It is very difficult to make a compelling case in favor of natural selection having targeted a particular trait. Of course, it is easy to spin a narrative that implicates natural selection, but biologists like Richard Lewontin, who has spent his life constructing compelling arguments (and deconstructing just-so stories), have repeatedly pointed out that it takes serious effort to catch natural selection in flagranti, so much so that in a famous essay (Lewontin Reference Lewontin, Scarborough and Sternberg1998) – curiously absent from Progovac’s book’s reference list – in the case of human cognition, one should give up trying. It’s just too difficult, with the little we can get from the fossil record.

In my opinion Progovac’s Evolutionary Syntax is an excellent illustration of what Lewontin said we should not do. I had high hopes when I took the book off the shelf. The blurb described the book as ‘a monograph [that] purports to stimulate interdisciplinary projects on language evolution’. The one-page preface indeed pressed all the right buttons, promising a high level of interdisciplinarity. Never mind the ‘details’, because ‘doing an interdisciplinary study of this kind inevitably leads to some loss of detail with each particular field, but my assessment is that any such loss is more than compensated for by the synergy among these fields, yielding insights that would never be possible without this kind of approach’ (preface). Unfortunately, the rest of the book did not live up to these expectations. To begin with, at the level of form, the book is extremely repetitive; footnotes in one chapter are repeated in the main text of the next chapter, to be taken on again in the conclusion and appendix. At the level of substance, here too I found the book pretty thin: by the end, one gets the feeling that we could have got the whole picture in a 30-page article.

The structure of the book is straightforward. Progrovac argues that syntax began with paratactic constructions consisting of two elements (reminiscent of intransitive small clauses in modern languages) (Chapter 2 and mainly Chapter 3), then things became more complex: a proto-coordination stage (Chapter 4), and, ultimately, a subordination stage. Progovac outlines scenarios that make each step advantageous, especially if looked at from the perspective of sexual/natural selection (and in so doing, tries to account for grammatical constraints like islands, in Chapter 5). Throughout the book, Progovac provides rich illustrations from modern languages that exemplify ‘proto-linguistic’ constructions (see especially Chapter 6 on compounds). Since she relies on minimalist concepts, Progovac favours the argument that her proposal demonstrates the possibility of a gradualistic model of language evolution even under the umbrella of the minimalist program.

Though she repeats this last point with regularity (‘One hypothesis that is not compatible with the findings [sic] in this monograph is that syntax emerged in all its complexity abruptly, as one single (minor) mutation/event.’ (208)), the point itself is not new. Clark (Reference Clark2013) succinctly, but compellingly, showed that gradualistic and saltationist scenarios are compatible with many theoretical frameworks, despite the rhetoric used by advocates of these frameworks. Progrovac is able to defend a gradualistic account because she makes certain theoretical assumptions, such as viewing the operation Move as distinct from Merge (the minimalist idea that movement is ‘internal merge’): ‘I do not consider move to be subsumable under merge’ (87 fn. 2). These assumptions allow her to set out an account of syntactic evolution that is consistent with those assumptions, but the theoretical choices remain poorly motivated. As she writes on page 193, hers is ‘a possible scenario’ – one among many – and this is precisely the main weakness of the book. Things could have been that way, but things could have been very different, and we have no way to tell.

Inspired by the notion of ‘living fossils’ in language, especially Jackendoff’s (Reference Jackendoff2002) interpretation of the term, Progovac analyzes data from currently spoken languages to argue that ‘proto-sentences’ had a certain form, but while reading the book I often asked myself what one can reasonably deduce about early evolutionary steps from current grammatical constructions. Too often, Progovac resorts to metaphors, such as ‘small clauses being transformed into sentences like the development of the heart’ (8 fn. 16), or ‘pidginization [is] comparable to the return to a simpler, unicellular mode of existence’ (xx). She also suggests that ‘Metaphorically speaking the small clause grammar can be seen as orange spots still lurking through the layers of the more recent, dominant black coloration of sentential/TP speech’ (55 fn. 29). Ultimately, as Progovac herself points out, grammatical fossils are just a metaphor: ‘what I hypothesize to be living fossils of language evolution are not identical to, but are closely comparable to what is considered to be[,] living fossils in biology’ (33 fn. 1). I have nothing against metaphors, as long as they are accompanied by solid arguments. On their own, they are far less effective.

Progovac argues for the view that single-word utterances, or small clauses, or intransitive clauses are ‘primitive’ structures, but I doubt that this is correct. Numerous arguments have been made for decades in favor of rich structures inside words and small clauses (e.g. Starke Reference Starke, Cardinaletti and Guasti1995), as well as the deepest layers of verb phrases (e.g. Ramchand Reference Ramchand2011). It would have been beneficial if Progovac had taken the time to address these arguments in the book.

