How much is phonological information worth: Evidence of functional dependency on phonology in the perception of word meaning

Researchers have not reached a consensus about the worth of phonology in visual word perception, especially in terms of whether or not word perception relies on phonological recoding. Previous work yielded data supporting three contrasting hypotheses: (1) Phonological recoding can be strategically controlled by the reader (the strategic hypothesis). (2) Phonological recoding is slow and affects only recognition of low-frequency words (the horse-race hypothesis). (3) Phonological recoding is obligatory in word recognition (the mediation hypothesis). This thesis examines the worth of phonology both in terms of its necessity in visual word perception and its functional relationship with the perception of word meaning. Specifically, the necessity and function of phonology were tested in three experiments using a semantic categorization task. Experiment 1 manipulated the proportion (50%, 18%, 0%) of homophonic words (e.g., "bare"--A kind of animal) in the experimental list to induce biased processing strategies. The results showed strong strategic bias against the use of phonological information in the presence of a large number of homophones. However, homophone interference persisted irrespective of homophone proportion and word frequency. Experiment 2 examined functional relationships between the perception of word meaning and phonology using forced responses at three (0, 200, and 400 msec) target-response SOAs (stimulus-onset asynchrony). The results showed homophone interference in all three SOA conditions although high-frequency words exhibited strong homophone interference only when SOA was 200 msec. In contrast, the magnitude of homophone interference for low-frequency words became strongest when SOA was 400 msec, suggesting a word-frequency-based variable time-window for detecting phonological effects. Experiment 3 examined whether homophone interference could be solely attributed to the speed and automaticity of visual-phonologic resonance using categorization and letter search tasks. Results suggest that mere speed and automaticity cannot account for the homophone effects in categorization. Results of three experiments taken together support the mediation hypothesis and suggest that the worth of phonology may largely lie in the functional dependency on phonology in the perception of word meaning.