Language as a Clue to Prehistory

[lpetrich]

Long Ranger 22 goes into Griffen 1988 "Germano-European" reinterpretation of the Proto-Indo-European stop-consonant voicings. I'd mentioned this solution earlier in this thread, but this article goes into a lot of detail, not only in IE, but also in Altaic.

Reinterpreting T, D, Dh as Th (aspirated), T, D -- much like Thai stop consonants.

Like *threi- (*trei-) "three", *twou (*dwou) "two", *dwer- (*dhwer-) "door"

In Germanic, the Th (aspirated) became Th (fricative) - and it survives best in English. Voiceless and voiced "th" didn't get turned back into a stop, /d/, as in most other present-day Germanic langs.

This reconstruction of Germano-European consonants was conducted without any particular knowledge of the Nostrastic theory. Nonetheless, the Germano-European reconstruction fits far more precisely into the Nostratic reconstruction of Kaiser and Shevoroshkin (1988) and of Bomhard (1992- also Bomhard and Kerns ms.) than do any of the Indo-European reconstructions. The Nostratic, Germano-European, and Indo-European reconstructions are summarized in table 2.

Traditional Indo-European	T	D	Dh
Germano-European	Th	T	D
Nostratic / Eurasiatic	T'	T	D

 

[lpetrich]

More from Long Ranger 17: Merritt Ruhlen reviewed Johanna Nichols's book "Linguistic Diversity in Space and Time."

I remember getting that book and being rather disappointed in it. It didn't seem to say very much. MR seems to agree. JN proposed an alternative to the comparative method: grammatical features: "head/dependent marking, complexity, alignment, word order, PP's, inalienable possession, inclusive/exclusive pronouns, plurality neutralization, noun classes, and numeral classifiers". What are these?

Head/dependent marking is where grammatical agreement is marked, either on the head of a phrase or on a dependent. Languages can also have both kinds together or neither. Consider:

"These houses are big"

The phrase "these houses" has dependent marking, "these" agreeing with "houses". The phrase "These houses are" has head marking, because "are" agrees with "these houses".

Alignment: nominative-accusative (intransitive subject like transitive agent), ergative-absolutive (intransitive subject like transitive patient/target).

Word order: a big subject. Overall, adjective-noun, genitive-noun, determiner-noun, adposition-noun, relative-clause-noun, ...

Inalienable possession: as opposed to alienable possession. Polynesian langs like Maori and Hawaiian use a or o depending on whether the possessor has control (alienable) or not (inalienable). Possession can also be obligatory, where one has to use a possessive with certain nouns, with "someone's / something's" for no specific possessor.

Inclusive/exclusive pronouns: some langs have different forms of "we" for the one addressed being included or excluded: "you and I" vs. "we without you".

"Plurality neutralization" - some langs don't have plural forms of nouns. Instead, one must uses plurality adjectives like "many".

Noun classes: grammatical genders, etc.

Numerical classifiers: counter words in eastern and southeastern Asian langs like Chinese.

Complexity? Too general. PP's? What's that an acronym of? I'd have to read the book to be sure.

"Most linguists will wonder about the feasibility of using typological traits at all in the investigation of genetic affinity, after Greenberg's demonstration of their absurd consequences in Africa."

 

[lpetrich]

Her main discovery: "four features - inclusive/exclusive oppositions, plurality neutralization, adpositional phrases, and inalienable possession - have distributions which take the form of global west-to-east clines, and noun classes may be a fifth"

Adpositions: prepositions and their reverse-position counterparts, postpositions.

However, "these distributions can be called west- to-east only in the sense that the Pacific behaves as though it were east of the New World. The areas of the Pacific are separated from the Old World by more miles of open ocean than the New World is, and it would be possible to weight the distance of a water mile so as to derive a more eastern adjusted geography for the Pacific"

One can look for these features in WALS Online - Features and one finds very patchy distributions, without clear evidence of what JN claimed to exist.