On page 185, Progovac notes that ‘a reviewer suggests that the saltationist view mentioned above does not necessarily exclude the possibility that vP, TP, and CP emerged later through grammaticalization processes, and that what emerged as a single mutation were the ‘‘design features’’ of language ... However, as I argued, these design features of syntax cannot be there without the functional categories in questions. ... the functional projections/features ... are the postulates of syntax on which all the other postulates rest’. Progovac is right, but only in so far as we stick to ‘lexicocentric’ models (see Boeckx Reference Boeckx2014). Such models cannot be assumed. They need to be argued for.

However, my biggest disappointment lies in the enormous gap between the promises of interdisciplinarity in the book’s preface and what Progovac actually does when she touches on other disciplines. She often writes that she (will) provide(s) ‘existing or potential corroborating evidence and testing grounds, from language acquisition, agrammatism, genetics, and neuroscience’ (8), and states on several occasions that ‘recent computational and brain-imaging work is consistent with the notions explored in this book’ (e.g. 127). The problem is that given the level of generality at which the argument is formulated, it is not too surprising that it is consistent with many possibilities. As she writes on another occasion, ‘the stages broadly outlined in this monograph are also consistent with the less likely possibility that ...’ (202). Being compatible or consistent with many possibilities is not a theoretical virtue.

Consider, the following passage, taken from page 21 (repeated on page 54); ‘It is also significant that the framework explored here can serve as a point of contact, an intermediary, between the fields of neuro-linguistics and genetics. To take one example, some recent experiments suggest that a specifically human FOXP2 mutation is responsible for increased synaptic plasticity, establishing better connectivity among the neurons in the brain [references omitted – CB]. If better synaptic plasticity is what facilitates the processing of more complex syntax, then one can hypothesize that the pressures to evolve the capacity for a more complex syntax could have contributed to the spread of the human mutation of e.g. the FOXP2 gene. Even at this preliminary level, one can appreciate the potential for synergy among the fields which can shed light on the evolution of human language: syntactic theory, neuroscience, and genetics’.

Geneticists like Simon Fisher, who spent considerable time working out the details of FOXP2 and trying to formulate linking hypotheses from genes to cognition, would not be moved by this passage. Are we sure that ‘better synaptic plasticity is what facilitates the processing of more complex syntax’? Progovac writes as if it is easy to make this case, but geneticists and neuroscientists know that it is far from being so. I, for one, fail to ‘appreciate the potential for synergy among the fields’. It’s just too vague to be testable. Progovac insists on this point throughout the book: ‘it is the evolutionary considerations like this that can provide the point of contact’ (21). In a footnote (fn. 31) on the same page, she writes ‘The exact contribution of FOXP2 and the other genes remains to be determined, as rightly pointed out by a reviewer, but I do not think this can be determined without some concrete input from linguists, and without some specific, testable hypotheses about language evolution. If these genes were even partly selected for some specific language abilities then, without hypothesizing what these specific abilities might have been, we will not be able to find out’. Anyone involved in research trying to link genes and cognition will find such statements extremely naive.

The same could be said about Progovac’s forays into neuroscience: ‘while the processing of full hierarchical structures is expected to show clear lateralization in the left hemisphere, with extensive activation of some specific Broca’s areas, the proto-structures, including root small clauses and VN compounds, are expected to show less lateralization, and less involvement of Broca’s area, but more reliance on both hemisphere, as well as, possibly, more reliance on subcortical structures of the brain’ (32; also 127, 170). Neurolinguists will likely smile at this passage, as it appears to rely too uncritically on the outdated Broca–Wernicke model. It’s now clear that at virtually all levels of complexity language processing relies on both hemispheres, and on subcortical structures. Moreover, these subcortical structures have co-evolved with the cortex, and it is just plain wrong to treat them as monolithic or more ancient (215).

It seems to me that Progovac vastly underestimates the difficulty of fruitful interdisciplinarity. On page 212 (fn. 1), she writes that ‘According to Poeppel and Embick (Reference Poeppel, Embick and Cutler2005) ... current neuro-linguistic research in the domain of syntax presents a case of cross-sterilization, rather than cross-fertilization. This is because, according to them, no meaningful correlates have been found, nor are expected to be found, between biological units of neuroscience (e.g. neurons, dendrites, axons) and the formal syntactic postulates such as Move, etc. The proposal of this book ... is that the missing piece needed to bridge the two vastly different fields is the consideration of the evolution of syntax. This appendix elaborates on that idea’. I side with Poeppel and Embick. I very much doubt we will be able to isolate ‘neuro-biological correlates of, for example, TP layering and VP shelling/transitivity’ (211).

Doing genuine interdisciplinary work is hard, bridging the gap between mind and brain is hard, connecting genes and cognition is hard, making a case for sexual/natural selection is hard. All of that is really, really hard. It doesn’t do the burgeoning field of evolutionary linguistics any good to trivialize these hard questions. We are at a stage where it is becoming possible to tackle these questions productively, combining various techniques from many disciplines. To do so, we should refrain from spinning just-so stories, and focus on articulating how-so stories.