Qui, Que, Dont, Où: French Relative Pronouns

Most guides present the French relative pronouns as five rules of roughly equal weight. They are not. Qui, que and où sit inside the hundred most frequent words of French in both speech and writing, so you acquire them whether you study them or not. Dont ranks 91st in books and 402nd in speech - and that single fact explains most of what learners find hard about it.

We measured that on Lexique 3.83, a public research corpus recording how often each French word appears in film subtitles and in books. What comes out is not a grammar problem. It is an input problem.

How common is each relative pronoun, really?

Here is the whole set, ranked in each register. Frequencies are per million words.

Three tiers fall out of this immediately. Qui, que and où are core function words in both registers - unavoidable, and learned early for that reason. Dont is a written word: four and a half times rarer in speech, relatively speaking, than on the page. And lequel and its contracted forms are effectively literary.

Why does dont feel so much harder than the others?

Because you have barely heard it. Take the three hundred most frequent words in written French and rank them by how far they fall when you switch to speech:

Only five words in that entire band decouple more sharply, and three of them - fit, eut, fut - are passé simple forms, a tense nobody speaks. Dont is a grammatical word that behaves statistically like a literary tense.

So the learner who builds French from conversation, video and app dialogue has met qui thousands of times and dont a few dozen. No amount of rule explanation compensates for that ratio; volume of reading does, which is the argument set out in how to learn French by reading. The mirror image of this problem is y and en, which look like top-twenty words and turn out, when you count what they are actually doing, to give you far less practice than their ranking promises. It is also why written French demands so much more vocabulary than speech - about 77% more for the same coverage.

What does each pronoun actually replace?

Every relative pronoun in French stands in for something specific in the sentence it collapses. Learn the slot, not the translation.

The reason English speakers hesitate is that English does none of this work. That covers subject and object alike, and can be dropped entirely - the train I take. French never drops the relative pronoun, so a slot that is invisible in your first language has to be filled explicitly in your second.

How do you choose between qui and que?

Look at the word immediately after the pronoun. That is the whole test.

• A verb follows → qui. Le train qui part à huit heures.
• A subject follows → que. Le train que je prends.

There is a second clue that costs nothing. Que elides before a vowel - qu’il, qu’elle, qu’on - and qui never does. If you see qu’, it was que.

Que also drags a consequence behind it that qui does not. Because it is a direct object placed before the verb, it triggers past participle agreement in compound tenses:

• Les lettres que j’ai écrites. - The letters I wrote. Two silent endings, visible only on the page.
• Les lettres qui sont arrivées. - agreement here comes from être, not from the pronoun.

This is one of the places where the written language carries information the spoken one deletes entirely, and it is why the rule survives so long as a writing error at B2 - the same mechanism at work in passé composé versus imparfait.

Où is the one place where French is simpler than English rather than harder, and learners still get it wrong. It covers place and time: la ville où j’habite, but also le jour où je l’ai rencontré. English switches to when there, so learners produce le jour quand, which does not exist. One pronoun, two jobs.

When exactly do you need dont?

Whenever the clause would otherwise contain de. That is the only rule, and it has three faces.

Verbs that take de. Parler de → l’homme dont je parle. Avoir besoin de → le livre dont j’ai besoin. Se souvenir de → le jour dont je me souviens.

Adjectives that take de. Être fier de → un travail dont je suis fier. Content de, sûr de, capable de behave the same way.

Possession, where English uses whose. La femme dont j’ai oublié le nom. - The woman whose name I forgot.

The practical consequence is that dont is not really a vocabulary item to memorise. It is a downstream effect of knowing which verbs and adjectives carry de, which is why it is best learned by storing the preposition with the verb from the start - the same habit that makes noun gender cheap if you attach the article early and expensive if you do not.

Then there is ce dont, for when there is no noun to point back to: c’est ce dont j’avais besoin. Same logic as ce qui and ce que, one step further along.

Is lequel worth studying at all?

Not as production, not before B2. The frequency table above shows why, and two structural facts explain the numbers.

After a preposition, French prefers qui when the antecedent is a person: l’ami avec qui je voyage. That removes half of lequel’s potential territory. And dont already absorbs every plain de, which removes most of the rest - duquel survives only inside longer prepositional phrases such as au sujet duquel or à cause de laquelle. A form squeezed from both sides ends up at rank 12,546 in speech, and that is exactly where we find it.

Recognise it when you read. Produce qui, que, où and dont accurately instead. That trade is what separates B1 accuracy from B2 accuracy, a distinction we take apart in B1 vs B2 in French.

In what order should you learn them?

The order our curriculum uses, which follows the frequency data closely:

At A2 the whole first block arrives together - qui, que, où, plus ce qui / ce que and a first pass at dont. At B1 a second dedicated block returns to dont and introduces lequel, auquel and duquel. At B2 the remaining complex forms - tout ce qui, à qui, nested clauses - close the system out.

The detail worth noticing is that dont is the only relative pronoun taught twice, at A2 and again at B1, with the same definition both times. That is not redundancy. It is the syllabus admitting that the first pass does not stick, for exactly the reason the corpus shows: between the two lessons, the learner will have heard the word perhaps a dozen times.

If you want the full sequence of what belongs at each level, it is set out in French grammar by CEFR level, and unit by unit across the 61 levels of the A1 to C1 curriculum - where the point of graded stories is precisely to put dont in front of you often enough that the rule becomes unnecessary.