published in Dhumbadji!, Vol. 3, No. 1, Summer 1996, (Letters to the Editor)
REVIEW OF ALEXIS MANASTER RAMER'S
Dhumbadji!, Vol. 2, No. 3, December 1995, pp. 16-17
Patrick C. Ryan
Professor Alexis Manaster Ramer concludes his article by modestly saying: "We are very far from being able to say the last word on the subject", a sentiment with which I can heartily agree.
Actually, I find the article marred by strange terminology, inadequate research, and flawed logic; points I will detail in the course of this review.
The theme, which might conceivably have been candidly included in the title of this article, is borrowing.
Professor Ramer borrows the research of Van Ginneken, Frei, Klimov, Pisani, and Winter, which purports to show that the Proto-Kartvelian etymon for the numeral "4", which he "posits" as *os1tx- ("or perhaps os1txw- or *os1txo"), was borrowed from Proto-Indo-European. Since "posit" means "put forward as fact or truth", I am sure that he is sure that one of these forms is factual but slightly at a loss to know which one of them he actually "posits".
We are not given a time-frame for this borrowing so we cannot really form a judgment about the necessity or desirability for speakers of Proto-Kartvelian to borrow a word for "4". Perhaps herds were very small in the PK homeland; and wealth was unknown (because uncountable!)
How shall we imagine the situation to have been? A people with a developed language who had no need to count? Or should we think that the PIE term for "4" struck them as more expressive?
Or should we suspect, as I do, that phonetically related terms for semantically related concepts are most easily explained by a common relationship, i.e. origin?
Allan R. Bomhard (and John C. Kerns) in The Nostratic Macrofamily (1994) — and in other works — has made a prima facie case for the derivation of PK from Nostratic, which also produced IE and Afroasian (AA). This possibility was not discussed or was allusion made to it. Before borrowing is assumed, I would think that common origin needs to be ruled out.
Next, Ramer denies the possibility of a metathesis of -k^t- in PIE *ok^to:(u)-, "eight" to -tx- in the PK form *os1tx-, "four", putting aside for the moment the discrepancy in meaning. Since metathesis usually involves consonant shifts that facilitate pronunciation — and in any language, -k^t- is probably going to be easier to pronounce than -tx- — Ramer is most certainly correct.
The PK *os1tx(w/o)- produces Georgia and Mingrelian otx-, Laz o(n)txo-, otxu-, and Svan woštx(w)-, all meaning "4".
The first very obvious characteristic of the PK forms quoted is that Svan, which seems to be showing the least modified form, can be reconstructed with -(w)-; and that Laz -u- points in the same direction.
Strangely (inadvertently?), the PIE form is never cited in this article in the form appearing in Pokorny (1922): ok^to:(u), showing a (n optional) -u that invariably represents PIE /w/ even though Ramer must be dimly aware of it because he cites Old Indian aSTau (sic!), in which he fails to indicate the long and accented a(á:).
Indeed, it is never made clear in the article why Ramer "posits" *os1tx- as the PK form, i.e. without the -w when the normal procedure is to assume the full form is the original form (Klimov's os1txw-), or is this something Ramer "posits" also?
There is, of course, a further totally neglected anomaly: the sporadic appearance of (n) in Laz o(n)txo-. No discussion of sporadic appearances of non-cognate -n- in Svan can be discovered in the article; and we should expect, at the minimum, that notice was taken of it.
By the normal rule, the full attested form is the original form, we might reconstruct PK *o(n)s1txw- based on the attestations cited in the article.
There is no problem with adding an -n- to the IE form, although it is nowhere attested in IE because a) a pattern of deletion of the nasal feature of dorsal nasals is abundantly apparent: e.g. IE 2. ank-, ang-, bend, shows Old Indian añcati and ácati, bends, curves; a root, from which Greek agkó:n, bend (of the knee, shoulder), elbow, is derived.
This last citation is of more than casual interest when we acknowledge (as Ramer does not explicitly do) that the expected rationale for number terminology is the configuration of the hand or fingers used in finger-counting, or alternatively, the body part touched (body-counting).
I do not know if Ramer is familiar with Georges Ifrah's One to Zero (1985), which explains these matters in satisfying detail.
Although the only thing we know about PIE and PK and AA and Nostratic finger-counting is what we can learn from analysis of the terminology, we can see in Ifrah that the Torres Strait Islanders used the shoulder as a body-counting touch-point for "8" while the Elemas of New Guinea used the elbow similarly.
On page 28 of his fascinating book (and I recommend it to Ramer), examples of various finger-counting methods are illustrated.
Illustration A-4 shows one method of finger-counting in which the little finger is bent while all other fingers are extended; it designates "4".
Interestingly, PIE has another root which is listed as wek-, we-n-k-, bend, in Pokorny, from which Old Indian váñcati, goes crookedly or slopingly, is derived.
Both these Old Indian examples show palatal dorsal (nasal) consonants; and please note that I prefer dorsal to indicate consonants articulated by the contact of the dorsum to a particular part of the mouth, usually the palate or the velum.
