Past participle agreement with avoir

The agreement of past participles conjugated with avoir is the single most-discussed rule of written French. It is the rule that French dictée exercises hinge on, the rule that adult French speakers second-guess in emails, and the rule that a working knowledge of will lift your written French immediately. The good news: the rule itself is short. The bad news: applying it requires you to identify direct objects on the fly and recognize when one has been displaced to a position before the verb.

This page lays out the rule, walks through the three syntactic contexts where it triggers, gives you the auditory test that distinguishes "matters in writing" from "matters in speech too," and works through enough examples to drill the pattern.

The rule

The past participle conjugated with avoir agrees in gender and number with a direct object that precedes the verb. If there is no direct object, or if the direct object follows the verb, the participle stays in its bare form.

That's the entire rule. Three things are doing real work:

1. Direct object — not the subject (with avoir, the subject never triggers agreement) and not an indirect object.
2. Precedes the verb — the default position of the direct object in French is after the verb, and in that default case, no agreement happens.
3. Avoir auxiliary — this rule is specific to avoir. With être, the participle agrees with the subject; see agreement with être and être + reflexive.

The default: no agreement

In an ordinary sentence with subject — avoir — past participle — direct object, nothing agrees. The participle is bare.

In every example above, the direct object (une pomme, trois lettres, Sophie, les valises) follows the verb. The participle stays in its base form: mangé, écrit, vu, pris. The fact that trois lettres is feminine plural changes nothing.

The crucial contrast with English: in French, the participle is passive to what comes after it. It only knows about what comes before it.

The three contexts that trigger agreement

A direct object can be moved in front of the verb in three syntactic ways. Each of them activates the agreement rule.

1. Object pronouns (clitics)

When a direct object is replaced by a pronoun — le, la, les, l' — the pronoun moves before the verb. At that point, the participle must agree with the noun the pronoun refers to.

The first-person and second-person clitics me, te, nous, vous can also be direct objects, and they trigger agreement just like le/la/les — but you have to know what gender and number the speaker is.

2. The relative pronoun que

When a noun is the antecedent of a relative clause introduced by que, that noun is the direct object inside the clause, and it precedes the verb (because relative pronouns sit at the start of their clauses). The participle agrees with the antecedent.

The contrast that makes this rule essential: que (object) versus qui (subject). Qui is the subject of its clause, not the direct object, and so it does not trigger participle agreement.

The agreement on arrivés in this last sentence is a consequence of être (subject agreement), not of avoir (preceding-DO agreement). With avoir, qui never causes agreement on the participle.

3. Fronted question word quel(le)(s) + N

When a direct object is fronted as a question phrase using quel/quelle/quels/quelles, the participle agrees with the noun in the phrase.

The same logic applies to combien de + N fronted as a direct object:

The auditory test: when does this matter in speech?

Most past participles end in a vowel (-é, -i, -u) or a silent consonant. For those, the masculine-singular form and the feminine-plural form are pronounced identically. The agreement is purely orthographic: it shows up in writing but not in speech.

Participle	m.sg	f.sg	m.pl	f.pl	Audible?
mangé	mangé /mɑ̃.ʒe/	mangée /mɑ̃.ʒe/	mangés /mɑ̃.ʒe/	mangées /mɑ̃.ʒe/	No
fini	fini /fi.ni/	finie /fi.ni/	finis /fi.ni/	finies /fi.ni/	No
vu	vu /vy/	vue /vy/	vus /vy/	vues /vy/	No

But for participles ending in a silent consonant (pris, mis, fait, dit, écrit, ouvert, peint, conduit), the feminine forms add an -e that makes the previously-silent consonant audible. Those agreements are heard.

Participle	m.sg	f.sg	m.pl	f.pl	Audible?
pris	pris /pʁi/	prise /pʁiz/	pris /pʁi/	prises /pʁiz/	Yes
mis	mis /mi/	mise /miz/	mis /mi/	mises /miz/	Yes
fait	fait /fɛ/	faite /fɛt/	faits /fɛ/	faites /fɛt/	Yes
dit	dit /di/	dite /dit/	dits /di/	dites /dit/	Yes
écrit	écrit /e.kʁi/	écrite /e.kʁit/	écrits /e.kʁi/	écrites /e.kʁit/	Yes
ouvert	ouvert /u.vɛʁ/	ouverte /u.vɛʁt/	ouverts /u.vɛʁ/	ouvertes /u.vɛʁt/	Yes

So when a French speaker says Quelles lettres as-tu écrites ? with the -t of écrites sounding clearly, that final /t/ is a marker of agreement. Saying écrit /e.kʁi/ in that sentence sounds wrong even to native ears.

