How many words do you need to speak French?

French has a problem the other big European languages have in milder form: the language people write and the language people speak have drifted apart, and a single frequency list cannot represent both at once.

Updated 13 October 2026 · 8 min read

This post is about what our own French list — 20,000 entries, ranked — can and cannot tell you about that. It includes one thing we measured and could not use, which seemed worth reporting rather than quietly dropping.

Two languages in a trenchcoat

The written-spoken gap in French is not a matter of register or slang. It reaches into the grammar, and the examples are ones any French course teaches without quite spelling out the consequence.

• A whole tense exists only in writing. The passé simple — il alla, elle fut — is standard in narrative prose and effectively absent from conversation, where the passé composé does the job. Learn French from books and you acquire a tense you will never hear; learn it from speech and you cannot read a novel. Just how rarely it is said is counted in the passé simple is nearly dead on screen.
• The negative loses half of itself. Written French negates with ne … pas. Spoken French drops the ne routinely: je sais pas. A frequency list built from writing puts ne very high; one built from speech does not.
• The first person plural changes word. Written French says nous allons. Spoken French overwhelmingly says on va, which is grammatically third person singular. Two different entries, same meaning, different corpora.
• Question formation differs entirely. Inversion — as-tu — against the spoken tu as … ? with intonation, or est-ce que.

None of these is obscure. Together they mean that the thousand most common words of written French and of spoken French are meaningfully different lists, in a way that is much less true of Spanish.

What our list cannot tell you

Here is the honest limitation, and it is worth stating because most sites in this space do not state any.

Our French list does not distinguish spoken from written frequency. It is a single ranked list. It cannot tell you whether ne is ranked where it is because of books, and we are not going to reverse-engineer an answer out of it.

Quantifying that gap properly needs corpora that keep speech and writing apart, and for French those exist.

Lexique pairs 15 million words from written sources with more than 50 million from film subtitles, drawn from nearly ten thousand films and television series — subtitles being the most accessible large sample of language that approximates conversation.

Lonsdale and Le Bras's frequency dictionary of French goes further and splits its 23-million-word corpus exactly in half: 11.5 million words of speech — parliamentary debate, phone calls, face-to-face conversation, interviews, film dialogue — against 11.5 million of writing.

And going back further, Français fondamental, compiled in the early 1950s by Georges Gougenheim's team, was built deliberately from recorded and transcribed speech rather than from text, producing a core list of 1,475 words. It exists precisely because the written frequency lists of the day were a poor guide to what people said.

None of those is our list, which is the limitation this section is about. But the method is well established and the answer is known.

We also could not measure inflection in French. For Spanish and German we counted how many entries are inflected forms of other entries, using the notes our lists carry. The French list's notes cover only plurals and reach fewer than one entry in ten, so the same count produces a number near zero — which would say something about our annotation and nothing whatever about French.

That is a reporting decision rather than a finding. French inflects its verbs about as heavily as Spanish does; our data simply cannot show it, so we are not printing a figure that would imply otherwise.

What the divergence looks like in practice

Lonsdale and Le Bras publish the comparison directly, and the two columns read like different languages.

Over-represented in spoken French: merci, bonjour and bonsoir, truc (thingamajig), mec (guy), flic (cop), and the everyday profanity — merde, putain. Interactional, disposable, and nearly absent from the literary lists. How often those words are said, and how rarely verlan like meuf and keuf turns up beside them, is counted in French slang you will actually hear.

Over-represented in written French, specifically literature and non-fiction: silence, visage (face), sourire (smile), âme (soul), épaule (shoulder). Nouns for describing people and scenes, which conversation has little use for because the people are in front of you.

That is the divergence in one glance, and it is a sharper illustration than any percentage. A learner who studies only written French acquires the second column and will be unable to buy a coffee.

What our list can tell you

French words are the shortest of the three. Mean entry length by band, measured across all three lists:

And it has the fewest long ones. Entries of twelve characters or more make up 2.8% of the 2,001–5,000 band in French, against 5.5% in German and 7.2% in Spanish.

