AbstractThe use of the mnemonic keyword method involves (a) the formation of an orthographic link between the form of a foreign word and that of a physically similar word in one's mother tongue (i.e. the keyword), and (b) the formation of a semantic link between the referent of the foreign word and that of the keyword. In a variation of this study method, English‐speaking learners were asked to study the grammatical gender of German nouns by substituting the concepts of a woman, a man, and an inanimate object for the feminine, masculine, and neuter genders, respectively. This more concrete representation of gender was then to be incorporated to an interactive image along with the referent of the keyword and that of the German noun. The three experiments reported in this paper yielded four main results: (a) instructing learners to recode the gender tag and to include its representation into an interactive mental image along with the referent of the German noun facilitated the recall of genders: (b) the facilitating effect of imagery instructions on the recall of genders was obtained only if the English translation was recalled; (c) the probability of correctly reporting the gender given that the English translation was not recalled was higher than the baseline for guessing if the learners were instructed to encode the gender before the translation, otherwise it was consistently near the baseline: and (d) in contrast with prior studies with French‐speaking subjects, the modified keyword method did not always enhance the recall of the familiar translation equivalents. The discussion bears mainly on the factors that might account for the difficulty of detecting a consistent keyword advantage in the recall of the familiar translat