After discussing her odd claims, MR continues:

There are problems with this book that go beyond its substantive fantasies. Despite the assertion on the cover jacket that this book "will be of crucial interest to linguists, archeologists, [and] population specialists", the book is essentially unreadable to anyone without a thorough background in linguistics and linguistic typology. Terminological jargon proliferates on every page, much of it of Nichols' own creation, ...

Like "macrogender". Is she referring to Bantu noun classes? It won't be very obvious to anyone who is not very into comparative and historical linguistics.

The reader who skips the typological sections (which is most of the book) will assume that these sections provide some evidence for this three-stage dispersal theory. In reality, there is not a scintilla of evidence for any of it, and the whole scenario seems simply to have been grafted onto the rest of the book because it fits the emerging archeological and genetic perspectives.

The book is really unreadable even for linguists, and every review I have seen of this book - all by linguists - mentions its "difficult" nature.

So she failed at what might have been some interesting research. She ought to have checked on how these typological features vary within language families, both recognized and putative.

 

[lpetrich]

Long Ranger 23 has a review of one of Vitaly Shevoroshkin's collections: "Dene-Sino-Caucasian Languages" (1991). I couldn't find any other reviews of the earlier volumes in the Long Ranger archives.

Reviewers Neile Kirk and Paul Sidwell are annoyed by VS's editorializing and his proposal that Salishan languages (Washington State and nearby) are Dene-Caucasian ones. "Shevoroshkin is doing a great job assembling these papers for publication, and we congratulate him, but his editorials do nothing for these volumes. He should let the contents speak for itself."

It has Sergei Starostin's comparison of North Caucasian, Yeniseian, and Sino-Tibetan. "Starostin's work represents classic reconstruction methodology, making it a pleasure to read. His reconstructions should be studied by all as lessons in how to get the job done by sticking strictly to the logic of reconstruction."

The reviewers say that the semantics are fairly easy for "bird" ~ "owl", but that it will be difficult to get the original meaning for a comparison of "kidney" in North Caucasian, "lung" in Sino-Tibetan, and "spleen" in Yeniseian. Also "oath, vengeance" in NC, "to curse, to insult" in ST, and "anger, to be angry" in Yen. But many of them are very reasonable, like "to freeze, ice" in Northeast Caucasian, "hail, rain with snow" in ST, "to freeze, to get cold" in Yen. Also, "hip, thigh" in Yen, "leg, calf of leg" in ST, "knee" in Lezgian (Northeast Caucasian).

"On the whole, in his comparisons of proto-forms with proto-forms, there are not the fanciful semantic developments postulated in the work of more reckless lumpers." Something like Goropius's "oak" ~ "protector from cold" (Latin quercus, Brabantic werd-cou)

After it is Sergei Nikolaev's paper doing a similar comparison with Na-Dene langs, at least Eyak-Athabaskan. At least as good as SS's comparison of Japanese with Altaic.

Then John Bengtson's papers, including Basque, North Caucasian, and Burushaski is a sort of "West SC". The reviewers say that JB should have checked some more etymological dictionaries, because there are plausible IE ones for Germanic "hair", "lamb", "dew", and "street" (*gatwôn).

 

[lpetrich]

Long Ranger 23 - John Bengtson proposes "Macro-Australic" - Australian, Indo-Pacific, Austric

Australian includes Macro-Pama-Nyungan -- Pama-Nyungan with some N Australian langs.

Indo-Pacific was proposed by Joseph Greenberg, and it contains Andamanese, Papuan, Tasmanian, Kusunda (W C Nepal), Nihali (C India).

Austric: Austro-Tai, Austroasiatic, Miao-Yao/Hmong-Mien

JB notes these comments by Vaclav Blazhek:

It is certainly possible, but there are also alternative explanations of Austric - Australian - Indo-Pacific coincidences:
(i) later AN [Austronesian] diffusion in Australia (Blust) and N. Guinea (Wurm)
(ii) Australian/Indo-Pacific substratum in AN (and Austro-Asiatic too?)
(iii) the heritage from proto-human.

Or at least Proto-Ex-Africa.

