This is an easy book to respect, a hard one to like, and a nearly impossible one to review. Easy: who wouldn't be impressed by a fat volume with 45 contributions by distinguished scholars on Sumerian, Epi-Olmec, and a whole lot in between? Hard: the book is not pretty and, more problematically, it is unclear who will be using it and for what purpose. And impossible: no one in the world would be competent to pronounce with even a modicum of authority on each and every chapter. When I was invited to try my hand at saying something about it, a member of BMCR's Editorial Board suggested that I might produce a brief "ALSO SEEN"-review, but that seems unfair to W(oodard) -- the book weighs 6.2 pounds! -- just as a blow-by-blow account of what I learned in the course of attempting to absorb over one thousand pages of fine print seems unfair to everyone else. The result is less a conventional review than some musings about what makes interdisciplinarity difficult.
What counts as an "ancient language"? The terminus post quem is of course Sumerian, whose oldest tablets date to around 3200 B.C.; the terminus ante quem is the 5th century A.D. -- "admittedly 'traditional'" (1), as W writes, thinking of A.D. 476, but he makes a good case at the beginning of the "Introduction" for stopping there. (In fact, the inspired decision to close the volume with Ringe's very fine chapter "Reconstructed Ancient Languages" gives the reader the opportunity to go back before the late 4th millennium B.C. and consider in a meaningful way the important question of what words sounded like before they were first written down.1) Even aside from those languages and scripts whose remains are mere disiecta membra (e.g., Sicel) and those that we understand not or hardly at all (e.g., Cypro-Minoan), there are still a remarkable number of languages attested over the course of the first 4,000 years of human logographic practice: Greek and Latin, of course, but also Elamite, Hurrian, Ge'ez, Lydian, Old Persian, Gothic, Armenian, Chinese, Mayan, ... How nice, then, to have access to such a range of knowledge between two covers.
But what kind of knowledge about these languages does W's book present? The answer is that each chapter is supposed to give a "meaningful grammatical description" (3). And the standard for "meaningful" here is extremely high, for these "mini-grammars" are very detailed indeed. For example, the main subsections in W's own 35-page chapter on Attic Greek are as follows:
1. Historical and Cultural Contexts; 2. Writing Systems; 3. Phonology [3.1 Consonants, 3.2 Vowels, 3.3. Phonotaxis, 3.4 Syllable Structure, 3.5 Vowel Length, 3.6 Accent, 3.7 Diachronic Developments (3.7.1 Obstruents, 3.7.2 Sonorants, 3.7.3 Combinatory Changes, 3.7.4 Laryngeals, 3.7.5 Vowels)]; 4. Morphology [4.1 Nominal Morphology (4.1.1 Noun Classes, 4.1.2 Adjectives, 4.1.3 Pronouns), 4.2 Verbal Morphology (4.2.1 Verbal Aspect, 4.2.2 Thematic Present Tense Stems, 4.2.3 Athematic Present Tense Stems, 4.2.4 Imperfect Tense, 4.2.5 Future Tense Stems, 4.2.6 Aorist Tense Stems, 4.2.7 Perfect Tense Stems, 4.2.8 Nonindicative Moods, 4.2.9 Verb Endings, 4.2.10 Infinitives, 4.2.11 Participles, 4.2.12 Verbal Adjectives), 4.3 Adverbs, 4.4 Compounds, 4.5 Numerals]; 5. Syntax [5.1 Word Order, 5.2 Clitics, 5.3 Post- and Prepositives, 5.4 Coordination, 5.5 Subordination, 5.6 Conditional Clauses, 5.7 Agreement, 5.8 Long-distance Anaphora]; 6. Lexicon; and 7. Reading List.
