• Welcome to the new Internet Infidels Discussion Board, formerly Talk Freethought.

Language as a Clue to Prehistory

The authors found that gender and noun classes were most likely inherited and that classifiers were most likely diffused, agreeing with previous results.

Finally, language speakers tend to stay in similar environments when they migrate (Nichols, 1992; Gray and Jordan, 2000; Ramat, 2012; Hock and Jospeh, 2019); therefore, if gender and noun class languages spread more by language expansion, we expect to find less variance within the natural environment surrounding their location. If classifiers spread more by feature diffusion, the spread is expected to be more independent of environmental factors.

...
The exact time of the spreads is unknown for most of our families, but given their reconstructed age range of 4000–8000 years (Nichols, 2008; Greenhill et al., 2017), the spreads most likely occurred during the mid-Holocene period. Therefore, we use mid-Holocene projections for the selected environmental factors (Derungs et al., 2018), which are shown in Fig. 3. The results are consistent with our hypothesis, as for each of the three environmental factors, classifier languages have the largest variance, followed by gender languages, and then by noun class languages (as shown by Quantile dispersion, Levene tests, and Conover tests.
 
 Noun class treats grammatical gender as a kind of noun class, and links to  List of languages by type of grammatical genders - they have a lot of variation.

Male vs. female, human vs. nonhuman, reasoning vs. nonreasoning, animate vs. inanimate, strong vs. weak, augmentative vs. diminutive, countable vs. uncountable, singular vs. plural, and many combinations of these.


Returning to climate and language, I composed an answer to Does climate play a role in the phonetic and grammatical development of languages? - Cult of Linguists - Quora
I'll repeat my explanation:

Greater abundance of consonants, both in consonant inventories and in syllables (consonant clusters, final consonants), is negatively correlated with temperature, humidity, precipitation, and vegetation. So people who live in warm and wet places will have mostly consonant-vowel syllables with not many consonants.

There is only a weak correlation of ejective consonants with altitude, but a stronger correlation with size of consonant inventory. Ejectives have a short pause between the consonant and the following vowel.

There is a similar correlation for lexical tone, with people living in warm and wet places having more tones in their language. Lexical tone is distinguishing words by relative musical pitch - high, low, rising, falling, etc. - and is best-known in Chinese.

As to why these correlations exist, some of the authors speculate that cold and/or dry climates are relatively hard on one’s mouth interior and throat and vocal cords and lungs, making speakers want to use fewer vowels and more consonants.
 
Evolutionary Dynamics Do Not Motivate a Single-Mutant Theory of Human Language | Scientific Reports - "One of the most controversial hypotheses in cognitive science is the Chomskyan evolutionary conjecture that language arose instantaneously in humans through a single mutation."

Then arguing that that is implausible. I could not follow their argument, but I agree on that implausibility.

One of the most influential theories of the human language faculty, articulated over decades by Chomsky and associates(1,2), proposes that (modern) humans are genetically-equipped with a unique computational capacity that specifically allows us to implement computations over hierarchically structured symbolic representations. According to the more recent formulation of this hypothesis, this capacity is enabled by a single syntactic operation known as Merge, which is the basis of our ability to represent complex grammars in a way that other species cannot(2). Put simply, Merge takes two linguistic units (say, words) and combines them into a set that can then be combined further with other linguistic units, effectively creating unbounded linguistic expressions. These, in turn, are claimed to form the basis for our cognitive creativity and flexibility, setting us aside from other species.
Seems like trying to reduce human language to the simplest possible element.
The key details of the evolutionary conjecture, presented concisely in Berwick and Chomsky(3), are as follows:

(1) “universal grammar” (a hypothesized innate basis for the computational capacities implied by their theoretical framework of language) can be pared down to little more than a basic combinatorial property known as Merge – this is the major conclusion of the so called ‘minimalist program’2;

(2) only (modern) humans have Merge;

(3) Merge is the result of a single mutation, perhaps provoking a rewiring in the brain yielding novel neural circuitry or network configuration that is missing in other species, and this only arose once in our lineage;

