The effects of NMT as a de facto dictionary on vocabulary learning: a comparison of three look-up conditions

Many language learners have reported using Neural machine translation (NMT) such as Google Translate as a reference tool for writing production or as a dictionary substitute to facilitate reading comprehension. Although increasing evidence has suggested that editing with NMT can help learners make improvement on the vocabulary level, no study has explored how consulting NMT for unknown vocabulary may impact the retention of word meaning compared to more conventional look-up conditions. Using a 3 × 3 quasi-experimental design, this study investigated the extent to which the use of NMT to learn unknown target words during reading comprehension activities yielded vocabulary retention compared to that of an online dictionary and how the findings varied by learners’ proficiency levels. The research questions focused on the comparison of the outcomes of three different look-up conditions: (1) NMT word search, (2) NMT sentence search, and (3) online dictionary search. ANOVA results showed that learners who performed online dictionary search and NMT sentence search demonstrated significantly higher levels of word retention than those who carried out NMT single-word search, and the same pattern was observed regardless of their proficiency levels.