The main goal of this important, thought-provoking work is ‘to set the stage for understanding the successful methods for establishing genetic relationships among languages’ (34), i.e. the techniques for classifying languages on historical linguistic (rather than, for instance, typological) grounds. Special attention is paid to ‘the methodology for investigating possible cases of remote linguistic kinship among languages not yet known to be related’ (394). As Campbell & Poser (henceforth C&P) observe, ‘[i]t is a general problem we have when dealing with prehistory that we are at the mercy of the evidence available’ (328). In order to achieve its objective of grouping languages ‘genetically’ (i.e. according to linguistic kinship), reconstructing the common proto-language, and, conversely, tracing the development from the proto-language down to the individual attested languages, the classical method of historical-comparative linguistics resorts to three main kinds of criteria or evidence: (i) basic vocabulary, (ii) systematic sound correspondences, and (iii) agreement of grammatical patterns, including shared aberrancy (4, 11, 31, etc). However, the comparative method can only reach as far back in time as these types of data allow. Given a sufficiently long lapse of time and because no structural or lexical feature that characterizes one individual language as opposed to another is immune to change, incessantly ongoing linguistic change will eventually wipe out all traces of the properties of a temporally remote ancestral language. Before the process has run its full course, though, there is a considerable time-span, for which the data that might attest kinship are more or less fragmentary or diffuse. In recent years, an increasing number of scholars have turned to exploring data of this sort in order to determine, if possible, larger genetic linguistic groupings, so-called macro-families, while resorting to assumptions and utilizing procedures that do not adhere to the requirements of traditional historical-comparative linguistics. Starting from a discussion of the emergence, characteristics and accomplishments of the classical comparative method, C&P in the volume under review critically scrutinize the procedures applied and the evidence used in the better-known proposals for macro-family groupings and submit that most, if not all, of them severely fail. Apart from the frontispiece and end matter, the volume contains all in all thirteen chapters. The principal author is Campbell, whereas Poser is responsible above all for chapter 5, parts of chapter 4, especially sections 4.8 and 4.11 (on Rask and Bopp, respectively), and an unspecified part of chapter 3 (ix). Considering the enormous wealth of material and topics in this book (historical, descriptive, areal-linguistic, typological, historiographical, and so forth), we can only discuss selected portions of it in this brief review.
The first chapter, ‘Introduction: how are languages shown to be related to one another?’ (1–12), furnishes a systematic overview and summary of what follows in the book. Chapter 2, ‘The beginnings of comparative linguistics’ (13–31), traces early attempts at classifying languages genetically, the gradual formation of comparative linguistics, the developing, though at the time still partial, understanding of the Indo-European (IE) language family, and the elaboration of the three primary and standard types of evidence for a genetic linguistic relationship that we have just referred to. The next, third, chapter, entitled ‘“Asiatic Jones, Oriental Jones”: Sir William Jones’ role in the raise of comparative linguistics’ (32–47), devotes itself to the earliest, most prominent, if largely only nominal, figure-head of comparative linguistics. Although not so indicated in the text, this chapter is a version of Campbell (Reference Campbell, Lakarra and Hualde2007) with moderate alterations, perhaps (cf. above) due to Poser. Here, C&P do away with several frequent misconceptions in linguistics about Jones. Most importantly, they point out that, already in the centuries prior to Jones’ famous 1786 Calcutta address, numerous scholars had discerned various connections between IE languages, that the relationship of Sanskrit to, e.g., Greek and Latin had long been noted, that the analytic practices adopted by Jones had existed before him, and that his procedures cannot easily be identified with the comparative method as understood in later centuries. They also underscore that the celebrated ‘philologer’ passage in the Calcutta address (‘The Sanscrit language, whatever be its antiquity, . . .’, etc.) is notoriously quoted out of context, for which reason parts of it are typically misinterpreted. To Jones’ credit in trying to create a linguistic genealogy, they cite his use of basic vocabulary as well as morphological and some phonetic correspondences, but round off with the shattering verdict that ‘Jones was far from being the initial discoverer of Indo-European relationships or the founder of the comparative method in linguistics’ (47). In chapter 4, ‘Consolidation of comparative linguistics’ (48–73), C&P proceed to outlining the contributions to this field by Gyarmathi, Hervás y Panduro, Adelung, Vater, Schlegel, Rask, Grimm, Humboldt, Bopp, Schleicher, etc., and contest, by means of a series of quotes from Max Müller onwards to Antoine Meillet, a pronouncement by Joseph Greenberg that the Neo-grammarians did not discuss methodology. In chapter 5, ‘How some languages were shown to belong to Indo-European’ (74–86), Poser sketches the history of research allocating Hittite and the sparsely documented Venetic (Lejeune Reference Lejeune1974) to IE and research proving Armenian to constitute a separate subgroup of IE. The broadly conceived chapter 6, ‘Comparative linguistics of other language families and regions’ (87–161), provides in turn an extensive, fascinating account of the history of establishing the Uralic (including Finno-Ugric), Semitic, Austronesian, Dravidian, and Sino-Tibetan language families as well as miscellaneous language families in America, Africa, and Australia.
