Zompist Bboard Again

bradrn wrote: ↑Tue Aug 31, 2021 2:00 amI strongly disagree with you here, so I’d be interested to know how you came to this opinion.

I've seen several proposals trying to link Basque with either PNC (Proto-North Caucasian) or PIE by way of regular sound correspondences. Unfortunately, most of the purported etymologies were wrong, rending useless the whole endeavour.

Talskubilos wrote: ↑Tue Aug 31, 2021 2:21 am

bradrn wrote: ↑Tue Aug 31, 2021 2:00 amI strongly disagree with you here, so I’d be interested to know how you came to this opinion.

I've seen several proposals trying to link Basque with either PNC (Proto-North Caucasian) or PIE by way of regular sound correspondences. Unfortunately, most of the purported etymologies were wrong, rending useless the whole endeavour.

Understandable. But surely you accept there is a difference between sound changes which have only very weak evidence and tenuous etymologies, as purported for Basque–??? links, and sound changes which each are attested from hundreds of cognates across multiple languages with clear and obvious semantic links, as suggested for IE?

bradrn wrote: ↑Tue Aug 31, 2021 2:29 amUnderstandable. But surely you accept there is a difference between sound changes which have only very weak evidence and tenuous etymologies, as purported for Basque–??? links, and sound changes which each are attested from hundreds of cognates across multiple languages with clear and obvious semantic links, as suggested for IE?

Of course so, but these "hundreds of cognates" also probably include some chance resemblances as well as loanwords/Wanderwörter.

Talskubilos wrote: ↑Tue Aug 31, 2021 2:33 am

bradrn wrote: ↑Tue Aug 31, 2021 2:29 amUnderstandable. But surely you accept there is a difference between sound changes which have only very weak evidence and tenuous etymologies, as purported for Basque–??? links, and sound changes which each are attested from hundreds of cognates across multiple languages with clear and obvious semantic links, as suggested for IE?

Of course so, but these "hundreds of cognates" also include some chance resemblances as well as loanwords/Wanderwörter.

The likelihood of chance resemblances certainly does goes up as you sample more words and languages; but it also goes down as you tighten restrictions on phonological and semantic similarities. Currently the range of phonogical correspondences accepted by mainstream historical linguists is very narrow: only words corresponding to each other by regular changes are accepted. The range of semantic correspondences is also restricted — a Greenbergian comparison of ‘udder’ to ‘suck’ to ‘throat’ (say) would never be acceptable. Finally, if you’re really dealing with a genuine sound change, the chance of a false positive actually goes down as you confirm that sound change in more and more words. All this goes to make chance resemblances extremely unlikely. (I will admit that a handful pop up now and again, but they’re still quite rare.)

Loanwords are also a factor, but they tend to be fairly easy to distinguish as they don’t follow the same sound changes as native words. Also, they tend to be rather rare in basic vocabulary, which helps a lot.

(And by the way, I meant that there were ‘hundreds of cognates’ for each sound change for each language. Probably thousands of cognate pairs in total.)

bradrn wrote: ↑Tue Aug 31, 2021 2:49 amThe range of semantic correspondences is also restricted — a Greenbergian comparison of ‘udder’ to ‘suck’ to ‘throat’ (say) would never be acceptable.

Unfortunately, semantic changes happen all of the time, so this approach is likely to discard true correspondences and include Wanderwörter.

bradrn wrote: ↑Tue Aug 31, 2021 2:49 amLoanwords are also a factor, but they tend to be fairly easy to distinguish as they don’t follow the same sound changes as native words. Also, they tend to be rather rare in basic vocabulary, which helps a lot.

I disagree on both.

Talskubilos wrote: ↑Tue Aug 31, 2021 3:23 am

bradrn wrote: ↑Tue Aug 31, 2021 2:49 amThe range of semantic correspondences is also restricted — a Greenbergian comparison of ‘udder’ to ‘suck’ to ‘throat’ (say) would never be acceptable.

Unfortunately, semantic changes happen all of the time, so this approach is likely to discard true correspondences and include Wanderwörter.

