Monday: the class knows all twenty words. Friday: half have vanished. It is not that students did not learn them. They stopped retrieving them.
The Swahili experiment
In a famous 2008 study published in Science, psychologists Jeffrey Karpicke and Henry Roediger asked students to learn 40 Swahili-English word pairs, such as mashua and boat. Some kept being tested on every word, even after getting it right. Others dropped words from testing once recalled.
A week later, the difference was huge. Students who kept practising retrieval remembered about 80% of the pairs. Those whose words were dropped from testing recalled just 36% and 33%. Extra studying of words they already "knew" made little difference. Being tested was what made memories stick.
The four-minute fix
You do not need a new scheme of work. Start each lesson with a four-minute retrieval quiz:
Mix old and new. Include words from last week and last term, not just today.
Keep it low stakes. No marks recorded, just self-checking.
English to target language. Producing a word is harder, and more useful, than recognising it.
Keep testing "known" words. That is the whole point of the study.
Ofsted's curriculum review for languages also stresses revisiting vocabulary and grammar in different contexts. A few minutes of recall at the start of every lesson is the cheapest, strongest tool in the cupboard, and it costs nothing but a mini whiteboard. It is also why every MELT unit ends with a retrieval page.
