Nicholas Evans is a specialist in Australian Aboriginal languages, as every one of his own dozen and a half reference entries reveals. The fragility of these and other varieties prompted the writing of his book, which he hopes will reach a mixed audience of linguists, language communities, and laymen. Evans feels that something irreplaceable is lost when a language dies, and he believes that “for certain riddles of humanity, just one language holds the key” (xviii). There are some things, he argues, that can never be directly transferred from one language to another: some “dimensions of reality” (179) are apparently accessible only via certain languages, and particular languages go hand-in-hand with “distinct thought-worlds” (180). The general point seems to be that some knowledge needs a quite particular linguistic vehicle. In this connection, Evans suggests that, despite heavy-duty criticism from philosophers and linguists, the Sapir-Whorf hypothesis is “making a comeback” (179), and he suggests that recent psychological experiments “allow us to catch thought alone, away from language's jealous embrace” (163), thus opening the door to more careful examination of the language-cognition relationship. While Evans acknowledges that those separate “thought-worlds” are “neither ultimately incompatible nor hermetically sealed from each other” (180), he tells us that his own experience in learning Kayardild forced him to “add a whole new channel of ongoing attention” (163). Finally, although Evans does not accept the simplistic relationship between biological and linguistic diversity that some others have found so appealing, he does write that “variety is the reservoir of adaptability” (18). Later on, he comments on the “cognitive benefits of multilingualism” (213–14).

This is all by way of introduction to the pointed discussion found in the final chapter. How big of a problem is language endangerment? How do we assess language decline and death? Is a language “safe” when it has lots of speakers or when it is an official medium of a nation state? (In this instance, incidentally, it is good to see that “nation state” is used correctly, signifying a coincidence of political and ethnonational boundaries.) Evans acknowledges that more documentation of threatened languages is needed and, indeed, cites Peter Ladefoged's suggestion that language shift is not always to be resisted. In this connection, he makes some insightful comments about the likely and appropriate roles of “outsiders” and “insiders.” He remarks, too, upon the difficulties attendant upon accurate and enduring archiving: modern electronic records do not, after all, come with lifetime warranties, and we may look back longingly at the permanence of Sumerian clay tablets.

Evans's book is a timely one, if for no other reason than that the current rate of language attrition is matched, as never before, with awareness, interest, and the capacity for fully fleshed documentation. It is also true that Evans presents a great amount of useful detail, together with many illustrations and examples. The basic message here is the simple and familiar one that the loss of languages is a tragedy worthy of greater attention by linguists and laymen. This seems unexceptionable enough, but the underpinnings of the thesis are at least arguable. The most basic problem, of course, is the assumption that there is rather more to the linguistic relativity hypothesis than most specialists are willing to concede. Evans might usefully have given some attention to distinctions between “weaker” and “stronger” versions of Whorfianism—the chief implication being that habit may prove a strong constraint without becoming hard-wired or unalterable. As well, it is a bit disingenuous to imply that psychologists have achieved any sort of unanimity in “catching thought alone” or, relatedly, that their work has reinforced a “strong” Whorfianism. Neither are the “cognitive benefits of multilingualism” a matter beyond psychological doubt. Finally here, the suggestion that “variety is the reservoir of adaptability” is clearly untrue for languages, where altered social circumstances and not any linguistic multiplicity per se drive change and adaptability.

None of this, incidentally, need undermine the work of those concerned to preserve and protect endangered languages. It is surely important, however, to remind ourselves that such work rests essentially upon pillars of preference and perceptions of morality and fairness. Not at all ignoble, of course, but rather different from the bases that are so often suggested.