The Retrieval Experiment: What Immediate Feedback Changed
The focused evidence on feedback timing comes from a laboratory experiment with Saudi undergraduate English-as-a-foreign-language learners. It asks a narrower question than the broad meta-analysis: when the learning material and short-answer retrieval task are held comparatively stable, does showing the correct answer after each response produce better retention than showing the answers after the whole test? Reconstructing the study makes its result useful without making it larger than its design.
A narrow comparison designed to separate effects
The study began with 190 volunteers from the English department of Umm Al-Qura University in Saudi Arabia. After 13 participants failed to attend the one-week and one-month delayed tests, the analysis included 177 learners. They were randomly assigned to four conditions: repeated studying, repeated testing without feedback, repeated testing with immediate feedback, and repeated testing with delayed feedback. The allocation was not designed to compare a fast digital system with a slow human teacher. It was designed to separate repeated exposure to material, retrieval practice by itself, and retrieval practice accompanied by two different feedback schedules.
The four conditions create a set of useful counterfactuals. Repeated studying shows what happens when learners encounter the passage and distractor work without having to produce answers. Repeated testing without feedback keeps the act of retrieval but removes answer information. Comparing those two conditions asks whether producing a response contributes anything beyond another exposure to the material. Comparing either testing condition with a feedback condition asks whether the correction or confirmation adds to retrieval practice. Finally, comparing immediate and delayed feedback asks whether the point at which the answer is shown changes the result. These comparisons do not isolate every psychological process, but they make the main alternatives visible.
The learning material was a 487-word prose passage selected for the participants’ level. A 15-item short-answer test assessed recall of important details and facts from the passage. Four mathematical distractor tasks were placed between learning and testing activities. Those tasks were not a second lesson; they created a separation between the passage and the answer so that performance would be less dependent on information still held in working memory. The outcome was scored out of 30: a fully correct response received two points, a partially correct response one point, and an incorrect or blank response zero.
The four conditions made different demands. Learners in the repeated-studying condition read the passage and completed the distractor tasks repeatedly, without retrieval tests. Learners in the no-feedback testing condition first read the passage and then completed three 15-item short-answer test trials, with a distractor task between trials and no answer information afterward. This condition measured what repeated retrieval could contribute without correction.
The two feedback conditions used the same repeated-testing structure. In the immediate-feedback condition, the correct answer appeared for ten seconds after each question. In the delayed-feedback condition, learners completed the test trial first and then saw the correct answers at the end, for 150 seconds in total—ten seconds per answer.