Lexical and sub-lexical representation in visual word recognition

Kichun Nam

Department of Psychology, Korea University

Korean sentences are spaced by Eojeols which consist of several morphemes and words. In this paper, the way of how Eojeols are comprehended and represented in mental lexicon was examined. In the first and second study, the primed lexical decision task was used to see whether Eojeols are represented in the full-list form or decomposed morphemes in mental lexicon. It was found that Korean Eojeols are represented in two forms: the stem plus grammatical pre-ending affix and the ending affix. It is reasoned that the representation structure of Eojeols in mental lexicon has evolved for the cognitive economy to reduce the memory size and to reduce interpretation ambiguities. In the third and fourth study, what the representation unit is in mental lexicon was investigated by using the form priming technique. It was found that the Eojeol representation unit in mental lexicon is the phonological syllable and this representation unit is common across visual and spoken word recognition. In the fifth study, the research question of whether there are sub-lexical units was examined. Based on the word length effect and the form priming by overlapping the syllable body, the letter and body plus coda units were proposed for the sub-lexical units. In the sixth study, in order to examine the validity of the proposed representation of Eojeols, the computational simulation was performed. The computational model simulated the human data very closely. Based on the computational model, the electronic dictionary to model mental lexicon was proposed. Also, it was examined how the morphologically complex Eojeols are comprehended and produced in the brain by using the fMRI technique. Based on the above studies, the Korean word recognition model was proposed, in which words and Eojeols in mental lexicon are represented by the phonological syllable sequences for the full-list and decomposed morphemes. Also, in this model, the feature, letter, and body plus coda units are used as sub-lexical units. Thus, the proposed model consists of the perceptual processing stage, the sub-lexical processing stage, and the lexical stage.