Tuesday, February 13, 2024

Proto-IE or Sprachbund? Dialogue with Josef G. Mitterer


Assorted retorts from yahoo boards and elsewhere: Proto-IE or Sprachbund? Dialogue with Josef G. Mitterer · Φιλολoγικά/Philologica: Indo-European Branches for I and II p. Plural, Pronouns

Q
How do we know that a Proto-Indo-European language really existed? What is the evidence?
https://www.quora.com/How-do-we-know-that-a-Proto-Indo-European-language-really-existed-What-is-the-evidence/answer/Josef-G-Mitterer


Josef G. Mitterer
Studied Romance Philology
Updated Tue 6.II.2024
As I’m the 39th user to answer this question I’ll try to make it short and efficient:

  1. ‘I brought’ means ἔφερον épheron in Ancient Greek and ábharam in Sanskrit.
  2. ‘I give’ means δίδωμι dídōmi in Ancient Greek and dadā́mi in Sanskrit.


The imperfect forms listed in the first point consist of three morphological segments: the augment, the verbal stem and the ending: é-pher-on, á-bhar-am. Now it is definitely conceivable that the verbal stem was borrowed from one language into the other, but it is really unthinkable that such a complex morphological formation would be adopted from one language into the other. — For instance, English was substantially influenced by Anglo-Norman, yet the “core” of the verbal morphology, the stem series of verbs (cf. sing—sang—sung etc.) were not touched and are still profoundly “Germanic” (cf. singen—sang—gesungen in German, syngja—söng—sunginn in Icelandic etc.). Another example: Basque has been influenced by Latin, then Spanish and French for over two millennia, yet there’s only a structural influence in verbal morphology, no material one.¹ These examples alone show that it is extremely unlikely that the forms é-pher-on, á-bhar-am have spread from one language to other languages.

Up to this point, however, it’s still thinkable that one language derives from the other. In fact, early Indo-European studies still assumed that Greek, Latin etc. didn’t descend from an unattested proto-language, but from Sanskrit. Luckily, it’s also easy to demonstrate that this cannot be the case. Let’s introduce two Latin words: quod ‘what’ and -que ‘and’. The equivalents in Sanskrit are kád ‘what’ and ca ‘and’, with a palatalized initial sound (/t͡ɕɐ/, somewhat similar to cha- in chant). — Why are there two different initial sounds in Sanskrit (k = /k-/ and c = /t͡ɕ/), while in Latin there’s only one (qu = /kʷ/)? If we presume that the Sanskrit vowel system with kád and ca is the “original” one, there’s no reason for this difference. However, if we presume that the Latin vowel system is “older”, it work’s out, because in this case we have a palatal vowel in que = kʷe. Now, it’s a very frequent phenomenon (also in English) that a consonant is palatalized when it’s before a palatal vowel. To say, (only) if we presume that the “original” form was *kʷe, both the Latin and the Sanskrit result are plausible:

  1. Latin: *kʷe = -que.
  2. Sanskrit: *kʷe > *ke > *ce > ca.


This means that the Sanskrit vowel system is an innovation and that *e (and *o) and *a merged in Sanskrit. Thus, Sanskrit can’t be “older” than Latin or Greek: it’s their “sister”, not their “mother”. Besides, this vowel shift also explains one part of the difference between the Greek é-pher-on and the Sanskrit á-bhar-am. The shift of final /-m/ to /-n/ in Greek is regular, too (cf., e.g., the Latin second-declension accusative taur-um and the Greek ταῦρον taûr-on ‘bull’), as well as the desonorization of the aspired occlusives (*bʰ > *pʰ).

These were two examples. Every single additional example would be an additional proof of the existence of Proto-Indo-European.

¹ The future form idatziko dut ‘I’ll write’ is structurally based on Ancient Spanish he de escribir, but it’s impossible to notice unless you understand the structure and meaning of the single elements.

Mon 5.II.2024

Hans-Georg Lundahl
Can you exclude the Sprachbund effect?

Obviously, a Sprachbund means speakers of sth ancestral to Greek and sth ancestral to Sanskrit once were neighbours, but so does the Proto-Language hypothesis.

