European languages before Indo-European

[WeepingElf]

I could open a new thread, but since we are talking about old languages and Europe...

Well, I'm interested in the old languages of Europe, before and after IE, and I want to find a good book with grammars for all those tongues. An encyclopedia, that is, with comprehensive and concise grammars (not fully developed, since that would result in a whole collection with many volumes).

Google Books has Encyclopedia of the languages of Europe and The ancient languages of Europe and I wonder if they're good resources. Anyone knows them or has any suggestions?

I don't know the first one, but I know the second one (The Ancient Languages of Europe). It treats those languages of Europe (excluding the Caucasus) that were spoken and written before about 400 AD, and of those only those we more or less understand: Greek, Latin and the other Italic languages, Continental Celtic, and Etruscan. I think it is worth reading.

With the exception of Basque, Etruscan and the Caucasian languages, the pre-IE languages of Europe are so poorly known that little can be said about them. In the Mediterranean, some were written, such as Iberian and Eteocretan, but the inscriptions are not yet understood. North of the Alps, none survived long enough to be still alive when writing came to their rescue, and the only evidence are old geographical names and apparent substratum effects and loanwords in the known languages - but those things are very difficult to interpret.

[Tropylium]

natural science Fachidioten

…oh, so that's where Finnish fakki-idiootti comes from! I've always thought the first part was from English "fuck", with just an odd restriction in meaning (it means pretty much the same as the German).

As I said earlier, Celtic shares a significant part of its lexicon with Pontic languages, specially Greek. For example, *gdonjo- 'man' is modelled after the same root than Greek khthó:n 'earth', khthónios 'underground'. Other words like *yekkā 'cure, treatment' or *yorko- 'roebuck' are also distinctively Pontic.

I see no reason to assume borrowing here, nor does any Indo-Europeanist worth his stripes.

Yep, overreliance on lexical distribution can be hazardous, as this can very, very easily represent shared retentions rather than shared innovations. I don't think there are any phonological isoglosses that would connect Celtic with Graeco-Armeno-Aryan? — while it does share a few with Italic.

…OTOH reliance on sound changes only can also lead to odd results: it would be possible† to posit an "Exo-Greek" grouping within IE, united by a number of non-trivial innovations: 1) loss of initial vocalic laryngeals; 2) merger of other vocalic laryngeals to a single *ə; 3) merger of syllabic resonant + laryngeal to a long syllabic resonant (whereas in Greek, *H e/a/o / {#,C}_C, and *R=H Rā/ē/ō)

†Disclaimer: supposing that Wikipedia: IE sound laws hasn't fuck'd these parts up. Any experts on here BTW, your help with checking the page would be appreciated)

[Octavià]

The coincidences with Italic seem to be due to a common "Danubian" or "Hesperic" substrate (me) or to contact influence (Schmidt).

So why exactly do you think this is better than thinking that there was an Italo-Celtic subfamily? And I remind you that a number of statistical phylogeny studies, which make no assumptions about the structure of the Indo-European family, group Italic and Celtic together.

Surely not the one by Gray & Atkinson (2003), which groups Italic and Germanic into a single node.

Which is based on glottochronology, a technique now considered unreliable. Gray and Atkinson aren't Indo-Europeanists. They are mathematicians who dabble in historical linguistics without understanding that field. That they managed to get the thing published in a peer-reviewed journal means nothing more than that the peer reviewers were natural science Fachidioten who were too impressed by the formulae to realize how phony the thing was. (EDIT: That the structure of the tree is doubtful is obvious from some errors in the branches, such as Polish being classified as East rather than West Slavic. If something like that could happen, it could also happen deeper in the tree.)

At least it was published online and it's available for free, something we can't say from many other articles/books.

As I said earlier, Celtic shares a significant part of its lexicon with Pontic languages, specially Greek. For example, *gdonjo- 'man' is modelled after the same root than Greek khthó:n 'earth', khthónios 'underground'. Other words like *yekkā 'cure, treatment' or *yorko- 'roebuck' are also distinctively Pontic.

I see no reason to assume borrowing here, nor does any Indo-Europeanist worth his stripes.

I didn't say anything about borrowing , but only that these words contradict the hypothesis of an Italo-Celtic node.

WHAT? Also, please don't abuse other people's terminology that way. You speak of "Europic", "Hesperic" and "Danubian" all the time, but refer to things that have nothing to do with what TaylorS and I mean when we use these words.

It's true that my own theory I've adopted somne of your terminology, and specifically the words "Hesperic" and "Danubian" (but not "Europic"), and in doing so I'm implicitly acknowledging your work. If you feel I've broken your "copyright", then I'll find other names for my own entities.

WHAT? There is too little evidence for an IE-Kartvelian relationship. Just a handful of similar-looking words - which are probably borrowings.

Really? In which direction? This is important.

You should compare languages and not just dictionaries.

I'm doing the same than other macro-comparativists/Nostraticists do.

What an entangled mess of dubious etymologies!

Thanks for your feeedback!

[Octavià]

With the exception of Basque, Etruscan and the Caucasian languages, the pre-IE languages of Europe are so poorly known that little can be said about them. In the Mediterranean, some were written, such as Iberian and Eteocretan, but the inscriptions are not yet understood. North of the Alps, none survived long enough to be still alive when writing came to their rescue, and the only evidence are old geographical names and apparent substratum effects and loanwords in the known languages - but those things are very difficult to interpret.

Yes, they're so for most people. But it doesn't mean they can't be studied.

This reminds me from the time when most European people were affraid to sail across the Atlantic.