(4) because it rests on a single mutation, Merge (as a formalization of a core aspect of language) has the additional virtue that its genetic basis could be very recent in evolutionary terms, in the sense that Merge has no external prerequisites and therefore the time window between no-Merge and Merge can be small, and this is not true of alternative hypotheses;

(5) finally, Chomsky and associates insist that this mutation be kept separate from any considerations of communication: according to their arguments, language is primarily advantageous as a means for internal thought and this is the only phenotypic consequence relevant to the fitness of a hypothetical single-mutant (Chomsky and colleagues take other aspects of language, such as the neural control over the vocal apparatus to produce speech, to be ‘peripheral’.)
 
A comparative wordlist for investigating distant relations among languages in Lowland South America | Scientific Data - some 501 meanings - "Reflecting the state of the art in computer-assisted approaches for historical language comparison, it can serve as a role model for linguistic studies in other areas of the world."

 Pano-Tacanan languages

Cultural influences on word meanings revealed through large-scale semantic alignment | Nature Human Behaviour
If the structure of language vocabularies mirrors the structure of natural divisions that are universally perceived, then the meanings of words in different languages should closely align. By contrast, if shared word meanings are a product of shared culture, history and geography, they may differ between languages in substantial but predictable ways. Here, we analysed the semantic neighbourhoods of 1,010 meanings in 41 languages. The most-aligned words were from semantic domains with high internal structure (number, quantity and kinship). Words denoting natural kinds, common actions and artefacts aligned much less well. Languages that are more geographically proximate, more historically related and/or spoken by more-similar cultures had more aligned word meanings. These results provide evidence that the meanings of common words vary in ways that reflect the culture, history and geography of their users.
Thus showing how precise semantic slots can be expected to be.
 
Mother Tongue – Journal of the Association for the Study of Language in Prehistory

Issue 24 has come out.

After the obituaries was Vaclav Blazhek's lexicostatistics family tree of the Andaman-Island languages, along with comparisons to previous published ones. They broadly agree, and VB estimates divergence dates starting with the first one, at 5000 - 4000 BCE -- close to Indo-European and Austronesian, two of the older well-estabilsihed families with archeological correlates.

Then discussing the "Indo-Pacific hypothesis", Greenberg's lumping and similar hypotheses.

Timothy Usher proposes:
  • Paleo-Sundic (West New Guinea and nearby) -- Kusunda (Nepal), Great Andamanese, Önge (Little Andamanese), North Halmahera, West Bird’s Head, Bernesu, Abun, Brat, ?Yawa
  • Old Oceanic -- Timor-Alor-Pantar & Trans New Guinea, East Papuan ("Paleo-Melanesian") & Tasmanian, Australian

Then "The Father Tongues L, R, and P".

Father Tongue? That's from often learning one's father's language in preference to one's mother's language.

The L, R, and P are Y-choromosome haplogroups.

That paper didn't seem to have any definite conclusions, though I note this:
Inside India a controversy has long waged amongst scholars and laymen, with one camp knowledgeable with regard to the overwhelming cumulative linguistic, archaeological and, most recently, molecular genetic evidence for an Indo-Iranian linguistic intrusion into the subcontinent from the northwest and the other camp opposing an ‘Aryan invasion’ and favouring an indigenous origin for the Aryans, with a handful of Western polemicists egging them on.

Nobody else seems to have such a nationalistic take on Indo-European history. Even the most nationalistic Britons don't question the Anglo-Saxon-invasion theory of how English was brought to the British Isles.
 
Mother Tongue – Journal of the Association for the Study of Language in Prehistory

Issue 24 has come out.

After the obituaries was Vaclav Blazhek's lexicostatistics family tree of the Andaman-Island languages, along with comparisons to previous published ones. They broadly agree, and VB estimates divergence dates starting with the first one, at 5000 - 4000 BCE -- close to Indo-European and Austronesian, two of the older well-estabilsihed families with archeological correlates.

