Language as a Clue to Prehistory

[lpetrich]

Back to "IE Secondary Products Terminology and the Dating of PIA".

Anatolian split off from the rest of IE some time around 4000 BCE, and their first archeological evidence is likely the  Cernavodă culture (4300 - 4000 BCE) on the lower Danube River.

The ancestors of the Tocharians split off later, as the  Afanasievo culture (3300 - 2500 BCE) of central Asia.

Proto-Core-Indo-European had no reconstructible words for such crops as barley, lentil, pea, chickpea and bitter
vetch, but it did have words for wheel and axle: *kwekwlos and *Hrot(H)os for wheel and *h2eks- > *aks- for axle.

*kwekwlos < *kwel- "to roll, revolve"
*Hrot(H)os < *Hret(H)- "to run" (only attested in Celtic)

From Tocharian A kukäl, B kokale "wagon", one infers that *kwekwlos was present in Indo-Tocharian.

PCIE had words for taming a horse, milking, tending livestock, different words for bull, cow, ox, different words for sheep, ram, lamb, and words for udder, butter, pasture, herdsman, cowherd, hornless, and wool.

"All these words suggest a pastoral society in which domesticated cattle, horses and sheep played an important role and were not bred exclusively for their meat."

Some of these words are also found in Anatolian, like cow, sheep, horse, pasture, and shepherd, and possibly also the spring season, *wes-r/n- "if this was the season when the herd returned to the open grasslands from a winter camp in a more forested area".

As to the other possibilities,

• Anatolian has a cognate with a different meaning
• Anatolian has a different word form with that meaning
• Anatolian has a logogram with no pronunciation clue for that meaning
• Anatolian has no uses of that word in the surviving texts

In the first one, *peh2- > *pâ- "to protect; to tend livestock" has a Hittite cognate that means "to protect, guard, keep (an oath)" and a Tocharian one that means "to protect, beware of, obey (rules)". Also, *demh2- "to tame" has a Hittite cognate that means "to (op)press".

In the second one, Hittite has a separate word form for udder, and a possibly-cognate one for wool. Separate borrowings from the same North Caucasian source? hulana vs. *ulh1neh1- > wlanâ (English wool, Latin lana, ...)

In the last two, Hittite words for milk and to milk are unrecoverable, as are for ox, lamb, ram, bull, and butter.

 

[lpetrich]

After discussing words for yoke and yoke pole, the article says "The speakers of Proto-Indo-Anatolian probably did not practise agriculture, because not a single agricultural term is shared between Anatolian and the other branches of Indo-European."

Mentioning a word for plow shared across much of Core IE: *h2erh3trom > *arotrom -- more-or-less "plowing thing" with "to plow" likely having an Anatolian cognate that means "to crush".

Then discussing the evidence of animal domestication.

Domestic bovines, sheep, goats, and probably horses were known in Dnieper-Donets II (5200 - 5000 BCE) and also Khvalynsk (4700 - 3800 BCE).

Animal traction? "The oldest archaeological evidence for animal traction comes in the form of direct archaeological remains of ploughs, wheeled vehicles, cheek pieces and nose rings, images and models of sledges, wheeled vehicles, ploughs and yoked animals, impressions of wheels, ploughs and threshing sledges on fossil soil, zoopathological traces on cattle skeletons, interpretation of cattle mortality patterns and burials of paired cattle ."

In Knossos in Crete, bovines used for traction were mostly female in the Neolithic, then mostly male in the Bronze Age. Evidence: "traction-related pathologies and sexable pelves". This may be due to castration of male bovines becoming common, from bronze knives outperforming stone ones in this act. Improved cutting tools may also have made possible the invention of the wheel.

Attested words for this act, like Latin castrâre, tend to be derived from more general words for "to cut", so the PIE word for this act is now lost. Was it also "to cut"?

 

[lpetrich]

Dairying? "While goat milking appears to have been practised throughout Neolithic Europe, sheep and cow milking have been argued to be post-Neolithic phenomena."

