The meaning of "une variation extraphonologique combinatoire"

Question

I'm reading Éléments de Phonologie française by G. Gougenheim and I don't understand the term variation extraphonologique combinatoire. Here's the context:

[...] il n'y a en français qu'un phonème l, bien qu'à côté de l'l sonore habituel il existe un l sourd chuchoté en fin de mot après une occlusive sourde ou un f (oncle, ample, souffle)[...]

Nous avons vu que l'opposition l sonore - l sourd est une variation extraphonologique combinatoire. Il y a aussi variation extraphonologique combinatoire dans l'opposition des noms de nombre six et dix avec s à la fin du groupe de mots (ils sont six; j'en vois dix) et des mêmes noms de nombre avec z en liaison (six hommes, dix enfants).

Can someone elaborate on this? If I understood this correctly, there is only one phoneme /l/ in French although it can be pronounced in different ways. Why wouldn't those two ls be two different phonemes?

Answer 1

This is a term particular to Gougenheim's own framework. A more theory neutral way of stating it would simply be "allophony".

Gougenheim distinguishes between oppositions phonologiques and variations extraphonologiques. The firsts are meaningful, functional distinctions (i.e. they differentiate phonemes), the seconds don't carry meaning and are outside the phonology of a language (hence extraphonologiques - in other words, they're phones).

He distinguishes between four variations extraphonologiques:

1. variations extraphonologiques combinatoires: Despite what the term "combinatoire may imply, he doesn't just mean assimilation by this (for example, the nasalisation of /g/ to [ŋ] in "longuement") but every phenomenon by which a phoneme is realised differently in a language (for example the different realisations of /p/ in English in [pʰ]it and in s[p]it).
2. variations extraphonologiques stylistiques: The different ways a phoneme might be realised by the same speaker in different registers of a language. For example, the same person might realise "l'autre personne" as /loːtpɛχ'sɔn/ in a casual conversation, /loːtχəpɛχ'sɔn/ when reading aloud and /loːtχəpɛχ'sɔnə/ when reading poetry aloud. Today we'd simply call it stylistic variation.
3. variations extraphonologiques individuelles: What we today call "free variation": something that individual speakers say in a few different manners (sometimes switching apparently randomly between variants). A good example in French would be whether the final consonant of fait, août or almanach is pronounced.
4. variations extraphonologiques concomitantes: Today we'd call that "sandhi": in French, elision and liaison are examples. Including sandhi phenomenons in variations is criticable and Gougenheim has been criticized for it.

For a recent review of Gougenheim's theories of phonology, see Peter Swiggers' 2016 paper, "De Prague à Strasbourg : Phonétique et phonologie du français chez Georges Gougenheim et Georges Straka"

---

And yes, that paragraph implies there's two (non meaningful ways to pronounce the phoneme /l/ in French, one that's voiced (like /z/ or /d/) and used in most contexts, and another that's voiceless (like /s/ or /t/) in the word final clusters /kl/, /tl/ and /pl/. Those are two different phones, but not two different phonemes because no two words can ever be distinguished by a voicing distinction of /l/ in French.

For them to be different phonemes (une opposition phonologique, in Gougenheim's words), you'd need them to potentially appear in the same place in a word or syllable, but that's not the case here. An example of an Language with such a distinction is Old English, where the only difference between land (modern land) and hlant (modern lant, a rarer word for urine) was the difference in voicing of l.