
Project PROC investigates whether stored forms of homophonous words and affixes differ in subphonemic detail. Results obtained in the first phase suggest that frequency-dependent word length differences between heterographic homophones (English thyme-time, German Saite-Seite) as reported by Gahl (2008) seem to be reliable and can be observed experimentally with well-controlled stimulus materials. By contrast, homographic homophones (English bank-bank, German Bank-Bank) did not show such effects. In comprehension, listeners do not seem to exploit word length differences for the disambiguation of homophonous words. We interpret these findings as speaking against the assumption of subphonemic detail (and hence different word forms) in lexical representations of homophones in general. For heterographic homophones, distinct orthographic representations may either induce separate phonemic representations or affect word length at later stages of word production.
In phase 2, project PROC will follow up on these findings in two lines of research. Using behavioral methods, Line 1 will clarify open questions with regard to frequency inheritance and frequency-dependent length effects of homophones. More specifically:
(a) Does frequency inheritance imply shared lexical word forms or can it also be observed between phonologically similar words (lexical neighbours)?
(b) Can word-lengthening effects be induced at later processing stages of word production?
Line 2 will investigate the relationship between lexical phonemic and graphemic word forms, more specifically, whether heterographic homophones have shared phonemic lexical representations or not. The two options predict different patterns of phonologically-mediated spelling errors. We will investigate the production and comprehension of spelling errors using corpus analysis, behavioural paradigms, and EEG.