Talk:Phonological history of French

Latin AU

According to the rules described on this page so far, Latin aucellus would have ended up as oseau (/o/). Something needs to explain how aucellus became OF aucel (/au/), thence oiseau. i.e. AU -> oi (/ua/) Fletpedia (talk) 21:24, 26 May 2014 (UTC).

See Wiktionary. The Old French form is oisel, not **aucel. The Latin ancestor could have been *avicellus instead, in which case the development might have been regular. I can't think of a parallel example (where -avi- stood in unstressed, or even better, immediately pretonic, first syllable) right now, though. User:Benwing is more savvy than I about Old French. --Florian Blaschke (talk) 15:44, 27 August 2014 (UTC)

Ah! There appear to be details not mentioned here or in Old French. It seems that oisel is like plaisir and intervocalic -s- is really /z/ (as still in Modern French) from /dz/, which behaves like a palatalised consonant. So, /ɔjzɛl/ < */ɔjdzɛl/ (early de-affrication of /dz/) < */aujdzɛl/ (monophthongisation) < */audzɛl/ (/dz/ ejects /j/ into a preceding open syllable) < */autsʲɛllu/ < */aukɛllu/ < */aukelluː/ ⟨aucellum⟩. So the i is a result of /j/-ejection, not from Latin -i-, which would have been syncopated anyway. --Florian Blaschke (talk) 16:20, 27 August 2014 (UTC)

Right. Single palatalized consonants in Gallo-Romance between vowels spat out a /j/ before them, and also after them if they were followed by a stressed open A or Ē, which resulted in A -> /ie/, Ē -> /i/. Thus plaisir /plaizir/ (OF) < plaidzir < plaidzʲieir < pladzʲeir < platsʲeːr < platsʲeːre < platsʲere < Latin placēre(m). I do think this is documented somewhere on this page. I'll take a look. Benwing (talk) 07:33, 29 August 2014 (UTC)

This is mentioned in "Table of vowel outcomes" second bullet point, but granted it could be clearer. Benwing (talk) 07:37, 29 August 2014 (UTC)

I was confused by the s (rather than c or z). Old French says that /dz/ is spelt as z, as in doze, which I expected to be present here too, but then I realised that /dz/ had already become /z/ at this point – onze, doze and treize could be special cases in that they had /ddz/ as the result of Gallo-Romance syncope, which was simplified to /dz/ along with the other geminates but not de-affricated (this appears to suggest that devoicing and /dz/ > /z/ actually preceded degemination). Moreover, the resulting diphthong /auj/ looked strange to me – I began to doubt whether /j/ was also ejected into a syllable ending in a closing diphthong, but since the rules only specify an open syllable for ejection to occur, which /au-/ certainly is, I figured /auj/ must be correct, seeing that the result is as expected. That was quite an aha! effect: simply following the rules faithfully and mechanically actually works out. --Florian Blaschke (talk) 23:54, 29 August 2014 (UTC)

/θ/

Second lenition: Same changes as in first lenition, applied again

And yet a couple of lines later we have "loss of θ" with no (other) origin for the segment given. So I gather it's not quite the same after all, and here /t/ lenited instead to [θ]? What about /p/ and /k/, then? Spirantization? Voicing? Nothing? --Trɔpʏliʊm • blah 22:26, 7 October 2014 (UTC)

The phoneme /θ/ arises from final devoicing of /ð/ after loss of final /e/, /ð/ arising from /d/ in the second lenition, itself from /t/ by the first. Analogously, after a vowel, /p/ is voiced to /b/ in the first lenition and to /v/ in the second, yielding /f/ when devoiced, as in chef < CAPUT. On the other hand, /k/ yields, via /g/, the approximant /j/, which is not devoiced, for example in OF lai < LACUM. --Florian Blaschke (talk) 14:17, 27 December 2014 (UTC)

virginis > vierge?