Josef G. Mitterer
Well, of course we know nothing with absolute certainty, and even people like Trubeckoj have surmised a Sprachbund effect (albeit, I think, with ethnological rather than linguistic motivation).

However, I don't know of any example of a Sprachbund that has anywhere near as much morphological material exchange as would be required for a PIE Sprachbund. If you look at the Balkan Sprachbund, for instance, you’ll find a number of common phenomena, yet the inflectional systems remain very stable: you won’t find something like Bulgarian endings in Romanian.

So I’d say the gap between Sprachbund phenomena we know and the parallelisms within PIE languages is enormous. I really can’t imagine they could be explained as Sprachbund effects.

Tue 6.II.2024

Hans-Georg Lundahl
“you won’t find something like Bulgarian endings in Romanian.”

Each of them already had case endings, and the personal endings of verbs are very similar, except for phonetics.

Obviously both have in common with each other and with Albanian the setup of adding definite articles as endings.

In some situations, you could not have a set up like the Indo-European 8 case system (like Latin, but three separate Ablatives) without borrowing material too — like if some lacked the cases beyond 2, 3 or 4. Germanic endings are sometimes not explicable in Indo-European terms, or the explanation is not verifiable.

Genitive plural bagme in Gothic, or the Weak Noun Declinsion, which looks a bit like a two case system, Nominative + Common Case. Not totally unrelated to Slavic Declinsion, I found. Bagme is not explicable. The explanation that these declinsions had endings that were lost due to sound change is possible but not verifiable.

Obviously the Balkan Sprachbund is very much younger than the Indo-European Sprachbund or Sprachbünder.

Making a difference in endings for Imperfect vs Simple Past is sth which could emerge from a Sprachbund between Finno-Ugrian and Semitic, the former having the present—past distinction, the latter having the imperfect—perfect distinction. It is lacking from the Northern-most families, Germanic and Balto-Slavic (or indirectly reintroduced, but in stems rather than nedings, for Slavic). The Indo-European verb also has an Ablaut that’s kind of Semitic, but at the same time endings that have been used as an argument for Indo-European being related to Fenno-Ugric in Nostratic. And the feminines in -a look a bit like Semitic feminines in -at. Note, in vocabulary, the possible Semitic influence would have been very much less prominent.

Thu 8.II.2024

Josef G. Mitterer
Of course, PIE reconstruction has gaps, but I still think they are only a very weak hint for a potential Sprachbund. Even apart from morphology (which in my opinion is the most direct evidence for a common proto-language), almost everything speaks against the Sprachbund theory. For example, it is extremely unlikely that the fundamental basic vocabulary (numbers, relationships, words meaning 'to go', 'to exist', 'to carry') is borrowed, and not just in isolated cases, but to an impressive extent.

I think the ablaut is a good example, too. We find such phenomena in many modern languages. It’s “natural” that vowels develop independently according to the stress, and it’s “natural” that these vocalic variations are secondarily grammaticalized.

"Making a difference in endings for Imperfect vs Simple Past is sth which could emerge from a Sprachbund between Finno-Ugrian and Semitic"

But aren’t we talking about an IE Sprachbund rather than a Sprachbund that includes IE and other families?

Hans-Georg Lundahl
“For example, it is extremely unlikely that the fundamental basic vocabulary (numbers, relationships, words meaning 'to go', 'to exist', 'to carry') is borrowed, and not just in isolated cases, but to an impressive extent.”

Depends on how like or unlike the vocabulary were in previous languages.

I’ve taken as an example the 1st and 2nd plural of Swedish, Italian and English, in a hypothetic Sprachbund setting, with no language learning facilities and before codified literature.

Swedish (nominative) vi, ni (obj case) oss, er
Italian (nominative) noi, voi (obj case) ci, vi
English (nominative) we, ye (obj case) us, you.

The Sprachbund would probably eliminate all forms on N- and on V-, and leave:

(nominative) chi, yi (obj case) os, yu.

I think counting too, getting common numerals would be of practical importance.

“an IE Sprachbund rather than a Sprachbund that includes IE and other families”

I’m not saying that the Semitic languages that we have were part of the IE Sprachbund, or that the Fenno-Ugrian languages that we have were so.