Now, this list is likely to look daunting even to the readers of BMCR, the vast majority of whom are comfortable with the language of Demosthenes and Plato. So imagine that you are a Classicist and have become interested in Greco-Aramaic interactions -- and reasonably enough, for this is a subject on which there is sure to be much research in the coming decades.2 What would you make of the 35-page chapter on Aramaic, which contains a separate (though short) section on "Assimilation and Dissimilation of /n/" and more than three pages -- most of it detailed notes to grammatical tables -- on bound personal pronouns?3 I can come up with five possible answers: (1) You already know Aramaic and so have no need to read it; (2) You do not yet know Aramaic but are linguistically extraordinarily gifted and have both the interest and the talent to internalize the information as presented here and then figure out how to go on to learn more; (3) You get some vague idea (or not) from the chapter of what you think you might be interested in and you then pay a visit or two or three to your colleague who specializes in Aramaic -- which, of course, you could have done anyway, provided you are lucky enough to have such a colleague; or (4) You get discouraged and give up on the whole enterprise. Most people, I suggest, fall into category (3) or category (4). Or into category (5): You think you really understand something on the basis of your reading (be it careful, be it superficial) and then you go off and, without much fuss, conduct "research" as though you were genuinely an expert in Aramaic linguistics and philology.
So these are the main problems I see with this encyclopedia, which seems to me a great thing in theory, but fundamentally misconceived. Cross-linguistic and -cultural work should be encouraged, but it is very difficult to think of who would be able to use W's book successfully. There just are not very many people who have the talent to abstract large amounts of generally usable information from a synoptic grammar, however good it may be. And even for those who can do this, an understanding of what is in a 35-page grammar is not the same thing as true understanding: many people style themselves as interdisciplinary, but (to coin a phrase) a little knowledge is a dangerous thing, and it is almost inevitable that even the most careful specialists in one field (Greek literature, say) will make mistakes in their early forays into another. The last thing I wish to do in writing these words is discourage boundary-stretching research -- quite the contrary! -- but it is worth remembering that the only thing worse than bad scholarship in one field is bad scholarship in more than one field.
One of the purposes to which scholars will most likely put the book is to aid in the study of geographical and intellectual areas where different languages, different grammars collide: Greek and Aramaic, for example. Interest in ancient bilingualism is, after all, running high.4 But the book disappoints in missing the chance to consider linguistic contact seriously. There are often nice, if scattered, remarks on lexical diffusion in the section entitled "Lexicon" that comes toward the end of almost every chapter, but the work would have been more immediately useful (though also even harder to produce) if there had been longer discussions of linguistic interaction in addition to the chapters on individual languages.5
A quick test of the book's success that should appeal to most of the present readers is to see how well the chapters that are indisputably Classical in focus deal with their subjects: "Attic Greek" and "Greek Dialects," both by W, and Clackson's "Latin." The chapter on Latin is very good, covering a lot of ground in just over 22 clearly written pages. There are, of course, some things about which one might quibble: for example, Clackson has the "Cambridge-school" reluctance to consider as a matter of course Latin(o-Faliscan) and the Sabellic languages (which Wallace treats in the following chapter) as genetically very closely related and belonging to the subgroup of Indo-European generally known as Italic (see pp. 789f.), and it was perhaps not a good idea for him to cite pontifex as an example of a compound ("lit. 