Question on PIE laryngeals

[linguofreak]

I was looking through the Wikipedia article on PIE today, and noticed that:

1. There are three sets of dorsal stops (/ḱ/ /k/ /kʷ/, etc).

2. There are three laryngeals.

3. One of the dorsal stop series was labialized.

4. h₃ was most likely labialized, as it o-colored adjacent vowels.

In light of the above, is it plausible to reconstruct pronunciations for the laryngeals such that each of the three shared a place of articulation with one of the dorsal stop series? Thus if we take the values of /ḱ/ /k/ /kʷ/ to have been [k] [q] [qʷ], might the laryngeals be reconstructed as: /h₁/ = [x] /h₂/ = [χ] /h₃/ = [χʷ] ?

If plausible, this would create a symmetry in the phonology of PIE by making the fricatives (mostly) match the stops in POA. OTOH, the fricative series would still lack /f/, so the symmetry still wouldn't be complete.

[TaylorS]

That's exactly my own thinking on PIE laryngeals..

[vec]

Having a q^w rather than a k^w seems odd too me.

Having k q k^w and x χ x^w seems like a better bet to me.

[alice]

This has occurred to me too. gsandi, who believes in four laryngeals, probably has other ideas though, and it'd be interesting to hear them in this context.

[Thomas Winwood]

I too prefer /x χ xʷ/, but it's the same basic principle and I'd reached the same conclusion independently.

So yes, they might.

[spats]

Having a q^w rather than a k^w seems odd too me.

Having k q k^w and x χ x^w seems like a better bet to me.

Not sure. Especially if the stops aren't entirely velar/uvular but rather prevelar and postvelar. Labial coarticulation is easier for dorsal consonants (hence the greater occurrence of sounds like kʷ, χʷ, qʷ than say tʷ or sʷ). Also, since palatal/bilabial coarticulation is relatively rare, a more fronted velar might partially block it. So you have k̟ k̠ k̠ʷ instead.

Also, from what I've read, there is some argument that /h/ and/or /?/ may also have been h₁, and at least some of the laryngeals may have been voiced. So... it's probably more complicated than that, though I would assume that at one point there probably was a full series of of fricatives at those points of articulation, whatever they were, but it evolved and merged with other sounds over time.

[WeepingElf]

That's exactly my own thinking on PIE laryngeals..

Mine too. It is difficult to ascertain, though, where exactly *ḱ and *k (and thus, *h1 and *h2 respectively) were pronounced, so I prefer to call them agnostically "front velars" and "back velars". It seems certain, though, that the PIE labiovelars were labialized back rather than front velars.

[spats]

It seems certain, though, that the PIE labiovelars were labialized back rather than front velars.

I'd tend to agree with you, but is there any hard evidence? If the labiovelars were of non-specific articulation but the labialization blocked palatalization in the Satem languages, then the result would have been the same, no? Or is there a descendant which preserves all three series and groups the dorsovelars with the labiovelars?

(Wikipedia indicates that some of the early descendants, specifically Proto-Anatolian, might have preserved all three series, but no information is given as to their realization.)

I should also add that I believe k'w (as opposed to k_w) did palatalize in Satem, so that's a pretty good argument against my devil's advocate above as well.

[AnTeallach]

This has occurred to me too. gsandi, who believes in four laryngeals, probably has other ideas though, and it'd be interesting to hear them in this context.

As I understand it gsandi doesn't really believe in the three dorsal stop series. But he does use some similar arguments in arriving at his preferred reconstruction of the laryngeals on his webpage:
http://www.tundria.com/Linguistics/pie-phonology.shtml

[Soap]

Words like kwon > čwon might have been due to the analogy of vowelless forms such as kwn > kun > čun. Still though I am skeptical of all k-q-qʷ type systems, as it seems unnatural to have qʷ without kʷ, and even more so Gʷ without gʷ.

[linguofreak]

Also, from what I've read, there is some argument that /h/ and/or /?/ may also have been h₁, and at least some of the laryngeals may have been voiced. So... it's probably more complicated than that, though I would assume that at one point there probably was a full series of of fricatives at those points of articulation, whatever they were, but it evolved and merged with other sounds over time.

