Slightly more interesting are the nasal diphthongs, as the less prominant component can be interpeted as a nasal glide (nasalized semivowel), which can become a nasal consonant (nasal stop). (I tried looking into the outcomes of the denasalization of Old Portuguese nasal vowels in Galician, but couldn't find much with specific details). For diphthongs ending in [ɪ̯̃], the corresponding nasal glide [ȷ̃] seems that it would clearly become /ɲ/; I have found several examples of languages with either [ȷ̃] as allophone of /ɲ/ or (in languages where most consonants have oral and nasal forms) with [ɲ] as a nasal counterpart of /j/. However, for [w̃], I found some alternations with /m/, and some with /ŋʷ/.
Most interesting though, was looking at what Quebec French does with the French nasal vowels /ɛ̃ œ̃ ɑ̃ ɔ̃/. First, there's a "clockwise" shift in the places of articulation, and diphthongization in /ɛ̃ ɔ̃/. Specifically, /ɛ̃/ raises and diphthongizes to [ẽɪ̯̃], /ɑ̃/ fronts to [ã] (or, according to one source, as far as [æ̃] in closed syllables), and /ɔ̃/ lowers and diphthongizes to [ɒ̃ʊ̯̃]. Most interesting, though, is what happens to /œ̃/. First, it is more centralized than oral /œ/, closer in place to non-elided /ə/, which is usually rounded (one source has Quebec /ə/ as [œ̈]). More notable, is that more than one work describes it as being "r-colored"; one describes it as [ɚ̃], another as [œ̃˞].
So, using these as a base, the most straightforward would be
/ɛ̃/ > [ẽɪ̯̃] > [eȷ̃] > /eɲ/
Similarly, I can have:
/ɔ̃/ > [ɒ̃ʊ̯̃] > [ɑw̃], with [w̃] becoming /m/ or /ŋʷ/; in the latter case, I'd expect merger with /ŋ/, but not before, pehaps, something like triggering rounding of a following unrounded vowel.
For /ɑ̃/, which remains monophthongal, the place for a following nasal consonant seems less clear. Based on the reading, /ɑ̃/ > [ã~æ̃] > /æŋ/ isn't implausible, but neither is [æ̃] > /æn/, or even just denasalization of [æ̃] > /æ/.
For all three of these vowels, a resulting coda nasal consonant would be subject to the usual nasal place assimilation to a following plosive or affricate.
I have not found details on how exactly the rhotic coloring of /œ̃/ is implemented, so a number of possibilities come to mind:
- If the rhotic nature is via retroflection, we can have /œ̃/ > [ɚ̃] > [əɻ̃] > /əɳ/. The problem with this is that my language has no retroflex consonants at this stage; thus, wouldn't [ɳ] be unlikely, and if present, likely to become /n/?
- The rhotic consonant is a uvular, /ʁ/, so I can have /œ̃/ > [ɚ̃] > [əʁ̃]. But what, then, can I do with the resulting [ʁ̃]? Possibilities:
- I could have it simply become a uvular nasal /ɴ/. But /ʁ/ is the only other uvular consonant, so I don't expect it to be stable, and would likely just become /ŋ/ in short order, yes?
- It could simply denasalize, [ʁ̃] > /ʁ/, so that /œ̃/ > /əʁ/ (Rhoticity "winning" over nasality).
- A third option was by considering having [ʁ̃] become /ʁ/+nasal stop. The first problem would be the place of that nasal (see /ɑ̃/ above; [ŋ], [n], [ɴ] all seem plausible). Second though, would be before a non-semivowel consonant (or word-finally), as this would require the nasal to be in the syllable coda with /ʁ/, and as currently set up, the phonotactics don't allow liquid+nasal coda clusters (nor nasal+liquid onsets; the nasals and liquids are treated as too close in sonority). A possible resolution, would be that before at least a voiced plosive, the resulting nasal, homorganic to the plosive, replaces the plosive: /œ̃b œ̃d œ̃g/ > [əʁ̃b əʁ̃d əʁ̃g] > [əʁᵐb əʁⁿd əʁᵑg] > /əʁm əʁn əʁŋ/. (In all other positions where producing a nasal consonant would be "blocked" by phonotactics, I'd expect denasalization [ʁ̃] > /ʁ/.)
- I could ignore the r-coloring of /œ̃/, and go with diphthongization instead (as one source reports for allophonically long [œ̃ː]): /œ̃/ > [œ̃ʏ̯̃] > /əɥ̃/ or /œɥ̃/. /ɥ̃/ would, clearly, go to /ɲʷ/, which would then almost certainly merge with /ɲ/ (though possibly with the same rounding effect as with the /ŋʷ/ < [w̃] possibility above).