Then discussing the "Indo-Pacific hypothesis", Greenberg's lumping and similar hypotheses.

Timothy Usher proposes:
  • Paleo-Sundic (West New Guinea and nearby) -- Kusunda (Nepal), Great Andamanese, Önge (Little Andamanese), North Halmahera, West Bird’s Head, Bernesu, Abun, Brat, ?Yawa
  • Old Oceanic -- Timor-Alor-Pantar & Trans New Guinea, East Papuan ("Paleo-Melanesian") & Tasmanian, Australian

Then "The Father Tongues L, R, and P".

Father Tongue? That's from often learning one's father's language in preference to one's mother's language.

The L, R, and P are Y-choromosome haplogroups.

That paper didn't seem to have any definite conclusions, though I note this:
Inside India a controversy has long waged amongst scholars and laymen, with one camp knowledgeable with regard to the overwhelming cumulative linguistic, archaeological and, most recently, molecular genetic evidence for an Indo-Iranian linguistic intrusion into the subcontinent from the northwest and the other camp opposing an ‘Aryan invasion’ and favouring an indigenous origin for the Aryans, with a handful of Western polemicists egging them on.

Nobody else seems to have such a nationalistic take on Indo-European history. Even the most nationalistic Britons don't question the Anglo-Saxon-invasion theory of how English was brought to the British Isles.
Maybe not Britons, but there certainly Greek nationalists who find it hard to believe that Greek could have come from anywhere else; there are Bulgarians who are convinced that, just as the written tradition of Slavic languages originated in the Balkans, so must have the Slavic languages themselves; there are, outside of Indo- European but within the same geographical region, Hungarians who insist the idea that the Hungarian language arrived with the Magyar invaders of the 9th century is an invention of the German speaking oppressors in the context of Austria- Hungary to belittle the Hungarians' proud heritage in direct continuation at least back to Avars and Huns, and possibly much further; some go as far as denying that the Uralic affiliation of Hungarian has any validity and claim a continous history stretching back to the early neolithic
 Körös culture, or that Hungarian is actually closely related to Sumerian.
 
Then Gregory Haynes on "Root Transformations in Proto-Indo-European" - transformations that don't change the overall semantics.

He had previously contributed to Mother Tongue about resonants being inserted into PIE roots between their consonants, and then functioning as semivowels.
They are: r, l, m, n, w, y, h1, h2, h3
MT 22: Resonant Variation in Proto-Indo-European
MT 23: Resonant Variations on Immortality

I couldn't find any good example with commonplace English descendants and good semantics.

*bhe(R)dh- almost makes it: English bid, bide, bind, blend, blind, blunder, Latin fidês "faith", fîdere "to have faith"
Semantically rather loose.

He also mentioned  ndo-European s-mobile - an initial s that is sometimes present, sometimes absent.

Like *(s)tawros "bull": English steer (castrated bull), German Stier, Latin taurus, Greek tauros

He proposes metathesis (reversing two sounds) of root consonants. An an example of metathesis, some English speakers turn "ask" into "ax".

He proposed Latin forma, Greek morphê "form" (both with otherwise obscure origin)
*bhormâ ~ *morbhâ

He also proposed voicing reduction: *D, *Dh > *T
Like *ghabh- ~ *kap-
*ghabh- > Latin habêre "to have", *kap- > Latin capere "to take, seize", English "to have"

He then gave many examples.
 
Next was a paper by Alexander Kozintsev on "The Dene-Caucasian Macrofamily: Lexicostatistical Classification and Homeland"

Proto-Indo-Europeans: The Prologue | Alexander Kozintsev - Academia.edu - a lengthy attempt to pin down the location of the Proto-Indo-European homeland. Proposes toward the end an initial homeland SE of the Caspian Sea, then splitting up south of the Caucasus Mountains, with the Proto-Anatolian speakers going into Anatolia, and the others going north to northeast of the Black Sea.

That would explain evidence of PIE-Semitic contacts like *(s)tawros "bull", *ker- "horn", *septm "seven", ...