This is why the rule matters in spoken French at all. For most participles, agreement is invisible — but for the consonant-final ones, native speakers hear it and judge accordingly.

What does NOT trigger agreement

The rule is narrow. Several things that look like they should trigger agreement do not.

Indirect objects don't count

The rule is direct object. Lui, leur (indirect-object pronouns) and en (partitive pronoun) never cause agreement, no matter how prominent they are in the sentence.

The en exception is worth pausing on. En refers to a quantity ("some of, three of"), and even when it's the apparent direct object, the rule treats it as a partitive that doesn't determine the participle's form. Many French grammarians fight about this; the standard position is no agreement with en.

Subjects never count, with avoir

This is the trap that English speakers fall into most often. With avoir, the participle never agrees with the subject — only with a preceding direct object. The subject's gender and number are irrelevant.

If you find yourself agreeing the participle with the subject, you've reverted to the être rule. Pause and check the auxiliary.

The verbs coûter, peser, mesurer, valoir, vivre, durer with adverbial complements

Some verbs take what looks like a direct object but is actually an adverbial complement of measure: les vingt euros que ça a coûté ("the twenty euros it cost"), les deux heures que j'ai dormi ("the two hours I slept"). These complements answer "how much / how long" rather than "what," and they don't trigger participle agreement.

The same verbs do trigger agreement when they take a real direct object: les efforts que ce travail m'a coûtés ("the efforts this work has cost me" — here efforts is a real DO).

A tour through edge cases

A few patterns that come up often enough to be worth listing.

Combien de + N

Treat as a regular fronted direct object — agrees normally:

Repeated participle forms with avoir

Some participles look like they shouldn't change, but check the consonant: cuit, conduit, écrit, peint, craint all add audible -e in feminine forms.

The participle of an impersonal verb

Impersonal expressions like il y a or il a fait (weather) don't have a real subject, and the participle stays bare in any case.

A practical drill: rewriting with a clitic

The fastest way to internalize the rule is to take a sentence with a postverbal direct object and rewrite it with a preceding clitic. Compare:

Reading the second member of each pair aloud and listening for the audible feminine -e is the best ear training there is.

Why this rule exists

A historical note that helps make sense of it. In old French (and old Italian, old Spanish, etc.), past participles of compound tenses agreed with the direct object in all positions, before or after the verb. Over centuries, French shifted: the agreement was lost when the direct object follows, and retained when it precedes. Italian preserved fuller agreement; Spanish and Portuguese mostly dropped it.

The shift was probably driven by phonology — the agreement endings on French participles are mostly silent, and the loss of audibility encouraged the loss of the rule itself in default word order. Where agreement remained audible (consonant-final participles in feminine forms: prise, mise, faite, dite), the rule survived. The current rule is a fossilized halfway point.

This explains the asymmetry: the rule isn't arbitrary. It tracks where agreement is still salient.

Comparison with English

English has no past-participle agreement of any kind in standard usage. I have eaten never changes form. So the avoir rule is genuinely foreign, and English speakers have a strong tendency to either skip it entirely or over-apply it (treating the subject as the trigger, by analogy with French être-agreement).

The transfer-error patterns are predictable:

• Forgetting agreement on a fronted DO (Les pommes que j'ai mangé — wrong; should be mangées).
• Adding agreement to the subject (Marie a mangée — wrong; should be mangé).
• Adding agreement when the DO follows the verb (J'ai mangées les pommes — wrong; should be mangé).

Drilling the three trigger contexts (clitic, que, fronted question) and the auditory test (consonant-final participles in feminine forms) is the path through.

Common Mistakes

Key takeaways

• With avoir, the past participle agrees only with a preceding direct object — never with the subject.
• Three contexts trigger the agreement: object pronouns (le, la, les, l', me, te, nous, vous), the relative pronoun que, and fronted question phrases (quel(le)(s) + N, combien de + N).
• Indirect objects (lui, leur) and the partitive en never trigger agreement.
• Most agreements are silent — visible in writing only. But participles ending in a silent consonant (pris, mis, fait, dit, écrit, ouvert, peint) gain an audible /t/, /z/, /s/ in feminine forms, making the rule audible too.
• The fastest way to internalize the rule is to rewrite postverbal-DO sentences with a preceding clitic and listen for the audible feminine ending.
• This is one of the trickiest rules in French and the one that defines well-edited written prose. Native speakers themselves trip over it; explicit drilling pays off.