An oddity is that words for goat in the IE langs vary quite a lot, making it difficult to reconstruct a PIE word for this animal. So the PIE speakers likely did not have had goats, though they did have bovines and sheep and pigs.

Sheep were originally domesticated for their meat, but the first evidence of wool production comes from SW Asia around 4500 BCE, when sheep became larger and more robust, with less distinction between the sexes, and 4000 BCE, the proportion of male sheep increases significantly. Lots of castrated ones?

In conclusion, "Speakers of Proto-Indo-Anatolian made use of yoked draught animals to pull a load, perhaps with the help of a sledge or a travois, but they cannot be demonstrated to have had knowledge of the wheel, the plough, dairying or wool production."

•  Samara culture (5200 - 4500 BCE) - near the Samara Bend in the Volga River
•  Khvalynsk culture (4900 - 3500 BCE) - between Samara Bend, Sea of Azov, North Caucasus, Ural River
•  Yamnaya culture (3300 - 2600 BCE) - between W Ukraine and the Ural River
•  Corded Ware culture (3000 - 2350 BCE) - between Belgium and Moscow

and

•  Dnieper–Donets culture (5000 - 4200 BCE)
•  Sredny Stog culture (4500 BCE - 3500 BCE) - in Dnieper–Donets area

 

[lpetrich]

Indo-European cereal terminology suggests a Northwest Pontic homeland for the core Indo-European languages - PMC

On the basis of the evolution of the subsistence strategies of consecutive stages of the protolanguage, we find that one or perhaps two cereal terms can be reconstructed for the basal Indo-European stage, also known as Indo-Anatolian, but that core Indo-European, here also including Tocharian, acquired a more elaborate set of terms. Thus, we linguistically document an important economic shift from a mostly non-agricultural to a mixed agro-pastoral economy between the basal and core Indo-European speech communities. It follows that the early, eastern Yamnaya of the Don-Volga steppe, with its lack of evidence for agricultural practices, does not offer a perfect archaeological proxy for the core Indo-European language community and that this stage of the language family more likely reflects a mixed subsistence as proposed for western Yamnaya groups around or to the west of the Dnieper River.

So increasing technology here also.

 

[lpetrich]

On Colexification among Basic Vocabulary | Journal of Universal Language
"We speak of colexification (cf. e.g., François 2008) when two distinct word meanings are expressed in one language by a single lexeme (i.e., are colexified)."

This includes polysemy (related meanings) and homonymy (coincident meanings).

The work was done with the  Automated Similarity Judgment Program (ASJP)'s word lists, a set of 40 highly-conserved meanings in many languages.

"In particular, it was found that shared colexifications corroborate the postulation of the Austric stock and the attribution of Sumerian to the Tibeto-Burman language family."

Author Vladimir Pericliev found these widespread colexifications.
Numbers: number of langs (number of families)

Polysemies:

• <mountain=stone> 140 (25)
• <I=we> 89 (31)
• <ear=leaf> 67 (14)
• <fire=tree> 64 (13)
• <horn=knee> 48 (3)
• <feather=hair> 44 (19)
• <bark=skin> 42 (21)
• <ear=hear> 41 (18)
• <drink=water> 40 (15)
• <man=person> 33 (18)

All of these are fairly obvious, with the exception of fire = tree and horn = knee. A mountain is a big stone, bark is the skin of a tree, one hears with one's ears and one drinks water, etc. In fact, in Italy there is a mountain called Gran Sasso "Big Stone".

Homonym sets:

• <come=dog> 78 (3)
• <louse=we> 59 (4)
• <name=tooth> 46 (7)
• <liver=two> 43 (3)
• <ear=fish> 32 (3)
• <ear=name> 35 (5)
• <see=we> 34 (3)
• <die=eye> 40 (2)

Not quite as common as the polysemies, at least by family.