How did Latin virginis /ˈwɪr.ɡɪ.nɪs/ develop into French vierge /vjɛʁʒ/? By the looks of the vowel chart, it seems to me that it should have developed into verge /vɛʁʒ/. Most other words I can think of follow the rules of this chart, so I'm just curious as to why "vierge" apparently does not. Thanks! 76.18.160.47 (talk) 20:59, 2 November 2014 (UTC)

This appears to be a special rule, according to CNRTL: "Virginem a été rendu par virgene, puis virge par élimination de la syll. finale; devenu vierge par développement de e entre i et r suivi de cons. (POPE, § 644 et § 500)." That is, e is inserted between i and r when a consonant follows. --Florian Blaschke (talk) 14:23, 27 December 2014 (UTC)

"Consonants" section

I created a section "consonants", but I am not sure if my sources are valid (most of it pieced together from the rest of the page and, alas, original research) but the developement of consonants from Vulgar Latin to French are unclear so I felt a section needed to be created for this. Should I put in the information? DjKarta (talk) 18:13, 9 July 2015 (UTC)

Unclear cases

Table still needs to explain some outcomes which, at the current moment, seem inconsistent...

Examples:

(1) cases of isolated words, like peu < PAUCUM (according to the table this should be paui [pwa] or perhaps pau [po] but peu seems inexplainable by the page atm) (2) a number of elements in the conjugation of of être and other irregular verbs (i.e. sius < SUM (+ STO???)-- unexplained at the moment) (3) unexplained inconsistencies in development of certain phonemes-- i.e. although the page gives the impression that 'c' before i, e [tsj] shifted to [s] late enough to be protected from lenition to [z] there are examples that seem to violate this--- oiseau [wazo] < AVICELLUM, raisin < RACE(:)MUM, etc. Another example is certain developments in [w>v] -- when it became word-final it shifted to [f] (boeuf < BOVUM, neuf < NOVEM), while intervocally various things seemed to have been able to happen -- in AVICELLUM it joins with the previous /a/ to make the diphthong /aw/, while the same did not happen in NAVIGARE, where it dropped entirely just after it protected the previous [a] from raising and the following [g] from (early) lenition, to make nager. These things are all explainable but it should be on the page with sources rather than the talk page with me typing OR. Then we have the case of boue < BAUA (Gaulish substratum), and the perhaps related case of taie < THE:CA (Latin < Greek). Page would probably do well to account for all of these. --Yalens (talk) 17:21, 26 January 2016 (UTC)

(1) The word peu is apparently from Proto-Romance *pōkū with "vulgar pronunciation" (an early rustic Latin diphthongisation possibly influenced by Umbrian), not *paukū. As mentioned in Romance languages#Latin diphthongs, this development is reflected in some other words such as queue too, and not only in French but also other Romance languages, though the phenomenon is inconsistent and unpredictable (this must have been a case of lexical diffusion within spoken Latin, or simply borrowing between different dialects or sociolects). Note, however, that Old French has poi, which seems to reflect the diphthong; it looks like the expected outcome to me, anyway.

(2) The -s in Modern French suis is etymologically spurious and was added orthographically after final -s had ceased to be pronounced. In Old French, it's still 'sui.

(3) For oiseau, see above. --Florian Blaschke (talk) 22:14, 25 July 2016 (UTC)

Another question that the page may need to handle with "au"-- it seems Gaulish-origin words with apparent original /au/ don't behave the same way as Latin words with /au/. For example, Gaulish *baua becomes French "boue" /bu/ not *"baue" /bo/. Was Gaulish /au/ treated as a different phoneme? That too seems a bizarre possibility as the most likely possibility, /ou/, would have likely become /ø/ "eu", not /u/ "ou". On a side note, the page doesn't really mention other early sound changes that mostly effected Gaulish words. For example, there is some apparent cases of unstressed intervocalic n-rhotacism. For example, Lingones becomes Langres, and it happens in a lot of other place names (even Londres for London < Londines?), though some of them have other oddities (Carnutes -> Chartres requires t/n metathesis either as a mistake in the Latin, or a proper change in the Gaulish or the French). Perhaps this mostly appears in Gaulish words because of their antipenultimate, rather than Latin's penultimate-or-if-light-then-antipenultimate stress, but I don't have a source on this at the moment. --Yalens (talk) 20:43, 14 March 2017 (UTC)

Found an explanation for a + w > ou in Pope, will add soon. --Yalens (talk) 21:09, 5 April 2017 (UTC)

When are these dated?

