The Great Proto-Indo-European Thread

[JounaPyysalo]

Kath,

I have a great confidence to the rules in question, because the related correspondences indicate that indeed there is a vowel PIE *ɑ = Neogr. *ǝ = de Saussure *A present in the correspondences in question.

I don't actually think there's too much doubt about the rules, because – to quote the our recently upgraded code, only improving the displayment of the rules – the very first four rules are the "colouring rules" of the Laryngeal Theory that basically everyone agrees upon these days:

define Rɑe›ɑa e -> a || ɑ _ ; # PIE *ɑe → ɑa | Møller's rule for PIE *ɑe | (SPIE §2.2.5.3)
define Rɑé›ɑá é -> á || ɑ _ ; # PIE *ɑé → ɑá | Møller's rule for PIE *aé | (SPIE §2.2.5.3)
define Reɑ›aɑ e -> a || _ ɑ ; # PIE *eɑ → aɑ | De Saussure's rule for PIE *eɑ | (SPIE §2.2.7)
define Réɑ›áɑ é -> á || _ ɑ ; # PIE *éɑ → áɑ | De Saussure's rule for PIE *éɑ | (SPIE §2.2.7)

The rest of it is just the same for the long vowels *ē and *ḗ, which – as no one has bothered to claim this before – I do myself:

define Rɑē›ɑā ē -> ā || ɑ _ ; # PIE *ɑē → ɑā | Pyysalo's rule for PIE *ɑē | (SPIE §2.2.10)
define Rēɑ›āɑ ē -> ā || _ ɑ ; # PIE *ēɑ → āɑ | Pyysalo's rule for PIE *ēɑ | (SPIE §2.2.10)
define Rɑḗ›ɑā́ ḗ -> ā́ || ɑ _ ; # PIE *ɑḗ → ɑā́ | Pyysalo's rule for PIE *ɑḗ | (SPIE §2.2.10)
define Rḗɑ›ā́ɑ ḗ -> ā́ || _ ɑ ; # PIE *ḗɑ → ā́ɑ | Pyysalo's rule for PIE *ḗɑ | (SPIE §2.2.10)

In this manner everyone is doing the same...

Jouna

[Richard W]

The fourth series is no longer accepted, which renders your point moot.

However, isn't the argument that Brugmannian *Th is an unnecessary hypothesis? Sequencing changes is not always easy, though I think it is commoner to project changes too far back.

[KathTheDragon]

The fourth series is no longer accepted, which renders your point moot.

However, isn't the argument that Brugmannian *Th is an unnecessary hypothesis? Sequencing changes is not always easy, though I think it is commoner to project changes too far back.

The simple fact is that there is independent evidence for a laryngeal immediately following the stop in virtually every case of an Indo-Iranian voiceless aspirate. A common example is sthita- < *sth₂-tó-, also Gr. στατος, La status, Eng. "stood" (with analogical full grade). There are a few apparent cognates where both Greek and Indic voiceless aspirates match voiceless stops, but there are probably alternate explanations.

[JounaPyysalo]

Kath & why not others,

In PIE Lexicon we use a set of revised IE sound laws in which the errors of the earlier suggestions have been removed to the effect that we are capable of generating the IE data in a nearly flawless manner. In the case of the root PIE *stɑh- ‘stehen’ we've corrected Saussure's overstated compensatory lengthening rule DS *eA -› aA -› ā and replaced it with the correct one, viz. PIE *eɑ -› aɑ -› a.

Consequently,

1. The zero grade of the root PIE *stɑh- appears in the stem RV. tasth- (pf.) ‘stehen’ (RV. tasthur [3pl]) with loss of PIE *ɑ (former ‘schwa’).

2. The *e-grade of the root, PIE *steɑh- appears in Cret. στανύω (pr.) ‘einsetzen’ : LAv. fra·stanva- (pr.) ‘vorankommen, Fortschritte machen’ (AIWb. 1604, frastanvanti), both from PIE *steɑhnwe/o- (i.e. without compensatory lengthening).