A question that strikes me is "what was the range of allophony?". Even without going into diachronics, for all we know, /h₁/ may primarily have been [x], but have had [h] and [ɣ] as allophones.

Another question is "Given this arrangement, what happened to /f/ in the fricative series? Did it ever exist? Can we solve any open problems by postulating an /f/ that disappeared after affecting its environment in some way?".

[WeepingElf]

It seems certain, though, that the PIE labiovelars were labialized back rather than front velars.

I'd tend to agree with you, but is there any hard evidence? If the labiovelars were of non-specific articulation but the labialization blocked palatalization in the Satem languages, then the result would have been the same, no? Or is there a descendant which preserves all three series and groups the dorsovelars with the labiovelars?

You are right; it is not entirely certain that the labiovelars were articulated as far back as the back velars, but why should they not? We know that they merged with the back velars in the satem languages, but in no IE language they merged with the front velars.

(Wikipedia indicates that some of the early descendants, specifically Proto-Anatolian, might have preserved all three series, but no information is given as to their realization.)

Luwian has *ḱ > ts, *k > k and *k^w > kw (Hittite has *ḱ > k).

I should also add that I believe k'w (as opposed to k_w) did palatalize in Satem, so that's a pretty good argument against my devil's advocate above as well.

It indeed did, e.g. Sanskrit aśva < PIE *eḱwos 'horse'.

EDIT:
@linguofreak: *f may have merged with *h3, especially if it was bilabial rather than labiodental.

[Jetboy]

Any particular reason for /k q qʷ/ instead of /c k kʷ/?

You are right; it is not entirely certain that the labiovelars were articulated as far back as the back velars, but why should they not? We know that they merged with the back velars in the satem languages, but in no IE language they merged with the front velars.

Don't a lot of the Romance languages?

[Thomas Winwood]

Any particular reason for /k q qʷ/ instead of /c k kʷ/?

Some people presume areal influence from PNC which had uvulars.

[Soap]

Any particular reason for /k q qʷ/ instead of /c k kʷ/?

Some people presume areal influence from PNC which had uvulars.

That and it's very rare for a language that has palatal stops to change them to velar unconditionally, whereas uvular --> velar is quite common. I found a counterexample once — Danish — but it's not clear whether proto-North Germanic actually had palatal stops or just palatalzied velars.

[linguofreak]

@linguofreak: *f may have merged with *h3, especially if it was bilabial rather than labiodental.

I was thinking along those lines. Are there any irregularities in the behaviour of h3 that could be explained by such a merger?

[Jetboy]

Any particular reason for /k q qʷ/ instead of /c k kʷ/?

Some people presume areal influence from PNC which had uvulars.

That and it's very rare for a language that has palatal stops to change them to velar unconditionally, whereas uvular --> velar is quite common. I found a counterexample once — Danish — but it's not clear whether proto-North Germanic actually had palatal stops or just palatalzied velars.

On the other hand, is it possible that PIE had palatalized velars for <ḱ>?

[TaylorS]

Words like kwon > čwon might have been due to the analogy of vowelless forms such as kwn > kun > čun. Still though I am skeptical of all k-q-qʷ type systems, as it seems unnatural to have qʷ without kʷ, and even more so Gʷ without gʷ.

I swear I have seen some Amerindian languages with qʷ but no kʷ

In my English descendant I have a series of sound changes that turn pharyngealized labials into labialized uvulars, but there are no labialized velars, so English /bre:k/ "break" becomes Mekoshan /qʷex/ "brekh".

[WeepingElf]

Any particular reason for /k q qʷ/ instead of /c k kʷ/?

You are right; it is not entirely certain that the labiovelars were articulated as far back as the back velars, but why should they not? We know that they merged with the back velars in the satem languages, but in no IE language they merged with the front velars.

Don't a lot of the Romance languages?

What I was referring to was a hypothetical development where *ḱ and *k^w merge while *k remains distinct. That happened nowhere. Of course, many Romance languages delabialized the labiovelars in some or all environments, merging them with the unlabialized velars, which, as these languages are "kentum", continue both front and back velars.