Using a lexical decision task, Wolter and Yamashita (2015) showed that collocations that exist only in L1 but not in L2 were not processed faster than collocations that only exist in L2 but not in L1 or a random combination of two words. This result seems to support the age/order of acquisition effects (Carroll & White, 1973) over Jiang's (2000) model of lexical acquisition for the processing advantage of congruent collocations. To replicate and extend Wolter and Yamashita (2015), the present study utilized both on-line (lexical decision task) and off-line (meaning recall test) measures. In line with Wolter and Yamashita (2015), results from the lexical decision task showed that there was no L1 activation effect by L2 (L1 Korean) learners, regardless of their L2 proficiency levels. The meaning recall test result, on the other hand, showed that L2 learners showed more tendency to rely on their L1, as suggested by Jiang's (2000) model. The discrepancy between on-line and off-line performances can imply that the findings of Wolter and Yamashita (2015) should be reexamined using different methods and techniques.