Looking at the putative long-range relationships, I find the methods inadequate. The author should have compared protolanguages as far as is possible to do so. This also means checking on colexifications in protolanguages vs. attested languages. Does one find developments like replacement of an earlier form of "we" with one that is the plural of "I"?

Also the ASJP list of meanings is small, and that risks coincidence.

 

[lpetrich]

Phonotactic Diversity Predicts the Time Depth of the World’s Language Families | PLOS ONE - that's sequences of language sounds (phonemes).

Every language has some subset of all possible sequences, like possible consonant clusters and what consonants can end a syllable.

The authors found the diversity of substrings or n-grams in a family and compared it to a putative age of that family, finding a correlation. They used only total number, when one could use methods that take into account different numbers of them, like 2 to power (number of bits to distinguish them) and the effective number of political parties. Formulas like:

\( \displaystyle{ 2^B \text{ where } B = - \sum_p p \log_2 p } \)
\( \displaystyle{ 1/P \text{ where } P = \sum_p p^2 } \)

The authors used the ASJP database, and used the families and calibration points in Automated Dating of the World’s Language Families Based on Lexical Similarity | Current Anthropology: Vol 52, No 6 with PDF file Automated Dating of the World’s Language Families Based on Lexical Similarity - shh768.pdf

They found similar dates, likely by fitting the same kind of model. I could not find the numbers of various sizes of n-gram for each of the families. I consider that model flawed, because it is a straight exponential model, assuming that each meaning has word forms that "decay" at the same rate. But some meanings are more stable than others, and I indeed find a better fit with the decline's exponential factor decreasing at greater ages.

 

[lpetrich]

I've found online John Bengtson's book on Euskaro-Caucasian / Vasco-Caucasian / Basque and North Caucasian.

I'm a bit reluctant to mention this since he seems to be almost the only one working on E-C, though Corinna Leschber has joined him in recent years. As a comparison, I note Papers from the 30th Annual Meeting of SEALS - JSEALS_Special_Publication_8_SEALSXXX.pdf

NEW EVIDENCE FOR AUSTRO-TAI AND OBSERVATIONS ON VOWEL CORRESPONDENCES
Alexander D. SMITH

at PDF page 82. It takes Weera Ostapirat's work and extends it, finding additional sound correspondences. So Austro-Tai is not just WO's work.

At researchgate.net and academia.edu is some of JB's work, so I looked there.

(PDF) On Criticism of S. L. Nikolayev/S. A. Starostin, A North Caucasian Etymological Dictionary

It has been noticed that the most comprehensive etymological dictionary of North Caucasian languages (NCED) has repeatedly been disparaged, and even totally ignored, due to alleged deficiencies of the lexicon. One also finds that some critics who dismiss this dictionary do not enumerate the shortcomings of the book themselves, but simply cite two reviews, one by J. Nichols and another by the late W. Schulze. Moreover, we find that the Nichols review has not been published and is not accessible to the examination of scholars, and the Schulze review, while citing a number of disagreements, is far from advocating the disregard of this volume but instead claims that it “belongs in the bookcase of anyone interested in etymological research”. It is also observed that more positive reviews, for example those by J. C. Catford (pre-publication), M. E. Alekseev, Ja. G. Testelec, and V. Chirikba (Čirikba) are overlooked by the detractors. These circumstances are examined in this paper, with the conclusion that the NCED, like any pioneering work, is not a permanent solution but a set of hypotheses that will have to be tested and modified, where necessary, over the coming decades. It does not deserve to be dismissed or ignored, but rather engaged with and discussed in the pursuit of better solutions to etymological problems of North Caucasian languages.

 

[lpetrich]

(PDF) On the quantification of Euskaro-Caucasian lexical matches (cognates) compared with Indo-European and Indo-Uralic Out of an "ultra-stable" list of 50 meanings, they find 7 very good comparisons: 2 two, 7 fire, 8 tongue, 12 what, 24 dry, 36 smoke, 40 star.

I've given their ranking by stability. Note that 5 ouf ot these 7 are in the most stable half.