[Octavià]

Yep, overreliance on lexical distribution can be hazardous, as this can very, very easily represent shared retentions rather than shared innovations.

I don't think the lexicon related to Chalcolitic inventions such as the wheel, copper, etc. originated from a relative small area (the Pontic Steppes) can be considered as "shared retentions". The "retention/innovation" dychotomy can also be interpreted as "substrate/superstrate".

I don't think there are any phonological isoglosses that would connect Celtic with Graeco-Armeno-Aryan? — while it does share a few with Italic.

Phonology is sensitive to areal influences. For example, Armenian has glottalic stops like Kartvelian but nobody would postulate a genetical relationship between them.

[Tropylium]

Yep, overreliance on lexical distribution can be hazardous, as this can very, very easily represent shared retentions rather than shared innovations.

I don't think the lexicon related to Chalcolitic inventions such as the wheel, copper, etc. originated from a relative small area (the Pontic Steppes) can be considered as "shared retentions". The "retention/innovation" dychotomy can also be interpreted as "substrate/superstrate".

The words you've mention'd as "Pontic", like *dʰǵʰom "land", don't seem to be inventions (that one has Italic and Balto-Slavic cognates too, besides). AFAIK the distribution of the wagon or metalworking terminology does not have systematic holes.

I don't think there are any phonological isoglosses that would connect Celtic with Graeco-Armeno-Aryan? — while it does share a few with Italic.

Phonology is sensitive to areal influences. For example, Armenian has glottalic stops like Kartvelian but nobody would postulate a genetical relationship between them.

This is valid for inventories, not for the specifics of development like the Italo-Celtic assimilation *p_kʷ *kʷ_kʷ, or the Graeco-Aryan development of syllabic *m, *n to a nasal vowel *ã.

(By contrast consider Romance vowel development — French has several front rounded vowels no dout by influence from Germanic, but these result from unconditional drift (†u †eu y ø), never umlaut as in Gmc. Romanian and Bulgarian likewise share the presence of /ɨ/, but the former gets it from *a *ə, the latter from *u *ʊ.)

[Morrígan]

I'm doing the same than other macro-comparativists/Nostraticists do.

Then no-one is going to take you seriously. These approaches are not considered methodologically sound, and for good reason. Performing comparisons at a greater time-depth requires raising the bar on proof of relatedness, not lowering it.

Armenian has glottalic stops like Kartvelian

Since when? Armenian has a three-way voiced/voiceless/aspirated contrast.

[Richard W]

That the structure of the [Gray and Atkinson] tree is doubtful is obvious from some errors in the branches, such as Polish being classified as East rather than West Slavic. If something like that could happen, it could also happen deeper in the tree.

It could also be a matter of what splits such a tree should be recording - how long has the North Slavic dialect continuum been dead? Note that Lusatian is shown as splitting off from the languages of the North Slavic continuum. As Frisian has remained part of the West Germanic continuum and English has not, perhaps it is right that Frisian should be assessed as more closely related to Flemish than to English.

(I conceive of the IE breakup being a dialect continuum falling part.)

The glottochronological element of G&A's reconstruction is subject to some form of internal correction - they do report, 'A likelihood ratio test with the extinct languages removed revealed that rates were significantly non-clock-like.'

As to the grouping in G&A, Figure 1c shows that if one removes 'doubtful cognates', Celtic suddenly groups with Italic rather than Germanic. One possibly doubtful cognate, not relevant to the comparison, is of English town (well represented across Germanic) with Irish dún 'fort' etc., possibly the source, in the sense of 'hill fort', of English down, only attested across Germanic in the Low German area.

[Mecislau]

It could also be a matter of what splits such a tree should be recording - how long has the North Slavic dialect continuum been dead? Note that Lusatian is shown as splitting off from the languages of the North Slavic continuum. As Frisian has remained part of the West Germanic continuum and English has not, perhaps it is right that Frisian should be assessed as more closely related to Flemish than to English.

I really see no justification whatsoever for Polish being grouped with East Slavic as opposed to the other non-Lechitic West Slavic languages. The division between West (incl. Polish) and East Slavic is one of the oldest in the Slavic family...

And what do you mean "Lusatian is shown as splitting off from the languages of the North Slavic continuum"; Lusatian (ie, Sorbian) is just another subbranch of West Slavic.

[WeepingElf]

[Gray & Atkinson 2003]
At least it was published online and it's available for free, something we can't say from many other articles/books.

Which says nothing about the quality of the work. All sorts of crap is available from the Net for free.

WHAT? Also, please don't abuse other people's terminology that way. You speak of "Europic", "Hesperic" and "Danubian" all the time, but refer to things that have nothing to do with what TaylorS and I mean when we use these words.

It's true that my own theory I've adopted somne of your terminology, and specifically the words "Hesperic" and "Danubian" (but not "Europic"), and in doing so I'm implicitly acknowledging your work. If you feel I've broken your "copyright", then I'll find other names for my own entities.

I find it a bit strange that you are using these names while you reject our hypotheses. What exactly do you refer to as "Hesperic" and what do you refer to as "Danubian"? So far, I got the impression that you use these names for branches of Indo-European proper, which is not what TaylorS and I use these names for. Using a name introduced by someone else usually implies that you refer to the same entity.

WHAT? There is too little evidence for an IE-Kartvelian relationship. Just a handful of similar-looking words - which are probably borrowings.

Really? In which direction? This is important.

I don't know; could be either direction. I don't have a list of these similar-looking words, much less compared them to a list of presumable Indo-Uralic cognates. If a word occurs in IE, Uralic and Kartvelian, it was probably borrowed by Kartvelian from IE.