One of the most central parts of the book is naturally chapter 7, ‘How to show that languages are related: the methods’ (162–223). Presenting numerous procedural caveats, it takes stock of the basic methodology for ascertaining genetic linguistic relationships. It forcefully recapitulates the fundamental methodological insight of historical-comparative linguistics that in order to make a hypothesis about genetic affinity at all plausible, we have to eliminate other potential causes for perceived similarities between the compared languages. Linguistic universals and typologically common traits cannot serve as arguments, nor can borrowings or, for that matter, onomatopoeia, sound symbolism, and nursery words. In particular, this chapter, like several passages elsewhere in the book, brings out very efficiently the great dangers posed by accidental similarity or chance resemblance in comparative research. C&P point here to the estimate that ‘5 to 6 percent of the vocabulary of any two languages may be accidentally similar’ (200, referring to Bender Reference Bender1969). Especially languages with small phoneme inventories and simple word or syllable phonotactics will inevitably exhibit many phonetic/semantic resemblances due to sheer chance. Furthermore, the recourse to short forms, i.e. ‘forms not sufficiently long to eliminate chance as a possible explanation for similarities perceived’ (375) will compound the problem (see also e.g. Ringe Reference Ringe1999). More generally, not only may individual lexical items in two or more unrelated languages exhibit surprising parallel sound/meaning correlations, but even whole sets of words that are semantically similar in the compared languages may display apparent sound correlations. To illustrate, the authors offer an entertaining tidbit of New Zealand Maori and Finnish phonetic/semantic similarities (175; some glossings slightly adjusted by reviewer):
At first blush, this set of words might appear to suggest the consonant correspondences k: k and r: r; yet, these correspondences are completely fortuitous, because it is not possible to carry them over to other word pairs that might be expected to have them and since we cannot demonstrate a systematic structural and lexical relationship between the two languages in other respects. Also a great variety of further issues are treated in this rewarding chapter.
The brief chapter 8, ‘The philosophical–psychological–typological–evolutionary approach to language relationships’ (224–233), is more of an aside, rehearsing the history of special ideological orientations towards language, and plays no great role for the ensuing discussion. In the substantial chapter 9, ‘Assessment of proposed distant genetic relationships’ (234–296), C&P examine critically the better-known proposals for macro-families, including Altaic (core language groups: Turkic, Mongolian and Manchu-Tungusic), Ural-Altaic, the so-called Nostratic, Eurasiatic (in Greenberg's view: IE, Uralic-Yukaghir, Altaic, Korean-Japanese-Ainu, Eskimo-Aleut, etc.), Amerind, and so forth. The unreliable procedures employed by the proponents of such macro-families, the lack of truly solid evidence, or both lead C&P to the assessment that none of these macro-families ‘meets the burden of proof’ (296). Chapter 10, ‘Beyond the comparative method’ (297–329), looks at recent, much publicized, approaches that try to overcome various limitations of the comparative method, above all, the fact that it cannot reach back in time more than some six to eight millennia. The first approach discussed is the type of large-scale, non-genetic, structural comparison as a guide to prehistory that is most prominently associated with Johanna Nichols’ work (e.g. Nichols Reference Nichols1992). On the basis of an examination of selected parts of this research, C&P express their dissatisfaction with most key features of her model. For instance, Nichols’ distinction between linguistic accretion zones with great linguistic diversity and spread zones with little diversity is ‘a misleading and unsupported abstraction and these notions should be abandoned’ (309). Similarly, in regard to another of Nichols’ cardinal concepts, her diachronically and areally stable features, C&P demonstrate that there is ‘nothing particularly stable about most of these putatively