 

[lpetrich]

Long Ranger 24 - Hal Fleming has these three names for versions of Nostratic: Muscovite, Bomhardian, and Greenbergian.

Global Etymologies by John Bengtson & Merritt Ruhlen

That issue also has "On the Nature of the Algonquian Evidence For Global Etymologies" by Marc Picard - finds errors in every one of the Algonquian citations. Were JB & MR working from Joseph Greenberg's book "Language in the Americas"? A book that has been abundantly criticized for containing oodles of errors.

Then Merritt Ruhlen responding to Lyle Campbell claiming that pronominal n-m is not very common in the Americas and that Eurasian m-t is surprisingly common. Ives Goddard responded to MR's claim of Almosan-Keresiouan n-m, complete with a table of the forms in the well-established subfamilies. Lots of variation: Algic n-k, Nootkan s-s, Salish n-n, Iroquoian k-s, Siouan d/w/m-y/n/r, ...

Issue 25 reverts to something like the newsletter's original form, mainly short blurbs about work in progress, reports on conferences, and the like.

There's a Long Ranger special issue that has a big list of putative borrowings in Vedic Sanskrit, and some discussion of them.

 

[lpetrich]

Now the Mother Tongue journal.

Issue 1 started off with R. L. Trask's criticisms of the Basque-Caucasian hypothesis. "On the whole, the work on Basque and Caucasian has been considerably more sober and careful than most of the other lines of inquiry." Then describing the large amount of vocabulary that Basque got from Latin and nearby Romance, and getting into Basque's phonology.

Words cannot start with a voiceless stop, so many borrowings' initial voiceless stops became voiced, though some later ones stayed voiceless. Latin cella > gela "room", medieval *kat- > katu "cat". Words cannot start with r: Latin rosa > arrosa "rose (flower)". Initial consonant clusters must be simplified: Latin pluma > luma "feather", Latin granum > garau "grain". /m/ was absent from pre-Basque, though the present language now has a lot of it. Except in eastern dialects, -l- or -n- was not followed by a voiceless stop: Latin tempora > denbora "time". Earlier -n- was lost, unless it was doubled: Latin annona > anoa "provisions". Between vowels, -l- > -r-, like Latin angelus > aingeru "angel".

Then he gets to his main business, criticizing Vasco-Caucasian proposed connections, including the segmentations of some of the roots.

 

[lpetrich]

Then John Bengtson responded to him. He conceded that WL Trask was likely right about some of the comparisons, that there are more likely Latin/Romance sources.

But corrections of this type make up a small fraction of Trask critique. The great majority of “errors” attributed to me and other vasco-caucasologists are simply cases where we disagree with (a) Trask’s “Pre-Basque Phonology,” and (b) Trask’s (and/or Michelena’s) pet etymological solutions (usually supposing Basque borrowings from a vague “Romance”).

Then pointing out "errors of fact", like North Caucasian roots having only one continent. That's common in Northwest Caucasian, but not in Northeast Caucasian, where two or more consonants are common. He proposes that Basque h is in some cases ancestral, with NC comparisons.

Also "errors of method", like claiming Basque-NC "arbitrary segmentation", like fossilized class prefixes. Prefixes in regular use int he NC langs. Basque ukondo ~ Proto-NC *q'HwantV

Then states that RLT's “Pre-Basque Phonology” is "A Hypothesis Built on Sand". Like proposing an alternative to pre-Basque having no initial voiceless stops. Instead, voiceless aspirated ones, with Latin voiceless unaspirated ones being turned into voiced ones. Also noting some voiceless ones being borrowed as voiceless ones.

'Cat' for example is a notorious Wanderwort: it is probably a loanword in both Latin and Caucasic, but the Basques may already have had the word before the Romans came.

The example I'd posted on.

