This study investigates realtime sentence production by speakers of two typologically different languages, English and Korean, in line with the assumption that the mechanisms underlying sentence production are contingent on the morphosyntax of an individual language. It is reasoned that the canonical word order of a language tailors the speakers' process of sentence productionhow they plan and construct sentential content incrementally, and that this in turn seems to influence their nonnative language learning and processing. In particular, because subjects are denoted by reference to tense in English while they are denoted by overt case markers in Korean, it is hypothesized that English speakers' message formation and linguistic encoding center on verbs, but Korean speakers elect a subject first and then draw a predicate accordingly. An online experiment was conducted with three groups: 25 English natives, 28 Korean natives and 25 Korean L1-English L2 learners, the results of which show that English L1 speakers are disposed to construct sentences based on verb information, but Korean L1 speakers on noun information; and the L1 patterns are persistent in L2 processing as well.