It seems to me 8th century changes and 5th century changes are mixed together here. Could someone date, however vaguely, the sound changes put in question on this page? 37.163.20.176 (talk) 15:23, 29 June 2017 (UTC)

You're correct, they are. One of the problems is that the period in question is the least clear, and there is some dispute of the chronology of things. If I get a moment (unlikely) I will try to add some sources. --Calthinus (talk) 18:45, 7 August 2018 (UTC)

Merger of "pot" and "peau"

In the 19th century there still was -- and perhaps in some dialectal accents there still is to this day -- a distinction between /ɔ/ and /o/ in word-final position, such that "pot" is /pɔ/ and "peau" is /po/. No real mention seems to be made of this distinction and its loss. 2.203.201.61 (talk) 20:30, 28 June 2019 (UTC)

See French phonology#Mid vowels. The rise of this distinction is not mentioned in the article because it arises organically from the sound laws listed here. Its loss is simply an omission. I've added it.  Done --Florian Blaschke (talk) 17:15, 22 September 2019 (UTC)

Hiatus contraction?

When did the contraction of a hiatus like in OF mëur /məˈyr/ > /myːr/ (Mod. mûr) 'mature' take place? Presumably in Late Old French or Middle French, but can it be dated any more precisely? --Florian Blaschke (talk) 20:40, 27 October 2019 (UTC)

Well afaik it's hard to say when exactly it started but we know much better when it was done. Originally the <ë> schwa was "countertonic" (secondarily stressed), but the first signs that it began to lose its syllabicity in the 13th century. It seems it didn't become "established" that the syllable was lost till the 16th though (if we want a more Neogrammarian usefulness standpoint, if we want a specific date, I'd say go with this one for convenience). Soon after that became the norm, any sign of the former schwa was lost, and the /y/ underwent compensatory lengthening (hence û). Cheers, --Calthinus (talk) 05:30, 26 March 2020 (UTC)

Disappearance of Old French /h/

@Nortmannus and Calthinus: I've just noticed that the disappearance of Old French /h/, originating in Germanic loanwords such as hache, is not mentioned in the list of sound changes. Do you know when that happened? --Florian Blaschke (talk) 02:54, 26 March 2020 (UTC)

I swear it was here. But if not I can add it. --Calthinus (talk) 04:36, 26 March 2020 (UTC)

Florian Blaschke Well actually it doesn't seem to have been here per my (lazy) scrolling through history. It's Late Middle French. If we want a cite, here is MK Pope [[1]] (see section 196, on page 94 -- Pope gets "updated" at times but she is kind of the "canon" and on this it appears that modern textbooks still use her view for educational examples for general discussion, see Trask's [[2]] sixteenth century... [h-instances] were already disappearing, in spite of the bitter complaints by purists about "h-dropping", and the new [h]s had disappeared by the eighteenth century -- so basically, Late Middle French, more or less). --Calthinus (talk) 05:00, 26 March 2020 (UTC)

@Calthinus: Thanks a lot! I was surprised that I couldn't find it myself in the article. Can you add it? --Florian Blaschke (talk) 08:08, 29 March 2020 (UTC)

 Done --Calthinus (talk) 15:13, 3 April 2020 (UTC)

du

I wonder how the development of Classical Old French del > Late Old French deu, du > Modern French du /dy/ fits into the chronology. If /u/ > /y/ and /o/ > /u/ had already both happened by the time /l/ was vocalised to /w/ on the way to Late Old French, how come deu (presumably /dɛw/? or /dəw/?) > du was still affected by /u/ > /y/? (It's also somewhat unclear how the diphthong eu became u. It's not a regular development in any case.) Either /u/ > /y/ had not actually happened yet, or at least /o/ had not been raised to /u/ yet, so the vowel system lacked a vowel /u/, and when deu /dVw/ lost its vowel and became something like /dw/, necessitating the vocalisation of the /w/, the result could not be identified with a back vowel and was vocalised to /y/ instead.

