Investigations into French Diacritics

This was originally two posts on cohost.

The rôle of the circumflex

In French, there are three broad reasons a word has a vowel with a circumflex on it:

  • To mark a historical vowel length, e.g. grātia → grâce, κῶνος → cône, δῐ́πλωμᾰ → diplôme — but not always!, e.g. ἀξίωμα → axiome;
  • To mark a change in spelling, particularly a dropped letter, e.g. ospital → hôpital, aage → âge, meur → mûr — but sometimes the original spelling existed to mark a vowel length, e.g. δῶμα → dosme → dôme;
  • To distinguish homonyms, e.g. mûr (ripe) vs. mur (wall), parait (parer) vs. paraît (paraître ← paroistre), chasse (chase) vs. châsse (case) — evidently some overlap with spelling changes.

A lot of these are accompanied by vowel changes in the pronunciation, where without the accent you would normally pronounce it as a different vowel. The â usually indicates an /ɑ/ rather than an /a/; the ô usually indicates an /o/ rather than an /ɔ/; the ê usually indicates an /ê/ in open syllables and an /ɛ/ in closed syllables.

However, according to Maurice Tournier’s « À quoi sert l’accent circonflexe ? » (What is the purpose of the circumflex accent?), there’s a few words that simply have no clear reason why they have the accent at all! Some of them I think are now explained by historical vowel lengths, e.g. dôme, but others don’t seem to be justified, apparently, meaning that the reason the vowel is pronounced that way is because it was spelt that way. To quote Tournier:

À cette première série, aujourd’hui incomplète, s’ajoutent tous les accents qui ne ne sont nullement issus d’une lettre étymologique disparue: âcre, aîtres, alcôve, âme, arôme, bêler, câble, câpre, chaîne, châle, châsse, diplôme, disgrâce, dôme, drôle, empêcher, extrême, fêler, flâner, frêle, geôle, grâce, infâme, maltôte, mânes, môle, pâle, pâtir, poêle, pôle, prêche, prône, râble, rêne, rôder, rôle, salpêtre, symptôme, suprême, trône, vêler.

Some of the ones with â, ê, or ô could plausibly be explained by the desire to enforce a particular pronunciation, but my arch nemesis poêle stands out in particular. I think it’s kind of cute, though, that Tournier ends his article with:

À quoi sert le circonflexe ? À se fair plaisir.

oe, oê, oë, oi

Browsing the Wikipedia page for French orthography, I found exceptionally upsetting exceptions.

  • RULE: A tréma on a vowel following another vowel makes the two vowels pronounced separately. EXCEPT: foëne /fwɛn/
  • RULE: Barring loanwords1, œ is /œ/ and oe is /oe/. EXCEPT: moelle /mwalø/, poêle /pwal/
  • RULE: oi is /wa/. EXCEPT: oignon /ɔɲɔ̃/

I was wondering how foëne, moelle, and poêle could have possibly arisen, so I got a TXT file of the entirety of the Littré dictionary and started by searching for words with in them. The most common usage seems to be as a precursor to , e.g. poëtepoète, aloësaloès, troënetroène, kakatoëscacatoès, poëre (obsolete); or to , e.g. goëlandgoéland, goëmongoémon, pekoëpekoé.

I did also find proëme, which once was also proème, but seems to now have settled on proême. I’ve noticed that Littré uses both è and ê in his phonetic transcriptions where modern French would pronounce as /ɛ/, so maybe which one gets used depends on which pronunciation dominated then, but I also have no idea how ê is supposed to sound different from è.

The second most common occurrence is in words where it makes a /wɛ/ sound, like foëne, boësse, and coësre2. However, what I’m searching for is uses of /3 as /wa/, which are:

  • boëtte, which is now boîte, the circumflex indicating a dropped s;
  • coëffe, which is now coiffe;
  • poêle, the only current usage of , it seems.

Interestingly, I also found poëslon from an excerpt by O. de Serres from the 16th century. I think this is a precursor to poêlon (une petite poêle), which would explain the circumflex to indicate a dropped s. moême is also listed as a variant to même, which does have a dropped s, but that’s pronounced /mɛm/.

There’s further mysteries, though: Littré lists as variants of cuy the spellings coëf and couët, and it’s unclear how these are pronounced. I would guess anywhere from /kɔe/ to /kɔɛ/ to /kuɛt/ to /kwɛ/. Another quotation he includes uses the word grailloët, and I haven’t been able to find what that word means or how it’s spelt now. I would guess /gʁajɔɛ/ for it.

In any case, I think it’s a little unprincipled to be analyzing uses of a particular orthography without having the context of historical vowel shifts that probably accompanied these orthographies. Given that I’m looking at one very specific orthography and a narrow set of vowels (/wa/, /wɛ/), it seems like it’ll be a lot of work to trawl through textbooks on historical French linguistics just to find this information…

Thankfully, I then found this online Dictionnaire de l’Académie française, and the really cool thing about it is that you can search for all historical versions of the dictionary, so you can literally see the spelling changes going back about four centuries! Here’s a table a few of the words above and their Officially Sanctioned Orthographies from a few of the earliest years of the dictionary.

Current 1694 1718 1740 1762
poêle (pan) poële poële poêle poêle
poêle (furnace) poële, poile poële, poile poêle poêle
poêle (dais) poesle, poësle4, poisle poesle, poisle poêle poêle
coiffe coeffe, coiffe coeffe, coiffe coëffe, coîffe coiffe
moelle moelle, moëlle4 moëlle moëlle moelle
boîte boiste boiste boîte, boête boîte

There are actually three etymologies for poêle, from Latin’s patella, pēnsilis, and pallium, respectively. In Old French, these seem to have been païella, poil, and paile. I guess the reason they converged onto such a bizarre orthography is because they were influencing one another.

Based on the 1740 edition, I think there was some recognition that the use of oe for /wa/ is unusual, and that they started spelling it as to indicate that pronunciation, but was then dropped later on, except keeping for the removed s.

Unfortunately, it doesn’t go back far enough for spellings of boîte before boiste, and Littré sources boëtte from Paré, a 16th-century surgeon. I might have to reference a Middle French dictionary for that. In fact, I suspect a lot of uses of oi were oe to begin with, and its pronunciation shifted from /wɛ/ to /wa/.

  1. Some loanwords are from Latin, similar to how UK spelling will retain œs in places, and are pronounced /e/ or /ɛ/; others are from German where originally there was an ö, and pronounced /ø/. 

  2. I’m excluding loanwords like floë /flo/ and broë /bro/. These two in particular seem to be Norwegian. 

  3. I’d also like to find uses of oe as /wa/, but this file uses simply oe where it should use œ, so it’s difficult to distinguish the two. 

  4. For some reason these entries are rather inconsistent, and uses both oe and in the same breath, although the former more often than the latter.  2