Returning to fossilized class prefixes, JB proposes Basque-NC

• u/o- I.sg. *u-
• i/e- II.sg. *y-
• bi/be- III.sg. *w-
• ar- IV.sg. *r-

JB proposes that Pre-Basque did indeed have /m/, something that would make possible this comparison for "tongue":

Basque mihi, mintz ~ Nakh mott ~ Dargi mez ~ Lak maz ~ Avar-Andi-Tsez michi, mits ~ Lezgian melz ~ NWC bza ~ NC *mêlts'i

For "nose", Basque muthur, mustur ~ Burushaski -multur

 

[Copernicus]

... Consider this rule that English speakers take for granted but seldom explicitly state: Order force: the old grammar rule we all obey without realising | Tim Dowling | The Guardian - "The rule is that multiple adjectives are always ranked accordingly: opinion, size, age, shape, colour, origin, material, purpose." That ordering is also used by speakers of other languages with adjectives before nouns, while for adjectives after nouns, the order is in reverse. Adjective Ordering Across Languages | Annual Review of Linguistics

This adjective-ordering is interesting. This is the first time I heard that not only do other languages have a standard order, but it's the same as English's order. Or, perhaps even more interesting, the exact(?) opposite order in adjective-after languages.

You show eight adjective classes, but I've seen up to eleven listed! The order can be memorized by memorizing a 12-word phrase. Shall we stage a contest to see who can come up with the best such phrase? Here are two such, using "A/An" for the quantity adjective.

An exquisite well-made miniature antique triangular ebony-colored French ceramic flat heating iron.

An impressive tall beautiful buxom young blond Swedish silky-skinned on-call working girl.

The rules governing word order are actually a lot more complex than Dowling's standardized order suggests. The interesting question to answer is what is going on when people deviate from that order. Many years ago, my friend Haj Ross was working on the arcana of word orders, and I had to opportunity to participate in some of his research and watch him work out the different factors--phonological, morphological, and semantic--that went into establishing such orders. His and William Cooper's results were reported on in the paper World Order. Very interesting stuff.

 

[lpetrich]

Then discussing Basque zilar (zirar, zidar, zilhar) "silver" ~ Proto-Germanic *silubran ~ Proto-Slavic *surebro ~ Lithuanian sidãbras ~ Proto-Berber *a-zraf ~ Akkadian sarpu "refined silver" ~ Greek sidêros "iron" -- seems like a wander word

Vaclav Blazhek: "Independently of Trask’s strict criticism, I am not so optimistic concerning a number of hopeful cognates connecting Basque with other Dene-Caucasian languages as my friends John Bengtson and Vjacheslav Chirikba are."

It's a credit to the editors of Mother Tongue that the collect some very skeptical views of Vasco-Caucasian, contributions more in line with RL Trask's contribution than with recognizable Vasco-Caucasian common ancestry.

Merritt Ruhlen:

If the central myth of twentieth-century historical linguistics has been the claim that the Indo-European family has no known genetic coimections with any other family, the first corollary of this myth has been the notion that the Basque language has no visible genetics links with any other language.

...
Though Trask concludes that the evidence coimecting Basque to other languages is “zero”, he also admits that there is nothing apparently wrong — at least from the Basque side — in roughly half of the 317 proposed etymologies. These 150 left-over etymologies, without any apparent defects, Trask dismisses as “vague resemblances . . . [that do not] constitute evidence for anything.”

... In each of the following etymologies Trask either has no serious objection to the Basque data, or his objections are so far-fetched that I have ignored them.

Responding to RL Trask's claim that the resemblance is arbitrary,

After all, Dene-Caucasian is just an apriori concoction of six arbitrarily selected families. If this is true, then Trask should be able to come up with equally cogent evidence connecting Basque with five other families. To my knowledge no one has ever attempted this.

Something like his tik-pal challenge with global etymologies, where tik = 1, finger, to point, pal = 2, pair. One ought to be able to find lots of examples of tik = 2, pal = 1.

But something like that has been done with automated comparisons using simplified phonologies. IE-Uralic passes, as does Turkic-Mongolic-Tungusic, though Korean and Japanese don't pass as additional members of that group.

 

[lpetrich]

Then Vitaly Shevoroshkin, noting "R. Trask’s impression that Proto-North Caucasian (PNC) was a language with 180 consonants is just one example of such misinterpretations of available data." Its reconstructors did that deliberately as a first step, and later found sound shifts that led to trimming that list by a sizable fraction.

