The Great Proto-Indo-European Thread

[Io]

I didn't find an Albanian thread but I saw the language is often mentioned here so I'll ask: is there an explanation that can make this sound change look any less bizarre than it is!?

LL, G

The letters ⟨LL⟩ and ⟨G⟩ are realised as a voiced velar fricative [ɣ] (also found in Greek: γάλα [ˈɣala], 'milk'). [..] In Piana degli Albanesi the tendency is to treat Italian loanwords differently from Sicilian, which results in the difference between llampjun, pronunced as [ʁampˈjun] (from lampione, 'lamp post'), and lampadhin, pronunced as [lampaˈðin] (from Italian lampadina). In the first example, the ⟨L⟩ becomes ⟨LL⟩ [ʁ] because it comes from Sicilian, whereas in the process of transference from the Italian ampadina to Arbëresh lampadhin, the ⟨l⟩ does not change but the ⟨d⟩ becomes [ð].

That's from the article on Arbresh.

[Pole, the]

I don't know, [l → ɫ → ɣ]?

[gach]

With the benefit of not having any expertise on the subject, some thoughts still come to mind:

- If the Italian loan is younger than the Sicilian one, it may have been saved from experiencing the velarisation shift.

- Or perhaps the Italian loan has entered a more literary register and thus had stronger pressure to remain close to its original shape.

- Lastly, what's the possibility of this having anything to do with two subdialects having different rules for handling loans?

[Vijay]

/l/ and /r/ are notoriously unstable when it comes to sound changes. Besides, /l/ > [ɣ] is not that different from /r/ > [ʁ].

[Zaarin]

I don't know, [l → ɫ → ɣ]?

Or ʎ > ʝ > ɣ, assuming <ll> was /ʎ/ at some point...

[Tropylium]

/l/ and /r/ are notoriously unstable when it comes to sound changes.

[citation needed]
Unstable enough to change into something else here and there, sure; but also often quite stable if given the chance.

At any rate, /ɫ/ to /ɣ/ doesn't seem that remarkable to me. /ɫ/ to /w/ is common as dirt (American English, Middle French, Polish, Veps, Permyak…) and we have here just one further step. And, since Albanian has no /w/, this could even have been two changes in a row: /ɫ/ > *w > /ɣ/?

Armenian also has this same thing: the consonant transcribed as ł in Classical Armenian becomes /ʁ/ in Modern Armenian.

There would surely be interesting details in why Sicilian /l/ gets loaned as /ɫ/ > /ɣ/ but Italian /l/ gets loaned as /l/ (e.g. perhaps Sicilian /l/ is phonetically more velarized), but I guess not all that much bizarreness anymore.

[jal]

There would surely be interesting details in why Sicilian /l/ gets loaned as /ɫ/ > /ɣ/ but Italian /l/ gets loaned as /l/

It would also be interesting to know why there are Sicilian loans at all, given tha Albania is not very close to Sicili.


JAL

[Vijay]

There would surely be interesting details in why Sicilian /l/ gets loaned as /ɫ/ > /ɣ/ but Italian /l/ gets loaned as /l/

It would also be interesting to know why there are Sicilian loans at all, given tha Albania is not very close to Sicili.


JAL

Arbëresh?

[jal]

Arbëresh?

Sorry, my bad, didn't look too closely. Thanks for setting me straight.


JAL

[Vijay]

No worries! I didn't look closely, either. I just figured that's probably the right answer in general for any question along the lines of "wtf does this part of southern Italy have to do with Albania?"

[Zju]

This article offers some pretty solid etymologies and it relies on *gʷ...gʷ dissimilating to *gʷ...h₃u - which points to dorsals and laryngals being the following:

<g gʷ h₃> = /ɡ ɡw ɣ/ or /ɡ ɡʷ ɣ/
<k kʷ h₂> = /k kw x/ or /k kʷ x/
<ḱ ǵ> = pre-velars
<ǵʰ gʰ gʷʰ> = breathy voiced or otherwise more marked in some way
<h₁> = /h/ or /ʔ/, not much debate here

Reasoning: evidence seems to suggest that h₂ and h₃ had the same PoA as the plain velars (the whole dissimilation thing), so if the plain velars were uvulars, PIE would have voiced and voiceless uvular fricatives, but no velar fricatives - talk about odd typology*.

OTOH the dissimilation may be something other than voiced stop to a voiced fricative. Your thoughts?

*Yea, I know it's PIE we're talking about, but still.

