Phonetic crime scene:
AEVUM rehearsal, November 2025

We are rehearsing ‘Air vif’, one of my compositions on a text by Paul Éluard. First line: ‘J’ai regardé devant moi’ (‘I looked ahead of me’). The sopranos come in. I stop them immediately.

‘Let’s go again. That “j’ai”: how do you pronounce it?’

Awkward silence in the ranks. Someone ventures:

‘Erm… “jé”? As in “été”?’

Another voice pipes up:

‘No, “jè”! As in “merci”!’

I smile. Here is a simple question that divides even experienced singers. I decide to consult my experts. My mother, an amateur singer for more than 30 years. Jean-Jacques, her choir director, who holds a doctorate from the CNSM (the Paris Conservatoire). My godfather, another seasoned singer. Béatrice, a soprano at the Opéra de Paris. Nicolas, a choir director with a master’s in modern literature from the Sorbonne.

Five experts. Five different answers.

My mother gives me one explanation. Jean-Jacques gives another, just as convincing. My godfather says something different again. Béatrice talks to me about vocal placement with fascinating precision: for her, ‘j’ai’ is a closed é leaning towards the œ of ‘œuf’. Not quite a pure ‘jé’, not quite an open ‘jè’. Something in between, a subtle shade that only a trained ear can perceive and reproduce. Nicolas, for his part, explains the phonetic rule to me precisely: the verb form ‘j’ai’ takes a closed é, but as soon as a consonant follows, you switch to an open è. And for forms ending in -ais/-ait, it is always an open è.

I hang up, slightly dazed. Each of them defends their own tradition, their school, their approach. And the strangest part? They’re all right. Or rather: they all hold part of the truth.

That’s odd, I say to myself. How can a simple ‘ai’ cause so much disagreement among professionals? How can there be five different answers to a question that seems so simple?

I need an anecdote to illustrate the confusion. I think back to a conversation with Nicolas. He tells me that in Lyon people naturally say ‘c’est vré’ for ‘c’est vrai’ (‘it’s true’). And he adds, with a little smile: ‘Phonetically, the Lyonnais are right. In theory, we should all pronounce it like that.’

Hang on. The Lyonnais are right? Then why don’t we all pronounce it like that?

That’s where our investigation begins.

The investigation begins: following the trail back to the Middle Ages

When the experts disagree, there’s only one solution: go back to the sources. Like a detective poring over old case files, I dive into the history of this pronunciation. And what I find fascinates me.

First clue: medieval texts (11th to 12th century)

The ‘ai’ we sing today was once a genuine diphthong, pronounced [ai̯]. Imagine: instead of saying ‘faire’, people in the Middle Ages said something like ‘fa-ire’, with two distinct sounds.

Where does that sound come from? From the Latin ‘facere’, where the ‘c’ became palatalised and turned into an ‘i’ that attached itself to the ‘a’. The result: [ai̯].

But here’s the twist: from the early 12th century, this diphthong begins to simplify into a plain [ɛ] (open è). The assonant texts of the period leave us clues: some poets already rhyme ‘ai’ with ‘e’, proof that they heard the same sound.

The body has been found. The diphthong died very early. But nobody told the grammarians.

Second clue: the Renaissance grammarians (16th century)

In 1531, Jacques Dubois still describes an ‘ai’ diphthong. But reading between the lines, you sense he is describing something he no longer really hears. It’s like a police officer describing a corpse as if it were alive.

Louis Meigret, in 1542, is more honest: ‘vous ne trouverez aucunes nouvelles de la diphtongue ay, mais tant seulement d’un e ouvert’ (‘you will find no trace of the diphthong ay, only an open e’). Translation: folks, that diphthong has been over for ages.

Jacques Peletier and Pierre de la Ramée confirm it: ‘ai’ is pronounced like ‘e’, generally open.

The crucial clue: the rhymesters aren’t fooled. Since the 12th century, they’ve been rhyming ‘ai’ with ‘e’. Poets have sharper ears than grammarians.

