This study reports on the findings of error analysis of a learner corpus consisting of 238 English argumentative essays written by Korean college students, specifically on errors that stem from confusion of lexical categories. A three-step error analysis was conducted following a procedure established by Rod Ellis in 1995. The results revealed 10 commonly confused lexical category dyads—coordinators or subordinators mistaken for adverbs, adverbs for conjunctions, gerunds for nouns, the adverb almost for either an adjective or a noun, the pronoun each other for an adverb, adverbs for nouns, conjunctions for prepositions, prepositions for conjunctions, the preposition to for an infinitive, and the adverbial for example for a preposition. Discussions of the causes of such confusion—homogeneous L1 translation, failure to correctly punctuate spoken language in writing, lack of L2 grammar knowledge, negative transfer of L1 morphosyntactic properties, and lack of L2 lexical knowledge—are delineated with specific examples. The limitations of the study and pedagogical implications are addressed.