The problem
Left column, the French word. Right column, its translation. Twenty lines, ten minutes, and the feeling of having worked well. But in conversation the word doesn’t come, or it comes in a sentence where it doesn’t fit. You know the word’s translation; you don’t know the word.
A translation is a label. What you need is a handle.
What the research found
In 1978, Slamecka and Graf showed something simple: we remember a word we produced ourselves (from a cue) better than a word we merely read. That’s the generation effect. Thirty years later, Bertsch and colleagues (2007) pooled 86 studies of it: the effect is reliable, and of medium size — d = 0.40. It isn’t a trick; it’s one of the most replicated regularities in memory.
For a foreign language, Laufer and Hulstijn (2001) proposed a way to predict how much of a word will stay: its involvement load. Three ingredients: need (do I need this word?), search (did I have to go and find it?), and evaluation (did I have to decide whether it fits in my sentence?). Copying a translation: low need, no search, no evaluation. Writing your own sentence with the word: all three at once. In their experiments, high-involvement tasks — writing a composition using the words — produced better retention than reading the word with its gloss.
In other words, what you experience as “slower” is exactly what makes the difference: to decide where the word goes, you have to think about what it means, what goes with it, what wouldn’t fit. Each of those decisions is one more handle.
Do this today
Open your list. For each word, write a true sentence about yourself that contains it: not “the baker sells bread” but “I buy my bread at the baker on my street, on Saturdays.” Say it out loud. If you’re unsure, keep the sentence and check later — the uncertainty is part of the work.
Tomorrow, cover the list and try to recover the word from your gapped sentence: “I buy my bread at the ___.”
How our lessons use it
The last step of “Je m’appelle Nadia” is called “Dis-le à ta façon”: three sentences of your own — “Je m’appelle… J’habite à… J’aime…” — to say out loud, change, and start again. You translate nothing; you produce, from a model, something that exists only for you. Our lessons have no translation column: meaning comes from the story, memory comes from your sentence.
Try it in Je m’appelle Nadia (A1): the last step is made for this.
The honest limit
The generation effect is solid, but weaker when the word is completely unknown and the cue too thin: producing anything at all doesn’t help. Hence the gapped template, which gives enough structure for your sentence to be possible. And an occasional translation isn’t a sin: it’s for understanding quickly. It isn’t for remembering.
Your questions
Is my sentence still useful if it has a mistake in it?
Yes. The effort of producing is what makes the word stay; the mistake gets fixed when you check, and a word corrected inside your own sentence is remembered better than a word copied out without a mistake.
One sentence per word is too long for twenty words.
Then do ten words with a sentence each rather than twenty without. Ten words kept are worth more than twenty recognised.
Are flashcard apps just translation?
Often, yes — French on the front, English on the back. They become far better if the back holds your sentence rather than the translation.
References
Bertsch, S., Pesta, B. J., Wiscott, R., & McDaniel, M. A. (2007). The generation effect: A meta-analytic review. Memory & Cognition, 35(2), 201–210. doi:10.3758/BF03193441
Laufer, B., & Hulstijn, J. (2001). Incidental vocabulary acquisition in a second language: The construct of task-induced involvement. Applied Linguistics, 22(1), 1–26. doi:10.1093/applin/22.1.1
Slamecka, N. J., & Graf, P. (1978). The generation effect: Delineation of a phenomenon. Journal of Experimental Psychology: Human Learning and Memory, 4(6), 592–604. doi:10.1037/0278-7393.4.6.592
You may also like
LE LABO · Nº 1
I forget everything by Friday — review on day 3, 14 and 45, not every night ↗
LE LABO · Nº 2