(PDF) Proto-indo-European-uralic comparison from the probabilistic point of view also has 7 good ones: 3 I, 5 thou, 6 who, 10 name, 15 to drink, 28 water, 45 to hear.

Also 5 out of 7 in the most stable half, though entirely different meanings. I find the probability of being entirely different for two selections of 7 out of 50 to be 0.323. For 5 out of 25, 0.292.

 

[Swammerdami]

Back to "IE Secondary Products Terminology and the Dating of PIA".

Anatolian split off from the rest of IE some time around 4000 BCE, and their first archeological evidence is likely the  Cernavodă culture (4300 - 4000 BCE) on the lower Danube River.

The ancestors of the Tocharians split off later, as the  Afanasievo culture (3300 - 2500 BCE) of central Asia.

Proto-Core-Indo-European had no reconstructible words for such crops as barley, lentil, pea, chickpea and bitter
vetch, but it did have words for wheel and axle: *kwekwlos and *Hrot(H)os for wheel and *h2eks- > *aks- for axle.

*kwekwlos < *kwel- "to roll, revolve"
*Hrot(H)os < *Hret(H)- "to run" (only attested in Celtic)

Anatolian did not inherit all of the words for wheel. This is logical since the estimated date for wagon development is after the Anatolian-IE_Proper split.

From Tocharian A kukäl, B kokale "wagon", one infers that *kwekwlos was present in Indo-Tocharian.

More definitely, ceramic doll wagons (children's playthings) have been found in Afanasievo excavations. (Any actual wooden wagons and wheels would have disintegrated, but the ceramic dolls are preserved.)

PCIE had words for taming a horse, milking, tending livestock, different words for bull, cow, ox, different words for sheep, ram, lamb, and words for udder, butter, pasture, herdsman, cowherd, hornless, and wool.

Yes. That the PIE People strongly emphasized stockbreeding is clear from their lexicon. IIRC there are half a dozen words for parts of a horse's body. This makes the decades-long refusal to accept the Gimbutas theory all the more baffling.

With the development of advanced wheel wagons, the Yamnaya people could become prosperous and semi-nomadic. This was the era when metallurgy was developing, and the high mobility of Yamnayan herders brought them into contact with both supply and demand sides of the traffic in tin and other metals.

I don't think it is known which Central Eurasian culture actually "invented the wagon" but Yamnaya took best advantage of it, and it was key -- along with secondary products, metallurgy knowledge (less important was their horsemanship?) -- to their successful expansion.

 

[lpetrich]

(PDF) The anthropological context of Euskaro-Caucasian

Shared words for: bovine, sheep, goat, pig, milk, to milk, butter, grains (wheat, rye, ...), legumes (bean, pea), land-raking tools, threshing tools, sieve, mill, to grind, flour.

No horses or donkeys or plows.

Proto-North-Caucasian had at least six words for metals -- "silver", "bright metal", "red copper, gold", "brass, gold", "lead", "silver, gold" -- and none of them have Basque cognates. There is one NC metal-related term with a Basque cognate, but its semantics include not only "brass, copper", but also several metal objects: bell, cowbell, belt buckle, hook, ...

This vocabulary is consistent with an ancestor of Basque having been introduced by Neolithic farmers, more specifically, the people of  Cardium pottery cutlure (6400 - 5500 BCE), named from its pottery having imprints of heart-shaped seashells. They are an offshoot of  Neolithic Greece (7000 - 3200 BCE), in turn descended from  Pre-Pottery Neolithic B (8800 - 6500 BCE) in the Fertile Crescent, in turn descended from  Pre-Pottery Neolithic A (10,000 - 8800 BCE) in the Levantine and Upper Mesopotamian parts of the Fertile Crescent, along with  Khiamian culture (9700 - 8650 BCE) -- the first farmers.

A date of 7000 BCE (9,000 BP) agrees with lexicostatistical calculations of the breakup time of Basque and North Caucasian.