'one who makes a bridge'"; 804) when both its original meaning and even its first element have been the subject of such intense debate.6 But these are very minor points and do not detract from the quality of Clackson's effort. As it happens, I even learned something about Latin morphosyntax from the chapter. On pp. 808f. Clackson discusses Livy 32.29.1 murus et porta de caelo tacta erant7 and explains how it is possible that neuter tacta can agree with masculine murus and feminine porta: "Agreement patterns in conjoined nominal phrases are ... interesting: in phrases where there is a single adjective but two conjoined nouns of different genders, the adjective will either be marked for agreement with the closest noun, or will be marked masculine or neuter. The choice between masculine and neuter is governed by the animacy of the referents: if the two conjoined nouns refer to animates, adjectives take the masculine; if inanimates, adjectives are marked neuter" (808). The use of the neuter here is indeed linguistically fascinating, and I confess that this grammatical possibility had never impressed itself in my head, despite its confident appearance in Allen & Greenough's and Bennett's grammars and in "Bradley's Arnold." However, I am not certain that what is in fact a rare option -- see now the description in "Der neue 'Menge'" -- deserves space in a work such as this.8
Unfortunately, I cannot be as positive about the two chapters on Greek, though much of my displeasure comes not from factual inaccuracy as such but from the simple observation that the editor of the encyclopedia himself displays astonishing sloppiness in his own contributions. The problems are largely of the kind it is not possible to exemplify in an on-line publication and not worth trying to explain in tortured prose: suffice it to say that in addition to various typographical mistakes of a familiar sort, macrons and other diacritical marks are sometimes there, sometimes not; there are missing and mistaken accents on Greek forms; and problems with and inconsistencies in the presentation of fonts are rampant, more so than in any other parts of the work. Someone who hands $170.00 over to "the oldest printing and publishing house in the world" deserves better. There are some substantive problems as well, beginning already with the second paragraph of the essay on Attic. W introduces the languages as follows: "Greek is a member of the Indo-European family of languages. It resides in that major subdivision of the family called centum ..., though its closest linguistic affinities are with the Indo-Iranian and Armenian languages, both members of the satem subset" (614). This statement is overtly contradictory: if language X belongs in a major linguistic subdivision a, then it by definition cannot have its closest linguistic affinities with two languages, Y and Z, in a different subdivision. If I were an alert student who came across this while trying to find something out about Greek, I would throw up my hands in despair. Furthermore, the fact is -- and I cannot stress this strongly enough -- that the hoary division of the Indo-European world into "centum languages" and "satem languages" (i.e., those that merge palatal and velar stops as velars [cf. Latin centum '100'] and those that merge velars and labiovelars and turn the palatals into quite distinct sibilants [cf. Avestan sat@m, the cognate of centum]), though possibly still useful in elementary classes on historical/comparative linguistics, is clearly inaccurate as a model of how the family actually developed and should therefore be abandoned.9 It is dispiriting to see in the chapter on Indo-European (co-authored by W; see fn. 1 above) a paragraph on the centum/satem-division in which the "conspicuous exception to [the centum/satem] distributional pattern" (536) is said to be Tocharian and in which one single paper is cited -- without any recognition that the major contribution of the paper in question, by H. Craig Melchert, is to argue conclusively against said division on the basis of evidence from Luvian.10
It would be tedious for me to go on at length about small factual errors and infelicities throughout the volume, and I list here merely five points where I scribbled an exclamatory note to myself.
p. 4 -- W ("Introduction") implies that Linear A might be Anatolian. Even a weak endorsement of this idea is unfortunate in a book that will be used as a major reference tool.
pp. 139 and 160 -- Huehnergard ("Afro-Asiatic") says that the Omotic languages are spoken by "about 3 million people"; Loprieno ("Ancient Egyptian and Coptic") says that there are "approximately one million speakers."
p. 536 -- Hoenigswald and W ("Indo-European") state that the reconstruction of the interdental fricative "thorn" in the Proto-Indo-European phonological inventory "remains problematic as another, more sophisticated solution has been proposed" -- but they do not give any indication of what this other solution actually is.11
p. 646 -- In the section on coordinating particles in Attic Greek, W writes as follows: "The conjunctive particle dé (δέ) is frequently used to introduce clauses and occurs in second position. Often a clause so introduced is coupled with a second clause marked by the particle mén (μέν), the two existing in a contrastive relationship ('on the one hand' ... 'on the other hand')." Hysteron proteron in a grammar?!