I am somewhat surprised by Ramer's characterization of PIE "*(sic! Are these somehow to be considered roots???)/k', g', g'h/" as "palatovelars". They are palatal dorsal stops else why would the tick (') be there? Ramer may not be aware that the term "palatovelar" was originally created to describe dorsals which often were pronounced with either one of two contact points: the palate or the velum.
Now there are points which Ramer makes that seem defensible; I do not think many will have a problem with reconstructing PK s1t as the cluster underlying the Kartvelian manifestations but the assertion of Ramer that the PK form, which I believe is correctly reconstructed as *wos1tx(w)-, is to be derived from Akkadian hamištu, "5", is the pure fantasy of an investigator who is determined to account for every similarity through borrowing, whatever the offense to common sense.
The two paragraphs purporting to explain how Akkadian hamištu, "5", is really to be interpreted as phonetically representing "[hawitu] or [xawitu]" will not satisfy any AAist or even general linguist; and trite examples like "Thus, innovative English spellings like lite and nite do not indicate that the words light [lait] and night [nait] have suddenly acquired a final vowel" will strike many readers of a journal devoted to philology as condescending and completely unconvincing.
Then, in another round of borrowing frenzy, Ramer tells us that the PKs borrowed "PIE *g'lHV-, daughter-in-law, sister-in-law'...(as) PK *-j1al-, and...PIE *mlg'-, milk' (itself a loanword from Semitic)...as PK *mL1j1e- milk'."
One can only wonder what (perhaps totally unattested) language may have been the original lender for these ubiquitous borrowings.
Now, Ramer candidly informs us that "The semantic mismatch between '4' and '8' also calls for some comment".
Bear in mind, he has told us on the previous page that "PK *xus1t- '5',...comes directly or indirectly from a form like Akkadian hamištu".
I would think that the "semantic mismatch" between "4", the previously assigned meaning for Ramer's PK *xus1t-, and his sudden assignment of the meaning "5" to the same word "calls for some comment" also. But, unfortunately, we do not have the benefit of Ramer's arguments for this sudden (and quite unexpected) transformation.
I say unexpected' because Ramer now launches into a paragraph or so on the relationship of "4" and "8"; perhaps, "5" was just a wayward thought which may better have remained lost.
Ramer borrows from Henning the idea that Avestan "ašti 'width of four fingers'" may be the singular form (as a hypothetical "4") for "PIE *ok^to:- as the dual of two (sic!): a hypothetical *ok^t-"; and then borrows from Johnson the idea that PIE "*penkwe '5',...denoted the whole fist", for which he claims to have found comparative support in Uralic and Altaic.
In view of Old Indian ásthi-, "bone" and Avestan ast(i)-, "leg, bone", the probable source of Avestan ašti is as a variant derivation from PIE ost(h)-, "bone". Four fingers is about the width of my leg. But, in any case, where is the (n) of Laz?
In the next paragraph, Ramer asserts that "all PK numerals above '3' are loanwords from Indo-European or Semitic", basing this rather wide claim on work which he has "(submitted)". Perhaps "submitted" is somehow a higher status of work than "manuscript" although I am not aware of the distinction. But if the reasoning is comparable to that which we see in this article, I am afraid that "manuscript" may be the appropriate designation.
Let us review. Ramer has told us that PK borrowed from PIE the base form "*ok't-", which he assumes on no evidence that I can see, meant *four, presumably at a time when PIE employed it for "4". Somehow or other, the PK form, which is borrowed from PIE, "came directly from a form like Akkadian hamištu", which meant "5".
Well, which came first — the Akkadian or the PIE?
It seems to be "4" of one and "5" of another. I confess I am not able to riddle out the meaning of these seemingly contradictory statements. Perhaps Ramer will submit a response to clarify matters for us all.
Next, Ramer tells us rather authoritatively that "the */x/ of PK *os1tx- has (sic!!!) to be a reflex of a PIE laryngeal" but the laryngeal in *ok'toH- represents the very dual suffix which is assumed to be responsible for the meaning "8".
Ramer's answer for this high hurdle that would have halted many a less talented jumper to conclusions is to "assume that the meanings '4' and '8' were interchanged in PK".
I do not believe that many readers will find this probable; and I, personally, find it incredible as an explanation for a relationship that is founded on so many other apparent contradictions. Something aside from fact that 2 x 4 = 8 must be presented for this thesis to have any merit.
I have read exchanges between Ramer and Bomhard on electronic discussion lists in which Ramer "deplores" Bomhard's wrong ideas" about Nostratic but having read this article, I can only hope that Ramer relents and reads Bomhard's work, which shows a more realistic approach to interlanguage comparisons than is shown here.
The only thing that I find almost inexcusable is Ramer's failure to identify as a numeral formant. Just a little research in Egyptian, with which he might not be familiar but certainly should have been done if AA comparisons are being contemplated, would have revealed that Egyptian numerals are regularly formed with -w. Gardiner (rev. 1973) reconstructs "1 w'(yw) 2 snw(y) 3 Hmt(w) 4 fdw 5 diw 6 srsw or sisw 7 sfx(w) 8 Hmn(w) 9 psD(w) 10 mD(w)...50 diyw", etc.