3. The long *ē-grade of the root, PIE *stēɑh-, appears in AV. nari·ṣṭā- (f.) ‘Scherz, Geplauder’ (EWA 2:22), Do. στᾱ- ≡ Li. stó- ≡ Lat. stā- etc.

J.

[Richard W]

The simple fact is that there is independent evidence for a laryngeal immediately following the stop in virtually every case of an Indo-Iranian voiceless aspirate.

How does that show that the change doesn't go back to PIE?

[KathTheDragon]

Kath & why not others,

In PIE Lexicon we use a set of revised IE sound laws in which the errors of the earlier suggestions have been removed to the effect that we are capable of generating the IE data in a nearly flawless manner. In the case of the root PIE *stɑh- ‘stehen’ we've corrected Saussure's overstated compensatory lengthening rule DS *eA -› aA -› ā and replaced it with the correct one, viz. PIE *eɑ -› aɑ -› a.

Consequently,

1. The zero grade of the root PIE *stɑh- appears in the stem RV. tasth- (pf.) ‘stehen’ (RV. tasthur [3pl]) with loss of PIE *ɑ (former ‘schwa’).

2. The *e-grade of the root, PIE *steɑh- appears in Cret. στανύω (pr.) ‘einsetzen’ : LAv. fra·stanva- (pr.) ‘vorankommen, Fortschritte machen’ (AIWb. 1604, frastanvanti), both from PIE *steɑhnwe/o- (i.e. without compensatory lengthening).

3. The long *ē-grade of the root, PIE *stēɑh-, appears in AV. nari·ṣṭā- (f.) ‘Scherz, Geplauder’ (EWA 2:22), Do. στᾱ- ≡ Li. stó- ≡ Lat. stā- etc.

J.

Oh. My. Fucking. God. How dense do you have to be to continue to claim that you are in possession of the Truth™, in a discipline like historical linguistics? I frankly don't care how amazing you think your solution is, but I am yet to be convinced it has any merits, coming out of a methodology that's on a similar level to mass comparativisitics. Your methods are wholly arbitrary and betray the whole point of the comparative method! I'm sorry you're receiving less than the praise you believe you deserve, but you come across as little more than a crackpot to me.

[KathTheDragon]

The simple fact is that there is independent evidence for a laryngeal immediately following the stop in virtually every case of an Indo-Iranian voiceless aspirate.

How does that show that the change doesn't go back to PIE?

Occam's Razor. In this case, the simplest hypothesis is that the specific sound change was an Indo-Iranian innovation. There is no need to project it back to PIE, since not a single other branch requires it. Doing so makes you guilty of the Sanskritocentricism that plagued early IE studies.

[JounaPyysalo]

Kath & why not others,

In PIE Lexicon we use a set of revised IE sound laws in which the errors of the earlier suggestions have been removed to the effect that we are capable of generating the IE data in a nearly flawless manner. In the case of the root PIE *stɑh- ‘stehen’ we've corrected Saussure's overstated compensatory lengthening rule DS *eA -› aA -› ā and replaced it with the correct one, viz. PIE *eɑ -› aɑ -› a.

Consequently,

1. The zero grade of the root PIE *stɑh- appears in the stem RV. tasth- (pf.) ‘stehen’ (RV. tasthur [3pl]) with loss of PIE *ɑ (former ‘schwa’).

2. The *e-grade of the root, PIE *steɑh- appears in Cret. στανύω (pr.) ‘einsetzen’ : LAv. fra·stanva- (pr.) ‘vorankommen, Fortschritte machen’ (AIWb. 1604, frastanvanti), both from PIE *steɑhnwe/o- (i.e. without compensatory lengthening).

3. The long *ē-grade of the root, PIE *stēɑh-, appears in AV. nari·ṣṭā- (f.) ‘Scherz, Geplauder’ (EWA 2:22), Do. στᾱ- ≡ Li. stó- ≡ Lat. stā- etc.