pp. 836f. -- Wallace gives five examples of "Sabellian vocabulary items [that] have solid etymological connections with languages in other branches of Indo-European but lack Latino-Faliscan cognates" (836), but three of the five do or may in fact have interesting correspondents in Latin: the word for 'son' seen in, e.g., Oscan puklui (dat. sg.) is found, with a slight semantic shift, in Latin pullus; the word for 'fire' seen in, e.g., Umbrian pir is fossilized in the Latin verb pur(i)go; and the word for 'community' or 'tribe' seen in, e.g., Oscan touto may perhaps underlie the Latin adjective totus.12
This encyclopedia has been in the works for too long, and that is surely one of the reasons it is not as good as one might have hoped.13 But of course producing it, even imperfectly, must have been quite an operation. Let me also be emphatic, then, that there are great things to be found in it: for example, Hale's chapters on Avestan (especially Young Avestan) and Pahlavi are admirably clear descriptions of two languages I have never found to be admirably clear; a number of contributors pay more than the usual lip service to syntax (e.g., Hale on Avestan again and Faarlund on Ancient Nordic, as well as Peyraube on Ancient Chinese); and Kaufman and Justeson's account of Epi-Olmec (with an appendix on Zapotec) is absolutely brilliant, cutting-edge stuff, the sort of thing that reminds this reader why he became a linguist in the first place. Every serious library should have a copy in its reference collection, and many serious scholars will enjoy dipping into it and finding out just how linguistically diverse the ancient world really was. The problem is simply that it is a book that will be easy to misuse, for it consistently provides both too much information and too little.
1. Note also the chapters "Afro-Asiatic" by Huehnergard (largely just about (Proto-)Semitic, however) and "Indo-European" by Hoenigswald and W ("with a discussion of syntax" by Clackson), each of which provides a summary of some major linguistic features of a prominent family (not of an attested language as such) and of the proto-language from which these features appear to have arisen. (Even though I am an Indo-Europeanist by profession, I found parts of the latter chapter extremely difficult to understand and would not choose to send a beginning student to it. The opening two paragraphs on the Comparative Method [see p. 534], for example, bear the unmistakable stamp of Hoenigswald, the greatest 20th-century theoretician of historical linguistic methodology [he died in 2003 and W dedicates the book to his memory] but not someone known for his easy style of exposition.) In general, the focus throughout the volume is (properly) on synchronic description, but diachrony rears her pretty head from time to time (a good example is the discussion of the origin of the Tiberian vowels on pp. 326-30 of McCarter's chapter on Hebrew).
2. See most recently David G. K. Taylor's excellent paper "Bilingualism and Diglossia in Late Antique Syria and Mesopotamia," in Bilingualism in Ancient Society: Language Contact and the Written Text, ed. by J. N. Adams, Mark Janse, & Simon Swain (Oxford: Oxford University Press, 2002) 298-331.
3. I would like to stress that I have no special complaints about Creason's chapter on Aramaic as such. I could just as easily have written about Latino-Punic or Armeno-Parthian interactions -- and the tenor of my comments would have been more or less the same since the chapters are all written according to the same basic template.
4. See especially the volume cited above in fn. 2 and J. N. Adams, Bilingualism and the Latin Language (Cambridge: Cambridge University Press, 2003). Eleanor Dickey has a review article of both ("Ancient Bilingualism") in Journal of Roman Studies 93 (2003) 295-302, and I have reviewed them in respectively BMCR 2004.02.19 and New England Classical Journal 31 (2004) 439-42.
5. As a curiosity, I note that three different authors choose to highlight in their respective sections on the lexicon manifestations of one single Wanderwort, the word for 'lapis lazuli' or 'dark blue enamel': Michalowski writes in "Sumerian" that "za3-gin2 'lapis lazuli' compounded with za 'stone' is of the same unknown origin as Akkadian uqnu or Hittite ku(wa)nna(s^)" (54); Huehnergard in "Afro-Asiatic" compares "?uk'niy- 'lapis lazuli' ... with ... Hittite ku(wa)nna-, Greek κϋάνος [sic: the word is actually proparoxytone]" (156); and W in "Attic Greek" calls Greek κύανος a "Hittite loan ... though itself likely of non-Hittite origin" (648).
