Uralic homeland and genetics and their implications for PIE

[Ryukendo]

When I look at all this stuff it seems like lots of people are switching over to an Eastern origin, or at least softening from the old "Parpola hypothesis" about Volosovo in the Volga. Kallio in his recent presentation on the CONTACTS conference last year (and presumably in a to-be-published manuscript) pointed to the shallowness of Uralic and the fact that all proto-sub-branches of the family look like PU, making all languages seem to split simultaneously from each other ~4000BP. The Saarikivi 2020 paper Nichols references seems to reach the same conclusion (Ref: Saarikivi JS. 2020. Uralic intermediate protolanguages: a descendent reconstruction. See Bákro-Nagy et al. 2020. In press... I can't find it yet, looks like still in the process of publication) Kummel in that conference also points to the preservation of some laryngeals in PIIr.

There seem to be a lot of in press articles that are not out yet. Looks like we're getting a flurry of activity around PU linguistics...

[Kristiina]

I disagree on most of her typological arguments.

Regarding the first set ”Typological similarities with North Pacific Rim languages”, you should first of all go to see how many non-finite verb forms Sanskrit possesses:
https://sanskritstudio.wordpress.com...ples-overview/
https://sanskritstudio.wordpress.com...it-infinitive/
https://sanskritstudio.wordpress.com...nskrit-gerund/
Moreover, my understanding is that PIE did dot use conjunctions either but plenty of nominal forms.

Extensive use of inflectional person marking is also seen in the west, e.g. in Caucasian languages, while in East Asia it is non-existant. To my knowledge personal pronoun roots that contain no semantic feature of person exist only in Nenets where new innovative forms have appeared (2.p. pidər°, 3p. pida). For the rest of the Uralic languages pronoun sets are very similar with IE pronoun sets and reflected on verbal inflection.

Johanna seems to claim that Uralic languages are head-marking, but according to Wals (2 articles written by Johanna herself), Finnish and Hungarian are dependent marking while Nenets is double-marking (https://wals.info/feature/23A#2/26.1/152.9). From the point of view of the whole-language typology Nenets, Hungarian and Finnish are all inconsistent while Ket and Nivkh and many American languages are head marking (https://wals.info/feature/25A#2/26.1/152.9). Not much support to her claim.

As for “relatively high frequency of flexible noun-verb roots”, I do not think that this argument is very relevant. Similarly, in European languages verbs are formed from nouns and nouns are formed from verbs. This is pretty universal. Moreover, English is extremely flexible as it uses exactly the same form as a noun and a verb. Noun incorporation in a verb is a more salient feature in the Pacific-Rim languages and it does not exist in Uralic languages.

The argument on “traces of nonaccusative alignment” is not relevant either, because nonaccusative languages pop up here an there (https://wals.info/feature/100A#2/18.0/149.1) and PIE itself shows very strong indications of ergative pattern.

Regarding the second set ”Typological similarities with Altaic-type”, the first features: “verb-final clauses, noun-final NPs, postpositions, possessor preceding possessed noun, synthetic compounds with the second element as head and primarily or exclusively suffixing morphology (inflectional and derivational)” are seen in the older forms of IE languages as well. These features have often been lost in the west. The following show the order of constituents in a NP in the world languages:
https://wals.info/feature/85A#2/16.3/152.9
https://wals.info/feature/86A#2/21.0/152.9
https://wals.info/feature/87A#2/17.9/144.7
https://wals.info/feature/85A#2/16.3/152.9
Areal patterns are obvious. In this respect, PIE seems to have been different from many Western European languages.

Excluding Ket, all North Eurasian languages are suffixing: (https://wals.info/feature/26A#2/22.6/152.9). As you see, most IE languages are defined as strongly suffixing as Altaic languages. Usually western Uralic languages do not have a strict word order, although eastern varieties tend to have it. On the other hand, many of you may know how strict word order is in Germanic languages.

Regarding Exponence of Selected Inflectional Formatives (https://wals.info/feature/21A#2/26.7/152.6), Johanna puts Finnish, Nenets, Russian, German and Greek in the same category of “case+number”, while Khalka, Turkish and Hungarian belong to the category of monoexponential case. Does this justify her claim that Uralic languages are like Altaic languages?

