Humans represent word knowledge in different modalities, including spoken and written forms. Language processing, such as reading comprehension, benefits from the synchronous retrieval of both forms. However, it is unknown how spoken and written words are best acquired and linked for subsequent synchronous retrieval. In the present study, we examine whether presentational timing affects word-form acquisition and linking. We test whether synchronous or asynchronous presentation of novel written and spoken forms leads to better retention. In two exploratory (Experiment 1 = 30, Experiment 2 = 48) and one pre-registered (Experiment 3 = 142) three-session experiments, participants studied Chinese words. During study, written (Pinyin) and spoken forms were presented simultaneously or asynchronously, with spoken preceding written forms and vice versa. In both exploratory experiments, we observed a recall advantage for spoken-first presentation at test, which required generating written forms when cued with the corresponding spoken forms. Experiment 3 assessed the robustness of these effects across tasks: the recall task used before and two recognition tasks (forced choice and match/mismatch). Replicating the earlier results, spoken-first learning enhanced later recall. However, recognition was unaffected by presentational timing. This dissociation suggests that only tasks with higher cognitive demands – such as recall – are sensitive to differences in presentational timing, whereas recognition tasks are not. The recall advantage for spoken-first aligns with prediction-based accounts of learning and highlights the importance of timing in linking novel word forms. These findings have implications for models of word learning and for instructional approaches in (foreign) language acquisition and rehabilitation.