The testing effect is the finding that retrieval practice can enhance recall on future tests. One unanswered question is whether first-test response mode (writing or speaking the answer) affects final-test performance (and whether finaltest response mode itself matters). An additional unsettled issue is whether written and oral recall lead to differences in the amount recalled. In three experiments, we examined these issues: whether subjects can recall more via typing or speaking; whether typing or speaking answers on a first test can lead to better final-test performance (and whether an interaction occurs with final-test response mode) and whether any form of overt response leads to better final-test performance as compared to covert retrieval (thinking of the answer but not producing it). Subjects studied paired associates; took a first test by typing, speaking, or thinking about responses; and then took a second test in which the answers were either spoken or typed. The results revealed few differences between typing and speaking during recall, and no difference in the size of the testing effect on the second test. Furthermore, an initial covert retrieval yielded roughly the same benefit to future test performance as did overt retrieval. Thus, the testing effect was quite robust across these manipulations. The practical implication for learning is that covert retrieval provides as much benefit to later retention as does overt retrieval and that both can be effective study strategies.Keywords Testing effect . Response mode . Retrieval practice Thousands of memory experiments have been conducted using recall as the criterion variable, whether in cued recall, serial recall, or free recall. One factor that varies throughout the literature is response mode-whether subjects write or speak their responses. In the first studies of memory, only spoken responses were used (Ebbinghaus, 1885/1964), but we suspect that the majority of recall studies today use typed or written recall, for ease of scoring.This article addresses two issues of response mode: how it may affect the amount recalled, and how it may affect later retention. Regarding the first-does response mode (speaking or writing) affect the amount recalled?-we can find few researchers who have seriously considered the question. Many studies have contrasted visual and auditory input, but few have investigated written versus spoken response mode. We suspect that most researchers assume that response mode does not matter, and thus use whatever procedure is handy for their purposes. As the bulk of the contemporary literature on long-term retrieval has predominantly used written or typed responding, it is important to address whether response mode can influence recall.The second, and perhaps more interesting, issue is whether response mode on a first test (speaking or typing) affects the benefit of that test on a second test. That is, if subjects study material and are then asked to speak their responses or to type them on a first test, will this difference...