I’ll admit that this is true when considering languages with thousands of years’ worth of separation — I am reminded here of how wheel and chakra are cognate. Luckily for us, semantic changes tend to occur in small steps, which means you can compare obvious cognates from closely related language to determine which semantic changes have occurred. Also, IE at least has a long history, which makes it possible to look at older texts and observe that two words with distinct semantics were more similar in the past.

bradrn wrote: ↑Tue Aug 31, 2021 2:49 amLoanwords are also a factor, but they tend to be fairly easy to distinguish as they don’t follow the same sound changes as native words. Also, they tend to be rather rare in basic vocabulary, which helps a lot.

I disagree on both.

Could you provide a counterexample please? Not, I should note, a counterexample accepted by you and you alone, but one which is accepted by mainstream historical linguists.

bradrn wrote: ↑Tue Aug 31, 2021 3:29 amI’ll admit that this is true when considering languages with thousands of years’ worth of separation — I am reminded here of how wheel and chakra are cognate.

This happens to be another instance of a Wanderwort in IE.

bradrn wrote: ↑Tue Aug 31, 2021 3:29 amLuckily for us, semantic changes tend to occur in small steps, which means you can compare obvious cognates from closely related language to determine which semantic changes have occurred. Also, IE at least has a long history, which makes it possible to look at older texts and observe that two words with distinct semantics were more similar in the past.

This can be done with e.g. Latin and Romance languages, but not with substrate loanwords or distantly related languages, for example.

bradrn wrote: ↑Tue Aug 31, 2021 2:49 amCould you provide a counterexample please? Not, I should note, a counterexample accepted by you and you alone, but one which is accepted by mainstream historical linguists.

I think that, for practical purposes, "historical linguists" is synonym to "Indo-Europeanists". Am I right?

Talskubilos wrote: ↑Tue Aug 31, 2021 3:35 am

bradrn wrote: ↑Tue Aug 31, 2021 3:29 amI’ll admit that this is true when considering languages with thousands of years’ worth of separation — I am reminded here of how wheel and chakra are cognate.

This happens to be another instance of a Wanderwort in IE.

Yes, this does indeed happen to be a modern Wanderwört. Generally speaking, such long-range loans have become much more common in modern times.

bradrn wrote: ↑Tue Aug 31, 2021 3:29 amLuckily for us, semantic changes tend to occur in small steps, which means you can compare obvious cognates from closely related language to determine which semantic changes have occurred. Also, IE at least has a long history, which makes it possible to look at older texts and observe that two words with distinct semantics were more similar in the past.

This can be done with e.g. Latin and Romance languages, but not with substrate loanwords or distantly related languages, for example.

I don’t see why not — you just work recursively. First, reconstruct the semantic changes in the lowest-level groups. That gives you a set of recent protoforms, which you can use to reconstruct the semantic changes in higher-level subgroups. Be persistent enough and you’ll eventually figure out the meaning of the original protoform.

(I’ll admit that this is very difficult to do, which is why most language families are so poorly-reconstructed. Luckily, we didn’t need to do this for IE, since we have primary sources for early languages like Gothic, Latin, Greek and Sanskrit which we can use to find the original semantics directly.)

bradrn wrote: ↑Tue Aug 31, 2021 2:49 amCould you provide a counterexample please? Not, I should note, a counterexample accepted by you and you alone, but one which is accepted by mainstream historical linguists.

I think that, for practical purposes, "historical linguists" is synonym to "Indo-Europeanists", am I right?

Yes, all ‘mainstream linguists’ that I know of (not even restricted to historical linguists!) accept IE as a valid family. But I’d reserve the term ‘Indo-Europeanist’ to people who actively study IE, rather than all people who merely accept it.

bradrn wrote: ↑Tue Aug 31, 2021 3:45 amI don’t see why not — you just work recursively. First, reconstruct the semantic changes in the lowest-level groups. That gives you a set of recent protoforms, which you can use to reconstruct the semantic changes in higher-level subgroups. Be persistent enough and you’ll eventually figure out the meaning of the original protoform.

(I’ll admit that this is very difficult to do, which is why most language families are so poorly-reconstructed. Luckily, we didn’t need to do this for IE, since we have primary sources for early languages like Gothic, Latin, Greek and Sanskrit which we can use to find the original semantics directly.)