You should compare languages and not just dictionaries.

I'm doing the same than other macro-comparativists/Nostraticists do.

Which exactly is the problem. Most macro-comparativists do the same thing as you - they just compare dictionaries, ignoring both the structures of the languages and what is known about the histories of the words compared. That is a pretty low standard, and as Goatface said, no-one is going to take you seriously.

With the exception of Basque, Etruscan and the Caucasian languages, the pre-IE languages of Europe are so poorly known that little can be said about them. In the Mediterranean, some were written, such as Iberian and Eteocretan, but the inscriptions are not yet understood. North of the Alps, none survived long enough to be still alive when writing came to their rescue, and the only evidence are old geographical names and apparent substratum effects and loanwords in the known languages - but those things are very difficult to interpret.

Yes, they're so for most people. But it doesn't mean they can't be studied.

Of course not. As you know, I do try to study them; but I am aware of the difficulties involved with that, and don't try to reach bold conclusions beyond what can be known.

[Octavià]

I'm doing the same than other macro-comparativists/Nostraticists do.

Then no-one is going to take you seriously.

Do you really mean nobody take seriously Nostraticists like Dolgopolsky or Bomhard?

These approaches are not considered methodologically sound, and for good reason. Performing comparisons at a greater time-depth requires raising the bar on proof of relatedness, not lowering it.

The reason why I've taken up macro-comparative linguistics is precisely to raise current standards.

Armenian has glottalic stops like Kartvelian

Since when? Armenian has a three-way voiced/voiceless/aspirated contrast.

You're right, but Armenian phonology has surely been influenced by Kartvelian, which has glottalics.

[WeepingElf]

I'm doing the same than other macro-comparativists/Nostraticists do.

Then no-one is going to take you seriously.

Do you really mean nobody take seriously Nostraticists like Dolgopolsky or Bomhard?

Goatface may have overstated, but most linguists consider the evidence adduced by people like Dolgopolsky or Bomhard insufficient. Look, D and B do not even agree on the sound correspondences, yet each comes up with several hundred "cognate sets". The two reconstructions are incompatible withe each other; at most one of them can be right - which probably means that both are wrong. A method which yields such false positives surely is unreliable.

It also tells a lot that both D and B claim that Proto-Nostratic was an analytical, isolating language, even though all the "branches" of Nostratic have rich inflectional morphologies. Why? Because the inflectional systems of the "Nostratic" languages do not match. Hence, they throw them away - and thereby discard what could be the strongest evidence for or against a relationship. They just compare dictionaries. Can't you see what is wrong about that?

These approaches are not considered methodologically sound, and for good reason. Performing comparisons at a greater time-depth requires raising the bar on proof of relatedness, not lowering it.

The reason why I've taken up macro-comparative linguistics is precisely to raise current standards.

If you want to raise current standards in long-range comparison, the first step you should do towards that is to comply with the standards of traditional comparative linguistics, which I don't see you are doing.

Armenian has glottalic stops like Kartvelian

Since when? Armenian has a three-way voiced/voiceless/aspirated contrast.

You're right, but Armenian phonology has surely been influenced by Kartvelian, which has glottalics.

Some Eastern Armenian dialects have the voiceless unaspirated stops realized as glottalics, probably indeed under influence from Kartvelian, yes. Whether the "Armenian sound shift" (*/t d dh/ > /th t d/ etc.) has anything to do with Kartvelian is uncertain.

[Octavià]

[Gray & Atkinson 2003]
At least it was published online and it's available for free, something we can't say from many other articles/books.

Which says nothing about the quality of the work. All sorts of crap is available from the Net for free.

OK, Then I'd like to know the references of those articles you don't consider to be "crap".

I find it a bit strange that you are using these names while you reject our hypotheses.

Not exactly. I accept part of them, namely the existence of extinct varieties related to the historical attested IE languages. What I don't accept is your proposed chronology and dialectal fragmentation of PIE.

What exactly do you refer to as "Hesperic" and what do you refer to as "Danubian"? So far, I got the impression that you use these names for branches of Indo-European proper, which is not what TaylorS and I use these names for. Using a name introduced by someone else usually implies that you refer to the same entity.

In my theory, these names designate several substrate languages spoken in some parts of Europe, probably since the Mesolithic. The historically attested IE languages are the result of the merger (in varying degrees) of these and other substrates with a Pontic superstrate, which very often is considered to be the real PIE.

WHAT? There is too little evidence for an IE-Kartvelian relationship. Just a handful of similar-looking words - which are probably borrowings.

Really? In which direction? This is important.

I don't know; could be either direction. I don't have a list of these similar-looking words, much less compared them to a list of presumable Indo-Uralic cognates. If a word occurs in IE, Uralic and Kartvelian, it was probably borrowed by Kartvelian from IE.

I think you've understimated the size of the list of IE-Kartvelian correspondences. Also, correspondences such as *paχ-ur- ~ za-px-ul- suggest the opposite direction.

Of course not. As you know, I do try to study them; but I am aware of the difficulties involved with that, and don't try to reach bold conclusions beyond what can be known.

I think you've got some prejudices about what "can be known" and what can not. In my analogy, nobody dared to cross the Atlantic until Columbus did it. PS: He had access to copies of the maps drawn by a Chinese fleet several decades ago (see Gavin Menzies' 1421).

[Octavià]

It also tells a lot that both D and B claim that Proto-Nostratic was an analytical, isolating language, even though all the "branches" of Nostratic have rich inflectional morphologies. Why? Because the inflectional systems of the "Nostratic" languages do not match. Hence, they throw them away - and thereby discard what could be the strongest evidence for or against a relationship. They just compare dictionaries. Can't you see what is wrong about that?