Xabier Zabaltza noted the efforts of Basque linguistic nationalists to replace Latin-Romance borrowings with coined words (neologisms) and dialectal words, something that can cause confusion when doing historical linguistics.

RL Trask responded. "It’s been particular fiin locking horns with my critics, but I’ve benefited from all the responses. Finally, I wish the proponents of the Dene-Caucasian hypothesis the very best of luck with their program. ... But I regret that I caimot accept as convincing the case which has been made so far."


Joseph Greenberg has an article on "The Concept of Proof in Genetic Linguistics" - but he doesn't give a summary of what he argued in it. But he mentions a false cognate, German Feuer /foyer/ and French feu /fö/, both meaning "fire". Like have-habere and deus-theos.

German Feuer < Proto-Germanic *fôr < PIE *péh2wr (> Greek pûr, Hittite pahhur, English "fire", ...)
French feu < Latin focus "hearth", like the other Romance words for "fire"

Replaced Latin ignis < PIE *h1ngwnís (> Sanskrit agni, Russian ogon', ...)

 

[lpetrich]

Review of a book, " The Emergence of Homo Sapiens and His Languages in Tropical Asia" by W. Wilfried Schuhmacher, Juan R. Francisco and, F. Seto. - reviewed by Robert Blust

Even in the iconoclastic and methodologically laissez faire world of long-range comparison, this little book, subtitled “Linguistic evidence for a Hoabinhian alternative to Engineer Anderson’s thesis”, must be regarded as a teratism.

Teratism - something with abnormal form or structure

This book has 5 pages of text and 88 of comparisons and 5 1/2 pages of unsorted references for the word forms

RB then quotes some of the "word salad" of the text, and tries to make sense out of what the authors are claiming. In Austronesian, what RB has expertise in, the authors made very scattershot comparisons, disregarding the work that has been done on comparisons and reconstructions.

In short, I have the usual complaint of those unimaginative enough to care about method: the authors have simply scanned dictionaries and assembled comparisons based on the method of “aha!” rather than the unfortunately more demanding and time-consuming Comparative Method. Finally, how any of this collection of random lexical observations relates to the thesis of the book (whatever that is, and I still am not certain) is completely obscure.

Also a review by Igor Diakonoff of "Indo-European and the Nostratic Hypothesis" by Allan Bomhard. Much of that review is about how well we understand Sumerian phonology - not very well, like in the usual transcription, 8 signs with value "du".

 

[lpetrich]

Mother Tongue 2 discusses Nihali (Nahali), a language in central India that is not very closely related to any other language: a language isolate. But it has oodles of borrowed vocabulary from neighboring languages.

Then Sergei Starostin on RL Trask's opposition to Vasco-Caucasian.

Bengtson has made one significant mistake — unfortunately, not at all uncommon among long rangers. He does not pay enough attention to the established tradition of Basque historical phonology and etymology, notably to the works ofMichelena, Trask, and others. Now this is the one thing that specialists in any field never forgive. Small wonder that Bengtson’s papers meet a violent opposition from the Vasconists’ side.

SS also took on RL Trask's list of English - Basque lookalikes, noting that many of them are borrowings from some shared source, most often Latin-Romance, like Basque musu ~ English muzzle and Basque pinu ~ English pine, both from Latin pînus.

Among his conclusions, "We do not as yet possess enough knowledge about phonological correspondences between Basque and other Sino-Caucasian languages, although recently J. Bengtson undertook some serious steps in this direction."

Also,

I am rather pessimistic about the future of Basque studies. As a matter of fact, I am very worried about the fate of long range comparison as such. Our field is right now very clearly broken in two opposite camps that cannot find common language. On one side, there is a small group of long rangers (“lumpers”), undertaking bold, but very often inaccurate, comparisons. On the other side there is the vast majority of narrow specialists (“splitters”) who are for the most part incapable of stepping beyond their own fiefdoms and who spend a lot of time and effort in trying to dismantle — against the obvious— any attempt at establishing deeper genetic links. Imagine how we would all profit from normal cooperation rather than from constant quarrels!