 

[lpetrich]

Yes. That the PIE People strongly emphasized stockbreeding is clear from their lexicon. IIRC there are half a dozen words for parts of a horse's body. This makes the decades-long refusal to accept the Gimbutas theory all the more baffling.

I've never heard of that one.

 

[lpetrich]

The Basque Language: History and Origin. | John D Bengtson - Academia.edu

A short version of his big book on Euskaro-Caucasian, a book that he has placed online:

BCR: Bengtson, John D. 2017. Basque and its Closest Relatives: A New Paradigm. Cambridge, Mass.: Mother Tongue Press/Association for the Study of Language in Prehistory. | Corinna Leschber - Academia.edu

I've downloaded it, and it it looks good.

Author John Bengtson is not stopping there. He is working on “Egyptian Loanwords in Basque” - like for 6 and 7.

Egyptian? Or some long-lost Afroasiatic language, one that got submerged by Phoenician, Latin, Berber, and Arabic.

 

[lpetrich]

Informal review of Heggarty et al. 2023 (Indo-European phylogeny) | Alexei S . Kassian and George Starostin - Academia.edu
by Alexei S . Kassian and George Starostin

responding to

Language trees with sampled ancestors support a hybrid model for the origin of Indo-European languages | Science
by Paul Heggarty et al.

Proposing an origin in what is now Armenia around 8000 BP (6000 BCE) with a population splitting off to what is now Ukraine and nearby around 7000 BP (5000 BCE).

Episode 1. Gray & Atkinson 2003 (Nature), “Language-tree divergence times support the Anatolian theory of Indo-European origin”.

Episode 2. Bouckaert et al. 2012 (Science), “Mapping the origins and expansion of the Indo-European language family”, “... decisive support for an Anatolian origin over a steppe origin”, “Indo-European languages originate in Anatolia”.

Episode 3. Heggarty et al. 2023 (Science), “Language trees ... support a hybrid model for the origin of Indo-European languages”, “New insights into the origin of the Indo-European languages”, “... significant breakthrough in our understanding of the origins of Indo-European”.

This is at least the third version of the Indo-European phylogeny published by the team led by Russell D. Gray & Quentin D. Atkinson.

...
Running ahead, the new lexical database, IE-CoR, which was essentially compiled in accordance with the philosophy and methodology elaborated within the Moscow School, is unquestionably impressive: the number of detected errors is modest (and, of course, some mistakes are always inevitable in such a large-scale project). On the contrary, the obtained tree classification is not only statistically weak, but also runs contrary to solidly established knowledge in some important points. Thus, it can hardly serve as a firm basis for further research.

The authors then criticize the IE-CoR.

Like "Introduction of explicit semantic specification of the concepts and diagnostic contexts." - for instance, in English,"who" is both an interrogative pronoun and a relative one:

Who is writing this post?
I am the one who is writing this post.

Like this: The Swadesh wordlist. An attempt at semantic specification by Alexei Kassian, George Starostin, Anna Dybo

I must note that Russian has the same problem as English with kto "who" - it is both an interrogative pronoun and a relative one there also.

 

[lpetrich]

"all the slots are filled with neutral and most commonly used words, parasitic synonyms are excluded"

"authoritative dictionaries and grammars are used (instead of anonymous Swadesh wordlists from Wikipedia, as in Bouckaert et al. 2012)"

I have to laugh at the idea of using Wikipedia as a primary reference, and I myself have contributed to that online encyclopedia.

"the involved forms are linked to the corresponding entries of online dictionaries (if available, e.g., Latin forms are linked to Lewis & Short’s Latin Dictionary)" -- which would certainly qualify as authoritarive.

They then get into other problems, including "Individual languages lack the list of sources used for wordlist compilation, i.e., which dictionaries and grammars have been used." and "Individual forms lack references to the sources." and "Comments on individual forms are too short." and "Forms lack morphemic division (affixes should be separated from the root)." Also such things as using only one of imperfective and perfective variants of verbs, when they ought to be treated as synonyms.

