Self-testing helped learners apply grammar in a new situation

In an artificial-language study, retrieval practice improved later memory and far transfer after one week, consistent with learning underlying rules.

Contents

In one sentence

Practising retrieval rather than simply restudying example sentences may help learners remember an unfamiliar grammar and apply its rules to new cases a week later.


What the researchers did

Opitz and Kubik studied whether the familiar benefit of self-testing can extend beyond memorizing simple lists. Participants learned an artificial language. One group repeatedly restudied grammatically correct example sentences; the other completed fill-in-the-blank tests after learning, which required them to retrieve missing material.

To test transfer, both groups judged whether new sentences were grammatical after a five-minute delay and again after one week. A week after learning, they also completed a final fill-in-the-blank memory test like the earlier one. The setup distinguished remembering trained material from applying a learned grammar to unfamiliar examples.


What they found

  • Retrieval practice produced a long-term memory benefit on the final fill-in-the-blank test.
  • It also produced a benefit on the far-transfer grammar-judgment test after the one-week delay.
  • The results are consistent with learners acquiring underlying grammatical rules rather than only memorizing examples.
  • The reported transfer benefit was delayed: the abstract specifically identifies the result after one week.
  • Self-testing may therefore help with more complex material than rote word lists.

What this means for learners and educators

  • After studying worked examples, close the notes and reconstruct a step, explain a rule, or complete a partial example from memory.
  • Include new examples in practice. A learner who can reproduce a familiar sentence may still not be able to use the underlying rule elsewhere.
  • Leave time before a check of application: this study examined both five minutes and a week, not only immediate performance.

Limitations and what we don't know yet

The study used an artificial language, which gives tight experimental control but is not the same as learning a natural language or a school grammar curriculum. The abstract does not state sample size, participant background, or effect sizes. The transfer test concerned grammaticality judgments, not conversation, writing, or long-term course achievement. The mechanism is supported by the pattern of results, but the study cannot show exactly what every participant represented mentally while learning.