A significant, but rarely posed, question in EFL writing is what new language the writers assimilate from the process. In this study, a group of intermediate EFL students in a state secondary school in Spain completed an L1(Spanish)-to-FL(English) written translation task on their own, with bilingual dictionary/glossary support. Three days later, this group received feedback on the vocabulary errors in their initial translations, consisting of giving the initial letters and dashes for the remaining letters of the correct word. They had to rewrite the initial translation with corrections, using the glossary support. A week after that, they were unexpectedly asked to translate the same text again without dictionary support or the previous translations, in a delayed post-test. The group looked up 158 items initially of which 148 or 94% were correct. Of these 148 items 106 or 72% were correctly recalled in the delayed post-test. Of the 136 items that were initially incorrect, 71 or 52% were also recalled. In total the students made mean gains of 13.6 vocabulary items. Not only is this result encouraging, but the analysis shows that a focus purely on errors is misleading as it does not reveal language gains. (Contains 5 tables.)