Maybe the vocalisation of [ɫ] to [w] had already happened in the 12th century or even earlier (perhaps significantly, ou > eu, presumably /ew/ or /øw/, as in louve > leuve, fell together with /ɛw/ as in cheveu, presumably into /øw/), and is simply not visible as [w] kept being spelled as l, though inverted spellings like *fel instead of feu should reveal this if it was really the case. Or, in any case, /u/ > /y/ (which is not easy to prove having happened at any point since it was concealed by the spelling; we can only conclude it must have preceded /o/ > /u/) happened after /l/ > /w/, by 1100 or not much later. In fact I did think that /l/ > /w/ had already happened in the 11th century and was concealed by the spelling. If /u/ > /y/ happened around 1100 and /o/ > /u/ in the 12th century, the chronology would still seem to fit. --Florian Blaschke (talk) 17:54, 4 September 2021 (UTC)

Explanation per Pope (1934: §843): Old French /e/ before an /l/ in preconsonantal position (>[ɫ] > [u̯] > [∅], merging with preceding vowel) was rounded and variably backed, resulting in either [ø] or [o]. Subsequently, unstressed [ø] raises to [y]† and [o] in general raises to [u]. Hence, for the word in question (< dē illum), two variants coexisted for a time: du~dou [dy~du]. Eventually the variation was resolved in favour of the former.

† Compare prudhomme, the first element of which is the same, etymologically, as preux.

The Nicodene (talk) 02:11, 17 September 2021 (UTC)

aider

I just realised that Modern French aider (like obsolete cuider 'to think, to consider') kept the d, which is surprising at first. However, the reason becomes apparent when the Old French form aidier – as well as cuidier – is considered: In Primitive or Archaic French, */ajˈdʲer/ and */kujˈdʲer/ or */ajˈdjer/ and */kujˈdjer/ (however you reconstruct these forms) were not affected by the second lenition, either because palatalised vowels failed to undergo the second lenition or because a following /j/ protected stops from being lenited (if one does not wish to reconstruct palatalised vowel phonemes at any point). Maybe this should be mentioned in the chronology. --Florian Blaschke (talk) 18:16, 4 September 2021 (UTC)

It's because a preceding /j/ blocks lenition; compare /ˈfakta/ > /ˈfajtə/ > /fɛt/ or (as mentioned in the article) /mejjeˈtate/ > /mejˈtate/ >> /mwatje/, with preservation of the original first /t/. I will answer your other question tomorrow. The Nicodene (talk) 09:09, 14 September 2021 (UTC)

But if a preceding /j/ blocks lenition, shouldn't it also have blocked the first lenition /t/ > /d/ in OF aidier and cuidier? --Florian Blaschke (talk) 14:07, 15 September 2021 (UTC)

As the article says: "The standard for central French was initially for lenition to occur before the unstressed vowel apocope". It should, however, say syncope, not apocope. The Nicodene (talk) 01:25, 17 September 2021 (UTC)

External link

I have found a video on YouTube about the phonological history of French (Why French sounds so unlike other Romance languages), but I do not know whether Wikipedia will accept it. תיל"ם (talk) 04:09, 21 November 2021 (UTC)

Certainly not usable as a reliable academic source. Nicodene (talk) 01:17, 21 September 2023 (UTC)

amor>amour

The expected outcome of latin amor (i.e. love) would have been **ameur, not amour (compare other Latin words ending with -or becoming -eur in French). So why do we have here an irregular derivation? Yassine Mehdi (talk) 18:17, 31 August 2023 (UTC)

@Yassine Mehdi ameur is in fact attested. Most likely amour reflects an old borrowing from Occitan, as love was a common theme in the lyric poetry of the troubadours; other, less likely, solutions have also been proposed. Nicodene (talk) 01:17, 21 September 2023 (UTC)

"Proto-French" ?

I don't really understand what Proto-French is, is it a intermediary step or something ? And to which date does it apply exaclty ? YanisBourgeois (talk) 21:19, 20 November 2023 (UTC)

I'm not sure – it seems to refer to a late prehistorical stage sometime between Gallo-Romance and Early Old French, 7th century perhaps, prior to the divergence of the northern dialects (Old Norman and Old Picard), but it's not really consistent (the change of /s/ to /h/ between vowel and consonant appears to have started only in the history of Old French, probably in the 10th or 11th century, since ⟨s⟩ is still written as such in Old French orthography even before voiced consonants, e. g. in isle). I have no idea what the source is for this "Proto-French" stage. --Florian Blaschke (talk) 20:54, 21 June 2024 (UTC)

OF leu, dame, damoisele

These Old French words appear irregular:

(1) OF leu 'wolf' seems to descend directly from Latin acc. sg. lupum 'wolf' – but wouldn't you expect -f at the end, like in chief from Latin caput or VL *capum 'head', nuef from Latin novem 'nine' or novum 'new', oef from Latin ovum 'egg', or buef from Latin bovem, from a prehistorical word-final */v/ that was devoiced? Maybe there was a rule that deleted word-final */v/ directly after /w/ at an early point before the devoicing, or so. There are, in fact, descendants of OF oef in several dialects that lack the final -f, though apparently no comparable f-loss for the other words.