stable features’ (310). The assumption in her early publications that languages keep splitting up into pairs at a constant rate is likewise refuted, as are several other claims. R. L. M. Dixon's punctuated equilibrium model (Dixon Reference Dixon1997), inspired by some thinking in biology, fares no better. The notions of equilibrium, with harmonious coexistence of numerous languages in an area, without major language change, but gradual convergence, and punctuation, involving a dramatic event that triggers substantial change and divergence, do not hold up to actually attested linguistic situations. As C&P show, contrary to the hypothesis, equilibrium may be coupled to diversification or exist without diffusion, and punctuation may involve diffusion. The chapter effectively demolishes both Nichols’ and Dixon's proposals. In chapter 11 (330–363), C&P address the question ‘Why and how do languages diversify and spread?’, examining, inter alia, David Nettle's application of the ecological risk hypothesis of economic theory to the spread of languages (the size of ethnolinguistic groups is held to correlate with economic risk: if provisioning is difficult, social networks will grow and their languages allegedly spread), Colin Renfrew's famous farming/language dispersal model, and Peter Trudgill's views of the role of language contact and isolation in producing simplicity or complexity in linguistic change. Again, C&P determine that the proposals are severely flawed with respect to how in actual reality the change, whether language spread or variation in complexity, takes place or not under the postulated favoring or disfavoring conditions. In chapter 12, with the innocent-sounding title ‘What can we learn about the earliest human language by comparing languages known today?’ (364–393), C&P dissect the endeavor by some writers, most notably, Merrit Ruhlen and John D. Bengtson, to trace all languages on earth back to a single proto-language, oddly labeled ‘Proto-World’. Proto-World researchers speculate that features of this primordial language still survive in actually attested ancient and modern languages. In the lexical realm, this means that they try to establish so-called global etymologies, using the procedure of mass or multilateral comparison of a restricted set of lexical items from different known languages. The typical working procedure of global etymology is the ‘casual inspection of lexical similarities’ (72, 395), mostly in living languages. However, the constraints for establishing similarity are not made sufficiently precise; hence, the approach is characterized by overwhelming phonetic and semantic looseness. The comparison of reconstructed forms tends to be neglected, and many other problems plague the enterprise. The argument that constitutes the ultimate death-blow to the lexical part of the endeavor, though, relates to the time-spans involved: ‘given the extremely long time since the origin of human language, absolutely all lexical items from that period have been replaced or changed beyond recognition in all languages’ (380). C&P cannot but conclude that ‘the search for global etymologies is at best a hopeless waste of time, at worst an embarrassment to linguistics as a discipline’ (393). From C&P's discussion, it plainly emerges that the methodology used by adherents of the ‘Proto-World’ hypothesis constitutes a return to a pre-scientific stage of language study.
The final chapter 13, ‘Conclusions: anticipating the future’ (394–403), sums up what has been achieved in the book and ends with a list of over a dozen linguistic genetic affiliations that have been more or less successfully established during the past hundred years or so. In addition, C&P provide an appendix of more than two hundred ‘Hypothesized distant genetic relationships’ (404–415), ranging from the conceivable to the totally unlikely ones, with selected relevant references. The following, extraordinarily comprehensive bibliography, occupying no less than 92 pages (416–507) with, at a very rough estimate, some 1700–1800 entries, constitutes a valuable resource for finding references on historical-comparative linguistics, its methodology, and genetic language classification. A detailed and helpful index of topics, authors, and languages mentioned (508–536) completes the book.