Transparent morpheme boundaries and minimal sandhi only applies to some Uralic languages. However, please take note on what Daniel Abondolo writes:
Because of its polysemy the term agglutinating is best avoided, certainly in connection with Uralic languages, as it is widely used in a variety of crucially different ways. It can be used to imply (1) relatively high levels of segmentability; or (2) of morpheme-shape consistency; or (3) of consistency of morpheme identity with grammatical function. No Uralic language achieves high levels in any of these areas, and many are quite low in two or all of them.

http://www.oxfordhandbooks.com/view/...0199935345-e-6

Johanna argues that Uralic and Altaic languages have simple syllable structure, but Wals contradicts her claims (https://wals.info/feature/12A#2/19.3/152.9) in that most Turkic languages are described as complex as most IE languages while some Uralic and Tungusic languages are defined less complex. If simple onset means no word initial consonant clusters, this feature is quite frequent in the world languages, e.g. in Spanish and Arabic.

In the end, I must say that I do not understand what she means with this: ”Context-free parsing of certain kinds; in particular, minimal or no effects of person hierarchies on the coding or interpretation of combinations of arguments, and minimal lexical classes determining inflectional paradigms”.

As for phonology, I agree that phonological similarities are obvious between Uralic and Altaic languages. However, phonology is usually areal, e.g. Finnish Swedish sounds more Finnish than Swedish.

Johanna is not accurate enough in her arguments. Her claims that are based on the typological arguments above are probably no reliable.

[Ryukendo]

I disagree on most of her typological arguments.

Regarding the first set ”Typological similarities with North Pacific Rim languages”, you should first of all go to see how many non-finite verb forms Sanskrit possesses:
https://sanskritstudio.wordpress.com...ples-overview/
https://sanskritstudio.wordpress.com...it-infinitive/
https://sanskritstudio.wordpress.com...nskrit-gerund/
Moreover, my understanding is that PIE did dot use conjunctions either but plenty of nominal forms.

Extensive use of inflectional person marking is also seen in the west, e.g. in Caucasian languages, while in East Asia it is non-existant. To my knowledge personal pronoun roots that contain no semantic feature of person exist only in Nenets where new innovative forms have appeared (2.p. pidər°, 3p. pida). For the rest of the Uralic languages pronoun sets are very similar with IE pronoun sets and reflected on verbal inflection.

Johanna seems to claim that Uralic languages are head-marking, but according to Wals (2 articles written by Johanna herself), Finnish and Hungarian are dependent marking while Nenets is double-marking (https://wals.info/feature/23A#2/26.1/152.9). From the point of view of the whole-language typology Nenets, Hungarian and Finnish are all inconsistent while Ket and Nivkh and many American languages are head marking (https://wals.info/feature/25A#2/26.1/152.9). Not much support to her claim.

As for “relatively high frequency of flexible noun-verb roots”, I do not think that this argument is very relevant. Similarly, in European languages verbs are formed from nouns and nouns are formed from verbs. This is pretty universal. Moreover, English is extremely flexible as it uses exactly the same form as a noun and a verb. Noun incorporation in a verb is a more salient feature in the Pacific-Rim languages and it does not exist in Uralic languages.

The argument on “traces of nonaccusative alignment” is not relevant either, because nonaccusative languages pop up here an there (https://wals.info/feature/100A#2/18.0/149.1) and PIE itself shows very strong indications of ergative pattern.

Regarding the second set ”Typological similarities with Altaic-type”, the first features: “verb-final clauses, noun-final NPs, postpositions, possessor preceding possessed noun, synthetic compounds with the second element as head and primarily or exclusively suffixing morphology (inflectional and derivational)” are seen in the older forms of IE languages as well. These features have often been lost in the west. The following show the order of constituents in a NP in the world languages:
https://wals.info/feature/85A#2/16.3/152.9
https://wals.info/feature/86A#2/21.0/152.9
https://wals.info/feature/87A#2/17.9/144.7
https://wals.info/feature/85A#2/16.3/152.9
Areal patterns are obvious. In this respect, PIE seems to have been different from many Western European languages.

