Semantic processing of Korean word and eojeol using a semantic priming task

This study examined whether Korean eojeols and words have similar semantic processing patterns. To this end, a semantic priming task with a 200ms SOA was used. Semantic relatedness (related vs. unrelated) was manipulated based on Word2Vec-derived semantic similarity, and prime type (word vs. eojeol) was also manipulated. Experiment 1 used accusative case markers to create relatively natural priming contexts, whereas Experiment 2 employed a wider range of case markers to construct less natural contexts. Both experiments showed reliable semantic priming effects, with faster responses for the related than the unrelated conditions across word and eojeol primes. Notably, no interaction between the relatedness and the prime type was found for either response times or accuracy in either experiment. These findings show that Word2Vec-based similarity elicits semantic priming and that Korean words and eojeols undergo the same semantic processing.