This study explores cross-linguistic differences in online sentence production, working on the theory that the grammatical characteristics of a language customize the speakers’ usual manner of sentence construction and utterance. It is reasoned that a contrast between English and Korean syntax has a direct bearing on that process: The sentential subject is licensed by a tensed verb in English, while it is by an overt morpheme in Korean. This gives rise to a substantial difference in the speakers’ speech patterns: English speakers’ message formation and linguistic encoding center on a verb, while Korean speakers elect or utter a subject first and then draw a predicate. An experiment is conducted on the hypotheses that (a) English L1 speakers are disposed to make sentences based on verbal information, so they would choose a subject that is conceptually more intimate with a given verb; (b) In contrast, Korean L1 speakers’ sentence construction is topic-oriented, so they would first consider the contextual properties of each referent such as shared familiarity and previous topicality; (c) Korean speakers’ L1 patterns would be persistent in their L2 English processing. The results and some implications for English teaching are discussed.