Processing backward translation at intermediate L2 proficiency

Abstract

Abstract This study opens a window into the mapping of the lexical structure of bilinguals as they reach intermediate-level L2 proficiency. A group of 32 Persian-speaking English learners participated in a translation recognition task, where they had to decide whether the sequentially presented English-Persian word pairs were translation equivalents. The Persian words in the experimental trials were not translation equivalents, but they either shared lexical or semantic features with the correct translations or shared phonological features with the preceding English words. For example, the Persian word /bɒːzuː/ in the word pair arm – /bɒːzuː/ was manipulated to create three categories of ‘no’ trials: (a) orthographically related to the translation equivalent (arm – /bɒːziː/ ‘game’), (b) semantically related to the translation equivalent (arm – /ʃɒːne/ ‘shoulder’), (c) phonologically related to the English word (arm – /ɒːrmɒːn/ ‘goal’). A significant interference effect was observed in response to the items in the first two categories, which indicates extensive utilization of both lexical and conceptual links in backward translation. However, the observed interference effect in the third category did not reach statistical significance, which calls into question the integrated activation of L1 and L2 prior to high level L2 proficiency. These results are discussed within the context of the Revised Hierarchical Model (Kroll & Stewart, 1994) and the Bilingual Interactive Activation Plus model (Dijkstra & Van Heuven, 2002).