Excluding Ket, all North Eurasian languages are suffixing: (https://wals.info/feature/26A#2/22.6/152.9). As you see, most IE languages are defined as strongly suffixing as Altaic languages. Usually western Uralic languages do not have a strict word order, although eastern varieties tend to have it. On the other hand, many of you may know how strict word order is in Germanic languages.

Regarding Exponence of Selected Inflectional Formatives (https://wals.info/feature/21A#2/26.7/152.6), Johanna puts Finnish, Nenets, Russian, German and Greek in the same category of “case+number”, while Khalka, Turkish and Hungarian belong to the category of monoexponential case. Does this justify her claim that Uralic languages are like Altaic languages?

Transparent morpheme boundaries and minimal sandhi only applies to some Uralic languages. However, please take note on what Daniel Abondolo writes:
Because of its polysemy the term agglutinating is best avoided, certainly in connection with Uralic languages, as it is widely used in a variety of crucially different ways. It can be used to imply (1) relatively high levels of segmentability; or (2) of morpheme-shape consistency; or (3) of consistency of morpheme identity with grammatical function. No Uralic language achieves high levels in any of these areas, and many are quite low in two or all of them.

http://www.oxfordhandbooks.com/view/...0199935345-e-6

Johanna argues that Uralic and Altaic languages have simple syllable structure, but Wals contradicts her claims (https://wals.info/feature/12A#2/19.3/152.9) in that most Turkic languages are described as complex as most IE languages while some Uralic and Tungusic languages are defined less complex. If simple onset means no word initial consonant clusters, this feature is quite frequent in the world languages, e.g. in Spanish and Arabic.

In the end, I must say that I do not understand what she means with this: ”Context-free parsing of certain kinds; in particular, minimal or no effects of person hierarchies on the coding or interpretation of combinations of arguments, and minimal lexical classes determining inflectional paradigms”.

As for phonology, I agree that phonological similarities are obvious between Uralic and Altaic languages. However, phonology is usually areal, e.g. Finnish Swedish sounds more Finnish than Swedish.

Johanna is not accurate enough in her arguments. Her claims that are based on the typological arguments above are probably no reliable.

I'm not qualified to comment on most of this stuff, but I really don't think she is as incompetent as you make her out to be, she thanks Holopainen, Saarikivi, Khanina etc in her manuscript while she was at uni Hensinki and they probably read through all that. And she literally invented the concept of head vs dependent marking in linguistic typology, I don't think shes wrong when she says what she does on that front at least...

I did a quick google search for two items on the list and found e.g. for inflectional marking of pronouns in West Caucasian languages, which she actually mentions in: https://www.degruyter.com/document/d...2017-0010/html; "Languages with primarily inflectional person include several East African languages, the West Caucasian language family, and many languages of the Americas, e.g., languages of the Salish, Siouan, Algonquian, Kiowa-Tanoan, and Keresan families.", alongside that she mentions the lowest level of this trait is found in e.g. Nakh-Dagestanian and in areas in Southeast Asia, like what you said. She talks about the general increase in this trait moving East and North away from Africa, not Southeast. The connection is to North America and the Pacific coast, so shes not contradicting you on that front.

About head- vs dependent-marked languages, theres only three Uralic languages coded in WALS for that and all are somewhat atypical (Finnish, Samoyed, Hungarian), here's what she had to say from her 1989 article that introduces this concept:
"An example of increased dependent-marking through boundary shifting is
found in the Uralic family, whose western members gradually change their type
from double-marking to dependent-marking-partly by restricting or losing
possessive affilxes, and partly by adding to the inherited case inventory through
accretion of postpositions (see Oinas for examples of the latter process... )"
Merljin de Smits says: "The consensus view on Proto-Uralic argument case-marking is that Proto-Uralic was a nominative-accusative, partially head-marking language... Proto-Uralic was head-marking in that, to some extent, object cross-reference on the verb may be reconstructed to Proto-Uralic: this co-reference took theshape of an element *-sV" (Janhunen 1982: 35)..."