In my view, reconstructed protoforms are useful tools for comparative purposes, but not a goal themselves, so proto-languages can't be mistaken as real entities. This also includes PIE.

Don't forget most languages spoken by humakind in ~40 kyears didn't left written records at all, so restricting ourselves to the study of written languages would result in a distorted perspective.

bradrn wrote: ↑Tue Aug 31, 2021 2:49 amYes, all ‘mainstream linguists’ that I know of (not even restricted to historical linguists!) accept IE as a valid family. But I’d reserve the term ‘Indo-Europeanist’ to people who actively study IE, rather than all people who merely accept it.

Of course, I accept IE as a valid family, but I don't agree with some of the dogmae and "traditions" of academic IE studies.

Getting back to the thread's topic, Basque native lexicon has a good number of nouns with final -i, namely begi 'eye', behi 'cow', ogi 'bread', (t)egi 'place', etc. which in some cases can be linked to the IE masculine -o- and probably would be parallel to -ā > -e for the feminine, as I explained before.

However, -i is extremely rare in Romance loanwords, which either end in -o* as in Spanish or -u as in Asturian and other poorly attested Romances. The latter are peculiar, because they often show vowel harmony -uCu, as in e.g. buru 'head' < Italoid (Lusitanian) *boro- < IE *g^wor- 'mountain'.

------------
*This would include ancient loanwords such as sendo 'strong, vigorous' < IE *swent- (English sound).

Nortaneous wrote: ↑Mon Aug 30, 2021 7:10 pm I haven't seen this, but the Yabêm thesis (*T and *Dh go back to the same series and the distinction developed out of transphonologization of an originally vocalic or suprasegmental contrast, like breathiness or a binary tonal opposition) seems promising as an explanation for why *TeT and *DheDh but not **TeDh or **DheT

Richard W wrote: ↑Mon Aug 30, 2021 3:54 pm Given the PIE root constraint, I don't think that *per and *bher being related should be a surprise. (Whether they are is another matter.)

Talskubilos wrote: ↑Mon Aug 30, 2021 4:35 pm This is where external comparanda fit in. For example, Semitic *gbl 'mountain' corresponds to both IE *ghebhōl 'head' (English gable, Greek kephálē) and *kapōl-o- 'head, skull' (Old English hafola, Sanskrit kapā́la-).

@Nort, Richard - that's the kind of internal reconstruction which needs further support - either by finding an up to now unknown IE language which confirms these theories (à la Anatolian with the laryngeals), or by external comparison in a wider group, like Boreal / Nostratic / Indo-Uralic, etc.
@Talskubilos: it's not wrong to look for connections outside IE. But if you cannot establish more than a couple of lookalikes, if you cannot show a system of sound correspondences between the various language families, you'll never be able to convince anyone either of loan relationships or of a deeper relationship.

hwhatting wrote: ↑Tue Aug 31, 2021 4:37 am@Nort, Richard - that's the kind of internal reconstruction which needs further support - either by finding an up to now unknown IE language which confirms these theories (à la Anatolian with the laryngeals), or by external comparison in a wider group, like Boreal / Nostratic / Indo-Uralic, etc.

Unfortunately, these proposed "macrofamilies" are based upon the std PIE model, so they suffer the same defects.

hwhatting wrote: ↑Tue Aug 31, 2021 4:37 am@Talskubilos: it's not wrong to look for connections outside IE. But if you cannot establish more than a couple of lookalikes, if you cannot show a system of sound correspondences between the various language families, you'll never be able to convince anyone either of loan relationships or of a deeper relationship.

We'll see.

Talskubilos wrote: ↑Tue Aug 31, 2021 3:57 am

bradrn wrote: ↑Tue Aug 31, 2021 3:45 amI don’t see why not — you just work recursively. First, reconstruct the semantic changes in the lowest-level groups. That gives you a set of recent protoforms, which you can use to reconstruct the semantic changes in higher-level subgroups. Be persistent enough and you’ll eventually figure out the meaning of the original protoform.

(I’ll admit that this is very difficult to do, which is why most language families are so poorly-reconstructed. Luckily, we didn’t need to do this for IE, since we have primary sources for early languages like Gothic, Latin, Greek and Sanskrit which we can use to find the original semantics directly.)