6. For an overview of the word's etymological possibilities, see Françoise Van Haeperen, Le Collège Pontifical (3ème s. a. C.-4 ème s. p. C.): contribution à l'étude de la religion publique romaine (Brussels: Institut Historique Belge de Rome, 2002) 11-45. The best linguistically based discussion is that of Marina Benedetti, I composti radicali latini: Esame storico e comparativo (Pisa: Giardini, 1988) 98-104, which begins, "Sull'etimologia di pontifex esiste una bibliografia vastissima." Benedetti, who does believe that ponti- is connected to pons, nevertheless argues strongly that the original meaning is 'facitore di cammini' rather than (with, e.g., Varro, Ling. 5.83) 'facitore, costruttore di ponti'; the same conclusion has now been reached by Juan Antonio Álvarez-Pedrosa Núñez & Juan José Carracedo Doval, "La etimología de latín pontifex," 'Ilu: Revista de Ciencias de las Religiones 5 (2000) 25-35, who adduce interesting material from Hittite ritual (I owe this last reference to Brent Vine). The most recent attempt to see a different first element entirely is Bernard J. Kavanagh's paper "Pontifices, Bridge-making and Ribezzo Revisited," Glotta 76 (2000, publ. 2002) 59-65, which picks up an old idea of F. Ribezzo that ponti- comes from a Sabellic form of 'five' (cf. Umbrian puntes).
7. Clackson translates this as, "The wall and gate had been touched from the sky," which is the sort of literal rendering that will make non-Classicists understand the famous rhyme that "Latin is a language, / As dead as dead can be: / First it killed the Romans, / And now it's killing me." Of course what Livy is really saying is that lightning struck.
8. See Allen and Greenough's New Latin Grammar for Schools and Colleges, founded on Comparative Grammar, ed. by J. B. Greenough et al. (Boston: Ginn, 1903; reprinted now by Caratzas and also -- in a 2001 edition updated by Anne Mahoney and with different pagination -- by Focus) 172 (para. 287.3), Charles E. Bennett, New Latin Grammar, 2nd rev. ed. (Boston: Allyn & Bacon, 1918; reprinted now by Bolchazy-Carducci) 154 (para. 235 B 2 b) β)), and 'Bradley's Arnold': Latin Prose Composition, ed. and rev. by J. F. Mountford (London: Longmans, Green, 1938; reprinted now by Bristol Classical Press) 49f. (para. 48 (b) and (c)). For much greater detail and nuance, see Hermann Menge, Lehrbuch der lateinischen Syntax und Semantik, new ed. by Thorsten Burkard & Markus Schauer (Darmstadt: Wissenschaftliche Buchgesellschaft, 2000) 325 (para. 257 (3)), who gives examples from Cicero and calls the choice of neuter plural "[s]eltener" than a singular that agrees with the nearer subject.
9. See now Michael Meier-Brügger, Indogermanische Sprachwissenschaft, 8th ed. (Berlin: de Gruyter, 2002) 130f. (~ Indo-European Linguistics, trans. by Charles Gertmenian [Berlin: de Gruyter, 2003] 130).