Yes, of course. This is precisely the main point where "splitters" (traditional comparative linguists) and "lumpers" (macro-comparative ones) disagree.

You must be aware that most morphemes evolved from specialized lexemes. For example, it's typical for formerly isolated personal pronouns to become part of verbal inflection, or for demonstratives to become articles, etc. This is why with such time depths one can't expect morphology (which traditional comparative linguistics lies upon) to be the same. However, I agree it must still be accounted for.

If you want to raise current standards in long-range comparison, the first step you should do towards that is to comply with the standards of traditional comparative linguistics, which I don't see you are doing.

See above.

Some Eastern Armenian dialects have the voiceless unaspirated stops realized as glottalics, probably indeed under influence from Kartvelian, yes. Whether the "Armenian sound shift" (*/t d dh/ > /th t d/ etc.) has anything to do with Kartvelian is uncertain.

Don't forget that Germanic has also the same sound shift. But I disagree with the last part. If we can establish a reliable correspondence between IE */d t dh/ and Kartvelian */t t’ d/, then also we can hypothesize the sound shift in Germanic and Armenian was due to Kartvelian influence.

[WeepingElf]

[Gray & Atkinson 2003]
At least it was published online and it's available for free, something we can't say from many other articles/books.

Which says nothing about the quality of the work. All sorts of crap is available from the Net for free.

OK, Then I'd like to know the references of those articles you don't consider to be "crap".

I cannot give references to every "non-crap" article I know, and it would be pointless. You said "at least it was published online and it's available for free", as if that was a reason to assume that it was good. I said that it isn't. It slipped by the peer reviewers because the latter were too impressed by the mathematical mumbo-jumbo to ask a real Indo-Europeanist on his opinion. If they had done, it may have been rejected.

I find it a bit strange that you are using these names while you reject our hypotheses.

Not exactly. I accept part of them, namely the existence of extinct varieties related to the historical attested IE languages. What I don't accept is your proposed chronology and dialectal fragmentation of PIE.

What exactly do you refer to as "Hesperic" and what do you refer to as "Danubian"? So far, I got the impression that you use these names for branches of Indo-European proper, which is not what TaylorS and I use these names for. Using a name introduced by someone else usually implies that you refer to the same entity.

In my theory, these names designate several substrate languages spoken in some parts of Europe, probably since the Mesolithic. The historically attested IE languages are the result of the merger (in varying degrees) of these and other substrates with a Pontic superstrate, which very often is considered to be the real PIE.

I see. So our main disagreement is that you place Hesperic (the OEH language) and Danubian (an intermediate between Hesperic and PIE proper) in the Mesolithic, while I opine for the Neolithic. That's fair. Also, we disagree about the time depth and subgrouping of Indo-European.

WHAT? There is too little evidence for an IE-Kartvelian relationship. Just a handful of similar-looking words - which are probably borrowings.

Really? In which direction? This is important.

I don't know; could be either direction. I don't have a list of these similar-looking words, much less compared them to a list of presumable Indo-Uralic cognates. If a word occurs in IE, Uralic and Kartvelian, it was probably borrowed by Kartvelian from IE.

I think you've understimated the size of the list of IE-Kartvelian correspondences. Also, correspondences such as *paχ-ur- ~ za-px-ul- suggest the opposite direction.

I have to admit that I know too little about lexical resemblances between IE and Kartvelian to draw reasonable conclusions. Can you point me to a list of apparent lexical correspondences? At any rate, the morphologies of IE and Kartvelian are different enough to be unsupportive of the idea of a close relationship. If two neighbouring languages have several hundred similar words but utterly different morphologies, I'd rather expect borrowing than descent from a common ancestor.

Of course not. As you know, I do try to study them; but I am aware of the difficulties involved with that, and don't try to reach bold conclusions beyond what can be known.

I think you've got some prejudices about what "can be known" and what can not. In my analogy, nobody dared to cross the Atlantic until Columbus did it. PS: He had access to copies of the maps drawn by a Chinese fleet several decades ago (see Gavin Menzies' 1421).

Knowing not to know is a virtue for a scholar, not a flaw. People who are sure that they know are more often than not people who have managed to deceive themselves.

[WeepingElf]

It also tells a lot that both D and B claim that Proto-Nostratic was an analytical, isolating language, even though all the "branches" of Nostratic have rich inflectional morphologies. Why? Because the inflectional systems of the "Nostratic" languages do not match. Hence, they throw them away - and thereby discard what could be the strongest evidence for or against a relationship. They just compare dictionaries. Can't you see what is wrong about that?

Yes, of course. This is precisely the main point where "splitters" (traditional comparative linguists) and "lumpers" (macro-comparative ones) disagree.

The difference between the "splitters" and the "lumpers" is that the former are more picky about the evidence. I agree with you that some "splitters" are too picky. For instance, I consider the evidence for a relationship between Indo-European and Uralic good enough to seriously consider such a relationship; yet, I would say that more research is needed.

You must be aware that most morphemes evolved from specialized lexemes. For example, it's typical for formerly isolated personal pronouns to become part of verbal inflection, or for demonstratives to become articles, etc. This is why with such time depths one can't expect morphology (which traditional comparative linguistics lies upon) to be the same. However, I agree it must still be accounted for.

Morphology can indeed change a lot, but it tends to be more stable than basic vocabulary, and much more stable than cultural vocabulary which cannot be older than the things themselves. Lexical comparison can easily be fouled by episodes of massive borrowings from a single source; that is less likely with morphology. Keep in mind that families such as Indo-European were originally established by morphological comparison.