[Zju]

On another note, has anyone attempted to reconcile *kap- 'seize, grasp' and *gʰebʰ-, *gʰabʰ- 'give', 'seize, take', 'have, hold'? It seems as if it has something to do with the PIE constraint against TVDʰ and DʰVT roots...

[Šọ̈́gala]

OTOH the dissimilation may be something other than voiced stop to a voiced fricative. Your thoughts?

In this model, what causes the different vowel coloring of h₂ vs. h₃?

Note that Carrasquer Vidal has suggested that *k (i.e. not *kʷ or *ḱ) had the same vowel coloring effect as h₂ does. So, this is suggestive that they had the same place of articulation. But, if that place of articulation is velar, then you'd have to explain why *kʷ does not also have the same result. Carrasquer Vidal suggests that *k was phonetically [q], which would make h₂ [X], a popular suggestion. My guess (per several comments above) is that h₃ would have to be further back from that, regardless of whether it was voiced.

[Šọ̈́gala]

On another note, has anyone attempted to reconcile *kap- 'seize, grasp' and *gʰebʰ-, *gʰabʰ- 'give', 'seize, take', 'have, hold'? It seems as if it has something to do with the PIE constraint against TVDʰ and DʰVT roots...

On the face of it, wouldn't the simplest thing be to assume either a morphological or phonetically conditioned alternation between voiced and tenuis sounds, perhaps starting from an originally tenuis form? I seem to recall seeing a list of similar alternations in PIE, but the only example that comes to mind is *ḱérd- "heart" vs. *ǵʰer- "bowels" (Gk. χορδή) and/or *ǵʰer- "enclosure" (Eng. "yard" and "garden"), one of the latter presumably underlying Skt. हृदय hṛdaya "heart".

[Zju]

OTOH the dissimilation may be something other than voiced stop to a voiced fricative. Your thoughts?

In this model, what causes the different vowel coloring of h₂ vs. h₃?

Note that Carrasquer Vidal has suggested that *k (i.e. not *kʷ or *ḱ) had the same vowel coloring effect as h₂ does. So, this is suggestive that they had the same place of articulation. But, if that place of articulation is velar, then you'd have to explain why *kʷ does not also have the same result. Carrasquer Vidal suggests that *k was phonetically [q], which would make h₂ [X], a popular suggestion. My guess (per several comments above) is that h₃ would have to be further back from that, regardless of whether it was voiced.

The different coloring is due to voicing. Assuming an a/ā/i/u initial vowel setup and [ɣ x] for /h₃ h₂/:
1.1. ā > ō, perhaps [ɑː] initially
1.2. Simultaneously with that: a > o / _h₃, h₃_, perhaps [ɑ]
2. o ō > o, perhaps [ɔ] at this stage from earlier [ɒ] < [ɑ], but it might had some residual length
3. Finally, a > æ > e / !_h₂, !h₂_ - that is, when not next to h₂.

Interesting that *k is suggested to have had the same coloring, as just in the next post I posted about *kap- 'seize, grasp' and *gʰebʰ-, *gʰabʰ- 'give', 'seize, take', 'have, hold'. It reminds me of some irregular NAE splits regarding the open vowels.
As for why *kʷ doesn't behave the same - rounding? Or it might not have been a phoneme in the protolanguage, as (correct me if I'm wrong) only Greek and P-Celtic show reflexes different than k or kw - so kw -> kʷ could have happened independently twice and earlier *w would block the coloring of *k.
And I'm still not comfortable with uvular and further back than uvular fricatives, but no velar one. Maybe *h₁ could have been [x] at an earlier stage then? Are there any U or PU borrowings (or cognates) that show any correspondance to *h₁?




Does anyone know how to search wiktionary for PIE words? It has a fair number of reconstructions, but they don't show up in regular searches or in translation sections, only in etymology ones, making searching for a PIE word with particular meaning quite a PITA.

[Šọ̈́gala]

Well, all of that makes a lot of sense to me, but is there any precedent for a voiced fricative producing distinctive vowel coloring? For that matter, is there precedent for rounding blocking a vowel-coloring effect?

About *kʷ, regardless of whether of not it's a unitary phoneme, there seem to be very few if any instances of *kw to compare it to. If you compare it to *ḱw, then naturally all of the satem languages have differing reflexes.

If you go with a traditional-model-plus-/q/ approach, where *ḱ *kʷ *k were phonetically *kʲ *kʷ *q, then you are in the position of not only having no velar fricative, but you have no plain (unrounded) velar stop, which even weirder. But people posit all kinds of unusual arrangements for PIE with the caveat that it was unstable and short-lived, which is why we see it changing into a range of different variants in the attested daughter languages. *kʲ *kʷ *q could very plausibly have developed starting from an earlier stage with *k and the vowels *a *e *i *o *u, then assuming that *a causes uvularisation, the front vowels cause palatalisation, and the rounded vowels cause rounding. Although I don't think I've ever heard anyone suggest it was quite that simple.

