ong & Alexand
aig. Stefan G fovin: Review of Joseph H, Greenberg (2002)
and of language contact. The contributors to this volume all demonstrate this
combination of skill, and each contribution can be taken in its own right as
a serious and stinmulating exercise in how to do historical detective work based
on linguistic evidence.
Reviewer's address
Geoltrey LJ Haig
Seminar lr Allgemeine und Vergleichende Sprachwissenschaft (SAVS)
Universitit Kiel
Olshausenste. 40
KIEL 2098 Germany
Esra: baig@emai.uni-kie. de
Indo-European and Its Closest Relatives: The Eurasatic language fam
ily: Volume 2: Lexicon. By Joseph H. Greenberg. Stanford, Californi
Stanford University Press, 2002. Pp. 216
Reviewed by Stefan Georg & Alexander Vovin (Bonn / University of
Hawai'i)
‘Two years after the publication of the first volume (Greenberg 2000) of Joseph.
H. Greenberg's (hereafter G) assessment of his “Eurasiatic” theory (which
claims the ultimate genetic relationship of Indo-European, Uralic, “Altaic’,
Korean, Japanese, Ainu, Nivkh/Gilyak, Chukchi-Kamchatkan, and Eskimo-
Aleut), the second and last installment of this work appeared, posthumously.
While Volume I dealt with grammar, ie. morphological evidence for the ex-
istence of this putative phylum, Volume II, which is considerably smaller,
presents lexical comparisons meant to strengthen G's point further. We have
dealt with the first volume in Georg & Vovin 2003, and there we were unable
to avoid a rather negative conclusion on both the likelihood of Eurasiatic as a
valid language family and on the particular arguments advanced by G. At the
same time we stressed that we do not think that every single comparative ob-
servation found in this book is necessarily invalid and that, on the contrary,
we do think that the investigation of “deeper” genetic relationships among the
language families of Eurasia may indeed be a worthwhile undertaking, whichjus /Besprechungen 21
could potentially lead to solid and, thus, exciting results, ifnot for “Eurasiatic”
(or “Nostratic”) as a whole, but certainly for some subsets ofthis grouping (as,
e.g, for Uralic-cum-Eskimo-Aleut (Seefloth 2001) and the possible addition
of Indo-European to such a phylum). However, we had to stress that, in order
to convince the linguistic community at large, any such endeavor will have to
adhere to rather strict principles and that consequently a potentially successful
demonstration of this or a similar hypothesis will rather have to concentrate
‘on few, but exceptionally carefully examined comparisons, rather than on a
long list of look-alike elements (muitum, nom multa). Here, we found G's ap-
proach mostly wanting, both in terms of methodology and of his handling of
the data. Many of the points made in our review article could easily be repeated
for the second volume. Sutfice it to say that, again, G's approach to the com-
parison of languages is antsToRICAL (ie., sacrificing the known and knowable
internal history of elements in favor of external similarities) and regards ~ ex
plicitly ~ the observation of regular phonological correspondences as of minor
importance, attitudes we are unable to subscribe to. Given the widely shared
assumption that systematic correspondences found in the morphological sys-
tems of languages usually count much more than lexical similarities when it
comes to demonstrating their relatedness, it may not be surprising that the
present second volume appears to us as an anticlimax to the first one.
Within the limited space of a short review it will be of course impossible to
demonstrate the shortcomings of the lexical comparisons found in detail, so we
will have to confine ourselves to a few points here. And, as with G's morpho-
logical comparisons, we do not think that aLL of them are necessarily wrong:
on the contrary, some etymological sets may indeed contain some items which
are truly shared by at least a subset of the “Eurasiatic” set of languages: some
of them may be the result of old and as yet undetected areal contacts, and, yes,
some may even be the remnants of lost protolanguages.