On Certain Aspects of Distance-based Models of Language Relationships, with Reference to the Position of Indo-European among other Language Families (2018) | Alexander Kozintsev - Academia.edu is much easier.

The Global Lexicostatistical Database - annotated 100-meaning Swadesh wordlists - though that paper used only 50 meanings.

AK tested out his algorithms - UPGMA, neigbor joining, and BioNJ - on the Indo-European languages, and it got mixed results, finding Italo-Celtic, Balto-Slavic, Indo-Iranian, and Greco-Albano-Armenian, and even a Western IE group: Germanic, Italo-Celtic.

But it placed Hittite near II and Tocharian near Armenian, unlike what other research shows:
(Hittite, (Tocharian, Core Indo-European))

Extending it to other Eurasian languages placed the Northwest Caucasian langs among the Northeast Caucasian ones, and it resolved all the generally-recognized families, but it had close to a polytomy in its center. That means very inadequate statistics.

The Altaic langs, Turkic, Mongolic, and Tungusic, had deep branches, almost at the center. Indo-Uralic is also that close, as is a group of Chukoto-Kamchatkan and Eskimo-Aleut: (Aleut, (Yupik, Inuit))

AK then used another algorithm, NeighborNet, one that can make connections between neighbors. This produced a sort of network in early IE, but the stronger subfamilies still came out of it.

Here, Hittite was closest to Albanian and Tocharian between Greek and Armenian. Italo-Celtic, Balto-Slavic, and Indo-Iranian appeared, with Germanic between Italic and Baltic.

Using NN on all the languages, he found a grouping of Altaic and CK-EA, a grouping closer to Indo-Uralic than to the others. Of the others, Kartvelian was the closest to these.

He did a k-means cluster analysis, but it tended to group together families represented by a small number of langs in this analysis. IE was often apart, since lots of subfamilies and ancient ones were used instead of PIE.

Then multidimensional scaling, fitting some points to one's distances, often 2D ones which can then be plotted. Also a minimum spanning tree, also based on those distances - which tree structure includes all the data points with shortest distance?

WIth this method, he finds a cluster of IE, Uralic, Altaic, Kartvelian, and Yupik-Chukchi (IY-CK). Semitic is somewhat distant from it, and North Caucasian more distant.
 
On the Homelands of Indo-European and Eurasiatic: Geographic Aspects of a Lexicostatistical Classification (2020) | Alexander Kozintsev - Academia.edu

He has a much larger sample of languages this time, though still with his 50-meaning wordlist for each.

He finds an Indo-Siberian cluster: IE, Uralic, CK, EA, an Altaic cluster that includes Korean and Japonic, a Eurasiatic cluster that includes Indo-Siberian and Altaic, a Nostratic cluster that includes Eurasiatic, Kartvelian, and Dravidian, and also Afroasiatic and North Caucasian clusters.

He then tries to find the homelands of these macrofamilies, something I consider very dubious.

(41) Asia or Africa? Locating the Afrasian homeland (2021) | Alexander Kozintsev - Academia.edu

His 50-meaning wordlist again.

I've found numerous hypotheses of the branching of the subfamilies of Afrasian (Afro-Asiatic), so there isn't a clear consensus, other than Omotic branched off first.

AK himself finds Omotic then Cushitic than Chadic then Semitic-Egyptian-Berber: Semitic (Egyptian, Berber)
 
Back to Mother Tongue and Alexander Kozintsev on the Dene-Caucasian languages.

He uses his 50-meaning list yet again.

He finds this clustering in multidimensional scaling, though in neighbor-net, it is weak.
  • Euskaro-Caucasian (Vasco-Caucasian): Basque - North Caucasian
  • Karasuk: Burushaski - Yeniseian
  • West DC: EC - Karasuk
  • Sino-Dene: Sino-Tibetan - Na-Dene
Na-Dene is a straight line of Athabaskan - Eyak - Tlingit

He also finds a split between Nostratic (Eurasiatic, Kartvelian, Dravidian) and Dene-Caucasian.