If IE emerged from a Sprachbund situation, any input from languages previous to that one would by definition by non-IE ones. They then (including stray strands of Uralic and Semitic, possibly) become IE by the situation.

“Even apart from morphology (which in my opinion is the most direct evidence for a common proto-language)”

Persian and Germanic share a past tense in -D.

Latin, Greek and Sanskrit share two past non-perfect and two perfects (present and past) as concept, but do not agree on endings.

Germanic languages do not have full personal ending paradigms with no needs for personal pronouns (except Gothic).

Lithuanian does not distinguish third person singular from plural.

Lithuanian and Greek have no third person -T, which Latin and Germanic share with Celtic.

“the fundamental basic vocabulary”

The Swadesh list gives only 25 % identity between English and Russian.

Josef G. Mitterer
"The Sprachbund would probably eliminate all forms on N- and on V-, and leave […]"

Here you are presupposing something that still needs to be proven. We don't know of any Sprachbund in which different pronouns can be proven to have originally coincided.

"If IE emerged from a Sprachbund situation, any input from languages previous to that one would by definition by non-IE ones. They then (including stray strands of Uralic and Semitic, possibly) become IE by the situation."

So you mean that, for instance, some Proto-Semitic and some Proto-Finno-Ugric variants (and maybe some others) formed a Sprachbund from which the Indo-European languages were to emerge, and some other Proto-Semitic and Proto-Finno-Ugric languages developed into the modern Semitic and Finno-Ugric languages?

"Latin, Greek and Sanskrit share two past non-perfect and two perfects (present and past) as concept, but do not agree on endings."

In which sense? I’d say there’s only one past non-perfect (the imperfect, alongside present tense, which is non-perfect by nature, but not past) and two perfective tenses (perfect and aorist).

"Lithuanian and Greek have no third person -T, which Latin and Germanic share with Celtic."

The final -T is often dropped. We have Italian (etc.) canta from Latin CANTAT, we have Belarusian чытае čytaje from Old East Slavic čitajetь, but we still have Sardinian cantat(a) and Russian читает čitajet. So if the Lithuanian-Greek vs. Latin-Germanic-Celtic difference is an argument for an IE Sprachbund, there are the same arguments for a Romance or Slavic Sprachbund, but we know there weren’t. Besides, I’m not sure for Lithuanian, but Greek phonotactics doesn’t allow final /-t/, anyway.

"Persian and Germanic share a past tense in -D.

"Lithuanian does not distinguish third person singular from plural."

I don't see how this is an indication of a Sprachbund.

"The Swadesh list gives only 25 % identity between English and Russian."

However, the actual question is if the relevant words have or had cognates in the other language, cf. fratello vs. hermano, sorella vs. hermana, ci vs. nos, vi vs. os, essi/loro vs. ellos, cane vs. perro, tavola vs. mesa, uccello vs. pájaro, bollire vs. hervir, trovare vs. encontrar, prendere vs. tomar, giallo vs. amarillo and many, many others between Italian and Spanish. And there it’s only two millennia, but we know that Italian and Spanish are not linked by a Sprachbund, but by a common ancestor.

Sat 10.II.2024

Hans-Georg Lundahl
"Here you are presupposing something that still needs to be proven."

Proof is not a question you pose about potentialities. For facts, you need proof. For impossibilities, you need proof, since they are a kind of necessities. But for potentiality, you only need to have no disproof.

"We don't know of any Sprachbund in which different pronouns can be proven to have originally coincided."

That was totally not the question. Your logic suffers from what you seem to regard as "scientific methodology", and I praise God I'm an amateur linguist and didn't undergo your training.

"So you mean that, for instance, some Proto-Semitic and some Proto-Finno-Ugric variants (and maybe some others) formed a Sprachbund from which the Indo-European languages were to emerge, and some other Proto-Semitic and Proto-Finno-Ugric languages developed into the modern Semitic and Finno-Ugric languages?"

That's one version of my hypothesis, yes. I'd not use "Proto" since for at least Semitic, I am as sure as for IE that different peoples spoke different languages right after Babel (Heber from Mitsraim, Iavan from Madai, standing for respectively Hebrews and Egyptians, Greeks and Medes-Medopersians).