Third clue: the 17th century officially buries the diphthong

In the 17th century, grammarians finally abandon the idea of the diphthong. Gilles Ménage (1694) settles the score with this fiction: ‘ai’ is a ‘false diphthong’, simply a spelling used to represent the vowel ‘e’.

But (and this is where it gets interesting) Ménage makes a distinction:

  • In general: ‘ai’ = open è [ɛ]
  • In certain verb forms ending in ‘-ai’: ‘ai’ = closed é [e]

This distinction would be formalised in the 20th century by Maurice Grammont. But it had already existed in practice since the 17th century.

The culprit identified: Grammont’s rule (20th century)

Maurice Grammont, in his treatise ‘La Prononciation française’, formalises what people were already doing by instinct. Here is his rule:

‘Ai’ is pronounced as a CLOSED É [e] in these specific cases:

  1. Verb forms in -ai: ‘j’ai’, ‘tu as’, ‘il a’ in the past historic
  2. Futures and past historics: ‘je chanterai’, ‘je chantai’
  3. A few rare forms: ‘sais’, ‘sait’, ‘vais’
  4. The adjective ‘gai’

For EVERYTHING ELSE, ‘ai’ = OPEN È [ɛ].

Simple, isn’t it? Not so fast.

The plot twist: modern usage contradicts Grammont

Back to 2025. I look it up in the Petit Robert. For ‘gai’, it accepts both pronunciations: [ge] AND [gɛ]. For ‘quai’: [ke]. So far, so good.

But for ‘sais’, ‘sait’ and ‘vais’? Modern usage prefers the open è [ɛ]. Apparently only the Comédie-Française keeps the closed é [e] for ‘je sais’.

In the future tense, the same story: ‘je chanterai’ (I will sing) tends to merge with ‘je chanterais’ (I would sing, the conditional) and to open towards [ɛ].

The plot thickens: the rules exist, but usage no longer really follows them.

The regional factor: Lyon is right

In Lyon, people naturally say ‘c’est vré’ for ‘c’est vrai’. And phonetically, the Lyonnais are right: according to the classical rules, we should all pronounce it like that.

But in Paris, people say ‘c’est vrè’. And nobody finds that odd.

This regional variation explains why my five experts gave me five different answers. Each of them speaks according to their own tradition, school and region.

What to remember when you sing

There you are, facing a score. ‘J’ai regardé devant moi’. How are you going to sing that ‘j’ai’? Here is your compass.

Rule no. 1: consistency above all

In a choir, uniformity matters more than ‘historical truth’. If your choir director says ‘we’re doing an open è’, you do an open è. Full stop. A vowel that isn’t unified breaks the ensemble sound and creates ugly dissonances.

Five people singing ‘jè’ + five people singing ‘jé’ = a choir that sounds out of tune. Even if, technically, everyone is ‘on pitch’.

Rule no. 2: follow the context

Baroque or classical repertoire: if you’re singing Rameau or Gounod, follow Grammont. Closed é for ‘j’ai’, ‘je chanterai’, ‘gai’. That’s the tradition of the period.

Contemporary repertoire: you have more freedom. But listen for what the composer had in mind. Some wrote their melodic line thinking of an open vowel, others of a closed one.

Romantic French mélodie (art song): everyday usage (open è) will often be more idiomatic.

Rule no. 3: a few certainties to reassure you

ALWAYS open è [ɛ]:

  • Imperfect: ‘j’avais’, ‘tu étais’, ‘il faisait’
  • Conditional: ‘je chanterais’, ‘tu viendrais’
  • When ‘ai’ is followed by a consonant: ‘faire’, ‘lait’, ‘maître’, ‘paix’

Probably closed é [e] if you follow the classical tradition:

  • ‘J’ai’, ‘tu as’ (present)
  • Future: ‘je chanterai’
  • ‘Gai’ (the adjective)

Grey area (ask your choir director):

  • ‘Je sais’, ‘tu sais’, ‘je vais’
  • Futures (the modern tendency is to open them)

Rehearsal scenes: concrete examples

Scene 1: ‘J’ai regardé devant moi’ (Paul Éluard)

You’re standing, score in hand. The choir director raises a hand. You breathe in. ‘J’ai…’

The classical option (Grammont): you pronounce ‘jé’ [e], a closed vowel, well forward, crystal clear. It carries a long way, it’s bright.

