Loanwords in (P)IE

[Talskubilos]

Ah, I forgot about that one. That pretty much rules out a Semitic Wanderwort here

I think it's semantically unconvincing, but if I'm not mistaken, IE *h₂erh₃- corresponds to Semitic *ħaruƟ-, where h₂ ~ ħ and h₃ ~ u. There's an interesting correspondence between IE "laryngeals" and Semitic post-velar consonants, which are also found in southern Eurasian families such as Kartvelian, NWC an NEC but absent in northern ones (Uralic, Altaic, etc.)

[WeepingElf]

Ah, I forgot about that one. That pretty much rules out a Semitic Wanderwort here

I don't think so, but Semitic is also important for the so-called "laryngeals", because they're absent from Boreal languages (Uralic, Altaic, etc.)

Proto-Uralic at least has a phoneme usually transcribed *x, whatever that may have been, but a dorsal fricative or approximant seems likely. Also, the PIE laryngeals may have been the outcome of conditional weakenings of stops. I would not give much on the absence of such phonemes in modern Uralic languages and the like - after all, the modern IE languages all have lost their laryngeals, too. (IMHO, PIE likely is a Para-Uralic language modified by a Para-NW-Caucasian substratum.)

[Travis B.]

(IMHO, PIE likely is a Para-Uralic language modified by a Para-NW-Caucasian substratum.)

The only problem with that is that practically all PIE-PU 'cognates' turn out to be loans, and generally from subbranches of IE (such as PII or PGmc) rather than PIE proper.

[Talskubilos]

Proto-Uralic at least has a phoneme usually transcribed *x, whatever that may have been, but a dorsal fricative or approximant seems likely. Also, the PIE laryngeals may have been the outcome of conditional weakenings of stops. I would not give much on the absence of such phonemes in modern Uralic languages and the like - after all, the modern IE languages all have lost their laryngeals, too. (IMHO, PIE likely is a Para-Uralic language modified by a Para-NW-Caucasian substratum.)

AFAIK, "laryngeals" in southern languages would correspon to Uralic *k. For example, NEC *χχHweje, whose oblique form is *χχHwej-rV 'dog' (NCED 26) respectively correspond to Uralic *koje 'man, person' and *koj(e)-ra > *kojra 'male (man, dog)'.

On the other hand, there a somewhat convoluted relationship between the IE numerals '4' and '8', of which Uralic is a node: *kʷet-wor '4' < *kʷet- 'to group in pairs' > Slavic *četá: ‘couple’, Baltic *kek(e)tā ‘troop’ (with reduplication) -> Uralic *kakta/kæktæ '2' -> IE *h₁oḱte-h₃u '8' with a fossilized dual suffix -h₃u. So I bet the "Boreal" (i.e. your Para-Uralic) part of IE retains the velar stop, but the "Southern" (i.e. your Para-NW-Caucasian) changed into a laryngeal.

[WeepingElf]

(IMHO, PIE likely is a Para-Uralic language modified by a Para-NW-Caucasian substratum.)

The only problem with that is that practically all PIE-PU 'cognates' turn out to be loans, and generally from subbranches of IE (such as PII or PGmc) rather than PIE proper.

I am aware of this objection; indeed, most of the lexical resemblances between the two families are quite clearly loanwords. But what do you make of such things as the pronouns? How likely is it that PU borrowed all its pronouns from PIE? That makes no sense to me, especially as northern Eurasia is full of languages with very similar pronouns. Did they all borrow all their pronouns from one language to the next? Isn't it more likely that these languages have a distant common ancestor? It is also such that the genes seem to play ball here. These language families are all spoken by people who descend, at least in part, from Ancient North Eurasians, a population that seems to have fanned out across northern Eurasia as it became more hospitable at the end of the last ice age. PIE, in particular, appears to have been spoken by a population that resulted from the mixing of ANE-descended Eastern Europeans and Western Caucasians. (Of course, genes say nothing about languages, but they can be used to trace ancient migrations which are a major factor in the spread of languages.)