J.

Oh. My. Fucking. God. How dense do you have to be to continue to claim that you are in possession of the Truth™, in a discipline like historical linguistics? I frankly don't care how amazing you think your solution is, but I am yet to be convinced it has any merits, coming out of a methodology that's on a similar level to mass comparativisitics. Your methods are wholly arbitrary and betray the whole point of the comparative method! I'm sorry you're receiving less than the praise you believe you deserve, but you come across as little more than a crackpot to me.

Kath,

The comparative method of reconstruction is the gold standard, not subject to historical linguistics, the latter rather a subset of the CM.

Also, patience, we are digitising the decision method of Indo-European etymology, which will not only confirm the etymologies presented, but enable their identification with an algorithm. Indeed there is Truth™ in comparative method of reconstruction.

J.

[Richard W]

The simple fact is that there is independent evidence for a laryngeal immediately following the stop in virtually every case of an Indo-Iranian voiceless aspirate.

How does that show that the change doesn't go back to PIE?

Occam's Razor. In this case, the simplest hypothesis is that the specific sound change was an Indo-Iranian innovation. There is no need to project it back to PIE, since not a single other branch requires it. Doing so makes you guilty of the Sanskritocentricism that plagued early IE studies.

You must add a parallel change in Greek, which adds an entity. There are hints in Latin, which others prefer to see as a -dʰlom instrument suffix.

I don't think Occam's Razor is reliable when it comes to parallel changes amongst related languages. One generally needs discrepancies to reveal that the changes are not a common inheritance; the likelihood of the change probably is a common inheritance. The vanishing of the unvocalised laryngeals in Indo-European languages is a set of parallel changes, but careful examination reveals that the changes were indeed parallel rather than inherited.

How surprising is the loss of a set of rare phonemes? One might naively think it was quite likely, and therefore Occam's Razor doesn't really help. Anyway, if you need Occam's Razor for an argument, then you haven't any evidence. You end up with 'conceivably, but not likely'.

[Pabappa]

Are you saying PIE could have had four stop series after all, based on the fact that in Greek PIE tH merges with PIE /dʰ/ whereas in most other branches it merges with /t/ ? There is a logic to that, but i;d think mainstream scholars must have some other reason why it wasnt phonemic in PIE.

[Richard W]

.... some other reason why it wasn't phonemic in PIE.

So far as I'm aware, aspiration wasn't phonemic in Classical Sanskrit, in as much as there were no clusters of plain stop + /h/ (or plain stop + /ɦ/) to contrast with. There may have been a contrast in Vedic Sanskrit, in that /D/+laryngeal made position but /Dʰ/ didn't - both are recorded as /Dʰ/. Now, the real reason for denying PIE *Th may be that it is more profitable to analyse it as /T/ + /h2/, but that doesn't imply that it wasn't real. And, of course, it is always possible that different speakers had different phoneme systems!

[KathTheDragon]

You must add a parallel change in Greek, which adds an entity. There are hints in Latin, which others prefer to see as a -dʰlom instrument suffix.

I don't think Occam's Razor is reliable when it comes to parallel changes amongst related languages. One generally needs discrepancies to reveal that the changes are not a common inheritance; the likelihood of the change probably is a common inheritance. The vanishing of the unvocalised laryngeals in Indo-European languages is a set of parallel changes, but careful examination reveals that the changes were indeed parallel rather than inherited.

How surprising is the loss of a set of rare phonemes? One might naively think it was quite likely, and therefore Occam's Razor doesn't really help. Anyway, if you need Occam's Razor for an argument, then you haven't any evidence. You end up with 'conceivably, but not likely'.

I was under the impression that Greek generally did not generate aspirates from *TH clusters, or if it did, the input was much more limited than Indo-Iranian. Although, I just remembered another reason why this must necessarily be an independant innovation: Brugmann's Law is blocked by PIE *TH clusters, proving that at the time of operation, it was still a cluster.

