The testing effect: why active recall beats rereading three times over

When we learn words, rereading them feels like the natural thing to do: look them up again, flip through ready-made cards with the translation on the back, run a highlighter over a phrase. Memory research shows this is the most widespread cognitive illusion in learning. Knowledge is not built while reading — it is built while trying to pull the information back out of your head.

The landmark experiment: reading versus testing

In 2006 Professors Henry Roediger and Jeffrey Karpicke of Washington University in St. Louis ran an experiment that upended teaching practice. Students were split into two groups:

• The reading group (study–study): reread the text four times.
• The recall group (study–test): read the text once, then in the remaining rounds tried to write down everything they could remember, with nothing to look at.

Straight after the session — five minutes later — the first group felt more confident and scored slightly higher. The fluency illusion made them believe they had learned it.

When the researchers tested both groups a week later, the result was dramatic:

• The students who had simply reread the text had forgotten more than 50% of it.
• The students who had actively retrieved it held on to more than 80%.

"The act of retrieving information from memory changes memory itself: the route to the knowledge becomes durable and resistant to forgetting. Testing is not a way to measure learning — it is the most powerful way to produce it." Roediger & Karpicke, Perspectives on Psychological Science (2006)

Why passive recognition is a trap

When you look at a card and see the answer on the back, or pick one of four buttons on a screen, the brain is using recognition. It is a very cheap neural operation: "ah yes, I have seen this word somewhere."

A live conversation has no buttons with hints. You have to perform retrieval on the spot: form a thought, find the word in long-term memory, attach the right ending and say it. Train recognition alone and real speech is guaranteed to trip you up.

How retrieval practice is built into a Lexy lesson

Every Lexy lesson deliberately trains retrieval without props:

1. The question back to Lexy: Lexy asks, you answer, and then you have to ask a question in return ("And what's your name? And where do you live?"). There is no prepared text in front of you — the shape of the question has to come out of memory.
2. Retrieval before the hint: the on-screen prompt is hidden by default. You see only the meaning of the phrase and try to produce it in the target language yourself. The hint opens only when you are genuinely stuck.
3. Synthesis in role-play scenarios: at A2–B2 you enter a scenario — ordering coffee, airport check-in, a hotel — where the dialogue is not linear and lines have to be built on the fly.