[Talskubilos]

I am aware of this objection; indeed, most of the lexical resemblances between the two families are quite clearly loanwords. But what do you make of such things as the pronouns? How likely is it that PU borrowed all its pronouns from PIE? That makes no sense to me, especially as northern Eurasia is full of languages with very similar pronouns. Did they all borrow all their pronouns from one language to the next? Isn't it more likely that these languages have a distant common ancestor? It is also such that the genes seem to play ball here. These language families are all spoken by people who descend, at least in part, from Ancient North Eurasians, a population that seems to have fanned out across northern Eurasia as it became more hospitable at the end of the last ice age. PIE, in particular, appears to have been spoken by a population that resulted from the mixing of ANE-descended Eastern Europeans and Western Caucasians. (Of course, genes say nothing about languages, but they can be used to trace ancient migrations which are a major factor in the spread of languages.)

The 1st personal pronouns in Etruscan mi and Basque ni < *mi are related to each other as well as to IE, but they don't tell us much about the rest of these languages.

[Travis B.]

What chance is it that 1st person /m/ is in the same category as /ma/ 'mother'?

[Talskubilos]

What chance is it that 1st person /m/ is in the same category as /ma/ 'mother'?

I guess they're much like linguistic universals and so they can be found in many Swadesh lists.

[WeepingElf]

Of course, a single pronoun can easily match by chance, especially considering how common the phoneme /m/ is. But entire pronominal paradigms are another matter, especially if the correspondence involves not just two but eight¹ language families occupying a large but contiguous area in which the people seem to be genetically related², too. And you won't deny that there are certainly linguistic relationships deeper than those established so far by the comparative method, do you? Or do you think like this creationist? So while I don't think Indo-Uralic or Mitian was proven, I do think these proposals are at least plausible.

[1] The eight Mitian language families, from west to east, are IMHO Indo-European, Uralic, Turkic, Mongolic, Tungusic, Yukaghir, Chukotko-Kamchatkan and Eskimo-Aleut.

[2] They all show Ancient North Eurasian ancestry, to varying degrees.

[Talskubilos]

Of course, a single pronoun can easily match by chance, especially considering how common the phoneme /m/ is. But entire pronominal paradigms are another matter, especially if the correspondence involves not just two but eight¹ language families occupying a large but contiguous area in which the people seem to be genetically related², too. And you won't deny that there are certainly linguistic relationships deeper than those established so far by the comparative method, do you? Or do you think like this creationist? So while I don't think Indo-Uralic or Mitian was proven, I do think these proposals are at least plausible.

[1] The eight Mitian language families, from west to east, are IMHO Indo-European, Uralic, Turkic, Mongolic, Tungusic, Yukaghir, Chukotko-Kamchatkan and Eskimo-Aleut.
[2] They all show Ancient North Eurasian ancestry, to varying degrees.

IMHO, classical genealogical trees aren't the best way of modelling linguistic relationships (especially long-range ones), so I focus on words instead, tracing their history through many millennia (remember the statement each word has its own history).

For example, IE *h₂rt-k´o- 'bear', where I'd replace t by a sibilant (possibly palato-alveolar) affricate in order to explain the various outputs t, s, š, kt, kṣ of the "thorny cluster" tk´ (k´þ in older reconstructions), has cognates in Altaic *karsi 'marten' (EDAL 733) > Turkic qarsa-q 'steppe fox' (with a suffix -q corresponding to Hittite -gga in hartagga-), NEC *χHVr[tɕ’]V (a Nakh-Dargwa isogloss) 'marten; otter' (NCED 1168) and Yenisseian *χa(ʔ)s (~ k-) 'badger'.

[WeepingElf]

The family tree model has its limits; the wave model describes a family like IE, which once was an expanding dialect continuum through which innovations could spread in all directions, resulting in overlapping isoglosses - as have been observed - much better. But that does not dispense of the notion of regular sound correspondences, and in order to judge whether an etymology is valid or not, one has to know these sound correspondences in the relevant languages. It is this way how we know, for instance, that Latin deus and Greek theos are not cognate, because a Latin /d/ does not match a Greek /th/. Your approach fails at this, and is a relapse into the kind of etymology about which a saying attributed to Voltaire goes that "consonants count for little and vowels for nothing at all", and is nicknamed "Kling-Klang-Etymologie" by German-speaking etymologists.

