Why Your Child Forgets Chinese Words by Friday

Every parent of a primary-school child in Singapore has seen it. Twenty words learnt for a dictation on Thursday, scored full marks, and by the following week half of them are gone. The instinct is to blame the child's memory or their attitude. The research says to blame the method.

Forgetting is normal, and fast

Hermann Ebbinghaus measured this in the 1880s, memorising lists of nonsense syllables and testing himself at intervals. What he found has been replicated ever since: most of what is learnt in one sitting is gone within days, with the steepest drop in the first day or two. The curve flattens only if the material is met again. A child who forgets by Friday is not failing; they are behaving exactly as the curve predicts.

Why re-reading feels like it works and does not

The standard way to revise a word list is to read it again. It feels effective because the words look familiar on the second read, and familiarity feels like knowing. It is not. In a widely cited 2013 review, Dunlosky and colleagues rated re-reading and highlighting among the least effective study techniques, precisely because they create that feeling without the memory behind it.

What works is the opposite of re-reading: closing the book and trying to produce the word. Roediger and Karpicke showed this in 2006 with a simple experiment. Students who studied a passage and then tested themselves on it remembered far more a week later than students who studied it twice — even though the second group felt more confident at the time. Pulling a word out of memory strengthens it; looking at it again does not.

The second half: spacing

The other robust finding is that when the practice happens matters as much as how. Cepeda and colleagues pooled hundreds of studies in 2006 and found that the same amount of practice, spread across several days, produces much better retention than the same practice done in one session. Spreading the sessions out is what matters; letting the gaps grow — a day, then a few days, then a few weeks — is the usual way to do it.

Put the two together and you have the whole method: test rather than re-read, and test again at increasing intervals. Nothing about it is new. It is simply not how most families revise, because cramming the night before a dictation produces a good score on the dictation, and the score hides the forgetting that follows.

A ten-minute routine that follows the research

1. Day 1: test, do not show. Say the word; the child writes it. Mark. Words the child got wrong go on a short list — this list is the whole point.
2. Day 3: test the short list again, plus five words from the full list chosen at random. Anything missed stays on the list; anything got right twice comes off.
3. Day 7: same again. The list should be shrinking. If it is not, the words on it need meaning work — a sentence, a picture, the radical — not more copying.
4. Day 21: a last pass on whatever is left. By now the list is usually three or four words, and those are the ones that genuinely needed the extra time.

Ten minutes a day, four sessions over three weeks, for one dictation list. The day-1 score will be lower than the child is used to, because the test comes before the cramming. The score three weeks later will be higher than it has ever been.

Three things that quietly undo it

• Copying the word ten times. After the third copy the hand is working and the mind is not. One test beats ten copies.
• Testing only the words the child knows. Confidence is pleasant and it does not build memory. The short list of misses is where the learning is.
• Skipping day 3 because the child seemed to know them on day 1. The first return is the one that matters most; it catches the words before the curve takes them.

This is the schedule AceLah runs automatically: every wrong answer comes back as a fresh question on the same point on day 3, day 7 and day 21. But the method does not need software. It needs a pencil, a list of misses, and a parent who tests instead of shows.