French orthography is doing some of this work — French spells a lot of sounds it no longer pronounces, but it does not pile on the derivational suffixes Spanish uses, and it does not compound the way German does. The practical effect for a learner is that French vocabulary arrives in smaller pieces.

Spoken French, measured

We cannot measure written French against spoken French from our own list, but we can measure one side of it properly. French film and television subtitles — 318 million words of the OpenSubtitles 2018 corpus — are the nearest large sample of speech there is, and they have counts.

Spoken French has the steepest frequency curve of any language we measured. Its first 100 word forms cover 56.7% of the dialogue, its first 1,000 cover 79.4%, and reaching 95% takes 13,909 forms, against 20,854 for Spanish. The comparison of all seven languages puts it at the top at every point.

That is the spoken half of the divergence above, measured. Conversation recycles a small set of pronouns and short words — je, pas, tu, ça, all among its twenty-five commonest — so hard that a thousand of them go a very long way. What subtitles cannot show is the written half. We would expect a written corpus to give French a flatter curve, though we have not measured it: the literary vocabulary in the second column above is exactly the kind of word that makes a tail long.

So what is the number?

Given the above, any single figure for French deserves a caveat attached, and it is this: ask which French.

• For conversation, the first thousand entries take you a long way, because spoken French leans hard on a small set of very frequent words and constructions. The obstacle is not vocabulary. It is that spoken French runs its words together and deletes sounds, so you will know the word and not hear it.
• For reading, you need more, and some of it is vocabulary you will never hear — the literary tenses, the written connectives, a formal register that speech has abandoned.
• For both, the honest target is higher than for Spanish, not because French has more words but because you are effectively acquiring two overlapping vocabularies.

The caveat that applies everywhere

Coverage is not comprehension. Knowing 95% of the words in a text does not mean understanding 95% of it, because the missing 5% carries the content while the words you know carry the grammar.

In French there is a second version of the same trap, specific to listening. You can know every word in a spoken sentence and still not segment it, because French does not put audible boundaries between words. Vocabulary size is not what is failing you there, and no amount of extra vocabulary will fix it.

The general argument is in how many words do you need to learn a language. The French course teaches from the list in frequency order, with recordings on everything, which is the part of the problem that actually responds to effort.

French's place among the languages English speakers find easiest, measured by time as well as by vocabulary, is in the easiest languages to learn.

The first 1,000 French words are also free to download, as a printable PDF and a CSV for Excel or Anki.

How that curve turns into weeks of study is in how long it takes to learn French, and the French words English already uses — and how common they really are in French — are in French words used in English.

How many French words do you already know? The two-minute French vocabulary test samples the same list band by band, with made-up words mixed in to catch guessing, and estimates your vocabulary size.

How big French itself is — from the Académie's 53,000 words to 500,000 in the Dictionnaire des francophones — is set out in how many words are there in French.

What those counts mean for watching a film, and how far French subtitles are from French speech, is in how many words you need to watch French films.

The hundred commonest word forms of all, each with its meaning, are in the 100 most common French words.

Sources

• Lonsdale, D. & Le Bras, Y. A Frequency Dictionary of French, Routledge — a 23-million-word corpus, half spoken and half written, with register-divergence tables.
• New, B., Brysbaert, M., Veronis, J. & Pallier, C. — the Lexique subtitle corpus, described in Brysbaert & New (2009), Moving beyond Kučera and Francis. springer.com
• Gougenheim, G. et al. — Français fondamental, CREDIF, early 1950s, compiled from recorded speech. wikipedia.org
• Nation, I. S. P. (2006). How large a vocabulary is needed for reading and listening? lextutor.ca
• Lison, P. & Tiedemann, J. (2016). OpenSubtitles2016: Extracting Large Parallel Corpora from Movie and TV Subtitles. Proceedings of LREC 2016. aclanthology.org
• Dave, H. FrequencyWords: frequency lists from OpenSubtitles 2018, content licensed CC BY-SA 4.0. github.com. The coverage figures are computed by scripts/frequency_curves.py and printed from its output rather than typed.
• The measurements of our own list in this post are reproducible from the published list.