[Talskubilos]

The family tree model has its limits; the wave model describes a family like IE, which once was an expanding dialect continuum through which innovations could spread in all directions, resulting in overlapping isoglosses - as have been observed - much better. But that does not dispense of the notion of regular sound correspondences, and in order to judge whether an etymology is valid or not, one has to know these sound correspondences in the relevant languages. It is this way how we know, for instance, that Latin deus and Greek theos are not cognate, because a Latin /d/ does not match a Greek /th/. Your approach fails at this, and is a relapse into the kind of etymology about which a saying attributed to Voltaire goes that "consonants count for little and vowels for nothing at all", and is nicknamed "Kling-Klang-Etymologie" by German-speaking etymologists.

Regular sound correspondences are precisely my primary working method, so please don't repeat this bullshit over and over.

[Talskubilos]

On the other hand, there a somewhat convoluted relationship between the IE numerals '4' and '8', of which Uralic is a node: *kʷet-wor '4' < *kʷet- 'to group in pairs' > Slavic *četá: ‘couple’, Baltic *kek(e)tā ‘troop’ (with reduplication) -> Uralic *kakta/kæktæ '2' -> IE *h₁oḱte-h₃u '8' with a fossilized dual suffix *-h₃u.

The Etruscan numeral cezp '8' is formed from the same root *g´hes- found in IE *g´hes-r- 'hand' plus a dual suffix *-p corresponding to *-h₃u.

[Lērisama]

Regular sound correspondences are precisely my primary working method, so please don't repeat this bullshit over and over.

This is news to me; you don't seem to put foreward your proposals in terms of regular sound correspondences. I think I speak for everyone when I say please show us these rather than asserting X is related to Y without evidence, which gets tiring really quickly.

[Talskubilos]

Regular sound correspondences are precisely my primary working method, so please don't repeat this bullshit over and over.

This is news to me; you don't seem to put foreward your proposals in terms of regular sound correspondences. I think I speak for everyone when I say please show us these rather than asserting X is related to Y without evidence, which gets tiring really quickly.

Perhaps I didn't explain myself well or you didn't read me carefully, but I don't do anything of the kind.
For example, I showed how "Borean" languages such as Uralic and Altaic keep word-initial velar stops while Southern ones, including Kartvelian, NWC, NEC, Semitic and part of IE fricativized them.

[Richard W]

What chance is it that 1st person /m/ is in the same category as /ma/ 'mother'?

Low. Looking at [url=https://wals.info/feature/136B#2/26.1/153.6|WALS Map 135.6[/url], its distribution looks quite clustered. They might be relatively easy to borrow - the Azerbaijani case has been presented as an example for a Turkish word being replaced by a very similar Persian woird, but I haven't dug into that example.

What does strike me as curious is a pairing of 1st person ///m/// and a velar obstruent. For PIE, we have /m/ and /h₂/ and also *egō (or whatever) while in Hungarian we have -m and -k. Now, they might be related, but I think the Hungarian -k is isolated in Uralic. I am also reminded of Proto-Austronesian *aku.

[Lērisama]

Regular sound correspondences are precisely my primary working method, so please don't repeat this bullshit over and over.

This is news to me; you don't seem to put foreward your proposals in terms of regular sound correspondences. I think I speak for everyone when I say please show us these rather than asserting X is related to Y without evidence, which gets tiring really quickly.

Perhaps I didn't explain myself well or you didn't read me carefully, but I don't do anything of the kind.
For example, I showed how "Borean" languages such as Uralic and Altaic keep word-initial velar stops while Southern ones, including Kartvelian, NWC, NEC, Semitic and part of IE fricativized them.

Where did you say this? I've read many sets of proposed word groupings, but it seems like you left guessing the correspondences you assume as an exercise to the reader, which is both easy to fix and seems a lot less crankish¹ than just giving lists of words and lettign us guess.