[WeepingElf]

Hi @ all,

I see that there is a lot of discussion on the glottalic hypothesis with assumptions more or less equaling to *T(h) T' D(h).

Now, I am not sure whether any of you has actually seen the consequences of such a theory in reality, and if not, please read the book of T.V. Gamkrelidze and V. V. Ivanov:
"Indo-European and the Indo-Europeans: A reconstruction and Historical Analysis of a Proto-Language and a Proto-Culture. Part I: The Text."
(Mouton de Gruyter Berlin-New York 1995.

In so doing you will notice that the output is very close to a total chaos as it is not nearly always obvious where the 'reconstructions' actually refer to.

J.

You do know that the glottalic hypothesis is really just a renaming of the three stop series? That is, traditional T == glottalic T, traditional D == glottalic T', traditional Dh == glottalic D. One-to-one correspondence. As far as real comparativistics goes, the two theories are absolutely equivalent, and the only real evidence either way is diachronic plausibility.

Yes. And in my humble opinion, the trajectories from PIE to the daughter languages become more complex if one assumes the glottalic hypothesis to describe the latest stage of PIE. Hence, while I consider it plausible that it describes some stage of PIE, I don't think this stage is the latest before break-up, but an earlier one, which works just as well.

[Richard W]

I was under the impression that Greek generally did not generate aspirates from *TH clusters, or if it did, the input was much more limited than Indo-Iranian. Although, I just remembered another reason why this must necessarily be an independant innovation: Brugmann's Law is blocked by PIE *TH clusters, proving that at the time of operation, it was still a cluster.

Now, what is Brugmann's law? One view is that it is, "PIE *o remains long in open syllables in Indo-Iranian." PIE *o in closed syllables could have been short by the break-up of PIE!

*TH clusters look very messy in general. I thought they were the origin of the aspirated Greek second perfects.

I found an interesting hand-out on the matter at https://doclib.uhasselt.be/dspace/bitst ... t%20th.pdf . The issue rumbles on, even if the majority deny PIE tʰ. I believe a majority also reject Brugmann's Law.

[KathTheDragon]

I can't see why people would deny Brugmann's Law, since there isn't really any viable alternative for all those ās where you otherwise wouldn't expect them. Apparent exceptions can be easily attributed to analogies of various kinds. Unless you can cite another explanation?

Assuming Brugmann's Law, it can easily be demonstrated that the immediate reflex of *h₃e remained distinct from *o, given that only the latter could undergo Brugmann's Law in IIr. and the latter was also lengthened under the stress in Anatolian. Based on these two environments (open syllables and stressed syllables), it's clear that *o was originally phonetically longer than *h₃e, but probably otherwise identical in quality (hence their otherwise total merger). Therefore, any kind of shortening in dialectal PIE must have taken place once the family was beginning to break up, and therefore, Late/Dialectal PIE at best.

It is an interesting hand-out you found, but I'm hardly convinced that the examples cited prove a whole series of voiceless aspirates. At best, there are possible instances of *tʰ, but none of the examples are clinching. Besides, there are clear indications that Greek and IIr. shared innovations (the verbal system, for one - all the other branches have rather different systems, iirc), so it's entirely possible that voiceless aspirates were another shared innovation. Until more convincing evidence to the contrary can be found, I'll continue to assume that PIE did not have voiceless aspirates.

[Richard W]

I favour Brugmann's law myself. Maybe there's an inherent problem with lengthening rules. Winter's Law has its problems, and I don't recall seeing an explanation of the late lengthening in English water and father or of the short vowel in the English noun wind.

Until more convincing evidence to the contrary can be found, I'll continue to assume that PIE did not have voiceless aspirates.

My point was that *Tʰ was unproven rather than disproven.

[KathTheDragon]

Sadly, it's very hard to produce an actual disproof of something.