The lack of -f in leu may also somehow have been analogically imported from the nom. sg. leus from Latin nom. sg. lupus, or obl. pl. leus from Latin acc. pl. lupōs, where the consonant seems to have been lost regularly before the following -s. Given that oef has the nom. sg. / obl. pl. form oés, this seems to explain the later f-less forms, too (though unlike uef, always with -f, leu never has -f – though maybe there are variants I'm not aware of). However, this did not happen in chief, despite the nom. sg. / obl. pl. form chiés, nor in buef, despite bués.

(2) The -a- in OF dame 'lady; woman' appears irregular, regardless of whether it is derived from Latin dominam 'mistress (of the house)' directly or from VL *domnam. Is there a special rule for nasal vowels that applies here?

(3) Apart from the same odd -a-, OF damoisele (apparently phonologically with /z/ from older /d͡z/ or /d͡ʑ/) has the added problem of the oi – if it's directly from VL *dom(i)nicéllam, you'd expect something more like *domesele, *damesele or *donsele (not sure what exactly), definitely not with -oi-, which indicates a stressed vowel. What I suspect is some kind of cross between something like *domeille or *dameille from VL *dom(i)nículam and *damesele (vel sim.; curiously similar to ModE damsel) as the regular outcome of VL *dom(i)nicéllam. It's possible that in this word, the -a- in dam- is somehow regular (it seems that dem- as in ModF demoiselle besides damoiselle isn't attested in OF, even though it might even be more expected), and was transferred from there to dame, where (as I suspect) the -a- is definitely not the regular expected outcome (wouldn't you expect something more like *dome?). --Florian Blaschke (talk) 23:20, 21 June 2024 (UTC)

The development of -oi- in damoisele comes from coalescence of -e- with an offglide -j- ejected from the following palatalized [zʲ] (< [dzʲ]). This is a different source than the ej/ei that developed by diphthongization from *[e] in stressed syllables, but I think that ej/ei from either source regularly developed to oj/oi, regardless of whether it was stressed or unstressed. Compare boisson < bibitiōnem (where the offglide j was ejected from a following sʲsʲ). Other examples of this development include voisin < vecinum, moisson < messionem, loisir < licere (although these are all word-initial syllables, so not perfect parallels to damoisele).--Urszag (talk) 00:12, 22 June 2024 (UTC)

Wow, thanks for the fast reply! I wasn't sure if the unstressed *-e- would have been preserved at all – you might have expected that it would be syncopated too, so I wasn't reckoning with a */dVmedzɛlla/ > */dVmejdzɛlla/ > */dVmejzɛla/ > dameisele. I suspect that the curiously archaic form of domnizelle in the Eulalia sequence is influenced by Latin, but I'm not sure. --Florian Blaschke (talk) 00:52, 22 June 2024 (UTC)

"/dz/ > /z/ unless final."

The rule "/dz/ > /z/ unless final." immediately follows the rule "Final stops and fricatives become devoiced." After final devoicing, how can final /dz/ still exist at all? Or does the devoicing rule not apply to affricates? That would be typologically odd.

For example, Latin pacem 'peace' ends up as */pajdz/ before final devoicing. If devoicing applies, the result is */pajts/, but if not, */pajdz/ remains as such. What we find in OF is pais, paiz, paix /pajs/. So it seems devoicing happened after all. However, this result is odd, too: Any deaffrication of /ts/ is only found late in OF according to the table. Or was it still pronounced /pajts/ earlier in OF? --Florian Blaschke (talk) 00:44, 22 June 2024 (UTC)

Pope says pacem would regularly yield pajts, and therefore derives pais from the nominative pax (p. 229, section 639). I've made some edits based on that source.--Urszag (talk) 14:19, 26 August 2024 (UTC)