G states explicitly that many of his etymologies are not new, and that he
relies on macrocomparative work done before him (especially on the works of
V. Illch-Svitych and Bomhard & Kerns); in this book, the focus is on “lan-
guages and families that have generally been excluded from the Nostratic fam-
ily, namely, Etruscan, Ainu, Gilyak, Chukotian, and Eskimo-Aleut” (8), Etr-
tuscan, however, surfaces only in two! (out of 437) etymological sets, BEFORE
1. And as we may note in passing, fora book with this ttl, the number of etymological
sets involving Indo-European strikes us as ow at 230 out of 437,16. Stefan Georg & Alexander Vovin: Review of Joseph H. Greenberg (2002
**hant (#35) and ive **to (#174). The first of these sets (Indo-European
cent) gives Etruscan handin ‘before’ This, however should rather be given
‘as ‘from before, since it contains the marker of the ablative, -@in, which of
course would have to be removed before any external comparison. The sec-
ond Etruscan word tur is indeed a verb meaning ‘to give, but the compari-
son with words meaning ‘gift’ like Greek dg#fov or Old Slavic dara is ham-
pered by the fact that the -r- here is not some obscure “root determinative’
but rather a fairly transparent heteroclitic nown-forming derivative (*-t/n-y
(Old Slavic dard, Latin donum) from the verb *deHs, whereas in Etruscan there
{sno reason to separate -1- from the root.
‘As one example for the methods employed we will have a closer look at
‘one etymological set, which is presented as especially important not only for
Euarasiatic itself, butalso for putative external connections of this macrophylum
with another one, namely G’s “Amerind” (Greenberg 1987). On p. 2 we learn
that “[t]he Eurasiatic family is most closely related to the Amerind family’, and
«(t)his is most strikingly shown by the root ma-n/ma-K/ma-r ‘hand, give, mea-
sure’ (..). A very similar root is characteristic of Amerind languages, but is not
found elsewhere to the best of my knowledge”. We will not dwell on the mer-
its of the “Amerind” comparison here (which involves, ia., Algonquian *mi:-
Shand’, Uto-Aztecan *ma: ‘hand? Quechuan *maki ‘hand’, and various words
for ‘give’ etc, see the table on p. 3), but we should check whether the Eurasiatic
etymology (#194) holds water. All “Enrasiatic” language families are present in
this comparison except Etruscan and Eskimo-Aleut,and we will examine them
one by one:
Proro-INpo-EuROPEAN *ms-r, "mo-nos, *me-, *me-te- ‘to measure,
*med ‘measure’, *mHr-/n-(th)- ‘hand’, This is a medley of very hetero-
geneous [-E roots and words which are united by little more than the
fact that they show an initial labial nasal. We have to distinguish, at least,
L-E *med- (‘measure, accounting for forms like Gothie mit-an), *meHy
‘measure’ and the noun *meHstis, mHotéis, which seems to be derived
from this root, but the laryngeal seems to pose a problem; this root ac
counts for, ia. Albanian mat, Sanskrit mati- etc.) and *meHr-, mHnés,
cone of the I-E words for ‘hand’ (cf. Greek apn, Latin manus, Albanian
marr ‘I take’ ete.). G's remark “[t]hat the proto-Indo-European form is
not an r/n-stem is shown by both internal and external evidence” is not
clear to us, but maybe it is inevitable when distinct roots are forcibly
united under the assumption that the meanings ‘measure’'give’/hand”Reviews / Comptes rendus / Besprechungen 217
are the “same” (the heteroclitic nature of *meHr- is, of course, beyond
reasonable doubt).
‘The “external” evidence for the Eurasiatic pedigree of this semantically and
formally vague pre-proto-root begins with:
Unatic *miye- ‘give, sll (cf. Finnish myy- sell, Nenets mi ‘id?), and
*mer3- ‘measure’ Both roots are taken from Rédei (1988.1:273, 275), G's
main source for Urali, but the ever-cautious Réel does not fail to note
serious difficulties with the second etymon (formal problems in Permic
and the strong possibility that Hungarian mér- “to measure’ may just be
a loan from Slavic, which, ifboth true, would cancel this Finno-Ugric ~
not Uralic - root completely). What we do not find in Rédei’s etymology
is Nenets mana ‘finger’, which may have come in “handy” for G to add
some more of the putative pre-proto-semantics to this root, There is no
such word in Nenets to the best of our knowledge, but maybe G is think
ing of mana ‘paw (of a bear or a wolf), fin (of seal); which does not look
tremendously close to the Nenets verb for ‘give sell, and is in all probabil-
ity a loan from Tungusic (sb).
Yoxaciin min. take’ (Kolyma) does exist, but one should add the much
better attested Tundra form men’ Giving’ and ‘taking’ are certainly re-
{ated concepts, and in some languages they are etymologically connected
as well, but we feel less convinced that this semantic latitude should be
allowed in an attempt to convince skeptics of the existence of a genetic
‘grouping?