[Tropylium]

in my humble opinion, the trajectories from PIE to the daughter languages become more complex if one assumes the glottalic hypothesis to describe the latest stage of PIE. Hence, while I consider it plausible that it describes some stage of PIE, I don't think this stage is the latest before break-up, but an earlier one, which works just as well.

The addition of earlier stages also amounts to "addition of complexity" to the chain of sound changes, though.

At best, there are possible instances of *tʰ, but none of the examples are clinching. Besides, there are clear indications that Greek and IIr. shared innovations (the verbal system, for one - all the other branches have rather different systems, iirc), so it's entirely possible that voiceless aspirates were another shared innovation.

Some cases might not even be common inheritance at all. There are numerous later potentially areal developments common between Greek and Indo-Iranian (*N̥ > a, *s > h, Grassmann's Law), and this may also involve vocabulary; say, early Scythian loans in proto-Greek?

Has anyone proposed reconstructing *dʰh₂ though? Devoicing to *th₂ would be an easy subsequent development, while this would also straightforwardly allow deriving voiceless aspirates in Greek.

The comparative method of reconstruction is the gold standard, not subject to historical linguistics, the latter rather a subset of the CM.

If I'm reading this right, this looks like it might be Jouna's biggest problem. It sounds like he is not trying to actually reconstruct history, but rather to build some kind of an axiomatized data compression algorithm that exists in vacuum.

The actual ontology is quite clearly comparative method ∈ historical linguistics ∈ sociolinguistics (and further ∈ sociology ∈ anthropology ∈ biology ∈ biochemistry ∈ …)

[KathTheDragon]

At best, there are possible instances of *tʰ, but none of the examples are clinching. Besides, there are clear indications that Greek and IIr. shared innovations (the verbal system, for one - all the other branches have rather different systems, iirc), so it's entirely possible that voiceless aspirates were another shared innovation.

Some cases might not even be common inheritance at all. There are numerous later potentially areal developments common between Greek and Indo-Iranian (*N̥ > a, *s > h, Grassmann's Law), and this may also involve vocabulary; say, early Scythian loans in proto-Greek?

Has anyone proposed reconstructing *dʰh₂ though? Devoicing to *th₂ would be an easy subsequent development, while this would also straightforwardly allow deriving voiceless aspirates in Greek.

Did Greek even have laryngeal aspiration?

[Tropylium]

At best, there are possible instances of *tʰ, but none of the examples are clinching. Besides, there are clear indications that Greek and IIr. shared innovations (the verbal system, for one - all the other branches have rather different systems, iirc), so it's entirely possible that voiceless aspirates were another shared innovation.

Some cases might not even be common inheritance at all. There are numerous later potentially areal developments common between Greek and Indo-Iranian (*N̥ > a, *s > h, Grassmann's Law), and this may also involve vocabulary; say, early Scythian loans in proto-Greek?

Has anyone proposed reconstructing *dʰh₂ though? Devoicing to *th₂ would be an easy subsequent development, while this would also straightforwardly allow deriving voiceless aspirates in Greek.

Did Greek even have laryngeal aspiration?

To be clear, I am not suggesting laryngeal aspiration as an areal Greek-II feature, I am suggesting loanwords that already had voiceless aspirates as an areal Greek-II feature.

[KathTheDragon]

Then wherefore the second line?

[WeepingElf]

in my humble opinion, the trajectories from PIE to the daughter languages become more complex if one assumes the glottalic hypothesis to describe the latest stage of PIE. Hence, while I consider it plausible that it describes some stage of PIE, I don't think this stage is the latest before break-up, but an earlier one, which works just as well.

The addition of earlier stages also amounts to "addition of complexity" to the chain of sound changes, though.

Sure, but less of that. I was speaking of the "classical" glottalic hypothesis with ejectives as proposed by Gamkrelidze and Ivanov (because implosives don't solve the *b problem at least), and positing that for Late PIE would mean adding parallel extra sound changes to each branch of the family. Positing it for a pre-stage would necessitate the "extra" sound changes happening only once, namely in Early PIE. And you agree with me that PIE did not fall from the sky fully formed but had a history itself that we may be able to reconstruct, no?

