, Andrea Martin
, and Joanna Morris
Eye movement experiments provide evidence that readers can use parafoveally acquired information about the initial syllable of a word to identify it once it is fixated. This suggests that readers hold syllable information in a phonological representation with multiple sublexical layers. The experimental evidence is inconsistent, however. Some of the variability in the findings may be attributable to differences in materials (e.g., clear evidence for syllable effects in English has been reported only for words with second syllable stress.)
To investigate the impact of stress pattern on the time course of syllable effects in word recognition, we conducted an event-related potential (ERP) experiment. Brain potentials were recorded while participants silently read single words preceded by syllabically congruent and incongruent 40 ms primes. A 100ms forward mask preceded the prime, which was followed by a 100ms backward mask, then by a target for 650ms. Participants read the words silently, making manual responses to interspersed filler items.
First Syllable CV Target Word demote Congruent Prime DE#### Incongruent Prime DEM###
An examination of the ERP response to CV-initial targets tested whether a prime that contains the target's first syllable boundary affects word recognition differently than a prime that violates the syllable boundary. Preliminary data from mid-line electrodes seem to suggest that syllable information is detected during lexical access of low frequency words. When CV-initial targets were preceded by incongruent (as opposed to congruent) primes, the congruent and incongruent waveforms significantly diverged around 200-250 ms, with an increasing negativity for the incongruent waveform. If this negativity can be interpreted as reflecting processing difficulty, its appearance in incongruently primed CV words is consistent with the eye movement evidence discussed earlier. The occurrence of the effect between 200 and 250 ms could suggest that syllable information is utilized relatively late in the lexical access process. The time-course and polarity of this ERP response is consistent with the detection of incongruent syllable information.