Pxoro-Atratc *matie ‘paw, thigh’ and *miala ‘measure’ (cf. Starostin et
al, 2003 *mitia and *miaia). The first of these etyma combines the Tun-
gusic source for Nenets mana discussed above (Ewenki mana, etc.) and a
very hypothetical Turkic *bA8, reconstructed on the basis of severely lo-
calized late forms like Balkan-Turkish maja ‘palm, Uzbek pajpog ‘camel’
foot’ and the like. The second word is backed up by Manchu miyalin “a
measure of capacity’ (and not one of length, as the putative connection
with ‘hand, ‘span’ and the like would suggest).
2. Once reasonable doubl is removed, such comparisons may of course renter the picture
this is the important difference between linguae quarunn affnitas est demonstata, and
Finguae quarum affitas est demonstraela, which is violated in almost every macrocom
pparativist work we are lamiliar withGeorg & Alexander Vovin: Review of Joseph H. Greenberg
Konean: while Korean mal (MK mail) ‘measure for liquids and grains,
18,039 litres indeed exists, alleged Korean mat ‘amount, measur is alex
ical ghost from Ramstedt (1949:140): there is only the MK bound noun
imi (modern Korean mankhum) meaning to the extent of, worth of It
remains unclear how this word would be connected with ‘hand’
Javanese: OJ masu ‘a measure, container for measuring grain’ is act
ally originally used to refer to a square-shaped container used to measure
irain, and not for the measure itself. Needless to say, square containers
do not square well with ‘hand’ either. Within Japonic masw is attested
‘only in Central Japanese, and not in other branches. This fact strongly
suggests that it isa loanword from the above-mentioned Korean word
‘The same can be said about the isolated Manchu word miyalin ‘measure
above: since it is not present in other Tungusic languages, it may well be a
loanword from Korean. Thus, even the alleged proto-“Altaic” provenance
of this word (Starostin et al.2003:922) is questionable.
[Anwu amoini ‘forearm is in all probability also a ghost word, because it
is only attested in one source (Batchelor 1938:31), a dictionary that is
plagued by non-existent words and forms. Ainu amunninfamunin ‘id?
certainly does exist, but the modern -nn- cluster points to the recon-
struction of prote-Ainu “amurnin ‘[forelarm (Vovin 1993:79). The word
has a restricted distribution in Ainu, being limited only to the area
of Southern Hokkaid6. Since reflexes of PA *tap ‘arm’ are much more
‘widespread in various and diverse Ainu dialects, it is highly likely that
Ainu *amurnin represents a local Southern Hokaids innovation with no
Eurasiatic provenance. In addition, this word looks like a compound in
Anu (but its exact morphological composition is unclear). Inthe set Anu
‘moka ~ omeka ‘gift’ also cited by G, the second form does not exist: the
only two forms attested are imeka in Yakumo and inmoka in other Ainu
dialects, This word frequently refers not just to any gift, but rather to a
gift of food leftovers that is given to a guest when he returns home. Any
connection with ‘hand’ willbe very vague at best.
Gunvax is represented by the verb imy- ‘give’ (where G is right that the
initial i- may be a petrified object marker) and nouns like tun'-min’ ‘in
ger’ (Which is attested in early sources only, cf. Grube 1892 furt-min and
Pisudski 1992 — whose data were collected before WW1-twtimun). How-
unlike the majority of other macrocomparativists, separates Korean and Japanese
from Altaic proper, but keeps botl in EurasiaReviews / Comptes rendus / Besprechungen
‘ever, as all other sources on Gilyak make clear, tun’ alone is the word for
“finger” and Sakhalin Gilyak mel does not mean ‘upper arm, but rather
“shoulder blade.
CrtuxoTian, finally, is represented by a root which is best given as
“*manyat- ‘hand; which, however, resembles the putative pre-proto-form
only in its first part.
Having mentioned these details, problems and shortcomings, we feel that no
point has been made regarding the historical reality of this ‘hand ete? etymon
in a Eurasiatic protolanguage. However, it cannot be ruled out that some items
from this plethora of vaguely similar words may indeed be ultimately related,
and not merely fortuitous resemblances. But the theory as advanced and de-
fended by G offers no possiblity to test it against any methodologically solid
yardstick, Its weakest point is, obviously, semantics. While it is of course pos-
sible that, in a given language, words for ‘hand” may be etymologically related
to words for ‘give, ‘measure, ‘sell; ‘paw, ‘foot, ‘sole’ and so on, it can of course
not be accepted that these meanings be treated as “the same” for the purposes
of language comparison. G chose a set of three basic meanings for this “ety-
mon’, but we could easily add to this list meanings such as ‘elbow, fingernail,
knee, grasp, seize, let go, punch, stroke, scales, heavy’ and so on ad infinitum
and inspect the welter of “Eurasiatic” languages in order to find (if only in
the most remote sources and most severely localized dialects) words with the
shape *mV-n, -k,-r conveying any one of these meanings. Who will doubt that
we will ind “evidence” in abundance?