In my view, reconstructed protoforms are useful tools for comparative purposes, but not a goal themselves, so proto-languages can't be mistaken as real entities. This also includes PIE.

Again, I’d be really curious to know how you came to this conclusion that protoforms aren’t real. In mainstream historical linguistics, ‘the protoform *x can be reconstructed from y and z’ is usually taken to be synonymous with ‘people used to say the word x, but then it changed over the years to become y in one descendent and z in another’, and I really can’t think of any other sane interpretation. (Of course, we may not have reconstructed it correctly, but that’s a different issue entirely.)

Don't forget most languages spoken by humakind in ~40 kyears didn't left written records at all, so restricting ourselves to the study of written languages would result in a distorted perspective.

Of course I agree with this! I never intended to suggest otherwise. I’m merely saying that the lack of written ancient languages can make reconstruction more difficult.

bradrn wrote: ↑Tue Aug 31, 2021 2:49 amYes, all ‘mainstream linguists’ that I know of (not even restricted to historical linguists!) accept IE as a valid family. But I’d reserve the term ‘Indo-Europeanist’ to people who actively study IE, rather than all people who merely accept it.

Of course, I accept IE as a valid family

Ah, thanks for confirming this! I was under the impression that you didn’t accept IE.

but I don't agree with some of the dogmae and "traditions" of academic IE studies.

…like what, exactly? I think that myself and others have been quite confused as to exactly what you do not accept.

bradrn wrote: ↑Tue Aug 31, 2021 5:27 amAgain, I’d be really curious to know how you came to this conclusion that protoforms aren’t real.

I said protolanguages, not protoforms.

bradrn wrote: ↑Tue Aug 31, 2021 5:27 amIn mainstream historical linguistics, ‘the protoform *x can be reconstructed from y and z’ is usually taken to be synonymous with ‘people used to say the word x, but then it changed over the years to become y in one descendent and z in another’, and I really can’t think of any other sane interpretation. (Of course, we may not have reconstructed it correctly, but that’s a different issue entirely.)

Cognacy doesn't necessarily imply common ancestry, as in the case of Wanderwörter.

bradrn wrote: ↑Tue Aug 31, 2021 5:27 am

but I don't agree with some of the dogmae and "traditions" of academic IE studies.

…like what, exactly? I think that myself and others have been quite confused as to exactly what you do not accept.

For example, I don't think PIE, as currently reconstructed, was a real language spoken by nomadic shepherds from the Pontic Steppes (aka Kurgans).

Talskubilos wrote: ↑Tue Aug 31, 2021 6:13 am

bradrn wrote: ↑Tue Aug 31, 2021 5:27 amAgain, I’d be really curious to know how you came to this conclusion that protoforms aren’t real.

I said protolanguages, not protoforms. … For example, I don't think PIE, as currently reconstructed, was a real language spoken by nomadic shepherds from the Pontic Steppes (aka Kurgans).

Hold on… I’m not sure how you can accept that all the various reconstructed protoforms of IE are real, but the protolanguage is not.

I mean, let’s go through this step-by-step. You accept that the IE language family is real; that is, that the various IE languages are all related. You accept that the reconstructed ‘PIE’ protoforms are all real; that is, that each set of cognate words from the various IE languages ultimately descent from a single word used thousands of years ago. But you don’t accept that these protoforms all came from the same language‽ How would it even happen, historically, that a bunch of words all ended up in exactly the same set of modern-day languages, but they didn’t originate in a single protolanguage? It beggars belief to suggest that a bunch of words from basic vocabulary like numbers and kin terms were all loaned into languages in every single IE subgroup with coincidentally identical forms.

bradrn wrote: ↑Tue Aug 31, 2021 5:27 amIn mainstream historical linguistics, ‘the protoform *x can be reconstructed from y and z’ is usually taken to be synonymous with ‘people used to say the word x, but then it changed over the years to become y in one descendent and z in another’, and I really can’t think of any other sane interpretation. (Of course, we may not have reconstructed it correctly, but that’s a different issue entirely.)

Cognacy doesn't necessarily imply common ancestry, as in the case of Wanderwörter.