Joseph Greenberg's hand-waving vs. refusal to accept that n-m pronouns could be an indicator of shared ancestry.

 

[lpetrich]

RL Trask noted old saying among historical linguists: “Look for Latin etymologies on the Tiber” - meaning at or near its speakers' homeland, and Sergei Starostin pointed out that it would be hard to find Latin's Indo-European connections if one was very strict about that. RLT responded by saying that SS was being too literal-minded about that. He listed some puzzles about Basque, puzzles that he says that Dene-Caucasian has not resolved.

Many Basque words start with vowels, and only a few consonants can start words. Fossilized prefixes?

Verbs have a variety of plural forms, sometimes prefixes, sometimes suffixes, sometimes apparently infixes. Variety like Slavic aspects or Proto-Indo-European ones (formed completely differently). Linguists recognize Aktionsart - lexical aspect - verbs having inherent aspects in their meanings. The Three Types Of Aktionsart has these examples:

• Activity (durative): walk, study, chat, play
• Accomplishment (durative-end): draw, build, drown, write
• Achievement (punctual): notice, recognize, win, find
• State (no change): have, seem, want know

I looked for sets of English verbs with similar meanings but with different aspects, without much success.

By comparison, singular vs. plural subject doesn't seem as worthy of such verb variations.

SS quotes RLT

... he appears to believe that such grave errors are no more than a minor irritant, and in no way an indictment of the procedure of trying to draw comparisons between languages one is ignorant of. My own view is that it’s hard enough to do comparative work on languages you know intimately, and that trying to work on languages you don’t know is a recipe for disaster.

then responds

I gave this long quotation because I think it is very important in revealing the positions where we all stand. My question to Trask would be: how many languages can one know intimately? Even the best polyglots, to my knowledge, rarely know more than a hundred languages, and of course not all of them “intimately”. I have rarely met a historical linguist intimately acquainted with more than a dozen languages.

That gives support to the bootstrapping strategy of macro-linguistics, to compare protolanguages of language families as much as possible.

 

[lpetrich]

Sergei Starostin:

I am, however, quite prepared to argue for some statistical considerations of my own. The principle that “to be considered related languages must have no less than 10% common most basic vocabulary” is purely empirical: I have done lexicostatistical research on a lot of linguistic families of Eurasia, and I can assure everybody that all languages that are usually considered related have more than 10% of cognates within the Swadesh wordlist. It is below that figure where the controversies start (Nostratic, Sino-Caucasian, Austro-Thai, etc., etc.).

He then mentions Sergei Yakhontov's test of borrowing on the Swadesh list of highly conserved meanings. He found 35 that are highly conserved, though in the less-conserved 65, he replaced 10 of the meanings with 10 others.

He finds the fraction of cognates in the more-conserved and the less-conserved parts and he then compares the two. Ancestry will give more in the more than in the less part, coincidences evenly, and borrowing more in the less than in the more part. I've come across this method being used to test the Austro-Tai and Sino-Austronesian hypotheses -- AT passed, SA failed.

"If we take Russian and English, for example, the percentage of cognates within the 35 word-list is 54% (19/35), while within the 65 word-list it is 25% (16/65)."

"This procedure was tested on a vast number of languages (Indo-European, Sino-Tibetan, Uralic, Altaic, Semitic and many others), and I must state that I do not know a single case where it does not work (such a case would be a widely accepted genetic relationship for which we would observe a different statistical correlation)."

Using this procedure on RL Trask's English-Basque list, he finds 3 (9%) on the more-stable part and 17 (26%) in the less-stable part. What one would get for borrowing, something that is indeed the case for most of these words.

To sum up: we need cooperation, but cooperation should be mutual. Long-rangers should be prepared to accept the results of research in any particular field, and specialists should be prepared to accept reasonable propositions and corrections from the outside.

Unfortunately, among the specialists at this point, there are few who are ready for concessions, and this is the real reason for my pessimism.

 

[lpetrich]

RL Trask responded again.

