Units of word recognition

Abstract

Last year at VSS, we showed that the word feature asymmetry, a high concentration of features leading to recognition in the upper half of words (Huey, 1908; Blais et al., 2009; Perea et al., 2012), cannot be accounted for by the distribution of features used in letter recognition and, thus, that words are not read by letters. We also reported a significantly smaller asymmetry towards the upper half of nonword trigrams. We hypothesised that constructs common to words and nonword trigrams were driving this asymmetry. Preliminary analyses identified two candidates: lexical trigrams and syllables. Our aim, here, was to investigate further these possible units of word recognition. Twenty observers were presented with two- and three-letter syllables (as defined in Lexique 3, New et al., 2001), as well as lexical bigrams and lexical trigrams (sequences of two or three letters found within words from Lexique 3 that are not syllables), and their non-lexical counterparts. Stimuli were blocked and presented in Arial lowercase, partially masked. The mask originated from the bottom or top and hid from 1/6 to 5/6 of the stimulus. We found that both syllabic stimuli, as well as the lexical bigrams and trigrams presented an asymmetry towards the upper half. We also found that none of these could be accounted for using the single letter data. Finally, syllables presented a significantly greater asymmetry than the lexical bigrams and trigrams. We will be discussing the implications of these findings for the units of word recognition.

Meeting abstract presented at VSS 2013