He then considers the question of the Dene-Caucasian homeland, something I consider as dubious as the Nostratic/Eurasiatic one. The most that I feel confident in is that they both had Central Asian / Siberian homelands.
 
When focused on the earliest proto-Indo-Europeans, it is common to refer to the language family as Indo-Hittite (or Indo-Anatolian) and divide that language family into two clades: Anatolian and Indo-European Proper. The Homeland of Indo-European Proper was Yamnaya and/or Sredny Stog, but what about Indo-Hittite? Some placed it further south nr Armenia with the Maykop civilization linguistically (and genetically?) ancestral to Yamnaya.

By coincidence it was just an hour ago that someone spontaneously shared a link with me to a very recent paper:
Its conclusion? Conventional thinking was correct: the East European steppes were the Homeland of Indo-Hittite. (The paper is mainly DNA analysis, and has much about Y-chromosome haplogroups so perhaps should have been posted in that thread.) The paper is a major treatise with 58 pages, lots of figures and about 100 authors. It will be a problem for me to even browse, but why not share this problem with others? 8-)

I've appended one of the figures just as a hint to the paper's scope.

F1.large.jpg
 
I'll look at *bhe(R)dh- more closely:
  • English "to bid" (offer, attempt) < PGmc *beudanan "to < PIE *bhewdh- "to be awake, aware"
    • "Buddha" < Pali buddha < Sanskrit buddha "awake, enlightened"
  • English "to bide" (abide (a + bide)) < PGmc *bîdanan "to wait" < PIE *bheydh- "to command, persuade, trust"
    • Latin foedus "agreement, alliance"
    • Latin fidês "faith, trust, confidence"
    • Latin fîdere "to have faith in, to trust, to have confidence in"
  • English "to bind" < PGmc *bindanan "to bind, tie" < PIE *bhendh- "to tie"
  • English "to blend" < PGmc *blendanan "to blend, mix, make murky" < PIE *bhlendh- "to blend, mix up, make cloudy, opaque"
    • English "blind" < PGmc *blindaz "blind"
    • English "blunder"

Turning to devoicing of stop consonants, that happened in Anatolian and Tocharian, and some Proto-Core-Indo-European speakers may have picked up some of these unvoiced variants, like *kap-, while still using *ghabh-.

Outside of Indo-European, Proto-Uralic and Proto-Dravidian had all-voiceless stop consonants.
 
By coincidence it was just an hour ago that someone spontaneously shared a link with me to a very recent paper:
Its conclusion? Conventional thinking was correct: the East European steppes were the Homeland of Indo-Hittite. (The paper is mainly DNA analysis, and has much about Y-chromosome haplogroups so perhaps should have been posted in that thread.) The paper is a major treatise with 58 pages, lots of figures and about 100 authors. It will be a problem for me to even browse, but why not share this problem with others? 8-)
I've read it. Interesting. They propose that Anatolian and Armenian speakers arrived from the east.

There has long been a controversy over which way the original Anatolian speakers got to Anatolia. From the east or from the west? From the west seems to be a more common opinion, and it is supported by the greater variety of Anatolian langs in western Anatolia.

As to Armenian, it may also have come from the west. There is some evidence of a "Balkanic" subfamily of Core Indo-European -  Paleo-Balkan languages - Armenian is well to the east of all of them except Phrygian
  • Armenian
  • Graeco-Phrygian:
    • Greek, Ancient Macedonian
    • Phrygian (W to C Anatolia)
  • Illyric
    • Albanian
    • Messapic ("heel" of Italy)
    • Illyrian (?) (~ ex-Yugoslavia)
    • Eastern-Romance substrate (?)
  • (uncertain relationships)
    • Thracian (?) (~ Bulgaria)
    • Dacian (?) (~ Romania)
    • Liburnian (?) (~ Dalmatia)
    • Paeonian (?) (~ N Macedonia)
    • Mysian (?) (NW Anatolia)
Among shared features are *aig- "goat" (Greek aix, aig-) and *dheh1s- "god" (Greek theos) Also, "Innovative creations of agricultural terms shared only between Albanian and Greek were formed from non-agricultural PIE roots through semantic changes to adapt them for agriculture. Since they are limited only to Albanian and Greek, they could be traced back with certainty only to their last common IE ancestor, and not projected back into Proto-Indo-European."
 