The modern option: you pronounce ‘jè’ [ɛ], an open vowel, closer to spoken language. It’s softer, more natural to a contemporary ear.

Both work. But not at the same time in the same choir.

Scene 2: ‘J’avais un ami’

No hesitation here. The imperfect ‘-ais’ is ALWAYS pronounced as an open è [ɛ]. ‘Javè’. End of story.

If someone in your choir pronounces it ‘javé’, stop them immediately. This isn’t a question of school or tradition. It’s just wrong.

Scene 3: ‘Je le ferai demain’

The future. Turbulence ahead. Two schools clash:

The classical school: ‘je le feré’ [e]. It follows Grammont, the lyric tradition.

The modern school: ‘je le ferè’ [ɛ]. By analogy with the conditional ‘je ferais’, the current trend in spoken French.

Your choir director will choose. You follow. It’s as simple as that.

Toolbox: the phonetic alphabet

To understand what we’re talking about, three symbols to remember:

  • [e]: closed é → ‘été’, ‘chez’, ‘nez’
  • [ɛ]: open è → ‘merci’, ‘lait’, ‘fête’, ‘mais’
  • [ə]: mute e → the ‘e’ in ‘premier’ that is barely pronounced

The difference between [e] and [ɛ]? It’s also a matter of vocal placement:

  • [e]: mouth more closed, higher sound
  • [ɛ]: mouth more open, lower sound

But here’s where it gets interesting: between these two vowels lies a whole continuum of shades. That’s what Béatrice was trying to explain to me when she said that for her, ‘j’ai’ was a closed é leaning towards the œ of ‘œuf’.

She wasn’t talking about a classic open [ɛ]. She was talking about a closed [e], slightly coloured, rounded, enriched by a touch of [œ]. A hybrid vowel, somewhere between the pure ‘é’ of ‘été’ and the ‘œu’ of ‘œuf’. Something imperceptible to the untrained ear, but fundamental to the vocal colour of a lyric soprano.

It’s this kind of nuance that makes pronunciation so hard to codify. You can’t reduce everything to [e] or [ɛ]. The human voice offers an infinite palette of colours between those two poles.

Epilogue: back in rehearsal

We go back to ‘Air vif’. ‘J’ai regardé devant moi.’

I’ve made my choice. We’ll sing a closed ‘jé’ [e], as Grammont prescribes for verb forms in -ai. It’s clear, it carries, it suits the neoclassical style of my composition.

But I’m careful not to tell the singers that ‘there is only one way to do it’. I explain: here is the classical rule, here is modern usage, and here is why I’m making this choice today. For this piece. In this context.

Tomorrow, on another text, I may well make a different choice.

What this investigation taught me:

  1. The experts don’t agree, and that’s normal. Language evolves, traditions vary, usage changes.
  2. The rules exist, but they aren’t absolute. Grammont codified a practice, not an eternal truth.
  3. Context comes first: baroque, classical, contemporary, each repertoire has its own logic.
  4. Consistency is king: better a united choir with the ‘wrong’ vowel than a divided choir with the ‘right’ one.
  5. Humility is essential: in the arts, as in learning, it’s better not to have too many certainties.

The richness of sung French lies precisely in this diversity of possible approaches. Don’t let anyone tell you there is only ‘one way’ to do it. The truth is more nuanced, more alive, more human than that.

To go further

Sources and references

This article draws on in-depth research in historical phonetics and lyric diction, in particular:

Online resources

To deepen your knowledge of the historical phonetics of French:

  • The Virga.org website offers exhaustive documentation on how French pronunciation has evolved in poetry and music
  • Le Grand Robert online provides a complete table of the International Phonetic Alphabet with many examples

Article written by Corentin Richard, composer and choir director, for The Choir Notebooks