Word translation at three levels of proficiency in a second language: the ubiquitous involvement of conceptual memory

Three groups of 20 unbalanced bilinguals, different from one another in second language (L2) fluency, translated one set of words from L1, Dutch, to L2, English (forward translation), and a second set of matched words from L2 to L1 (backward translation). In both language sets we orthogonally manipulated 3 word characteristics: word imageability, word frequency, and cognate status. Similar data patterns were obtained for the 3 proficiency groups, suggesting that word translation was qualitatively the same in all 3. Especially, word imageability affected performance similarly across the 3 proficiency levels. This suggests that our 3 groups did not differ in the degree to which they involved conceptual memory in the translation process. Further, contrary to the "asymmetry" model of word translation, conceptual memory appeared to operate as much in backward translation as in forward translation. The combined data indicate that, at least beyond the initial stages of L2 fluency, "concept mediation" is a universal process in translating words between 2 languages. We can account for the few, generally small differences between the fluency groups in terms of the idea that some subsets of words lag in L2 knowledge development.