How many words do you need to watch French films?

Fewer than for Spanish, Italian, Portuguese or German. In 318 million words of French film and television subtitles, 1,076 word forms cover 80% of everything said, 4,876 cover 90% and 13,909 cover 95%. Spanish needs 7,577 forms for the same 90%. The catch is that those figures describe French subtitles, and French subtitles are tidier than French speech. A learner with enough words can read a French film long before they can hear one.

Updated 12 October 2026 · 6 min read

The count

These are word forms, counted in the OpenSubtitles collection of film and television subtitles: est, sont and était are three rows. French needs roughly two thirds as many as Spanish or Italian, and of the languages we measured only English, which barely changes its words at all, needs fewer.

What each level is like in practice: English-language films run at about 79 words of dialogue a minute, by the count in the study cited below. So knowing 90% of the words means about 7.9 unknown words a minute, 95% means about 3.9, and 98% means fewer than two.

What the research says is enough

The standard studies are of English. Webb and Rodgers analysed the scripts of 318 films and found that the 3,000 most frequent word families, plus proper nouns, covered 95.76% of the words, and 6,000 families covered 98.15%. A word family is a base word with all its forms, so être with est, sont, était and the rest is one item, not dozens.

That is why their numbers are smaller than the table's. For French, read the two together: the form counts tell you how the first few thousand words pay off, and the family counts are the better guide to the top end. The top end of a raw word list is inflated in any case. Of the 834,768 different forms in the French list, 47% occur exactly once, and most of those are names, foreign words and typing mistakes. That is why 98% appears to need 2.9 times as many forms as 95%.

Why French needs fewer forms

French dialogue leans on a very small set of very common words. Its first 64 forms are half of everything said, and the comparison of all the languages we measured puts French first at every point; that is the subject of which language has the steepest frequency curve.

So the vocabulary is the kind part of French. The hard part is the next section.

What you read is not what you hear

Subtitles are written French. Three counts show how far that is from the French in the actor's mouth.

One written word in 15 is a clipped one. French shortens je, le and la, ce, de, que, ne, te, me and se before a vowel and joins them to the next word: j'ai, l'ami, c'est, qu'il. Those nine clipped forms alone are 6.8% of all the words in French subtitles. On the page the apostrophe shows you where one word ends. In speech nothing does: je l'ai is two syllables with three words in them.

The subtitles keep the ne that speech drops. Written French makes a negative with two words, ne … pas. In everyday speech the ne is routinely left out: je sais pas for je ne sais pas. In the subtitles, ne and its clipped form n' appear 97 times for every 100 pas. Ne also goes with jamais, rien and plus, so that is not every negative, but it is the great majority of them: the subtitlers write the ne far more often than actors say it. If you read French subtitles while listening, you will often read a word that was never spoken.

On beats nous. For "we", spoken French mostly uses on, with the verb in the he-and-she form: on va, not nous allons. In the subtitles on is said 8,661 times per million words and nous 4,727, which is 1.8 to one, and nous has a head start because it also means "us". Even tidied-up French has mostly gone over to on.

There is a fourth gap that a word count cannot show at all: liaison. Les amis is said as one run of sound with a z in the middle that belongs to neither word on the page. Knowing a word in writing is no guarantee of catching it in a film, and in French less so than in Spanish or Italian.

The casual layer

Film French also uses a layer of informal vocabulary that textbooks hold back: ouais for oui (758.6 per million words against 3,147), truc for "thing", mec for "guy", boulot for "job". They are few and very common, so they are quick to learn; they are counted in French slang you will actually hear. Going the other way, the literary past tense that fills French novels has almost vanished from the screen, as measured in the passé simple, nearly dead on screen.

So how many words?

• To start, with French subtitles on: about 2,000 words. The first 1,076 forms are 80% of the dialogue and 3,000 are 87.1%.
• To follow most scenes: about 3,000 word families, the research figure for 95%. In forms that is somewhere between the 4,876 that cover 90% and the 13,909 that cover 95%.
• To relax without subtitles: 6,000 families or more, and a trained ear. The second part takes longer than the first.

Because the ear is the bottleneck, the order of work matters more for French than for most languages:

1. Learn words with their sound from the first day, not from the page.
2. Watch with French subtitles, and notice where the subtitle and the voice differ. The missing ne is the easiest one to catch.
3. Replay short scenes without subtitles once you know every word in them. That is practice at hearing, which no word list gives you.

Coverage is not comprehension in any language; the general argument is in how many words you need to learn a language.

Where to start

The 1,000 most common French words are free and in frequency order, and the two-minute French vocabulary test estimates how many you already know. The French course, Verbavia (our own course), teaches from that list with recordings, free on the web. Why written and spoken French have drifted apart is in how many words you need to speak French; whether the rest of the language is as kind as its vocabulary is in is French hard to learn; and the same question for Spanish, where the difficulty is the verbs and not the ear, is in how many words you need to watch Netflix in Spanish.

Sources

• Webb, S. & Rodgers, M. P. H. (2009). The Lexical Coverage of Movies. Applied Linguistics, 30(3), 407–427. doi.org. The words-a-minute figure is the films' 2,841,887 running words divided by their running time of 602 hours, both as the paper reports them.
• 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. Every count is computed by scripts/frequency_screen.py and printed from its output rather than typed. The clipped forms counted are c', l', j', d', qu', n', t', m' and s', each a separate entry in the French list.