It seems to me there could easily have been more than one phoneme that is *h₁, since all it has to do is act like a laryngeal and do no vowel coloring (and when syllabic merge with the most common vowel sound in Hellenic). Perhaps there were /ʔ/ /h/ and /x/ all distinct originally (I don't know what the Uralic evidence has to say about that, though).

Does anyone know how to search wiktionary for PIE words? It has a fair number of reconstructions, but they don't show up in regular searches or in translation sections, only in etymology ones, making searching for a PIE word with particular meaning quite a PITA.

I'm not aware of any way to search roots, but note that they still have nouns all on one big page: https://en.wiktionary.org/wiki/Appendix ... pean_nouns

[Howl]

OTOH the dissimilation may be something other than voiced stop to a voiced fricative. Your thoughts?

In this model, what causes the different vowel coloring of h₂ vs. h₃?

Note that Carrasquer Vidal has suggested that *k (i.e. not *kʷ or *ḱ) had the same vowel coloring effect as h₂ does. So, this is suggestive that they had the same place of articulation. But, if that place of articulation is velar, then you'd have to explain why *kʷ does not also have the same result. Carrasquer Vidal suggests that *k was phonetically [q], which would make h₂ [X], a popular suggestion. My guess (per several comments above) is that would have to be further back from that, regardless of whether it was voiced.

I would suggest the bilabial fricative (ɸ) for reconstructing h₃. It is not in the velar region, but why does it have to be? It is less marked than the uvular fricative (X) proposed for h₂, which explains why h₂ has more direct reflexes (the 'h' in Hittite). And it would definitely cause rounding. The development of the vowel coloring would look something like this: ɸɛ -> ɸœ -> ɸɔ

[Pole, the]

Well, all of that makes a lot of sense to me, but is there any precedent for a voiced fricative producing distinctive vowel coloring?

Yes, see German /x/ vs /ʁ/.

For that matter, is there precedent for rounding blocking a vowel-coloring effect?

Yes, see English /ɛr ɪr ʊr → ɜr/ vs /ɔr → ɔr/.

If you go with a traditional-model-plus-/q/ approach, where *ḱ *kʷ *k were phonetically *kʲ *kʷ *q, then you are in the position of not only having no velar fricative, but you have no plain (unrounded) velar stop, which even weirder.

The hypothetical chain shift /k kʷ q/ → /kʲ kʷ q/ → /kʲ kʷ k/ doesn't strike me as particularly implausible.

[Zaarin]

then you are in the position of not only having no velar fricative, but you have no plain (unrounded) velar stop, which even weirder.

It's not unattested. Many languages of the PNW have no plain velar stop.

[Šọ̈́gala]

Well, all of that makes a lot of sense to me, but is there any precedent for a voiced fricative producing distinctive vowel coloring?

Yes, see German /x/ vs /ʁ/.

Those would appear to have different places of articulation.

For that matter, is there precedent for rounding blocking a vowel-coloring effect?

Yes, see English /ɛr ɪr ʊr → ɜr/ vs /ɔr → ɔr/.

Well, that's a case where the roundedness of the vowel itself blocks vowel coloring. It's beyond controversy that some PIE vowels were immune to coloring. I'd curious to see an example where the roundedness of the consonant neutralizes vowel coloring.

[Šọ̈́gala]

I would suggest the bilabial fricative (ɸ) for reconstructing h₃. It is not in the velar region, but why does it have to be? It is less marked than the uvular fricative (X) proposed for h₂, which explains why h₂ has more direct reflexes (the 'h' in Hittite). And it would definitely cause rounding. The development of the vowel coloring would look something like this: ɸɛ -> ɸœ -> ɸɔ

For a convincing account of why h₂ and h₃ cause vowel coloring, one would want to explain why they are different from other PIE consonant phonemes. If h₃ causes coloring to /ɔ/ because h₃ is rounded, then why don't other rounded consonants have the same effect? You'd think that at least/w/ would if not the stops. That's why I have tended to assume that h₂ and h₃ both have unique places of articulation, although h₂ could well be same as for *k (as well as *g and *gʰ???).