Checking on four of the langs in IE-CoR revealed only a few errors.

 

[lpetrich]

"The genealogical tree of the IE family offered by Heggarty et al. is not convincing. Some important nodes have very low probability, the tree is poorly resolved (see below), and some important nodes contradict our historical knowledge."

Heggarty et al. published a DensiTree, a lot of trees overlaid on each other. Trying to find a consensus tree gives

Anatolian, Tocharian, Albanian, Indo-Iranian, (Greek, Armenian), (Balto-Slavic, (Italic, (Celtic, Germanic) ) )

Instead of (Balto-Slavic, Indo-Iranian), and instead of Anatolian branching off, then Tocharian.

BSl = Balto-Slavic, IIr = Indo-Iranian, WEu = West European (Italic, Celtic, Germanic)

BSl shows specific affinity with both IIr and WEu, which suggests that either within the pair of BSl-IIr or that of BSl-WEu strong contacts took place in prehistoric times. Linguistically, BSl-WEu ties produce the impression of being due to contact origin rather than common inheritance (sharing of specific cultural vocabulary; irregular phonetic correspondences, etc.). On the other hand, the BSl-IIr pair looks like a typical ancient clade with minor post-split contacts. This remains beyond the scope of the present review, but we suspect that in their analysis, Heggarty et al. mistook certain contact-induced innovations between BSl and WEu as inherited retentions.

 

[lpetrich]

In order to preventively defend their results from criticism by "traditionalist Indo-Europeanists", Heggarty et al. include several sections of a more theoretical nature into the Supplement, such as an attack on the method of “linguistic paleontology” (pp. 19-21) which, to us, seems highly undeserved and, in places, bordering on the irrational.

AK and GS then argue that concrete meaning > abstract meaning is much more common than the reverse, meaning that "wheel" > "abstract circle" is much more likely than "abstract circle" > "wheel".

Heggarty et al. ask "Linguistic Paleontology — Did Speakers of Proto-Indo-European Know the Wheel?" - trying to account for the Anatolian langs not having cognate words for "wheel", when that's easy to account for is the ancestors of the Anatolian speakers split off before some early IE speakers either invented the wheel or acquired it.

AK and GS: "Overall one could say that Heggarty et al. 2023’s tree in many respects is even less reasonable than the IE trees by Gray & Atkinson 2003 and Bouckaert et al. 2012."

 

[lpetrich]

(PDF) The "Nostratic" roots of Indo-European: From Illich-Svitych to Dolgopolsky to future horizons by George Starostin, Mikhail Zhivlov, and Alexei S. Kassian

They seem very critical of how slow progress has been in Nostratics since the pioneering work of Vladislav Illich-Svitych and his colleagues, and how little mainstream acceptance it has gotten.

The one major methodological breakthrough achieved by Illich-Svitych (even regardless of whether one accepts or rejects his practical results) was his ability to show how reconstructed protolanguages, usually believed to be the “end” of the reconstruction, such as Proto-Indo-European, Proto-Uralic, or Proto-Dravidian, may actually be substituted for historically attested languages and compared between themselves, as part of a recurrent process, in strict formal accordance with the classic comparative method. This has elevated macro-comparative linguistics from mere speculation to the status of a discipline that is at least capable of generating falsifiable hypotheses, and thus deserving of scholarly attention.

I agree that that is a good method. It uses in a very concise way all the work on the history of well-established language families.

It is, however, not a matter of debate that, since the original publications of Illich-Svitych, not a single piece of research on Nostratic has appeared that the “international community of Nostraticists” would unanimously value as equal to, let alone surpassing the value of Illich-Svitych’s pioneering enterprise, adding it to the figurative “golden fund” of important discoveries in the field. Not a single correspondence has been made more precise, not a single new grammatical morpheme has been reconstructed, not a single new lexical etymology has been suggested in such a way that would convince every single supporter of the Nostratic hypothesis of its truthfulness, or would make it look undeniably more probable than any possible alternative.

