Are some languages better than others? The question posed in the title of this book is, as Dixon acknowledges, one that not many linguists would ask. And indeed, the book is not written for linguists, but for a general audience. To some extent, its provocative title serves as a pretext for showing the reader a glimpse of the diversity of spoken human languages, with sections illustrating cross-linguistic variation in features such as case, tense, negation, possession, evidentiality, and definiteness. (Signed languages are not mentioned at all.) Many of the examples come from languages Dixon has worked on in his long and distinguished career as a descriptive linguist and fieldworker, in particular Dyirbal (indigenous to the Australian state of Queensland) and Jarawara (indigenous to the Brazilian state of Amazonas). But Dixon's purpose here is not solely descriptive, and he argues (in chapter 1) that the question in the title is one that should be taken seriously as part of the scientific study of language.

Linguists have shied away from asking this question, Dixon suggests, because of its racist history; in distancing ourselves from colonialist prejudices about the relative worth of languages, we have retreated into a dogmatic insistence on equality. But, he contends, “[i]f linguistics is to be regarded as a natural science […] then evaluation must be an element in its modus operandi” (p. 8). For Dixon, evaluation is one of the four goals of any science, the other three being description, explanation, and prediction. I am skeptical about the inclusion of evaluation as a defining property of science – does a zoologist ask whether some animals are better than others? Does an astronomer try to identify the best stars? Zoologists' research might help a rancher decide which animals are the best to raise for a particular purpose in a particular climate, and the work of astronomers might tell a navigator which stars are the most useful to steer by, but these are examples of how scientific knowledge may be applied, and not necessarily part of the practice of science itself.

In any case, if one wants to ask whether some languages are better than others, one must first define the criteria of evaluation. Dixon addresses this question promptly in section 1.1, and returns to it in chapter 9, “Better for what purpose?” The functions of language he identifies are essentially social ones: the list in section 1.1 comprises “assisting in the process of belonging”, “enabling cooperative endeavour”, “reflecting social organisation”, “display[ing] emotions”, “convey[ing] information”, “aesthetic expression”, “scholarly thought and argumentation”, and “proselytisation” (pp. 2–4). But the grammatical features he discusses in the rest of the book are almost entirely irrelevant to the practical matter of choosing a language that will most effectively serve these purposes. A language's suitability for social goals depends on its social properties, not its structural ones. If I want to convey information or emotion, or to convince an audience of some proposition, what I care about is what languages my intended audience knows, and what social values they assign to them – not which language has the optimal number of tenses or cases. A person who is campaigning to be Prime Minister of Canada needs to be able to make their case in English and French, not because English and French have formal properties that are particularly suited to political persuasion, but simply because they are the official languages of Canada, and the ones with the largest numbers of speakers. The factors that make a language better or worse for the purposes Dixon identifies are external to the language itself.

One exception might be the function of “reflecting social organisation”: presumably Dyirbal kinship terms do a better job of reflecting what familial relationships are important in Dyirbal society than those of other languages, and Korean honorific markers are particularly well suited to reflecting Korean norms of politeness and social structure. But such statements seem almost tautological, and the observation that each language is good at reflecting the organization of the society in which it is spoken does not get us any closer to saying whether one language is better than another in any general sense.

By suggesting that some languages might be better than others for communicative purposes, Dixon challenges Jakobson's (Reference Jakobson and Brower1959: 236) dictum that “languages differ essentially in what they must convey and not in what they may convey.” In support of that challenge, though, the book offers only a single anecdote about a difficulty in translation. In section 8.2 (pp. 183–187), Dixon recalls being asked by some heritage speakers of Dyirbal for help in translating a speech into Dyirbal from English. The English speech included the sentences “Jim Robertson was our chief. He was an honest man.” Dixon explains that Dyirbal has no expression for ‘chief’, because there were no chiefs in traditional Dyirbal society (the role having been imposed by white settler policemen), and none for ‘honest’, because honesty was assumed of everyone (and translating the sentence as ‘He never lied’ would wrongly imply that Robertson had been accused of lying).