Ramer cited "*penkwe 5'"; and, although he overlooked the usually reconstructed (u) of *ok^to:(u)-, which actually supported his abortive comparison (-u of Laz otxu; -(w) of Svan woštx(w)), surely he is aware of the (u) of IE *dwo:(u), "two". Would he expect assume to assume that *dwo:(u) is the dual of a totally unattested *dwo, "*one"; or perhaps "penkwe" is the dual of an unattested *penk, "two-and-a-half"?
Well, I think it is clear that Ramer's article has not moved me to conviction.
Since I have also discussed matters with Ramer electronically, I will do what Ramer failed to do (in my opinion) in our discussions, and offer up my own interpretation of the facts to the sacrificial knife of Ramer's sharp wit.
Nostratic is a reconstructed language that is the antecedent of IE, AA, and (P)K, as Bomhard as shown in many publications over many years.
The relationships among the words discussed is not borrowing but derivation from Nostratic roots that were, at least, originally common to all.
We should expect differences in numeral terminology because counting methods were subject to preference and improvement.
The (P)K term for "4", correctly reconstructed as *wo(n)s1tx(w) is based on a Nostratic *wank(y)-, "bend" (IE *wenk^-); and tells us that the Kartvelian-speakers used a finger-counting method that involved the bending of one finger to indicate "4".
The -t of the PK is Nostratic -t and IE -t, a former of ordinal from cardinal and collective numerals. The final -(w) is probably Nostratic -*wa, a suffix indicating numerals generally.
I cannot, at present, offer a reasonable suggestion for the significance of (P)K -x- but I do accept the possibility that it may represent a guttural (which I will not specifiy as laryngal or pharyngal) that is reponsible for length of o: in IE *ok^to:u and *dwo:u, with a significance that remains to be determined.
The IE word for "8", *ok^to:u is composed of the same final elements as the (P)K but the first element is seen in IE 2. *ank(^)-, *ang(^)-, bend (Nostratic *ank(y)), indicating a body-counting method for IE, and designating either the 'elbow' or the 'shoulder'.
However, the Nostratic root underlying (P)K wo(n)s1txu- may be present in IE also but in a disguised form and with a different meaning.
Ifrah illustrates the body-counting system of the Torres Strait Islanders, and shows the (right) wrist as "6", which is also "6" for the Papuans and Elemas.
Though "6" is usually reconstructed as *swek^s, there are some IE languages which show *wek^(s) : *uk^(s): e.g. Armenian vec. and Greek héx.
To judge by Armenian gangur, "flexus", IE *we-n-k-, "bend", may have once designated the "bendable" joint of the wrist though the -s remains to be explained (perhaps as a multiplicative suffix). I ask no one to accept this purely speculative suggestion.
I am not saying that IE exclusively had a terminology for numerals based on body-counting since I see traces of finger-counting in other numerals but I assert that Proto-Language "1" and "2" were also part of a body-counting system: PL na, "nostril", "1 (of two)"; and PL fa, "palm", "2" (possibly also PL no, "abdomen", "nine?"). IE oi-no(s), "1" is based on a simple combination of PL ?e-$e, "that (yonder)" + na, "1", while Old Indian e:ka, "one" is simply PL ?e-$e, "that (yonder)" + khe, "other" + fa, "two, number".
Ramer's exercise here is a perfect example of what mischief can be perpetrated when someone who should know better substitutes assumption for analysis and imagines borrowing where only genetic relationship exists.
Bomhard, Allan R. and John C. Kerns. 1994. The Nostratic Macro-family A Study in Distant Linguistic Relationship. Trends in Linguistics: Studies and Monographs 74. Berlin, New York City: Mouton de Gruyter
Gardiner, Sir Alan. 3rd edition, revised 1973. Egyptian Grammar: being an Introduction to the Study of the Hieroglyphs. London: Oxford University Press
Ifrah, Georges. 1985. From One to Zero: A Universal History of Numbers. (translation of Histoire Universelle des Chiffres) New York: Viking Penguin, Inc.
Pokorny, Julius. 1959. Indogermanisches Etymologisches Wörterbuch. Volume I. Bern and Munich: Francke Verlag
Ryan, Patrick C. 1990. Pre-Nostratic "Pronouns" Early Noun Substitutions. Mother Tongue 11, September 1990.
1993. An Inquiry or Thought Paper. Mother Tongue 19, Spring 1993.
1994. Proto-Language "He" and "It" IE -l/-n Nouns. Dhumbadji! Vol. 1, No. 4. Winter 1994.
1996. Merritt Ruhlen's Two Books on Language Origins. (Review). Eurasian Studies Yearbook. Vol. 68 (1996). Berlin, Bloomington, London, Paris, and Toronto: Eurolingua