As becomes clear from the preceding overview of the volume, this is a weighty work of very wide coverage that goes right to the heart of the current issues in genetic language classification on a large-scale basis. While the adherents of most contemporary macro-family proposals should be deeply disturbed by its contents, there is for the average historical-comparative linguist little of a major principled sort to disagree with in this excellent book. As it elegantly demonstrates, strict adherence to the comparative method is a must. Moreover, much of its methodological discussion is of interest to related areas of inquiry such as investigations of presumed pre-historic language ‘mixing’, involving adstrata, substrata, superstrata or the like (e.g. works like Vennemann Reference Vennemann2003, who argues for a pre-historic Vasconic substratum and an Afroasiatic or ‘Semitic’ superstratum in Germanic, or, from another angle, Morvan Reference Morvan1996, who seeks to pinpoint a North-Eurasian or Ural-Altaic substratum in Basque). By the nature of their overall objective, C&P will have to cover some historical-comparative ground that is familiar to many of their intended readers, but this is always done in an engaging and illuminating fashion. Moreover, each chapter is written in such a way that it can be read separately, a measure that introduces some slight, but tolerable repetition between different parts of the book. In addition to chapter 3, based on Campbell (Reference Campbell, Lakarra and Hualde2007), substantial other portions of the volume draw on the impressive stream of previous publications by the first author in this area (cf. 162n1, 243n4, 276n19, 298n2, 330n1). In fact, the proportions of text contributed by the two authors are conspicuously unequal. Occasionally, the reader might think of minor emendations of data or forms cited. To give just one example, in the comparison of English, Hindi, and Maori forms following Swadesh's 100-word list (382–384), the information may be added that Maori hēki ‘egg’ (as opposed to the indigenous hua ‘fruit; egg’ or huamanu ‘egg’, lit. ‘birdfruit’; hua < Proto-Polynesian *fua ‘fruit’) and hāona ‘horn’ are both loans from English, that Maori waewae is the regular word for ‘foot’ (also ‘leg’), whereas pū and take, which are also listed, refer to ‘foot’ in the meaning ‘base, foundation’ of, e.g. a mountain or hill, or that the normal Maori verb for ‘come’ is haere mai, haramai, while the cited heke and kuhu (383) rather convey meanings like ‘descend; migrate’ and ‘insert, enter’, respectively (cf. Williams Reference Williams1971, Ngata Reference Ngata1993). In the discussion of potential Eurasian linguistic relationships, a reference to Robbeets’ (Reference Robbeets2005) large-scale investigation of conceivable linguistic parallels in certain Asian languages or language groups would be in order. Similarly, in connection with the listing of works groping in vain for a genetic affiliation for the linguistic isolate Basque (405–406), Trask's (Reference Trask1997:358–415) extensive, devastating appraisal of such undertakings definitely deserves to be mentioned. Technically, C&P's volume is handsomely produced. For a work of this size, inconsistencies, inadvertencies and misprints are extremely rare (e.g. R. L. Trask, i.e. Robert Lawrence (Larry) Trask, happens to be renamed Richard Larry on pages 500, 534; on page 78n6 read Hrozný; on page 514 DeLancey; on page 515 Egerod, Søren; and so forth).
To sum up, this work constitutes an impressive, forceful and convincing plea for maintaining sound, time-honored methodology in genetic linguistic classification. The scope and limitations of the classical comparative method are clearly set out, as are the severe methodological and empirical problems inherent in virtually all current attempts to establish spectacular long-range linguistic relationships. One of the many major insights, long an integral part of the equipment of the historical-comparative linguist, but skillfully elaborated and expounded in various places in the volume, is how easy it is in genetic language classification to be misled by accidental similarities or chance resemblances between two languages – fortuitous coincidences that will inevitably arise just because of the way the linguistic structure and vocabulary of human languages are universally configured. The book deals with a global range of languages, it is well organized and written in a refreshing, accessible style. In view of its broad coverage, it has an obvious function not only as a text for reading, but also as a reference work. Being an obvious textbook candidate for intermediate and advanced classes and seminars on general, IE, Finno-Ugric and any other historical-comparative linguistics, a less expensive paperback edition would be highly desirable. As fascinating and exciting as comprehensive, this work is a masterly synopsis of the state of the art, contemporary issues, and sound argumentation in its area and a valuable guide for conducting further work in genealogical linguistic classification.