Whether or not her arguments are accurate will depend on statistical data, and not individual or family-level exceptions--for a lot of WALS features you get only three or two (typically Finnish and Hungarian, sometimes Samoyed) Uralic languages, and if you see her presentation she has a much larger number of Uralic datapoints (really a lot more datapoints than WALS for Northern Eurasia in general). Like you said she was involved in the WALS database, and its very likely data she will present in Grunthal. et. al. will be partially from WALS (i.e., actually the same data you are using), but possibly expanded in some way. As for the coding, I highly doubt Uralic linguists as reviewers or readers will let her get away with miscoded data, especially not from a contributor to WALS.

[CopperAxe]

Reviving this thread, what do people make of this publication?

I havent been able to read it yet and given that I'm not much of a linguist I wouldn't be able to properly assess the findings, but the abstracts seems really in line with what I and many others are seeing in the archaeological record and population genetics from the past and present, and when linguistics, archaeology and genetics all fit together like three peas in a pod I think it is very telling.

[Kristiina]

The problem is that there is a lot of variation among languages in one family. A part of the languages point to one direction, and others to other directions. Decision on what is the original feature is often subjective. If a linguist wants to prove an eastern origin, he/she claims that the eastern variant is original, or vice versa. In the IE case we can be much more certain of the original form as we have old texts. In the Uralic case, we must use the word 'may' too often.

The other thing is that she seems not pay enough attention to what we know on the structure of PIE. On many aspects, it is similar with Uralic languages (order of constituents, SOV, case system, verbal agreement type, nominal verb forms).

IMO, there is a continuum across Eurasia: IE languages - Uralic languages - Micro-Altaic languages - Paleo-Siberian languages (the most heterogeneous group)

Uralic languages share traits most of all with languages that are their geographic neighbours.

I know the structure of more Uralic languages than compared on WALS in the links that I posted. I put the links because they make clear to all what is involved here.

[Kristiina]

On the head marking issue, I would still like to make a comment.

Go to see the examples on head marking on WALS: https://wals.info/chapter/24
You see that genitive case means dependent marking: dog's tail.

If you go to see the Proto-Uralic website, you see that genitive case is reconstructed to Proto-Uralic: https://en.wikipedia.org/wiki/Proto-Uralic_language

How can a proto-language be head marking, if it has a genitive?

Among Uralic languages, Ugric languages are head-marking and lack genitive. On the map we see that Ket is head-marking. I would not be surprised if for example Okunevo language was head-marking.

On verbal person marking (https://wals.info/feature/102A#2/8.4/178.6), I would like to add that it exists in Ket, Ainu and Eskimo languages (and in Basque, Georgian and NW Caucasian group, among others) and to a minor extent in half of the Uralic languages (Erzya, Ugric, Samoyedic). It is probably a Paleo-Siberian trait. However, among the Uralic languages it needs not come from America or the Pacific Rim as it may have been widespread in the languages around the Yenisei area.

This is from Wals:
"To give another example, in Khanty (Ob-Ugrian, Uralic; Siberia) as well as various other Uralic languages (e.g. Hungarian, Nenets, Mordvin, Mansi, and Selkup) there are person suffixes for the A which are preceded by suffixes for the P. The latter, however, indicate only number and not person, as shown in (13)...
Given that Map 102A is about the marking of person, markers of gender and/or number markers such as those in Archi and Khanty have not been considered."

https://wals.info/chapter/102

The only Uralic language that uses extensively inflectional person marking is Mordvin. Is this enough to state that Proto-Uralic used extensively inflectional person marking?

[Ebizur]

The problem is that there is a lot of variation among languages in one family. A part of the languages point to one direction, and others to other directions. Decision on what is the original feature is often subjective. If a linguist wants to prove an eastern origin, he/she claims that the eastern variant is original, or vice versa. In the IE case we can be much more certain of the original form as we have old texts. In the Uralic case, we must use the word 'may' too often.