He appreciates Vladislav Illich-Svitych's procedure of working from reconstructed protolanguages -- all except Altaic securely established. But working from the individual members of families he finds much more problematic.

He objects to Sergei Starostin's comparison to biological taxonomy, noting the quality of data available. But it must be noted that macroscopic-feature taxonomy was not nearly as successful as genetics-based taxonomy, where we now have the whole genomes of a large number of species, and large numbers of gene and protein sequences for many more. This has enabled resolution of whole areas that were previously not resolved, and they corrected some widely-held but erroneous notions. This is especially dramatic in microbe taxonomy, where we went from crude hand-waving to a very detailed phylogeny.

Instead, the remorseless processes of phonological change, grammatical restructuring and lexical replacement gradually but surely extinguish all evidence of the origin of every language, and hence all evidence of a possible common origin for some languages. Linguists may reasonably differ over the maximum time depth across which a common origin may be recovered, but everybody agrees that such a time depth exists (except, of course, for the Proto-Worlders, whose ideas I, like most others, dismiss as absurd beyond comment). Moreover, contact between languages provides another mechanism for obscuring ancestry and there is no parallel to contact in the biological world. Classifying languages fundamentally different from classifying organisms.

Macroscopic organisms, yes, but prokaryotes especially have exchanged numerous genes with each other, to the point that some biologists have doubted whether one can construct a meaningful phylogeny of them -- the phylogenies of their genes are too discordant for that.

He then reasserts that no Basque-Caucasian comparisons have been found.

Then an exchange between RLT and Merritt Ruhlen.

 

[lpetrich]

Mother Tongue 3 has attempts to find the closest relatives of Nihali (C India) and Kusunda (C Nepal). "Single, isolated languages are traditionally the hardest to classify since they offer so little to work with, and the more distant the relationship, the tougher the problem." The authors have some speculations, but nothing definite.

Harold Fleming noted "One cannot substitute 'smell' for 'nose', or 'ear' for 'hear' if one proceeds Swadeshly." That is, sticking to the Swadesh 100-word list. But the Swadesh 207-word list and some other such lists also contain "to hear" and "to smell", though as separate entities. He also noted English "dog" displacing "hound" and the Proto-Semitic word for "Moon" surviving in Amharic as "month". That has happened elsewhere, like in Latin: luna "Moon" vs. mensis "month".

He is right that sticking to that list may be too strict for long-distance relationships, but if one wishes to expand the semantics, one ought to be cautious and stick to closely-related meanings, like nose ~ smell and ear ~ hear.

Germanic words for "to hear" are derived from Proto-Germanic *hauzijanan, and the Greek word is akou-, both derived from PIE *h2kh2owsye- (*akowsye-), from *h2ek- (*ak-) "sharp" and *h2ows- (*ows-) "ear" and -ye- (verb from noun suffix) -- "to be sharp-eared"

PIE had *klew- for "to hear" as its more usual word, with descendants like Proto-Slavic *slushati.

 

[lpetrich]

Next was some articles on Sumerian, the first language ever written down. Igor Diakonoff: Sumerian ~ Munda (of India)?, John Bengtson: Sumerian ~ Dene-Caucasian? Allan Bomhard: Sumerian ~ Nostratic?

Then Paul Benedict's recommendations for long rangers. Use the oldest forms available, both attested and reconstructible. Be sure to include what the reconstructions are for, like for some subfamily of whatever family one is working with. If one doesn't have reconstructions, then cite every available form, whether attested or partial reconstructions.

Then an editorial of what is good for Mother Tongue. Writing should be clear, it should be brief, and it should be elegant, in that order. I can't help but mention some problems I've had with some of the longer articles, like not stating summaries or conclusions.

"As regards editorial policy, I recommend that, as far as possible, we transcend exclusively in-group discussions of linguistic subclassification." - like subgroupings of Indo-European.

Our articles should appeal not only to linguists with "long range" interests but also, and most obviously, to prehistorians, including those scholars whose interests overlap prehistory, such as historians, archeologists, and anthropologists. Beyond these, we should appeal to as many as we can of investigators in the humanities, social studies, and life sciences whose interests include the origin and development of that most distinctively human behavior pattern: language.