[Tropylium]

So, on a whim I've extracted the list of PIE(ish) verb roots from LIV

Looking thru the root-final clusters, a few Weird PIE Things:
1) There are no examples of root-final *-sT-, but about five of *-sD-.
2) Examples of root-final *-PH- are quite rare other than *-th₂- (which is common even in three-consonant clusters).

On the last — given that /tʰ/ tends to be the most stable aspirated consonant, I wonder if it's possible that PIE had voiced aspirates after all, and they just generally shifted to something else? E.g. *kʰ > *h₂, or *pʰ > *f > *h₁. And I suppose even *tʰ breaking to *th₂ might be possible.

[Pabappa]

Typo for "voicelcess" i assume?

ANyway, I thibnk there is still a lot of work to be done in PIE. And it will possiblt never end, unless someone digs up a 5000 yrs old cuneiform script inside a cave in Goergia that turns out to be written in pure PIE. Anything is possible. -sD- could have been allophinically /sT/ and the nsomehow switched in the daugher languages due to a "lenis becomes voiced" shif tor something. I cant really say anuthing intelligent, other than that I think PIE scientistis rely too much on theories of unconditional changes. It could very well be that the sound changes involved are so messy that we'll never be able to figure out what hte true PIE phonology was. Just look at Old English to Middle English for example, there are like fifty rules each with a bunch of tiny excveptions. Multiply that by ten and ytou have the amount of sound cvhanges that likely existed from PIE to the early attested languages.

[KathTheDragon]

So, on a whim I've extracted the list of PIE(ish) verb roots from LIV

Looking thru the root-final clusters, a few Weird PIE Things:
1) There are no examples of root-final *-sT-, but about five of *-sD-.

There are in fact 7 listed, 5 with final *-sd- and 2 with *-sg-. Virtually all of them are suspect in one way or another: *masd- probably does not exist, *rasd- is Italo-Celtic only, and may well be a compound, *pesd- is onomatopoeic, *ǵʰeisd- and *h₂eisd- are extensions of *ǵʰeis- and *h₂eis-, *mesg- is conceivably an extension of *mes-, leaving only *resg-.

2) Examples of root-final *-PH- are quite rare other than *-th₂- (which is common even in three-consonant clusters).

On the last — given that /tʰ/ tends to be the most stable aspirated consonant, I wonder if it's possible that PIE had voiced aspirates after all, and they just generally shifted to something else? E.g. *kʰ > *h₂, or *pʰ > *f > *h₁. And I suppose even *tʰ breaking to *th₂ might be possible.

LIV lists 14 roots in final *-th₂- of which 4 are IIr. only (*wyeth₂-, *ḱneth₂-, *g(ʷ)renth₂-, and probably *ḱreth₂-), one is an extension (*preuth₂- < *preu-), two are actually the same root (*peth₂-), and one may not exist (*h₃wath₂-), leaving 7. But note that LIV assumes that only *h₂ aspirates a voiceless stop in Indo-Aryan, so aspirates are automatically reconstructed as *Th₂. However, it is not the case that only *h₂ triggers aspiration, so it is quite possible that at least some of these should be *-tH- instead. As for other final *-TH- sequences, there are 9 in LIV, of which two are the same, and the root does not in fact end in a laryngeal (*KrepH- = *k(ʷ)RepH- -> *krep-). The remaining 7 are *sekH-, *kʷerpH-, *peth₁-, *terk/ḱH-, *meiḱh₂-, *reikh₂-, and *h₂eḱh₃-. So while *-th₂- isn't as overwhelmingly more common as it first seemed, roots in final *-tH- are slightly more common than other stops. A quick survey suggests that this is actually quite reasonable, in the case of *t versus *ḱ. The truely interesting gap is in roots ending in *-pH-. Why only one?