For many languages that possess a traditional writing system, major changes in typology are visible in the historical record, both in regard to grammar (e.g. the loss of the noun class/gender system and the decay of the marking of case on nouns and the marking of person and number of the subject and mood on verbs in English) and phonology (e.g. the shift from a very simple syllable structure in the earliest records of Mongolian as well as reconstructed proto-Mongolic to the fairly complex syllable structure of modern Khalkh). The Korean language appears to have gone full circle in regard to phonotactics: proto-Korean most likely had a simple CV(N)-type syllable structure, but by the time of the earliest records in hangeul script, the language had developed complex syllables allowing consonant clusters (of up to three consonant phonemes) in both the onset and the coda of a syllable and complex nuclei with glides and diphthongs. Most of the phonotactic complexity of Middle Korean has been lost in the Korean language as it is spoken today, though traces of the former consonant clusters may be found in a few N+N compounds (e.g. Middle Korean ʨoh "foxtail millet (plant), Setaria italica" + psɔl "hulled grain, [esp.] hulled grains of rice" > Modern Korean ʨopsal "foxtail millet, hulled grains of Setaria italica," Middle Korean pje "rice (plant), Oryza sativa" + psi "seed" > Modern Korean pjɤpsi "seed rice, grains of rice with the hull and germ intact").

[Kristiina]

Whether or not her arguments are accurate will depend on statistical data, and not individual or family-level exceptions--for a lot of WALS features you get only three or two (typically Finnish and Hungarian, sometimes Samoyed) Uralic languages, and if you see her presentation she has a much larger number of Uralic datapoints (really a lot more datapoints than WALS for Northern Eurasia in general). Like you said she was involved in the WALS database, and its very likely data she will present in Grunthal. et. al. will be partially from WALS (i.e., actually the same data you are using), but possibly expanded in some way. As for the coding, I highly doubt Uralic linguists as reviewers or readers will let her get away with miscoded data, especially not from a contributor to WALS.

I checked that paper. She seems to argue that Uralic languages belong to Pacific-Rim cluster. Does this mean that she suggests that Houtaomuga individual spoke pre-Proto-Uralic 5000 BC and Baikal Neolithic 6000 BC Proto-Uralic?

In her typological analysis she focuses on two parametres: realization of causative alternation and inflectional person. In the first graph we see that Uralic languages are as causative a Indo-Iranian languages, although Hindi beats all. In the second graph we see that Uralic languages are less causative than Mongolic, Turkic and Paleosiberian languages. The great conclusion is that Uralic languages cluster with Pacific Rim languages. On p. 31 we see that Proto-Uralic used more inflectional person than PIE (which is a presumption). Then we see that percentage of inflectional person reaches its maximum in America. The conclusion is that Proto-Uralic homeland is in the east. However, we see that there are languages that use even more inflectional person to the west from Uralic languages. Uralic languages are probably on the same level as Micro-Altaic languages. This situation again seems to justify that Uralic languages belong to Pacific Rim languages.

The last part of her article is dedicated to fur trading hypothesis. We now have Iron Age samples from Estonia. Are Iron Age Estonians fur traders? We also have ancient Hungarian genomes. I do not think that anybody thinks that they were fur traders. I definitely want to get relevant ancient DNA before I formulate any theories about the expansion of Uralic languages. I do not think that what we have now (Baikal Neolithic, Houtaomuga and Ymyiakhtakh) are relevant for Proto-Uralic.

[Kristiina]

I'm not qualified to comment on most of this stuff, but I really don't think she is as incompetent as you make her out to be, she thanks Holopainen, Saarikivi, Khanina etc in her manuscript while she was at uni Hensinki and they probably read through all that. And she literally invented the concept of head vs dependent marking in linguistic typology, I don't think shes wrong when she says what she does on that front at least...

No, I do not think that she is incompetent in typology, but she is too eager to invent theories which she then tries to prove with aggregation of disparate data.

[Ryukendo]

I definitely want to get relevant ancient DNA before I formulate any theories about the expansion of Uralic languages. I do not think that what we have now (Baikal Neolithic, Houtaomuga and Ymyiakhtakh) are relevant for Proto-Uralic.

What is your view on Kra001?