Also,

I further urge my colleagues to avoid vituperation, not only with regard to conventional "short range" linguists but also when expressing disagreement, however well-justified, with fellow long-rangers.

Then proposing publishing language family-tree diagrams. I agree on publishing diagrams and tables and the like.

 

[lpetrich]

In fairness to short rangers, as they might be called, they have likely seen a lot of long-range crackpottery. I've tangled with an advocate of some of that in another forum. He proposed that Hebrew and Greek are two dialects of the same language, that Hebrew is Greek. He used such arguments as showing off maps of Greek and Phoenician colonies in the Mediterranean Sea some 3000 - 2500 years ago. I responded by pointing out all the languages on the Pacific Rim, though the present-day Mediterranean would also do. The Numbers List I'll do only 1, 2, 3.

Present-day Mediterranean: Spanish uno, dos, tres, Catalan un, dos, tres, French un, deux, trois /aN, dö, trwa/, Italian uno due, tre, Serbo-Croatian jedan, dva, tri /yedan, dva, tri/, Albanian një, dy, tre, Greek ena, dhio, tria, Turkish bir, iki, üç /bir, iki, ütS/, Arabic wâhhid, ithnân, thalâtha, Hebrew akhat, shtayim, shalosh. Kabyle (Berber) yiwen, sin, tlata (from Arabic; other Berber: krad)

I tried to explain to him that Hebrew is much closer to Arabic than to Greek, but without any success. He said that one shouldn't compare to langs away from the Mediterranean Sea.

Mediterranean of 3000 - 2000 years ago: Gaulish ônos, duô, treis, Latin ûnus, duo, três (Old Latin oinos), Greek heîs, duô, treîs, Aramaic hhad, trân, tlâtâ, Hebrew ahhat, shtayim, shâlôsh, Egyptian wu\\aw, sinuwwaj, khamtaw

My example at the time was the present-day Pacific Rim. Maori tahi, rua, toru, Indonesian / Malay satu, dua, tiga, Tagalog (Filipino) isá, dalawá, tatló, Chinese (Mandarin) yî, èr, san, Korean hana, tul, set, Japanese hitotsu, futatsu, mittsu, English one, two, three /wan, tû, thrî/, Spanish uno, dos, tres

 

[lpetrich]

Roger W. Wescott:

One issue on which my views differ from those of most of my MT colleagues is that of the nature and antiquity of phonemes. I doubt that any proto-language prior to Common Indo-European had prosodic and vocalic as well as consonantal phonemes. I assume that the speech-forms of Proto-Nostratic consisted exclusively of consonants.

With vowels for making the words pronounceable. Thus being much like Kabardian, a NW Caucasian language with vowel phonemes /a/, /â/, /(schwa)/, but with numerous allophones for following different consonants.

He also proposes vowel variations for expressing shades of meaning, which makes vowels somewhat phonemic.

English is far from that. Consider bat, bet, bait, bit, beet / beat, bot / bought, boat, boot -- differing by vowel but being semantically unrelated.

He then proposed consonant ablaut / apophony, like PIE *kap- "to take" > Latin capere, English have, and *ghebh- "to give" > English

"An example of meaningful apophony provided for Proto-Global by Alfredo Trombetti is kor, 'eagle' (i.e., 'high flying bird') as against gor, 'crow' (i.e., 'low flying bird')." - a century ago, this Italian linguist proposed a lot of long-range etymologies like these ones. Proto-Global ~ Proto-World ~ Proto-Human ~ Proto-Sapiens

RWW cited
pet 'wing' ped 'foot'
kel 'warm' gel 'cool'
lewk 'bright' lewg 'dark'
plew 'flow' bhlew 'overflow'
wekw 'speak' weghw 'vow'
wer 'water' wel 'wet'
tewk 'thigh' doyk 'toe'

I'd have to check them out. Some of them are indeed correct, though wer "water" should be wed.