The problem
You are at the bakery. You know « pain », « croissant », « deux », « s’il vous plaît ». You even revised the numbers last night. And yet, when the woman behind the counter looks at you, nothing joins up. The words are all there, each in its own box, but the sentence never arrives. You point, you smile, you leave.
This is not a vocabulary problem. You have been learning bricks, and never walls.
What the research found
Linguists call them “formulaic sequences” or, more plainly, chunks: groups of words a native speaker produces in one go, such as « je voudrais », « ça fait combien », « il y a », « c’est pas grave ». A large share of everyday speech is thought to be made of these chunks, reused as they are, rather than sentences built word by word at the moment of speaking.
Frank Boers and his colleagues (Boers et al., 2006) wanted to know whether learning chunks changed anything when people actually spoke. They followed 32 students of English in Belgium over 22 hours of class. Half worked with authentic materials in which the teacher constantly drew attention to ready-made word combinations; the other half followed a conventional, grammar-focused course. At the end, examiners who did not know who had followed which course interviewed everyone. The chunk group was rated as more proficient speakers, and the more chunks a student used, the higher their oral rating.
Why does it work? Because a chunk pulled out whole costs your brain almost nothing. It frees attention for everything else: listening, choosing the next word, breathing. Merrill Swain (Swain, 1995) showed from another angle that speaking is not only the result of learning but one of its engines: it is when you try to produce a sentence that you notice what is missing. Chunks give you something to produce right away, even with three words of vocabulary.
Do this today
Pick one situation from tomorrow: ordering a coffee, introducing yourself, asking a price. Write down three whole chunks, not three words: « Je voudrais un café, s’il vous plaît. » « Ça fait combien ? » « Je m’appelle Léa et j’habite à Lyon. » Say each one out loud five times, in full, without stopping in the middle. The goal is not to understand every word; it is for the chunk to come out in one go, like a single long syllable.
How our lessons use it
In Je m’appelle Nadia (A1), the story is written in chunks you can reuse as they are: « je m’appelle », « j’habite à », « le samedi, j’aime ». The last step, “Dis-le à ta façon”, asks exactly that: take those chunks and say three sentences about yourself, out loud. Not a list of words, three whole sentences that come out in one piece. If you would rather start with your ears, the shadowing version of « L’ascenseur » on our YouTube channel does the same thing through listening: one sentence, a pause, you repeat the whole chunk.
The honest limit
The Boers study involved 32 advanced students of English, not beginners in French, and it is a single study, not a meta-analysis. It measured an impression of fluency judged by examiners, not grammatical accuracy. And nobody has shown that learning chunks is enough on its own: at some point you will want to change them, and that is where grammar takes over. Chunks get you started; they do not take you all the way.
Your questions
What exactly is a chunk?
A group of words that always travel together, like « je voudrais », « ça fait combien ? » or « il y a ». You learn it in one piece, without breaking it down into grammar.
How many chunks do I need?
Few, but well. Five chunks you can say without thinking beat fifty you only recognise on the page. Add one or two a week.
Does this replace grammar?
No, it comes before it. Once « je voudrais un café » comes out on its own, the grammar of the conditional becomes an explanation of something you can already say.
References
Boers, F., Eyckmans, J., Kappel, J., Stengers, H., & Demecheleer, M. (2006). Formulaic sequences and perceived oral proficiency: Putting a Lexical Approach to the test. Language Teaching Research, 10(3), 245–261. doi:10.1191/1362168806lr195oa
Swain, M. (1995). Three functions of output in second language learning. In G. Cook & B. Seidlhofer (Eds.), Principle and practice in applied linguistics: Studies in honour of H. G. Widdowson (pp. 125–144). Oxford University Press.
You may also like
LE LABO · Nº 12
My pronunciation is a mess — 45 seconds of shadowing a day, then we'll talk ↗
LE LABO · Nº 13