Professional translators deal with problems like this all the time, and one can imagine various solutions in different contexts. In a conversation between Dyirbal speakers who also knew English, the speaker might simply borrow the English words chief and honest, in the same way that an English speaker might use Japanese words to say something like “The samurai adhered to bushidō.” Or, if the English terms were unfamiliar to the listener, the speaker might provide a longer explanation in Dyirbal, sacrificing the concision of the English text but still conveying its meaning. The situation Dixon describes is a somewhat artificial one, in which the point is not really to transmit a specific meaning in Dyirbal (which the audience of the speech could not be expected to understand), but rather to say some laudatory words about Robertson while using the language to performatively highlight the speaker's Dyirbal identity. If there were still a robust community of fluent Dyirbal speakers, the language might have developed standard expressions for concepts from the adjacent white culture, and there would certainly be people who could compose a speech in Dyirbal that would serve the intended purpose. But the deliberate suppression of Dyirbal language and culture has left very few speakers and considerable language attrition among those who remain (Schmidt Reference Schmidt, Seliger and Vago1991; Dixon Reference Dixon2015: ch. 16). If Dixon's point here is simply that there are ideas that can be expressed more concisely in one language than in another, this is certainly true, and could equally well have been illustrated with examples of Dyirbal words that have no simple translation into English. But in the context of the question asked in the title of the book, the anecdote seems to invite the reader to infer that Dyirbal has less expressive power than English, even though a single example cannot possibly support any such conclusion, and unfair as it may be to compare a dying language with a thriving one. Frustratingly, Dixon does not report what advice he gave the speakers; did he come up with a solution, or did he just tell them that their language couldn't do what they wanted it to?

Although Dixon claims that the endeavour of evaluating languages is a scientific one, the book reflects his own preferences about the distinctions a language should make, rather than proposing any more objective metric. Occasionally he sounds like a peevish prescriptivist, as when in two separate places (pp. 160–161, 195) he decries the use of the word gender to refer to a property of people (rather than a grammatical property of words). Thus on page 195: “The implication is that one should now say: ‘Noun Mädchen ‘girl’ in German is of neuter gender but refers to a person of female sex gender’. This is daft.” Although Dixon calls it “recent” (p. 160), this use of gender is not all that new; one usage guide published more than half a century ago wryly cautions writers that “[Gender] is not a substitute for sex (but, then, what is?)” (Bernstein Reference Bernstein1965: 199). Nor is it daft: it is very useful to be able to talk about sociocultural gender as a phenomenon connected to, but distinct from, both grammatical gender and biological sex (McConnell-Ginet Reference McConnell-Ginet and Corbett2014).

More central to the premise of the book are the preferences Dixon expresses in chapter 10, “An ideal language”. Here, he identifies some linguistic features as basic necessities, others as desirable but inessential “luxuries”, and still others as undesirable. For example, the “minimal acceptable” personal pronoun system has a two-way number contrast, but a three-way singular–dual–plural system is “ideal”, and larger sets of number contrasts “are, of course, useful but fall within the basket of luxuries” (p. 225). Homonymy is bad (pp. 216–217); reduplication is good (pp. 220–222); the ideal syllable template is (C)V(C) (p. 224). Some of these judgements have an obvious functional basis (e.g., homonymy causes ambiguity), but many of them are given without explanation (why is three the optimal number of numbers?). Dixon acknowledges that the features in this chapter “simply reflect my opinion” (p. 215), and he invites readers to revise the list to suit their own preferences and priorities. This is all very inclusive and democratic, but what happened to the notion that evaluation is part of science? The scientific enterprise of descriptive linguistics can furnish the data to allow someone to choose a language that suits their preferences, but there is no scientific basis for saying that one language is better than another.

The reader may be either relieved or disappointed to discover that Dixon does not in fact claim that any specific language is better than any other. The very brief chapter 11, “Facing up to the question”, does not face up to the question. Instead, it delegates it: “It is up to you, the reader, to decide” (p. 245). The reader is invited to make their own version of the list of desiderata in chapter 10, weight them according to taste, and then use them to assign scores to a pair of languages and see which one comes out ahead. “This book is”, Dixon says, “a hypothesis awaiting confirmation” (p. 246). But it is not a testable hypothesis in any scientific sense; Dixon merely expresses a vague expectation that a language that scores higher than another on the reader's version of the criteria in chapter 10 will also be better at “[p]resumably some – preferably most, probably not all” of the functions described in chapters 1 and 9 (p. 246). Having put forward the “dangerous” proposition that “it may turn out that one language can be shown to be slightly (never more than that) superior” (p. 1), Dixon ultimately neither affirms nor refutes it, nor does he offer a credible method for testing it.

Who should read this book? Linguists would be better off reading Dixon's scholarly descriptive publications. Non-linguists may learn many interesting facts about grammatical variation, but would be well advised not to take the book's framing premise too seriously. Perhaps the best audience for it would be constructors of artificial languages, who might take its list of features as a source of inspiration for their own creative efforts, rather than as a basis for assigning scores to natural languages; on the other hand, this audience is probably better served by Rosenfelder (Reference Rosenfelder2010) and Petersen (Reference Petersen2015). The most positive thing I can say about this book is that it may, as Dixon intends, inspire debate; I am less confident that the discussion will be a constructive one.