When I say ‘the protoform *x can be reconstructed from y and z’, that reconstruction is of course only possible if we have already confirmed that y and z are not loanwords. (It works the other way round too: if we can’t reconstruct a protoform by reversing regular sound changes, they either aren’t cognate or one of them is a loanword.)

bradrn wrote: ↑Tue Aug 31, 2021 7:07 amHold on… I’m not sure how you can accept that all the various reconstructed protoforms of IE are real, but the protolanguage is not.

I mean, let’s go through this step-by-step. You accept that the IE language family is real; that is, that the various IE languages are all related. You accept that the reconstructed ‘PIE’ protoforms are all real; that is, that each set of cognate words from the various IE languages ultimately descent from a single word used thousands of years ago. But you don’t accept that these protoforms all came from the same language‽

That's right.

bradrn wrote: ↑Tue Aug 31, 2021 7:07 amHow would it even happen, historically, that a bunch of words all ended up in exactly the same set of modern-day languages, but they didn’t originate in a single protolanguage? It beggars belief to suggest that a bunch of words from basic vocabulary like numbers and kin terms were all loaned into languages in every single IE subgroup with coincidentally identical forms.

Numerals and kinship words are only a small subset of the +2000 lexical items reconstructed for PIE.

bradrn wrote: ↑Tue Aug 31, 2021 5:27 amWhen I say ‘the protoform *x can be reconstructed from y and z’, that reconstruction is of course only possible if we have already confirmed that y and z are not loanwords.

How could we do that?

Talskubilos wrote: ↑Tue Aug 31, 2021 7:15 am

bradrn wrote: ↑Tue Aug 31, 2021 7:07 amHow would it even happen, historically, that a bunch of words all ended up in exactly the same set of modern-day languages, but they didn’t originate in a single protolanguage? It beggars belief to suggest that a bunch of words from basic vocabulary like numbers and kin terms were all loaned into languages in every single IE subgroup with coincidentally identical forms.

Numerals and kinship words are only a small subset of the +2000 lexical items reconstructed for PIE.

But do you accept at least that that sort of basic vocabulary must be inherited from the common ancestor of all IE languages?

bradrn wrote: ↑Tue Aug 31, 2021 5:27 amWhen I say ‘the protoform *x can be reconstructed from y and z’, that reconstruction is of course only possible if we have already confirmed that y and z are not loanwords.

How could do that?

Well, there’s no way to prove a negative for sure. But as I said, if you can reconstruct a protoform and every single one of its descendants are consistent with the expected sound changes, that certainly provides strong evidence against loaning.

bradrn wrote: ↑Tue Aug 31, 2021 7:26 amBut do you accept at least that that sort of basic vocabulary must be inherited from the common ancestor of all IE languages?

Something of the kind, although not exactly so. In fact, some of these words were borrowed from other languages.

Talskubilos wrote: ↑Tue Aug 31, 2021 7:33 am

bradrn wrote: ↑Tue Aug 31, 2021 7:26 amBut do you accept at least that that sort of basic vocabulary must be inherited from the common ancestor of all IE languages?

Something of the kind, although not exactly so. In fact, some of these words were borrowed from other languages.

Worded like that, as a general principle, I don't think anyone has a problem with that, including standard IE historical linguistics. It's when you assert that for specific words (like "horse") where there is disagreement. So this needs to be discussed word by word.

Talskubilos wrote: ↑Tue Aug 31, 2021 7:33 am

bradrn wrote: ↑Tue Aug 31, 2021 7:26 amBut do you accept at least that that sort of basic vocabulary must be inherited from the common ancestor of all IE languages?

Something of the kind, although not exactly so. In fact, some of these words were borrowed from other languages.

Which words, exactly? I’d like some specific examples.

But I’m starting to believe that the disagreement between us is mostly one of terminology. When I refer to ‘PIE’, I simply mean the common ancestor of all the IE languages. This language may well be entirely different to that reconstructed by various researchers; I accept that this may be the case, though I think it unlikely. I consider the statement ‘PIE does not exist’ to be a trivial falsity, since I define PIE only as the most recent common ancestor of the IE languages, and we all agree that such an ancestor exists.