If you want to raise current standards in long-range comparison, the first step you should do towards that is to comply with the standards of traditional comparative linguistics, which I don't see you are doing.

See above.

I still don't see why your methodology is an improvement over the standards of more traditional historical linguistics. In fact, what you are doing is a disimprovement.

Some Eastern Armenian dialects have the voiceless unaspirated stops realized as glottalics, probably indeed under influence from Kartvelian, yes. Whether the "Armenian sound shift" (*/t d dh/ > /th t d/ etc.) has anything to do with Kartvelian is uncertain.

Don't forget that Germanic has also the same sound shift. But I disagree with the last part. If one can establish a reliable correspondence between IE */d t dh/ and Kartvelian */t t’ d/, then one can hypothesize the sound shift in Germanic and Armenian was due to Kartvelian influence.

Sure, Germanic has a similar shift. Kartvelian influence? Hardly. It would be quite ridiculous to assume, barring further evidence, that Kartvelian languages were ever spoken in northern Central Europe. Also, I don't see how a presumed correspondence between IE and Kartvalien, whch does not even match the Armenian sound shift, should count as evidence for a Kartvelian influence on Armenian!

[Octavià]

I cannot give references to every "non-crap" article I know, and it would be pointless.

Well, it was actually Goatface who mentioned them, so it's up to him to provide such references.

You said "at least it was published online and it's available for free", as if that was a reason to assume that it was good. I said that it isn't. It slipped by the peer reviewers because the latter were too impressed by the mathematical mumbo-jumbo to ask a real Indo-Europeanist on his opinion. If they had done, it may have been rejected.

Not I know yours. Thank you very much

I see. So our main disagreement is that you place Hesperic (the OEH language) and Danubian (an intermediate between Hesperic and PIE proper) in the Mesolithic, while I opine for the Neolithic. That's fair. Also, we disagree about the time depth and subgrouping of Indo-European.

I'm affraid OEH wasn't a single language but rather a continuum of several varieties, as shown by isoglosses like *χ (= H₂) > zero/s as in al-/sal- 'to spring, to flow'. This is also found in *χEmV (˜ ħ-) ‘warm’ (Dolgopolsky's ND 2586) > *sam-/*sum-ro- 'summer' vs. *χēm-ºr- 'heat (of the day)' > Greek hēméra, Armenian awr 'day'.

Knowing not to know is a virtue for a scholar, not a flaw. People who are sure that they know are more often than not people who have managed to deceive themselves.

I think many people just don't want (e.g. by fear) to know by themselves, so they have to relie on what "knowledgeable" people tell them. This precisely what happens in most religions, but also (unfortunately) in science.

[Octavià]

The difference between the "splitters" and the "lumpers" is that the former are more picky about the evidence. I agree with you that some "splitters" are too picky. For instance, I consider the evidence for a relationship between Indo-European and Uralic good enough to seriously consider such a relationship; yet, I would say that more research is needed.

I think most lexical isoglosses between Uralic and IE are due to borrowing rather than common inheritance. Although this doesn't exclude a genetical relationship, it puts it far part in time.

Some people have also pointed out the similarities between Uralic and Dravidian, so in the framework of Nostratic theory, an Uralo-Dravidian would be possible.

Morphology can indeed change a lot, but it tends to be more stable than basic vocabulary, and much more stable than cultural vocabulary which cannot be older than the things themselves. Lexical comparison can easily be fouled by episodes of massive borrowings from a single source; that is less likely with morphology. Keep in mind that families such as Indo-European were originally established by morphological comparison.

This is precisely why current models fail to explain the huge gap between Anatolian and the rest of IE languages.

I still don't see why your methodology is an improvement over the standards of more traditional historical linguistics. In fact, what you are doing is a disimprovement.

Why?

Some Eastern Armenian dialects have the voiceless unaspirated stops realized as glottalics, probably indeed under influence from Kartvelian, yes. Whether the "Armenian sound shift" (*/t d dh/ > /th t d/ etc.) has anything to do with Kartvelian is uncertain.

Don't forget that Germanic has also the same sound shift. But I disagree with the last part. If one can establish a reliable correspondence between IE */d t dh/ and Kartvelian */t t’ d/, then one can hypothesize the sound shift in Germanic and Armenian was due to Kartvelian influence.

Sure, Germanic has a similar shift. Kartvelian influence? Hardly. It would be quite ridiculous to assume, barring further evidence, that Kartvelian languages were ever spoken in northern Central Europe.

No, what it would be actually ridiculous is to assume that (Proto-)Germanic was always spoken there.

Also, I don't see how a presumed correspondence between IE and Kartvelian, which does not even match the Armenian sound shift, should count as evidence for a Kartvelian influence on Armenian!

No, it matches it:

IE Armenian Kartvelian
d t t
t th t’
dh d d

See it?

[WeepingElf]

The difference between the "splitters" and the "lumpers" is that the former are more picky about the evidence. I agree with you that some "splitters" are too picky. For instance, I consider the evidence for a relationship between Indo-European and Uralic good enough to seriously consider such a relationship; yet, I would say that more research is needed.

I think most lexical isoglosses between Uralic and IE are due to borrowing rather than common inheritance. Although this doesn't exclude a genetical relationship, it puts it far part in time.

Some people have also pointed out the similarities between Uralic and Dravidian, so in the framework of Nostratic theory, an Uralo-Dravidian would be possible.

I do not and cannot claim to know, but to me, Uralic bears a closer resemblance, especially in morphology, to Indo-European than to Dravidian, a family whose relationship to IE and Uralic seems doubtful to me.