10. See H. Craig Melchert, "PIE Velars in Luvian," in Studies in Memory of Warren Cowgill (1929-1985): Papers from the Fourth East Coast Indo-European Conference, Cornell University, June 6-9, 1985, ed. by Calvert Watkins (Berlin: de Gruyter, 1987) 182-204 (similar conclusions were reached independently by Anna Morpurgo Davies & J. D. Hawkins, "A Luwian Heart," in Studi di storia e di filologia anatolica dedicati a Giovanni Pugliese Carratelli, ed. by Fiorella Imparati [Florence: ELITE, 1988] 169-82). Important further remarks on the question that take account also of Lycian are to be found in Melchert's paper "New Luvo-Lycian Isoglosses," Historische Sprachforschung 102 (1989) 23-45, at 23-32, and see also his Anatolian Historical Phonology (Amsterdam: Rodopi, 1994) 53f. and passim. For an overview of the "theory" of centum vs. satem, see Johann Tischler, "Hundert Jahre kentum-satem Theorie," Indogermanische Forschungen 95 (1990) 63-98; an interesting but to my mind wrong-headed approach to the matter is found in Andrew L. Sihler, "The Myth of Direct Reflexes of the PIE Palatal Series in Kati," in Studies in Honor of Jaan Puhvel. Part One: Ancient Languages and Philology, ed. by Dorothy Disterheft, Martin Huld, & John Greppin (Washington, D.C.: Institute for the Study of Man, 1997) 187-94. There is, by the way, nothing odd about the "centum language" Tocharian except that it was spoken far to the east while most of the "centum languages" are western; what really is odd if one believes in the reality of the centum/satem-division is that there may be some "mistreatments" (i.e., centum for satem or vice versa) in Armenian, Albanian, and perhaps elsewhere as well. I note that the terms "centum" and "satem" show up in passing in a few papers aside from the one on Indo-European: on p. 676 (Jamison, "Sanskrit"), on p. 742 (Hale, "Avestan"; note his pointed remark, "For those who hold that the centum-satem division ... is a matter of subgrouping, ...," which rightly casts doubt on the whole idea), on p. 780 (Brixhe, "Phrygian"), on p. 842 (Wallace, "Venetic"), and on p. 905 n. 1 (Jasanoff, "Gothic").
11. For more information, see, e.g., pp. 72 and 105f. in both the original German edition and the English translation of Meier-Brügger's handbook (see fn. 9 above) and the references cited there.
12. Latin pullus and Oscan puklu- are clearly very closely related, whether or not one accepts that they are exact cognates, both going back to Proto-Indo-European *put-lo-: according to the more usual view, pullus derives rather from a would-be variant put-slo- (see, e.g., Gerhard Meiser, Historische Laut- und Formenlehre der lateinischen Sprache [Darmstadt: Wissenschaftliche Buchgesellschaft, 1998] 119); alternatively, Benedicte Nielsen in her 1998 Copenhagen M.A. thesis suggests a scenario whereby the expected preform *put-lo- might be right after all (her idea has been noted most recently in two papers in Transactions of the Philological Society 100 , "Nominal Composition in Sabellic and Proto-Italic" by Frank Heidermanns [185-202, at 188 n. 7] and "The Compound as a Phonological Domain in Indo-European" by Jens Elmegaord Rasmussen [331-50, at 343]). On R. Thurneysen's old idea that pur(i)go comes from a collocation *pur ago 'carry fire,' see Gregory Nagy, Greek Mythology and Poetics (Ithaca: Cornell University Press, 1990) 180 (~ "Six Studies of Sacral Vocabulary relating to the Fireplace," Harvard Studies in Classical Philology 78  71-106, at 105); Nagy re-emphasized the etymology in his presentation "The Iguvine Tables and Ritual Language: Teaching Culture through Formula and Variation" at the 136th Annual Meeting of the American Philological Association [Boston, January 2005]). The prehistory of totus is very much disputed, with some scholars suggesting -- often cautiously -- that it goes back to something that meant 'of the whole tribe' (see, e.g., Calvert Watkins, The American Heritage Dictionary of Indo-European Roots, 2nd ed. [Boston: Houghton Mifflin, 2000] 92) and others vehemently denying this (above all Stefan Zimmer in many publications, beginning with "Zur Etymologie und zu den ältesten Belegen von lat. totus," Glotta 63  221-25).
13. Colleagues have talked about the work's imminent appearance for many years; I read drafts of several of the chapters long in advance of publication and am in fact -- full disclosure -- explicitly thanked for commenting on Eska's "Continental Celtic," something I did in the summer of 1996. (Tuite notes that his "Early Georgian" was mostly written in 1996, and there are indications of a date of completion long before 2004 in the chapters on "Etruscan" and "Epi-Olmec" as well.)