The authors then compare the great success of historical linguistics with Indo-European, with its numerous well-established etymologies and sound correspondences.

 

[lpetrich]

Comparing 50 highly-stable word forms across Indo-European, one finds around 20 - 25 matches, some of them not very obvious when one only compares only those words in pairs of langs, while IE and Uralic have only 7 phonetically-clear matches.

Then discussing various schools of thought about Nostratics: the "Charleston School" of Allan Bomhard, the "Haifa School" of Aharon Dolgopolsky, and the "Moscow School" of successors of VIS in that city like Sergei Starostin, Vladimir and Anna Dybo, Sergei Nikolaev, and GS and AK. They might also have mentioned the "Palo Alto School" of Joseph Greenberg.

AB places a lot of emphasis on regularity of sound correspondences, though at the expense of similar precision of semantics, resulting in a lot of phonetically good but semantically bad etymologies.

AD built on VIS's work, but he listed oodles of synonyms: head 5 (primary meaning), 5 (figurative meaning), brain 3, eye 7 (pure 2, eye / to see: 7), nose: 4, tongue 5, tooth: 7, hand 10 (not counting finger, palm), foot 10 (not counting leg, sole), penis 6 (pure 3, penis/tail 3).

Part of the problem is the limited distribution of some forms, something that AD neglected to take into account. AD also has implausible semantic connections like liver ~ meat.

GS and AK then get to the Moscow School, their home turf, criticizing the Nostratic database in TOB Databases

 

[lpetrich]

Before turning to final conclusions, we would like to briefly mention yet another serious problem facing Nostratic studies – that of the robustness of family-level reconstructions. If several proto-languages are to be compared, we expect that the reliability of phonological and lexical reconstruction will be approximately the same within Proto-Indo-European, Proto-Uralic etc. Unfortunately, this is not the case.

Then discussing problems with reconstruction of Proto-Uralic. Consonants are in good shape, vowels not so much.

Then mentioning the Global Lexicostatistical Database (GLD) project A preliminary result is rejection of the East - West split proposed by VIS:

West: Indo-European, Kartvelian, Afroasiatic; East: Uralic, Altaic, Dravidian.

Instead, the GLD project finds:

• Core Nostratic, Eurasiatic: ( (IE, Uralic), (Chukotko-Kamchatkan, Eskimo-Aleut) ), Altaic
• Nostratic: Core N, (Kartvelian, Dravidian)
• North Borean: (Macro-N, Afroasiatic), Dene-Caucasian
• Borean: NB, Austric

Most importantly, restricting our comparanda to the 50-item set has allowed to clearly delineate between Nostratic and the “non-Nostratic” language families of the Old World, such as North Caucasian or Austronesian – families whose “optimal” 50-item sets show no links to Nostratic whatsoever.

 

[lpetrich]

AK and GS opt for quality rather than quantity, proposing reconstruction with semantic slots, for lack of a better word, and plausible semantic shifts between slots.

Semantic slots are well-defined in some cases, like numbers and pronouns, and not so well in others. Body parts may seem well-defined, but many languages have the same word for arm and hand, leg and foot, and/or finger and toe. In fact, some languages call toes "foot fingers" (Spanish dedos del pie).

For colors, there is the problem that low-tech people often have a very limited set of generic color words. One can use definitions like

• white: snow, new teeth, paper
• black: charred stuff, coal
• red: blood
• yellow: egg yolk
• green: plant leaves that are not dried up
• blue: sky

adding to Kassian's and Starostin's color examples.

Here are some cases of indefinite semantics:

• Body of water: puddle, pool, pond, lake, sea, ocean
• Strip of flowing water: brook, creek, stream, river

However, "water" is one of the most stable meanings, comparable to "name", personal pronouns, and small positive integers.

In my lists of highly-stable meanings, I find these numbers of memberships in distinct lists:

• 14: water
• 9: rain
• 3: ice, lake, river, sea, snow
• 2: foam