We do not feel the need to scrutinize the “Amerindian” set given by G, since
we have no doubts that the data given there are of the same quality. But, as in
Georg & Vovin 2003, where we compiled a mock list of “Penutian cognates”
to many of G's morphological comparanda to show that his method is unable
to provide us with speciric and INDIVIDUAL-DEFINING evidence, We tried to
take up the implicit challenge of G's dictum that this “root” does not exist else-
where, Well, if we allow a certain laxness in outer shape and concentrate on
the anatomical part of the meaning we think that the (prefixed) root *g-wan
“hand, wrist’ comes close enough. This word is found in a significant subset of
non-contiguous Sino-Tibetan languages (Old Chinese, Kuki-Naga, Himalay
ishs Matisoff 1985:434). And one could possibly also mention Proto-Tai *mwi
A2 ‘hand’ (Li 1977:75, 265) or even better Proto-Kadai *mja A ‘hand’ (Ostapi-
rat 2000221). We will not, however, claim on this evidence that Sino-Tibetan20. Stefan Georg & Alexander Vovin: Review of Joseph H. Greenberg (2002)
(or, for that matter, “Sino-Caucasian” or even “Dene-Caucasian”) and Kadai
are “most closely related” to “Eurasiatic” and “Amerind”.
Progress is possible in comparative linguistics and language classification,
but here, as elsewhere, no shortcuts are available, only those which eventually
lead into a cul-de-sac.
References
Batchelor, John. 1938, An Ainu-English-Japanese Dictionary. Taye: hvanami Shoten,
(Georg, Stefan & Alexander Vovin. 2003, “From Mass Comparison to Mess Comparison
Diachronica 20:2.331-362
Greenberg, Joseph H. 1987, Language in the Americas, Stanford: Stanford University Press.
Greenberg, Joseph H. 2000. bndo-Europeas and its Closest Relatives: The Furasiaic language
family: Volume 1: Grammar. Stanford Stanford University Pres.
Grube, Wihelm. 1892. Dr, Leop. x Schrencks Reisen und Forschungen int Amurlande.
‘Anang 2um I, Bande. Este Lieferung: Linguistsche Ergebnisse, 1: Gilikisches Worter-
verzrichniss nets gram, Benwerkurigen. St Petersbueg.
Li, Fang-Kuci, 1977, A Handbook of Comparative Tai. Honolulu: The University Press of
Havel
Matisott James A. 1985, "Out on a Limb: Arm, hand, and wing in Sino~Tibetan’
‘of the Sino-Tibetas Area: The state ofthe art (= Pacific Lin
by Graham Thurgood, James A. Matisott & David Bradley, 421-450, Canberra: The
‘Australian National University.
‘Ostapirat, Weera. 2000, "Proto-Krs' Linguistics ofthe Tibeto-Burman Avea 231.221
Disudski, Bronista, 1992. Sakhalin Nivhgu Glossary, prepared for print and edited by
Aleksandr B. Ostrovskiy & Allred E. Majewicz. Steszew: International Institute of
Ethnolinguistic and Oriental Stu
Redei, Kitoly. 1988, Uralisces Etymologisces Worterbuch. Baral 1. Uralsce wnd fnnisch
ugrische Schicht. Wiesbaden: Otto Harrassowitz,
loth, Uwe 2001. "Die Entstchung polypersonaler Paradigmen im Uralo-Sibirischen’
Zentnalasiatsche Studien 0163-191
wrostn, Sergei, Anna Dybo & Oleg Mudrak, 2003. Eiymolegical Dictionary ofthe Abie
Languages. Leiden: Brill
Vowin, Alexander. 1993. A Reconstruction of Provo-Ainu, Leiden E.}. Bri
Reviewers’ addresses
Stelan Georg
Heerstrase 7
Bow D-53111 Germany
E-mail: Georg-Bonnet-onlinede