Languages that are poorly preserved are sometimes called Trümmersprachen (German: "ruins languages", sg. Trümmersprache). Of Balkanic, we have inscriptions in Phrygian and Messapic, Thracian words mentioned by some ancient authors, and personal and place names from what is presumed to be Illyrian.

I read somewhere that more has been written on these languages than what survives of them.


Returning to the Balkanic languages, here are the three surviving ones:
  • Greek, written from ~1400 to ~1200 BCE, then from ~ 800 BCE
  • Armenian, written from ~ 400 CE, lots of borrowings from Iranian
  • Albanian, written from 1462 CE (first surviving example), lots of borrowings from Latin/Romance and South Slavic
 
Last edited:
That paper adds to the evidence of dispersion from the Pontic-Caspian Steppe - Ukraine and S European Russia - around 4000 - 3000 BCE, a dispersion of Proto-Indo-European-speaking people.
Here are the oldest language families with archeological correlates of their protolanguages and expansions.
  • Pama-Nyungan - 3000 BCE - N Australia
  • Indo-European - 4000 BCE (all), 3000 BCE (core) - Pontic-Caspian steppe (Ukraine, S Eur Russia)
  • Austronesian - 4000 BCE - Taiwan
  • Benue-Congo (Wikipedia Volta-Congo) - 4500 BCE - where the Benue River flows into the Niger River (W C Nigeria)
  • Transeurasian (Macro-Altaic) - 7000 BCE - West Liao River (NE China)
  • Euskaro-Caucasian - 8000 BCE - S Anatolia

I'm skeptical of Alexander Kozintsev's work for two reasons. He used a rather small list of meanings: 50 of them. I also could not follow how he made his archeological identifications. I couldn't find anything that stands out in them as much as (say) the first farmers in some place.

But those criticisms aside, he has some very interesting methods, like alternatives to tree structures.
 
When Scientists Drift into Pseudoscience by geologist Steve Dutch
The paranoid tone of pseudoscience occasionally surfaces in the works of orthodox scientists; not, as one might expect, in the works of scientists who are trying to bring about a scientific revolution, but more often than not in the works of scientists who are resisting change.
Then noting anti-drift geologists getting unhinged in the late 1960's and early 1970's, as continental drift became generally accepted.
There was,for example, the familiar tactic of downgrading findings to the level of"hypothesis" or "speculation". And of course there were charges of ulterior motives and unscientific conduct.

...
The reaction of the Lamont scientists to the anti-drift articles was not so much anger as amused, open-mouthed incredulity that any scientist would actually put his name on something like that.
Like some anti-drifters claiming that drifters use flat maps rather than spherical geometry.
By about 1970, geologic opinion was solidly behind continental drift. Geology not only let the anti-drifters air their views, it actually bent the rules in their favor. Taken as a whole, the anti-drift literature must surely rank among the worst rubbish ever printed in a major scientific journal.

...
One major anti-drift paper is "filled with assertions that mostly are incorrect, misleading, or irrelevant", and contains "many denunciations of individual and collective drifters as lacking in honesty and knowledge".
SD then continued with "How a Real Scientist Opposed Plate Tectonics"
Continental drift had its opponents in the U.S.S.R., too. Soviet science was a good deal more authoritarian and hierarchical than science elsewhere, and at the top of the heap in Soviet geology was the tough and brilliant V.V. Beloussov, who for a long time was a severe critic of continental drift. Yet Beloussov stands not only head and shoulders, but knees and ankles above the other critics we have read. Beloussov thoroughly understood the evidence that was being used in support of drift, and he realized that he not only had to point out errors in the drift theory, but come up with some other interpretation that worked better than continental drift. He made a serious effort to try, and if he did not succeed, it was not for lack of effort or ingenuity, it was because he was backing the wrong side.
 Vladimir Belousov
 
Steven Dutch: "So here goes with a list of some academic disciplines where the rhetoric suggests that some participants have crossed the line into pseudoscience. In every case it's the conservatives."