Morphology can indeed change a lot, but it tends to be more stable than basic vocabulary, and much more stable than cultural vocabulary which cannot be older than the things themselves. Lexical comparison can easily be fouled by episodes of massive borrowings from a single source; that is less likely with morphology. Keep in mind that families such as Indo-European were originally established by morphological comparison.

This is precisely why current models fail to explain the huge gap between Anatolian and the rest of IE languages.

Indeed, the morphology of Anatolian is different enough from the standard model of PIE to make an earlier stage of PIE necessary to account for it. You may call this a "failure" of the current model; I'd rather say that the model needs to be amended to explain the differences.

I still don't see why your methodology is an improvement over the standards of more traditional historical linguistics. In fact, what you are doing is a disimprovement.

Why?

Because you just compare dictionaries, disregard morphology, and use ad-hoc sound changes! Your "cognate sets" are so small that no meaningful regularities can be observed.

Some Eastern Armenian dialects have the voiceless unaspirated stops realized as glottalics, probably indeed under influence from Kartvelian, yes. Whether the "Armenian sound shift" (*/t d dh/ > /th t d/ etc.) has anything to do with Kartvelian is uncertain.

Don't forget that Germanic has also the same sound shift. But I disagree with the last part. If one can establish a reliable correspondence between IE */d t dh/ and Kartvelian */t t’ d/, then one can hypothesize the sound shift in Germanic and Armenian was due to Kartvelian influence.

Sure, Germanic has a similar shift. Kartvelian influence? Hardly. It would be quite ridiculous to assume, barring further evidence, that Kartvelian languages were ever spoken in northern Central Europe.

No, what it would be actually ridiculous is to assume that (Proto-)Germanic was always spoken there.

Certainly, the branch of IE later to become Germanic came from the east, and the Germanic sound shift may have happened further east than where the Germanic languages enter the light of history, perhaps in what is now Poland. But that is still far from the Caucasus! I really don't see how one can plausibly appeal to Kartvelian in order to explain anything in the post-PIE development of Germanic.

Also, I don't see how a presumed correspondence between IE and Kartvelian, which does not even match the Armenian sound shift, should count as evidence for a Kartvelian influence on Armenian!

No, it matches it:

IE Armenian Kartvelian
d t t
t th t’
dh d d

See it?

No. They don't match. It is /t/ (< PIE *d) that is glottalized in Eastern Armenian dialects (and presumed to have been glottalized in PIE by Gamkrelidze & Ivanov), not /th/ (< PIE *t). Also, distant relationships between IE and Kartvelian, if any, do not play a role for such contact situations. The speakers of the languages certainly weren't aware of that relationship!

[WeepingElf]

I see. So our main disagreement is that you place Hesperic (the OEH language) and Danubian (an intermediate between Hesperic and PIE proper) in the Mesolithic, while I opine for the Neolithic. That's fair. Also, we disagree about the time depth and subgrouping of Indo-European.

I'm affraid OEH wasn't a single language but rather a continuum of several varieties, as shown by isoglosses like *χ (= H₂) > zero/s as in al-/sal- 'to spring, to flow'. This is also found in *χEmV (˜ ħ-) ‘warm’ (Dolgopolsky's ND 2586) > *sam-/*sum-ro- 'summer' vs. *χēm-ºr- 'heat (of the day)' > Greek hēméra, Armenian awr 'day'.

Certainly, Hesperic was a family of related languages, not a homogenous, uniform language. I have to admit that I currently don't know where the isoglosses run, which is why I am going to map the Old European hydronymy. But don't hold your breath for it. Whether the examples you give are valid or not, remains to be seen.

Knowing not to know is a virtue for a scholar, not a flaw. People who are sure that they know are more often than not people who have managed to deceive themselves.

I think many people just don't want (e.g. by fear) to know by themselves, so they have to relie on what "knowledgeable" people tell them. This precisely what happens in most religions, but also (unfortunately) in science.

I definitely do not fear to know! I am eager to know. But I know that I don't know, and that the methodology must be sound if you really want to know and not just deceive yourself (and perhaps others).

[Tropylium]

Armenian has glottalic stops like Kartvelian

Since when? Armenian has a three-way voiced/voiceless/aspirated contrast.

You're right, but Armenian phonology has surely been influenced by Kartvelian, which has glottalics.

Some Eastern Armenian dialects have the voiceless unaspirated stops realized as glottalics, probably indeed under influence from Kartvelian, yes. Whether the "Armenian sound shift" (*/t d dh/ > /th t d/ etc.) has anything to do with Kartvelian is uncertain.

Kortlandt has shown (1984: "Proto-Indo-European glottalic stops: The comparative evidence") that a glottalic stage must be presumed even for precedessors of some Armenian dialectal systems that do not have glottalics currently. In particular, there is a dialect group that flips classical *d *t /t d/, and since it also retains classical *tʰ separate, this requires some fourth (or fifth, considering the /dʰ/ in some areas) intermediate to be possible. His interpretation is that classical *t was /tʼ/ which in this dialect (and possibly, also the others that spontaneously changed voiced stops to voiced as well) underwent /dʼ/ /d/. As independant evidence, in erly loanwords from Russian, voiceless unaspirated stops (no Slavic variety as aspiration in stops I think) are substituted by aspirated stops, not voiceless unaspirated (though he does not mention what the dialect to be in contact with Russian would have been; if it were /d tʰ/ dialects, that would explain that as well).

None of this rules out early Kartvelian (or hell, even Semitic, depending how arcaic the neighboring Aramaic varieties may have been) influence on proto-Armenian, however.

I agree that Kartvelian influence in Germanic is ridiculous, but I've seen a number of folks who try to postulate an Eastern origin for Germanic to work this in… I would say the similarity can be accounted by slightly modified Glottalic Theory and some simple phonetic universals:
0) PIE had an unstable voiced glottalic series.
1) Most branches resolv'd this in favor of deglottalizing, usually merging them into plain voiced stops (but Italic, Greek and Indo-Iranian by adding aspiration to the plain voiced stops, to retain the contrast).
2) Armenian and Germanic were two branches to opt for resolution by devoicing. This resulted in the voiceless stops becoming aspirated (or they may have been allophonically aspirated originally; it's difficult to say).
3) Germanic later spirantized its aspirates (as did Italic, Iranian and Greek).
In other words, we expect *dʼ to become either *tʼ or *d, which is the easiest explanation for why we could have two separated *tʼ-areas. (We do not expect to see a far-reaching areal change *tʼ *dʼ (:> *d), so shared retention cannot be evoked here.)

BTW, while I haven't taken any particularly close look at Uralo-Dravidian, my initial impression has been that it's about as plausible as Hungaro-Sumerian (ie. a bunch of *pućka). The comparisions have a large degree of semantic leeway and appear to be assembled by perusing the UEW side-by-side with a Tamil dictionary. I have no dout many would fall apart immediately if we had a detail'd reconstruction of proto-Dravidian (which, AFAIK, we don't.)

[WeepingElf]

Armenian has glottalic stops like Kartvelian

Since when? Armenian has a three-way voiced/voiceless/aspirated contrast.

You're right, but Armenian phonology has surely been influenced by Kartvelian, which has glottalics.

Some Eastern Armenian dialects have the voiceless unaspirated stops realized as glottalics, probably indeed under influence from Kartvelian, yes. Whether the "Armenian sound shift" (*/t d dh/ > /th t d/ etc.) has anything to do with Kartvelian is uncertain.

Kortlandt has shown (1984: "Proto-Indo-European glottalic stops: The comparative evidence") that a glottalic stage must be presumed even for precedessors of some Armenian dialectal systems that do not have glottalics currently. In particular, there is a dialect group that flips classical *d *t /t d/, and since it also retains classical *tʰ separate, this requires some fourth (or fifth, considering the /dʰ/ in some areas) intermediate to be possible. His interpretation is that classical *t was /tʼ/ which in this dialect (and possibly, also the others that spontaneously changed voiced stops to voiced as well) underwent /dʼ/ /d/. As independant evidence, in erly loanwords from Russian, voiceless unaspirated stops (no Slavic variety as aspiration in stops I think) are substituted by aspirated stops, not voiceless unaspirated (though he does not mention what the dialect to be in contact with Russian would have been; if it were /d tʰ/ dialects, that would explain that as well).

I am not convinced by everything Kortlandt wrote, but I think he is someone who should know what he is doing, and it is not at all implausible. The standard model of PIE stop phonology is indeed not without typological problems, though not to the extent that it could not have existed that way. IMHO, Proto-Europic had */t t' d/, which shifted to */t d' d/ in Early PIE.

None of this rules out early Kartvelian (or hell, even Semitic, depending how arcaic the neighboring Aramaic varieties may have been) influence on proto-Armenian, however.

I agree that Kartvelian influence in Germanic is ridiculous, but I've seen a number of folks who try to postulate an Eastern origin for Germanic to work this in… I would say the similarity can be accounted by slightly modified Glottalic Theory and some simple phonetic universals:
0) PIE had an unstable voiced glottalic series.
1) Most branches resolv'd this in favor of deglottalizing, usually merging them into plain voiced stops (but Italic, Greek and Indo-Iranian by adding aspiration to the plain voiced stops, to retain the contrast).
2) Armenian and Germanic were two branches to opt for resolution by devoicing. This resulted in the voiceless stops becoming aspirated (or they may have been allophonically aspirated originally; it's difficult to say).
3) Germanic later spirantized its aspirates (as did Italic, Iranian and Greek).
In other words, we expect *dʼ to become either *tʼ or *d, which is the easiest explanation for why we could have two separated *tʼ-areas. (We do not expect to see a far-reaching areal change *tʼ *dʼ (:> *d), so shared retention cannot be evoked here.)

This makes sense.

BTW, while I haven't taken any particularly close look at Uralo-Dravidian, my initial impression has been that it's about as plausible as Hungaro-Sumerian (ie. a bunch of *pućka). The comparisions have a large degree of semantic leeway and appear to be assembled by perusing the UEW side-by-side with a Tamil dictionary. I have no dout many would fall apart immediately if we had a detail'd reconstruction of proto-Dravidian (which, AFAIK, we don't.)

In other words: a classical example of dictionary comparison. Dictionaries are no doubt useful in language comparison - if they are the right kind, namely etymological dictionaries which list the oldest accessible forms of the words. The stuff used by tourists and language learners is pretty useless in historical linguistics; also, you have to take the phonology and grammar into account.

[Octavià]

Indeed, the morphology of Anatolian is different enough from the standard model of PIE to make an earlier stage of PIE necessary to account for it. You may call this a "failure" of the current model; I'd rather say that the model needs to be amended to explain the differences.

I'd rather say rebuilt.

I still don't see why your methodology is an improvement over the standards of more traditional historical linguistics. In fact, what you are doing is a disimprovement.

Why?

Because you just compare dictionaries, disregard morphology, and use ad-hoc sound changes! Your "cognate sets" are so small that no meaningful regularities can be observed.

