An unjustifiably long and geeky return.

[wwidsith]

From a linguistic point of view, it's really a fascinating time to be in France right now. The whole language seems to be going through a lot of big changes on a basic level, and things like grammar and pronunciation are relatively big issues

Comments

[ruakh]

> In normal speech nowadays, triggering the subjunctive (which happens after souhaiter que) automatically overrides tense considerations, so that you would usually use what is historically the subjunctive present, J'aurais souhaité qu'il reste.

I don't know if that's quite the best way to explain it. Even in normal speech, the passé du subjonctif functions more or less as a tense (even though it's structured as an aspect); and conversely, even in older forms of French, when the imparfait and plus-de-parfait du subjonctif were in active use, they really followed a sequence-of-tenses of rule rather than indicating the tense of their own clause. For example, one would write « Il demanda que je vinsse demain » ("He asked that I come tomorrow"), even though there's nothing imperfect about tomorrow.

> My French teacher at school still distinguished maître from mettre because of vowel-length: […]

Where was your French teacher from? I didn't learn any contrastive vowel-length distinction in school, and the TLFi gives those words as [mεtʀ̭] and [mεtʀ] (respectively); but I've been given to understand that Quebeckers maintain such distinctiosn. (BTW, I have no idea what the difference between [ʀ̭] and [ʀ] is supposed to be. Do you know?)

[wwidsith]

Yes, of course there was always a sequence-of-tenses thing going on -- but now? Does it really still function more or less as a tense? I never hear it function at all. To me, once you have to use the subjunctive, that is all you need to know. But that's just how I think of it and you're right that it's certainly not a technical analysis.

My teacher was from Nantes. She taught us a few words distinguished by length of /ε/, maitre and mettre was one pair and the other one I remember was tette and tête. No longer true. I think the R diacritic is distinguishing between voiced and unvoiced? Not sure

[ruakh]

I don't really understand your first paragraph. Maybe I didn't express myself well. In older French, you had to ask yourself two questions: (1) what tense was the main clause in? and (2) what tense is this clause in relative to the main clause? The former selected between présent and imparfait (or between passé and plus-que-parfait), and the latter selected between présent and passé (or between imparfait and plus-que-parfait). In current normal French, you no longer ask question #1, but you still ask question #2. "I'm glad y'all came" is « je suis heureux/se que vous soyez venu(e)s », not « je suis heureux/se que vous veniez ».

The reason I balked at "triggering the subjunctive […] automatically overrides tense considerations" is that question #2 seems more like "tense considerations" to me than question #1 does, but I guess I can see it the other way as well.

The /ε/-length stuff is interesting, I'll have to look further into that!

[wwidsith]

Oh yeah that's obviously the case, and you're right I didn't really make that very clear.

[muckefuck]

Sure that's the right diacritic on [ʀ̭]? Devoicing would be shown with an open circle, i.e. [ʀ̥], syllabicity with a vertical stroke, i.e. [ʀ̩]. The only use I know for the carat below in this context is indicating apicodental articulation in Americanist notation (i.e. a trill with the tongue against the teeth, which would be completely foreign to normative European French pronunciation).

[ruakh]

It's right, at least for some value of "right": I literally copied and pasted it here from the TLFi. (http://www.cnrtl.fr/lexicographie/maître, at the very bottom.) And I've found the TLFi to be very good; even though it's a digitized version of a print work - see http://fr.wikipedia.org/wiki/Trésor_de_la_Langue_Française_informatisé - I've never encountered anything that seems to be a scanno. But I suppose there's a first time for everything, so I can't say I'm sure that it's right for an appropriate value of "right".

I think you're on-target with your "syllabicity" idea. I wish I could be sure, though. I just tried looking up some words that I thought would show how it treats syllabic consonants - lèvre, -isme - and found that it seems to use an open circle for them!

Also - I just noticed that although the TLFi itself doesn't give [ε:] in its entry for maître, it mentions a 1973 dictionary that does so. And I see that the TLFi gives [ε:] in its entry for lèvre, so it does believe in a phonetic long ε, whether or not it believes in a phonemic one . . . languages are too hard, I should just give up now. :-P

[muckefuck]

Well, that's freakin' annoying: they don't give a key for this implementation of IPA. At least they cite Martinet-Walter as their source for that transcription, so I can scare up the print volume and see what it says.

[ruakh]

I assume that the print TLF must have lots of front-matter and such, but all that I can find online are these two prefaces.

[muckefuck]

I've have a look at the print because the trip to the Martinet-Walter cleared up nothing. Their transcription isn't even phonetic, it's phonemic (i.e. /mɛ(:)tr/), so I don't even know who introduced that [ʀ̭]. I'll go back and have a look at the print TLF.

[muckefuck]

Okay, mystery solved: According to the section on pronunciation in the prefatory matter of the print edition, the carat below is used as a sign of "désonorisation", i.e. devoicing. And you're right, the usual IPA symbol for this, the ring below, has been repurposed to show syllabicity. Why anyone would want to deliberate confuse matters this way, I have no idea. Bloody French!

[ruakh]

Bloody French, indeed! :-P

Thanks for your research. I'm so impressed. I've wondered about the TLFi's phonetic notations for years, but I've never once tried to track down a print copy.

. . . but wait, so does this mean that the R is devoiced in maître, but not in mettre? I never realized that such a distinction existed. I always thought that R-devoicing was a very shallow phonetic feature, completely determined by context (and not even present at all in some dialects). Another thing I'll have to look into!