"I’d say there’s only one past non-perfect (the imperfect, alongside present tense, which is non-perfect by nature, but not past) and two perfective tenses (perfect and aorist)."

I am using "perfect" as it is used in English grammar. Aorist is past in English grammar, it is not perfect present (aka perfect) and it is not perfect past (aka pluperfect). What you seem to mean by "perfect" vs "non-perfect" is in English grammar "simple" vs "continuous" (exception, habits and repeated actions are in English grammar simple rather than continuous).

"The final -T is often dropped."

I know that explanation.

"Besides, I’m not sure for Lithuanian, but Greek phonotactics doesn’t allow final /-t/, anyway."

Nor does Slavic (your examples reflect original -ti, not original -t).

I know that explanation. I also know that the possibility of an explanation is not proof that it is the right one, like you pretended I didn't know about the hypothesis of harmonising pronouns.

That the explanation is there proves the IE by PIE hypothesis possible on this side of the equation, but that it is needed proves that there is a lack of proof for it, on this issue.

"And there it’s only two millennia, but we know that Italian and Spanish are not linked by a Sprachbund, but by a common ancestor."

Romance languages have sth like 70 or 80 % lexical identity, not sth like 25 %. We know that a very shortlived Sprachbund between Arabic and Persian resulted in Persian grammar losing cases and in 30 % or so lexical similarity.

Since you brought up kinship words, there is exactly one family or “branch of IE” which has all the (supposed) PIE words for father, mother, son, daughter, brother, sister. It’s Germanic.

EDIT: I missed the remark “I don’t see this as an indication of a Sprachbund” — the thing is we have a structural difference between different sets of IE languages. Germanic verb morphology has more in common with Finnish than with Sanskrit. It’s you who pretend that a Proto-language is a proven, I pretend that a Sprachbund is possible and from some other considerations (Biblical chronology, identities in Genesis 10) preferrable. For me it’s sufficient that a Sprachbund is viable. It’s the weaker hypothesis, therefore the default.

L.D. 11.II.2024

Josef G. Mitterer
"Proof is not a question you pose about potentialities. […]"

Okay, so I should have said it’s something highly speculative.

"That was totally not the question. […]"

So what was the question? In fact, I sometimes find it hard to understand what you want to say with your examples.

"right after Babel […] Biblical chronology, identities in Genesis 10"

I would not bring the Bible into contact with historical linguistics. Well, you didn't use it as a linguistic argument, anyway, but I personally wouldn't use the Bible as a reason to consider a Sprachbund phenomenon either.

"I know that explanation. I also know that the possibility of an explanation is not proof that it is the right one"

Yes, it’s a question of plausibility. We know nothing about the past with absolute certainty.

"That the explanation is there proves the IE by PIE hypothesis possible on this side of the equation, but that it is needed proves that there is a lack of proof for it, on this issue."

Can you please be a bit more concrete about this and that?

"Romance languages have sth like 70 or 80 % lexical identity, not sth like 25 %."

Yes, but they evolved from Latin less than 2000 years ago, and the Romance languages have been influenced by Latin for most of their history and thus “held together” in a certain sense.

"We know that a very shortlived Sprachbund between Arabic and Persian resulted in Persian grammar losing cases and in 30 % or so lexical similarity."

How do we know that the loss of cases in Persian is a result of the contact with Arabic? While as for the lexicon, again, it doesn’t affect the very basic vocabulary and, again, there’s no material Sprachbund effect in morphology, even if we accept that the loss of declension is due to the contact with Arabic. — And the Spanish-Arabic contact didn’t leave any trace in Spanish grammar.

"Since you brought up kinship words, there is exactly one family or “branch of IE” which has all the (supposed) PIE words for father, mother, son, daughter, brother, sister. It’s Germanic."

What do you conclude from this?

"Germanic verb morphology has more in common with Finnish than with Sanskrit."

I’m not so sure about that. The “strong” stem series fit exactly into the PIE ablaut system which, obviously, is somewhat “obscured” due to the Sanskrit vocalism. And the verbal endings are obviously related to each other. Besides, French verbal morphology has little in common with the Latin verb, before all if you disregard the historical spelling and the “traditional” classification of pronouns.

