Avoir vs Être in the Passé Composé: The Data

Four compound tenses in five take avoir. The other one in five takes être - and the closed list you memorised on day one covers only 60.6% of those.

That gap is the whole problem with how this is taught, so we measured it. Across 928,000 words of French prose checked against Lexique 3.83, we identified 9,625 compound tenses and sorted them by auxiliary, by verb, and by whether the mnemonic would have helped.

A word on method first. Our corpus is nineteenth-century prose, which narrates in the passé simple rather than the passé composé, so we counted every compound tense - passé composé, plus-que-parfait, future perfect, conditional past. The auxiliary choice is identical across all of them; only the tense of the auxiliary changes. That gives a much larger sample and the same answer.

How often does each auxiliary actually fire?

Avoir is the default in a strong sense: guess it blindly and you are right four times in five. But one in five is not a rounding error - it is roughly one sentence per paragraph, and because the auxiliary sits at the front of the verb, getting it wrong is the most audible mistake available in the tense.

The verbs that most often sit behind avoir are exactly the ones you would hope: être itself heads the list at 756 occurrences - j’ai été - followed by faire, dire, avoir, voir, pouvoir, prendre and donner. All eight are inside the first two hundred words of French, which is worth noting for a practical reason: the avoir half of this tense becomes automatic very early simply because the verbs that trigger it are the ones you use constantly.

Does the mnemonic cover what you need?

This is where the teaching goes wrong. Split the être cases:

Two of every five être auxiliaries belong to a reflexive verb, and no mnemonic contains them, because the class is open. You can make a reflexive out of almost any transitive verb - se voir, se dire, se perdre, se permettre all appeared in our count - so there is no list to learn.

The commonest reflexives in the sample make the point on their own: se faire, se passer, se tromper, se mettre, se trouver, se arrêter, se perdre. Not one of them is a verb of movement, and not one of them appears in any version of the acronym. They are ordinary transitive verbs that happen to be pointing back at their subject, which is precisely why the class cannot be enumerated and has to be handled by rule.

That inverts the usual order of teaching. The acronym is finite, feels learnable and gets the attention; the reflexive rule is a single sentence, has no exceptions, and covers a class you cannot enumerate. “Every reflexive verb takes être” is worth more than the entire list, and it is usually taught as a footnote to it.

Which verbs on the list actually matter?

Not the ones you would expect. Ranked by real frequency:

Those top five carry 57.8% of all uses of the list; the top ten carry 81.6%. The tail is very long and very thin - monter 23, descendre 20, retourner 11, survenir 8, apparaître 7, redescendre 1 - and four members of the standard list (intervenir, provenir, renaître, ressortir) did not appear once in 928,000 words.

Two things stand out. Mourir is second, which surprises learners who meet it late as a piece of vocabulary rather than as a grammar item. And monter, descendre and retourner - the verbs that consume the most classroom time because they can also take avoir - sit near the bottom of their own list.

What about the verbs that take both?

They are real, and they are smaller than their reputation. Monter, descendre, sortir, rentrer, retourner and passer switch to avoir when they take a direct object, and the meaning switches with them:

The rule is clean: no object, être; object, avoir. It is worth knowing and it is worth roughly the ten minutes it takes, rather than the disproportionate share of a syllabus it usually receives.

Does être plus a participle always mean a compound past?

No - and this one costs reading comprehension rather than production. Alongside the 2,039 genuine compound tenses with être, our scan found 2,877 further sequences of être plus a past participle that were passives or states, not compound pasts at all: est fini, est assis, est obligé, est forcé, est perdu.

So more than half of all “est + participle” sequences in French prose are not the passé composé. Il est fatigué is a state, la porte est fermée is a passive or a state, and neither is “he has tired” or “the door has closed”. Learners who have just been taught that être plus a participle equals a past tense will misread all of them.

The tell is the verb. If it is on the closed list or reflexive, it is a compound past. Otherwise, être plus a participle is describing how something is, not what somebody did.

What is the agreement rule, and does it matter?

It matters in writing and is almost entirely silent in speech, which is exactly why it survives as an exam favourite.

With the closed list, the participle agrees with the subject: elle est venue, ils sont partis, elles sont restées.

With avoir, it agrees with nothing - unless a direct object comes before the verb, which is what makes les lettres que j’ai écrites agree. That interaction belongs to the relative pronouns, where it does most of its damage.

With reflexives, it agrees with the reflexive pronoun only when that pronoun is the direct object. Elle s’est lavée agrees; elle s’est lavé les mains does not, because the hands are the object and the pronoun has become indirect. Ils se sont parlé never agrees, because parler takes à.

That last rule is genuinely difficult and genuinely late. Our A1 to C1 curriculum puts the auxiliary choice at A2 and the reflexive agreement rule in the final grade, which is the right split: you need the auxiliary immediately and the agreement only when you start writing.

So what should you actually learn?

In this order. Reflexive means être - one sentence, no exceptions, and it covers two fifths of the hard cases. Then the ten verbs that matter: venir, mourir, arriver, rester, devenir, partir, entrer, tomber, sortir, revenir, plus their compounds, which is 81.6% of the list’s real usage. Then avoir for everything else, which is right four times in five.

Leave the transitive exceptions and the reflexive agreement rule until you are writing, and let the rest arrive through reading - the argument for which, and the level at which it starts working, is set out in how to learn French by reading. Where this sits in the wider sequence is in French grammar by CEFR level, and the choice this tense forces against the imperfect is in passé composé vs imparfait.