> No one can deny that Joseph Greenberg was one of the great anthropological
> linguists of the twentieth century (W. Croft, Lg. 77 [2001]: 815–30]). Trained as an
> Africanist, his first contributions were to bring order from the chaos of African language
> classification—he insisted on eliminating appeal to speakers’ ‘racial’ characteristics in
> language grouping, and excluded typological data from consideration. (On my one meeting
> with G, introduced by C. F. Hockett after G’s LSA presidential address in Chicago in
> 1977, I asked whether his typological studies were done because of their intrinsic interest,
> or in order to clarify his genetic endeavors: emphatically the latter, he said.) The importance
> of his work on Afroasiatic languages, on typology, and on universals has never waned. But
> G was not content to wrest order from the chaos of African languages; he gradually
> englobed the entire world, and the climax of his labors is the announcement of a Eurasiatic
> superphylum containing Etruscan, Indo-European, Uralic-Yukaghir, Altaic, Korean-
> Japanese-Ainu, Gilyak, Chukotian, and Eskimo-Aleut.
> G’s African results were published between 1949 and 1954, in articles reprinted
> with revisions in 1955. Not until 1959 did he hit upon what would later be called ‘mass
> comparison’, and the ensuing reduction from 12 African language families to 4 phyla
> (presented in the 1963 Languages of Africa) is increasingly questioned as specialists
> acquire further data (neither Nilo-Saharan nor Khoisan seems likely to survive intense
> scrutiny). ‘The Indo-Pacific hypothesis’ (Current trends in linguistics 8 [1971]: 808–71)
> sank virtually without a trace (R. Blust, Lg. 54 [1978]: 475–77, anticipated most types of
> criticism that would be leveled a decade later), but then the eagerly awaited Language in the
> Americas (1987, adumbrated as early as 1953 and 1960) was delivered into the hands of
> Americanists, and their critique was devastating. Aside from the notorious ‘should be
> shouted down’ (Current Anthropology 27 [1986]: 488), with its evocation of the
> censorious reaction to Immanuel Velikovsky that undeservedly turned him into a media
> star in 1951, which has been quoted ever since to ‘prove’ that professional linguists are
> hostile to exciting new ideas, a panel spread over several numbers of the International
> Journal of American Linguistics 58 (1992) detailed problems of methodology and of
> misinterpreted data. G’s reply (IJAL 62 [1996]: 131–64; similarly those in CA 27: 492f.,
> 28 [1987]: 664–66, and Lg. 65 [1989]: 107–14) defended the data but as to methodology
> mostly referred to the treatment in Essays in linguistics (1957, chap. 3), which includes
> probably the first of many irrelevant invocations of the discovery of Indo-European as the
> pattern for mass comparison. G never came to terms with the fact that IE is unique among
> the world’s language phyla in the shallowness of its branching and the depth of its
> attestation over multiple families. No other phylum has the luxury of such ancient diversity
> as Hittite, Vedic and Avestan, and Mycenaean—or even Homeric, Italic, and Gothic from
> more recent centuries.
> The title of the present volumes was derived from that of Orrin W. Robinson’s Old
> English and its closest relatives: A survey of the earliest Germanic languages (Stanford,
> 1992; cf. Lg. 69 [1993]: 836–39), perhaps in an attempt to suggest its content was similar.
> This only does a disservice to the earlier book. Vol. 1, Grammar, comprises 3 chapters: 1,
> ‘The historical background’ (1–23), surveys prior ‘long-ranger’ proposals and tries to
> show that the Eurasiatic defended here isn’t all that different from the Nostratic of Illich-
> Svitych and his Moscow school, despite G’s exclusion of Afroasiatic, Dravidian, and
> Kartvelian. 2, ‘Some aspects of the comparative phonology of Eurasiatic’ (24–60), does
> not discuss comparative phonology at all: it shows or suggests that all the branches of
> Eurasiatic had some form of vowel harmony. These harmonies are of various types, and
> no attempt is made to show how all of them could have derived from some common
> ancestor; and is not the presence of vowel harmony one of those ‘typological criteria’
> whose ‘exclusion’ G called ‘central’ (Word 45 [1994]: 24)? 3, ‘Grammatical evidence for
> Eurasiatic’ (61–239), discusses, in less than one to a few pages each, 72 putative
> morphemes found in more than one branch (the last few occurring in Altaic only). Again,
> there is no attempt to display any coherent paradigms, nor any attempt at showing their
> development through the millennia.
> Vol. 2, Lexicon, offers 437 ‘etymologies’ (in alphabetical order of their glosses,
> rendering utterly otiose the ‘Semantic index’ [205–10]), ‘focus[ing] on those that involve
> languages and families that have generally been excluded from the Nostratic family’ (9);
> unlike in LIA, a ‘phonetic’ form is provided for each one. From the ‘Phonetic index’
> (211–16) the putative phonemic inventory of Eurasiatic can be extracted:
> p  t       t‘    c       k       q               i                               u
> b  d                       g                               e               o
>    s               c*                                              a
> m  n                       N
> v  r       l
> w  y                               h
> (most forms are CVC; a few consonant clusters occur). It also reveals that as many as 5
> independent etyma could have exactly the same shape (and sets of 2, 3, or 4 are frequent):
> from ter are derived Proto-Altaic(!) *t‘e¯r´u ‘dung’, *tÌa¯re ‘narrow’, *t‘era ‘hair, wool’,
> *tÌorku ‘star’, and *t‘Ìaru ‘curse’ (the first four corresponding respectively to Middle
> Korean t@¯l ‘dirty’, tj@la- ‘short’, t@lhi ‘hair’, and tal ‘moon’). Neither of these findings
> seems prima facie likely.
> Prominent at the end of vol. 1 chap. 1 is ‘The question of Etruscan’ (22f.), which
> claims ‘I have cited Etruscan grammatical elements in Chapter 3’—but I cannot find a
> single mention of Etruscan anywhere therein, and the reference list (283–311) includes not
> a single source by an Etruscanist. A total of two Etruscan forms is given in vol. 2, tul
> ‘stone’ < *tul and tur ‘give’ <  *to. Does this suffice to join an ill-understood language to a
> superphylum?
> Supposing that G is correct in his postulation of Eurasiatic and that all his data are
> impeccable, he has conspicuously failed to carry out what was previously an important part
> of his program: subgrouping language families (cf. Essays chap. 4, LIA, CA 27: 479, 28:
> 651, etc.): the 11 branches of Amerind fall into 6 groups of 1, 2, or 3—but the 8 branches
> of Eurasiatic are simply presented as coordinate. G’s previous classifications are
> publications of what others would consider the base, not the apex, of their work: suggested
> groups of languages, which might or might not be confirmed through the real work of
> comparative linguistics, the demonstration of systematic correspondences (ideally, of
> shared morphological innovations; R. Hetzron, Lingua 38 [1976]: 89–104). The Eurasiatic
> material does not achieve even the level of the Amerind.
> Also absent from these volumes are tables like the useful ones in LIA appendix C
> showing the distribution of morphemes among the ‘etymologies’. A liberal scattering of
> typos in the English prose, and the identification in the reference list of vol. 2 (196–202) of
> ‘Greenberg 2000a’ as LIA rather than as vol. 1, do not bolster confidence in the accuracy
> of the data included. [Peter T. Daniels, New York, NY]