On Jun 23, 2013, at 3:07 PM, Alex Fink <[log in to unmask]> wrote: > On Sat, 22 Jun 2013 12:16:05 -0500, Eric Christopherson <[log in to unmask]> wrote: > >> I remember reading in descriptions of at least two languages that, in those particular languages, vowels separated by glottal sounds (i.e. [?] or [h]) often undergo total assimilation or, in at least one of those languages, metathesis. > > Not much of concrete value here, but: > >> 1. Is this fairly common cross-linguistically? > > I've noticed the pattern, too. Though your case of metathesis (assuming you mean of the vowels) surprises me; I can't think of even an irregular example of V-V metathesis right now. That was in Saanich (Timothy Montler's grammar). As I recall, stress accent comes into play there too -- the V-glottal-V metathesis always involves a schwa which would be stressed and a less neutral vowel, and there seems to be some statistical disfavoring of stressed schwas there, although they are still pretty common. There is also ablaut causing alternation of schwas and fuller vowels, so it's really complex. > >> 2. Is there any reason to suppose transglottal interactions of these sorts would be more likely (either in specific languages or crosslinguistically) than in pure sequences of two vowels? > > Like what you and Roger have said, it is often sensible to view [? h] as pure phonation (and thus "not quite consonantal" if consonants require an oral obstruction) -- this is so in many feature systems, but also just articulatorily, as the supra-glottal part of the vocal tract doesn't have anything to do in making [? h]. As such, if [? h] occur adjacent to a vowel, the tongue body (and rounding, etc.) can just retain their vocalic position during the [? h]; and if they occur adjacent to tw˛ vowels, one each side, then the vocalic gestures can reach through the [? h] and influence each other. > > None of that, however, makes these interactions *more* likely than interactions of two directly adjacent vowels. I don't know, and would be really surprised to learn, of any language with V+glottal+V assimilations that lets (phonemic) vowel sequences surface unharmed. OK. I was doubtful, but it would be really helpful for this one diachronic conlang of mine. I was thinking maybe something of the glottal stricture could influence the vowel qualities (but I understand that's really rare compared to the influence of uvular or pharyngeal or epiglottal stricture). Could I plausibly fudge and say that VV sequences would end up with one of the Vs as a glide, and say vowel/glide sequences *don't* assimilate? > >> 2a. Or two vowels separated by some other kind of consonant? > > Well, that's your everyday umlaut / vowel harmony / etc. I don't know of any case of umlaut that is blocked by glottals. I would imagine that the reason we don't see t˛tal assimilation across arbitrary consonants (as we do across glottals) very often is because that would be too harsh a loss of contrastiveness. Ah -- actually I was asking if there would be any reason that V-glottal-V assimilation would be more likely than general VCV assimilation. My guess is yes, because in reality the two vowels really are adjacent.