[Howl]

I would suggest the bilabial fricative (ɸ) for reconstructing h₃. It is not in the velar region, but why does it have to be? It is less marked than the uvular fricative (X) proposed for h₂, which explains why h₂ has more direct reflexes (the 'h' in Hittite). And it would definitely cause rounding. The development of the vowel coloring would look something like this: ɸɛ -> ɸœ -> ɸɔ

For a convincing account of why h₂ and h₃ cause vowel coloring, one would want to explain why they are different from other PIE consonant phonemes. If h₃ causes coloring to /ɔ/ because h₃ is rounded, then why don't other rounded consonants have the same effect? You'd think that at least/w/ would if not the stops. That's why I have tended to assume that h₂ and h₃ both have unique places of articulation, although h₂ could well be same as for *k (as well as *g and *gʰ???).

But this is a weakness of the whole laryngeal theory. You have these magical laryngeals that do vowel coloring and then mostly disappear in the daughter languages. And you have a lot of other consonants, none of which do any vowel coloring at all. So these laryngeals really had to be different from the ordinary consonants. But there is only a limited number of places in the mouth to articulate from. And it is really hard to come up with consonants that are not remotely like the ones PIE already uses (like velars and probably also uvulars).

Even Carrasquer Vidal's 'qa' is not really vowel coloring, since there still has to be a 'qe' -> 'ke'. So what he does with 'qa' is vowel preservation. But maybe the whole point of the laryngeals was also some form of vowel preservation.

[Šọ̈́gala]

But this is a weakness of the whole laryngeal theory. You have these magical laryngeals that do vowel coloring and then mostly disappear in the daughter languages.

It's definitely a tricky business positing phonemes that have no segmental reflexes. But, on the other hand, is it really so surprising that a language would have one or more consonant phonemes that are easily dropped? (see per Janhunen below). In fact, I'd wager that this is more likely to happen if they are causing assimilation effects in neighboring vowels. This reduces the semantic load on those consonants themselves, which makes people more willing to elide them.

And you have a lot of other consonants, none of which do any vowel coloring at all. So these laryngeals really had to be different from the ordinary consonants. But there is only a limited number of places in the mouth to articulate from. And it is really hard to come up with consonants that are not remotely like the ones PIE already uses (like velars and probably also uvulars).

I would suggest uvular for the a-coloring consonants (probably including *k), which leaves epiglottal~pharyngeal for o-coloring.

Even Carrasquer Vidal's 'qa' is not really vowel coloring, since there still has to be a 'qe' -> 'ke'. So what he does with 'qa' is vowel preservation. But maybe the whole point of the laryngeals was also some form of vowel preservation.

I'm not 100% sure what your background assumptions here are. I can't give a full account of MCV's theory, but what I'm picturing is something like the traditional model with a couple tweaks, basically *k = [q] and the basic ablaut vowel was originally /ə/ or /a/ rather than /e/. In that case, there would be no instances of *qe-. There would be *ḱa, *kʷa, *qa (or *ḱə, *kʷə, *qə) > *ḱe, *kʷe, *qa, comparable to *h₁a, *h₂a > *e, *a. You're right that the reflex *a after *q or *h₂ is a vowel preservation rather than a shift in that scenario.

Janhunen http://www.sgr.fi/sust/sust253/sust253_janhunen.pdf has some typological observations about "laryngeals", which he defines loosely as velars and postvelars other than velar stops and nasals. He suggests they are typically labile in terms of place of articulation and easily dropped. If PIE had at two or more stops/fricatives/approximants in postvelar places of articulation, that would be a phonological system pretty similar to, say, a normal Semitic language (not commenting on a genetic or areal link to Semitic, just pointing out the precedent). If those consonants later lenited to Ø, that doesn't seem like a very unusual development.

If we go with the idea that *k and *h₂ had the same place of articulation, it's striking that *h₂ is so common and *k so rare. I wonder if original [*q] became a fricative generally and was preserved as a stop either sporadically or under some unknown conditions.

[Salmoneus]

If we go with the idea that *k and *h₂ had the same place of articulation, it's striking that *h₂ is so common and *k so rare. I wonder if original [*q] became a fricative generally and was preserved as a stop either sporadically or under some unknown conditions.

If it really were true that *k caused colouring (is it? how solid is that?), then this might be explained by fricative mergers?

If we started off with /k q x X/, it's not unreasonable that the two fricatives might merge without the plosives merging, and that this merger might be to uvular. Indeed, even if we start with /k x q/, the /x/ could back all by itself. As you point out yourself, laryngeals are unstable and prone to moving (and I think merging).

[KathTheDragon]

is it? how solid is that?

Imo it's not even a thing. The evidence usually cited in support is that there are many instances of *a adjacent to *k - which is true - but then it's controversial whether *a even exists, and you still find *e adjacent to *k anyway.