Reading French ordinal numbers aloud: 1er, 2e, XXe siècle
1er, 1re, 2e, 2nd, 21e, XXe siècle, primo: a written ordinal has one right reading and several wrong ones. The French forms that trip up speech synthesis, and how a well-prepared engine handles them.
An ordinal states a rank, not a quantity. In French, "2" and "2e" look almost identical, but one reads "deux" and the other "deuxième", and the whole difference sits in a superscript letter that an unprepared voice treats as a character to spell out. You hear the mistake at once: "deux e", "deux ème", "vingt et un e". It is the kind of slip that does not break the sentence but instantly signals a machine reading, at the exact moment your content is trying to earn trust.
Ordinals turn up wherever a text ranks, numbers or places things in time: an article about the XXe siècle, a page mentioning the 3e édition, a report listing "1°, 2°, 3°", an address on the 5e étage, a league table, a graduating class. Unlike cardinals, ordinals mix digits and letters, and that mixture is exactly what trips a voice predicting sound character by character. Here are the forms that genuinely cause trouble, the reading you should expect, and the mistake you hear on generic voices.
The table of ordinals that fool the machine
| Written form | Expected reading | Common error |
|---|---|---|
| 1er | premier | "un er", "un e r" |
| 1re | première | "un re", "un r e" |
| 2e | deuxième | "deux e", "deux ème" |
| 2d, 2nd | second | "deux d", "deux n d" |
| 2de, 2nde | seconde | "deux de", "deux n d e" |
| 3e | troisième | "trois e" |
| 21e | vingt et unième | "vingt et un e" |
| XXe siècle | vingtième siècle | "x x e siècle", "vingt siècle" |
| Xe République | cinquième République | "x e République" |
| 1°, 2°, 3° | primo, deuxièmement (or un, deux, trois) | "un degré, deux degrés" |
| n° 4 | numéro quatre | "n petit o quatre" |
| Louis XIV | Louis quatorze | "Louis quatorzième", "Louis x i v" |
| Ve arrondissement | cinquième arrondissement | "v e arrondissement" |
Each line is a test you can run against any tool, ours included, in two minutes. The right-hand column is not a hypothesis: these are the readings you actually hear on voices with no explicit model of French.
Why an ordinal resists
The difficulty comes from three places at once. First, the French abbreviation is not stable in real writing: "2e" or "2ᵉ" is correct, but you also meet "2ème", "2eme" and "2°" in live text. A voice has to recognise that these variants all mean the same rank, then rewrite them as "deuxième". Second, gender and number change the reading: "1er" is "premier" but "1re" is "première", "2d" is "second" but "2de" is "seconde". The superscript letter carries all the information, and if the engine ignores it, it loses the whole word.
The third trap is the Roman numeral used as an ordinal. "XXe siècle" reads "vingtième siècle", but "Louis XIV" reads "Louis quatorze", a cardinal, not an ordinal. Nothing in the writing says which to apply: context decides, a monarch on one side, a century or a Republic on the other. A rule that treated every Roman numeral the same way would be wrong half the time. That is why Roman numerals get their own handling, covered in reading Roman numerals aloud.
Good preparation means normalising the text before synthesis: recognising that a pattern is an ordinal, identifying its gender and number, then rewriting it into the full form the voice will read correctly. It is rule work, not timbre, and it is the heart of the approach set out on the French text-to-speech page: what matters happens before any sound comes out. A beautiful voice fed raw text reads "le 1er" as "le un er"; an ordinary voice that is well prepared reads it "le premier".
What WeDispatch does, without overpromising
The common forms in the table, arabic ordinals, gender and number marks, "1°, 2°" enumerations, "n°", and Roman numerals in their usual context, are handled by the normalisation applied to every article. That is the baseline we treat as given, not as a paid extra.
What remains is the genuinely ambiguous. "IX" on its own can be an ordinal or a cardinal depending on its neighbours; "2d" is sometimes an author's shorthand for "second", sometimes a typo for "2e". In those cases, two answers. First, well-typed source text removes the ambiguity: spelling out "deuxième" when the rank carries meaning is good writing practice, audio or not. Second, for a specific reading that must hold across all your articles (a numbered proper noun, a collection, a house label), the pronunciation lexicon lets you fix the expected form once, with nothing to regenerate.
We would rather say it this way than claim a machine always guesses. It does not guess: it applies solid rules across the vast majority of cases, and leaves you in control of the rest.
The test to run before you choose
Take a real article from your site, not a demo text. Find the places where a rank carries information: an edition, a floor, a century, a number, a ranking. Have the tool you are evaluating read those passages, and listen only to those points. It is faster and far more predictive than a general listen, because these forms recur every day in your content.
Ordinals are cousins of numbers, the other great revealer, covered in reading numbers aloud in French, and of dates, where "1er" keeps coming back, handled in reading French dates aloud. A "deux e" heard instead of "deuxième" rules out more candidates, on its own, than most people expect.
Give your articles a voice with WeDispatch
This blog is itself voiced by WeDispatch. Curious how it sounds on your content?
Book a demo