Enjoyable, but doesn’t solve the mystery
I have read Thom Scott-Philips’ book with great pleasure, but also with a very critical eye. It is extremely well written— I have read most of it during long train rides and had no difficulties concentrating on it. For someone who is as easily distracted as myself that says quite something. I also felt myself almost persuaded by the arguments, but in the end there are many things with which I disagree.
I do agree that pragmatics and inference are extremely important in language, and that they perhaps deserve more attention than they usually get. However, I also do think that Thom presents too much of a straw man argument. Most serious students of language evolution have been convinced of the importance of pragmatics and inference for a long time (although, given the complexity of the subject they may not have made it the focus of their research). The position that Thom appears to be arguing against is perhaps the syntax-centred, Chomsky-inspired formal view of language. However this position is of decreasing importance in both linguistics and the study of language evolution. I definitely do not agree that the ostensive-inferential model proposed by Thom solves all problems of language evolution.
It may help explain how communication started, but it really doesn’t explain why language is the way it is (Thom’s discussion of combinatorial structure notwithstanding, the ostensive-inferential model does not explain how language became combinatorial), and I was quite baffled to find nothing about acquisition in the book. I also felt extremely uncomfortable with the part where the evidence for cases where there appears to be no theory of mind but there does appear to be language was discussed. I think Thom could have been much more self-critical there and now runs the risk of falling into the pit of confirmation bias. I also think there is much more continuity between the code model of communication and the ostensive inferential model. For instance one can imagine a continuum between a pure code model, with fixed signals and meanings, and more ostensive-inferential models where meanings are progressively less well-defined and need to be acquired from context. Again, the lack of discussion of the role of acquisition is a weakness of the book. I also find that Thom overstates the importance of ostension and inference. A lot of language (and even non-linguistic communication) is extremely stereotyped. Come to think of it, Thom’s favourite example of the raised and tilted coffee cup is pretty conventionalised as well, and actually a good example of the outcome of (culturally evolved) ontogenetic ritualisation. There may be a role for ostension in establishing conventions through cultural evolution, but I am not convinced it plays a very prominent role in everyday language use. In the end, if we have learned anything in the last twenty years of studying language evolution, it is that there is no single explanation for humans having language, and language being the way it is. On the contrary, it seems clear that a number of factors have conspired, aided with a dose of coincidence, to give humans language and other (intelligent) animals not. Inference and ostension have no doubt their role to play, and I think the book makes a good point of arguing it. However, I also think the book is old school language evolution in the one-man-one-idea tradition of Pinker, Bickerton, Dunbar or Mithen, to name just a few. Such theories provide food for thought and may help to attract students to the field (in itself extremely important) but the real progress is made by painstaking and self-critical research of more detailed questions (a point also made by Richard Moore on this site). Of course, Thom has made contributions of this nature as well, and for me his book would have been better if it had reflected this approach to language evolution more.