¹ The other thing that strikes me as crankish is your very high confidence that your findings are certainly correct, rather than a possibility.

[Travis B.]

From looking at a similar case in the Americas, there is a similar pattern of n-m in western North America and Mesoamerica, also crossing multiple families that have not been demonstrated to be related. What is the chance that this sort of thing is areal rather than genetic?

[Ketsuban]

For example, IE *h₂rt-k´o- 'bear', where I'd replace t by a sibilant (possibly palato-alveolar) affricate in order to explain the various outputs t, s, š, kt, kṣ of the "thorny cluster" tk´ (k´þ in older reconstructions), has cognates in [bluuuuuurgh]

*h₂ŕ̥tḱos is very transparently a nominal incorporating the root *h₂retḱ- "damage, destroy" which has Indo-Iranian reflexes (e.g. Sanskrit rákṣas "damage, injury", Avestan rašah "damage"). You do not need to replace anything in the reconstruction to "explain" the outputs; thorn clusters (named because early reconstructions contained a thorn, not because they were "thorny") have been extensively studied specifically because Anatolian and Tocharian provided answers to the question of how they evolved (metathesis, assibilation, sometimes both) in the other branches of Indo-European.

This is why people find discussions with you taxing: the brazenness which which you disregard actual Indo-European reconstructions comes off like you don't believe there were languages ancestral to the Indo-European family, and the whole thing is a shadow created by a cabal of Semitic- and Uralic-speaking magicians. Identifying loanwords is great, but "I don't believe pigs are habitual diggers" and "I think Indo-European speakers were imbeciles who had to be taught how to count by Semitic speakers" are bad reasons to assert a loanword relationship.

[WeepingElf]

There are some things that look like proper comparative method at first glance but aren't. One is the technique used by Nostraticists, both the "Moscow school" and Allan R. Bomhard. These people give sound correspondences and cognate sets for the families they compare, which look nice. But that's deceptive. What they do is to set up reasonable-looking correspondences between the phonemes of the various families and then trawl the families for words that happen to conform to them. With many of the families containing dozens of languages, and matches in one language each in just two of the compared families considered "good enough", it is hardly surprising that they find something. Languages are big, each having thousands of words, so it is no wonder if something matches by mere chance.

How such a procedure can go wrong can be shown by comparing Greek and Armenian that way. Both languages have the same three types of stops - voiceless, voiced¹ and aspirated - and using this technique, one would surely match each type in Greek with the same type in Armenian (the Nostraticists indeed prefer such trivial correspondences), and there are probably some words that seem to match that way (any post-shift Iranian loanword in Armenian which has a cognate in Greek, and whose PIE etymon doesn't involve voiced aspirated stops² would fit!) - but alas, we know that the stop types really do not match that way, as they have been permuted in Armenian by the Armenian sound shift. It tells a lot that both the Moscow school and Bomhard have found hundreds of "cognates" despite using different sound correspondences - they can't both be right, which probably means that both are wrong.

Real comparative work involves finding cognate candidates and then reconstructing the proper sound correspondences from what can be observed in the words resembling each other. This is much more difficult, and gives no results if the languages are unrelated or the relationship is too deep in the past - after about 10,000 years, the comparative method simply runs out of cognates, with correspondence sets so numerous (because of the large number of conditioned sound changes accumulating) and also so small (most of them containing only one single item) that they can no longer be distinguished from chance resemblances. (Also, words are simply lost.) Kind of like how the C14 dating method fails with samples older than 50,000 years because too little C14 is left in them. We can then only catch hints at relationships, such as similar pronouns, but can't prove the case any more by the comparative method because there are too few cognates left.

[1] There is reason to assume that the Old Armenian voiced stops were actually breathy-voiced, inherited intact from PIE. This has been reported for some Eastern Armenian dialects, and it would make sense of the Western Armenian shift where the voiced stops merged with the aspirated stops.

[2] In words with such stops, the PIE voiced aspirated stops appear as aspirated stops in Greek but as voiced stops in Iranian and thus Armenian, spoiling our neat correspondence.