I disagree. Don't forget I study substrate languages, which by definition only survive in the form of loanwords. Working with small datasets might be painful, but by no means "ignorable".

Certainly, the branch of IE later to become Germanic came from the east, and the Germanic sound shift may have happened further east than where the Germanic languages enter the light of history, perhaps in what is now Poland. But that is still far from the Caucasus! I really don't see how one can plausibly appeal to Kartvelian in order to explain anything in the post-PIE development of Germanic.

Certainly, I don't have enough evidence to say Kartvelian (or perhaps an extinct relative of it) was a substrate/adstrate to Proto-Germanic, but the fact Armenian underwent the same shift than Germanic remain to be explained.

IE Armenian Kartvelian
d t t
t th t’
dh d d
See it?

No. They don't match. It is /t/ (< PIE *d) that is glottalized in Eastern Armenian dialects (and presumed to have been glottalized in PIE by Gamkrelidze & Ivanov), not /th/ (< PIE *t).

In that case, it would be Kartvelian *t’ which would correspond to IE *d (Armenian /t/) and Kartvelian *t to IE *t (Armenian /th/).

This means Armenian and Germanic stop system is closer to the one postulated by the glottalic theory, which would be valid for Eurasiatic/Nostratic but not for PIE. Thank you for helping me understand the matter

[WeepingElf]

Indeed, the morphology of Anatolian is different enough from the standard model of PIE to make an earlier stage of PIE necessary to account for it. You may call this a "failure" of the current model; I'd rather say that the model needs to be amended to explain the differences.

I'd rather say rebuilt.

IMHO, it is still largely appropriate for Late PIE; for Early PIE, the Anatolian data are of paramount importance, and a new model must be built. Some scholars such as Jay Jasanoff are working on that; many, however, still try to reduce Hittite to the standard model. Academic inertia.

I still don't see why your methodology is an improvement over the standards of more traditional historical linguistics. In fact, what you are doing is a disimprovement.

Why?

Because you just compare dictionaries, disregard morphology, and use ad-hoc sound changes! Your "cognate sets" are so small that no meaningful regularities can be observed.

I disagree. Don't forget I study substrate languages, which by definition only survive in the form of loanwords. Working with small datasets might be painful, but by no means "ignorable".

Sure. We don't know, for instance, the grammars of the languages the Old European hydronymy came from. But it seems that you, like many long-rangers, apply dictionary comparison to languages whose grammars are known, and that's problematic.

Certainly, the branch of IE later to become Germanic came from the east, and the Germanic sound shift may have happened further east than where the Germanic languages enter the light of history, perhaps in what is now Poland. But that is still far from the Caucasus! I really don't see how one can plausibly appeal to Kartvelian in order to explain anything in the post-PIE development of Germanic.

Certainly, I don't have enough evidence to say Kartvelian (or perhaps an extinct relative of it) was a substrate/adstrate to Proto-Germanic, but the fact Armenian underwent the same shift than Germanic remain to be explained.

Yes; but when a phenomenon occurs in a language that in all likelihood had no contact with Kartvelian at any time, doesn't that speak against holding Kartvelian responsible for it? I think the matter can be resolved without bringing Kartvelian into play (except for the recent glottalization in Eastern Armenian dialects). Have you read Tropylium's post?

IE Armenian Kartvelian
d t t
t th t’
dh d d
See it?

No. They don't match. It is /t/ (< PIE *d) that is glottalized in Eastern Armenian dialects (and presumed to have been glottalized in PIE by Gamkrelidze & Ivanov), not /th/ (< PIE *t).

In that case, it would be Kartvelian *t’ which would correspond to IE *d (Armenian /t/) and Kartvelian *t to IE *t (Armenian /th/).

This means Armenian and Germanic stop system is closer to the one postulated by the glottalic theory, which would be valid for Eurasiatic/Nostratic but not for PIE. Thank you for helping me understand the matter

Indeed. The glottalic theory assumes that PIE was like the Eastern Armenian dialects in this regard; and Bomhard assumes that that system was inherited from Nostratic. This is the main point where Bomhard disagrees with Dolgopolsky, BTW. For my opinion, read my reply on Tropylium's post.

[Octavià]

I agree that Kartvelian influence in Germanic is ridiculous, but I've seen a number of folks who try to postulate an Eastern origin for Germanic to work this in…

Honestly, none of us know anything about the substrate or substrates from which Germanic emerged, so these kind of "considerations" look unscienfical to me.

I would say the similarity can be accounted by slightly modified Glottalic Theory and some simple phonetic universals:
0) PIE had an unstable voiced glottalic series.

"Voiced glottalic stops"? As the glottis is closed, voice is only possible if they're implosive, that is, the air is sucked into the mouth rather than expelled from the larynx.

BTW, while I haven't taken any particularly close look at Uralo-Dravidian, my initial impression has been that it's about as plausible as Hungaro-Sumerian (ie. a bunch of *pućka). The comparisions have a large degree of semantic leeway and appear to be assembled by perusing the UEW side-by-side with a Tamil dictionary. I have no dout many would fall apart immediately if we had a detail'd reconstruction of proto-Dravidian (which, AFAIK, we don't.)

Wikipedia says this:

Uralo-Dravidian
The theory that the Dravidian languages display similarities with the Uralic language group, suggesting a prolonged period of contact in the past is popular amongst Dravidian linguists and has been supported by a number of scholars, including Robert Caldwell, Thomas Burrow, Kamil Zvelebil, and Mikhail Andronov. This theory has, however, been rejected by some specialists in Uralic languages, and has in recent times also been criticised by other Dravidian linguists such as Bhadriraju Krishnamurti.