His examples: Extreme Paleolinguistics, Communication with other Species, Quasicrystals, Humans in North America (before 10,000 years ago), Biblical Artifacts (there are some real ones amidst the oodles of fakes)

 Quasicrystal - a structure with "impossible" symmetries like fivefold symmetry. They get such symmetries by being aperiodic tilings, making them halfway between crystalline and amorphous states.

Communication with other species - that was tried roughly half a century ago, without much success. Chimpanzees, orangutans, and gorillas can learn lots of sign-language signs, but they cannot do much more than that, mostly two-sign phrases.
 
The dolphin researchers lack the cultural awareness that perhaps they should play a Mariah Carey song.
 
Looking for stuff on  Nilo-Saharan languages I found Besprechungsartikel . The classification of Nilo-Saharan on Roger Blench: Nilo-Saharan opening page

Comparing two rival reconstructions of Nilo-Saharan, by Marvin Lionel Bender and Christopher Ehret. MLB seems to take a much more comprehensive approach than CE, citing many more sources.
Nilo-Saharan reconstruction is heavily dependent on unpublished and semi-published materials. Bender has always been scrupulous about citing these and as far as possible, publishing them, even where they do not meet modern standards of transcription. Ehret, on the other hand, has no discussion at all of the quality of sources or indeed which citation is linked to which source. As a consequence it is often difficult to know where his citations come from, when they differ from the obvious source.
The two have very different phonologies, with CE having many more consonants than MLB. RB repeats Robert Blust's warning about "proto-form stuffing", projecting every descendant feature onto a protolanguage.

About semantics, he notes:
Bender (1996a)
p. 77 belly, inside, liver, outside, intestines, heart
p. 79 follow, enter, exit, hunt, chase, dance, return, rise, turn
p. 137 earth, country, land, ashes, down, dust, mud, sand, charcoal

Ehret (2001)
No. 1129 husk, shell, fur, to slide under, to shovel up, hair pulled out in fright, tweezers, hair, feather, to remove
No. 1134 to descend, to deflate, to be asleep, to trim lamp
No. 1262 to drip, blood, sap, cold, cataract, tear, river, dew
Some of them don't seem very semantically related.
The source of this generous attitude to cognacy is the great difficulty in finding evidently related forms across the whole range of Nilo-Saharan. But perhaps we may need to accept the total number of true PNS forms will always be small, as a consequence of the great time-depth of the family. A parallel may be Greenberg's claims for the unity of Amerind, the great phylum he proposed to unite most of the languages of the Americas. Greenberg (1987) put forward common glosses that involved dramatic semantic shifts which were not accepted by Amerindianists; some fifteen years on, they still remain in the wilderness.

Then,
There is an important difference that relates to our historical understanding of the phylum. Ehret includes a number of reconstructions that imply agriculture and livestock production were known to the speakers of PNS; Bender does not. This in turn is linked to Ehret's unique conviction that Nilo-Saharan is a tight phylum with a large number of traceable proto-forms and thus of limited historical time-depth. Bender (1996a) offers no immediate estimate of the time-depth of Nilo-Saharan although he does comment that it 'agrees very well with the Khartoum Mesolithic-Neolithic back to 30,000 years B.P.'. It is unclear how seriously we are meant to take this, but the point that Nilo-Saharan vocabulary is pre-agricultural is well taken. If indeed great internal diversity is characteristic of Nilo-Saharan then an antiquity much greater than agriculture in Africa is thereby implied. If so, there should be no convincing reconstructions of agricultural terms at the level of the proto-language. Phonology, semantics and cultural reconstruction thus become all of a piece.

MSB vs. CE on NS seems almost like Orel-Stolbova and CE on Afrasian.
 
Back
Top Bottom