Hans-Georg Lundahl
"I should have said it’s something highly speculative. ... So what was the question?"

My view on IE as originating in a series of Sprachbünder, is like that of Trubetskoy (the founder of Balkan linguistics) very rightly described as highly speculative.

I think it can stand comparison with PIE logically, as to probability of being true. It definitely does not compare to Brugmann et consortes as far as detailed and argued in detail solutions are concerned. How my theory stands is about how PIE stood in the time of Bopp or Grimm.

Now, on the Balkan, take the free standing plural pronouns of Bulgarian, Romanian and Modern Greek:

ние вие те noi voi ei εμείς εσείς αυτοί

N = N ние = noi, V = V вие = voi, T = T те = αυτοί

No conflict of meaning, so no need to harmonise. This is because the languages were already all of them IE before entering the Balkan Sprachbund.

If we add Albanian, ne, ju, ata gives another N = N, another T = T.

"I would not bring the Bible into contact with historical linguistics."

Would you bring Julius Caesar and Alcuin of York in contact with it?

"Can you please be a bit more concrete about this and that?"

The fact that the explanation is there, proves that the hypothesis "IE metagroup arose by PIE proto-language" is possible on this side of the equation, namely the fact that some families have and some lack -T in the third person singular.

The fact that the explanation is needed, proves that the hypothesis "IE metagroup arose by PIE proto-language" is not fully proven, on this side of the equation, namely third persons in or not in -T (and cognates).

"Yes, but they evolved from Latin less than 2000 years ago,"

I get what you are saying. Slavic and Germanic split from each other twice as long ago. Supposedly. Therefore less lexical identity is to be expected. First of all, see the point I made on previous point. But second, you did not hear the argument out. The 25 % similarity is below what we have between an undoubted case of Sprachbund paired languages, Persian and Arabic.

I e, when the numbers are that low, the other explanation is very certainly in the competition, again, on this side of the equation, namely lexical similarity.

"the Romance languages have been influenced by Latin for most of their history and thus “held together” in a certain sense."

Which illustrates the potential of Sprachbünder. I think the example with 70 % similarity is Romanian ... which was not so held together.

"How do we know that the loss of cases in Persian is a result of the contact with Arabic?"

How do we know that Middle English loss of cases is a result of Anglo-Saxon contact with Norse? We don't, but it's a clear option, since in both cases, the case systems were different, in N—AS in endings, in P—A in syntax.

"as for the lexicon, again, it doesn’t affect the very basic vocabulary"

Can you document that? I couldn't read a Swadesh list in Arabic script if my life depended on it.

"And the Spanish-Arabic contact didn’t leave any trace in Spanish grammar."

Unless the definite article comes from it. Latin, mother language of Spanish as well as of French, didn't have one. By the time of Chant of St. Eulalia, I think the Strassburg Oaths lack the article, though that could be a posh mannerism, Arabic has been able to influence all Ibero-Romance and some Gallo-Romance languages since 711, or even 670, if you go back to North Africa, which is 210 years.

"What do you conclude from this?"

That the idea "basic vocabulary doesn't change" doesn't fit, and especially not with a PIE hypothesis.

"The “strong” stem series fit exactly into the PIE ablaut system which, obviously, is somewhat “obscured” due to the Sanskrit vocalism."

Which brings me to the fact that the weak stem series doesn't. Just as the weak declinsion doesn't show a full 4 case or 5 case declinsion, unless by paradigmatic comparison with strong declinsion. And "obscured" in Sanskrit = not proven. The point I was making though is, Germanic like Finnish has a conjugation for present and for ONE past. Not for past-continuous vs past-simple times prior action or not.

"And the verbal endings are obviously related to each other."

I think Sanskrit has more personal endings in common with Finnish than with Germanic.

"French verbal morphology has little in common with the Latin verb, before all if you disregard the historical spelling and the “traditional” classification of pronouns."

Very good. You have once again a) showed the PIE hypothesis is possible, given enough time, and b) on certain sides not proven. When a thing is possible but not proven, and another thing is possible but not proven, is the default the Academic consensus, or is the default the weaker claim?

No comments: