Wednesday, June 10, 2015

The Language Myth - Book Review

Linguistics professor Vyvyan Evans recently published a new book that has at least one group of linguists in a state of frenzy: The Language Myth: Why language is not an instinct. The book's blurb sums up its content:
Some scientists have argued that language is innate, a type of unique human 'instinct' pre-programmed in us from birth. In this book, Vyvyan Evans argues that this received wisdom is, in fact, a myth. Debunking the notion of a language 'instinct', Evans demonstrates that language is related to other animal forms of communication; that languages exhibit staggering diversity; that we learn our mother tongue drawing on general properties and abilities of the human mind, rather than an inborn 'universal' grammar; that language is not autonomous but is closely related to other aspects of our mental lives; and that, ultimately, language and the mind reflect and draw upon the way we interact with others in the world
Evans' grounds his motivation in the idea that there are a variety of false claims about how language works ("myths") deeply rooted in our culture's background knowledge as well explicated in introductory text books. He goes further to claim that these false claims have been pushed by a small number of pre-eminent scholars whose fame and influence have caused these false claims to be taken more seriously than they deserve on their face.

By all rights, I should be a good audience for this book. I was trained as a linguist in a department that was openly hostile to the language instinct doctrine that this book argues against (see my post about that experience). 

The book is organized by two principles. First, each chapter starts by stating one false claim and providing a description of why it was proposed as an explanation of how language works. Second, each chapter then deconstructs the myth into component claims and shoots holes in each one. 

The Good
Evans does a service to the lay audience by pointing out that that deep divisions exist within the filed of linguistics. Too often non-experts assume a technical field is homogeneous and everyone agrees on the basic theories. This is simply not true of linguistics. 

Evans also does a service to his audience by stepping through the logic of refutation. His point-counterpoint style can be detailed at times, but I appreciate a book that doesn't treat its readers like third graders (I'm looking at you Gladwell).

For me, the standout chapter was 5: Is language a distinct module in the mind. This chapter is devoted to neurolinguistics and here Evans is at his sharpest when leading the reader through his point-counterpoint about brain regions and functionality.

The Bad
Evans fails to do justice to the myths he debunks. He was accused of creating straw men (and addresses this somewhat in the introduction), but ultimately I have to agree. Evans does not provide a fair description of arguments like poverty of the stimulus

Evans quickly shows his bias and directly attacks just two people: Noam Chomsky and Steven Pinker (and to a lesser extent, Jerry Fodor). Evans wants to debunk general notions that have crept into the general public's background beliefs about language, but what he really does is rail against two guys. And worse, he often devolves into a detailed point-counterpoint with just one book, Pinkers' 1994 The Language Instinct. Any reader unfamiliar with that book will quickly get drowned by arguments against claims they never encountered. As an exercise, I would recommend Evans re-write this book without a single reference to Chomsky, Pinker, or Fodor. I suspect the result will be a more effective piece of writing. 

Lest some Chomskyean take this review wrongly, let me be clear: I think Chomsky is broadly wrong and Evans is broadly right. But even though I believe Pinker is wrong and Evans is right, I find Pinker a far superior writer and seller of ideas. And that is a serious problem. 

Evans would have been better off throwing away the anti-Chomsky rants and simply write his view of how language works. A book on its own terms. Instead he comes across as your drunk uncle at Christmas who can't stop complaining about how the ref in a high school football game 20 years ago screwed him over with a bad call. This might actually be true, but get over it. 

I feel Evans has taken on too much. Each myth is worth a small book itself to debunk properly. This is partly what leads to the straw man arguments. Efficiency. A non-straw man version of Evans' book would be 3000 pages long and only appeal to the three people in the world who know enough detail about both Chomsky and functionalist theory to properly understand all that detail. So I *get* why Evans chose this style. I just think Pinker is better at it. Ultimately Evans alienates his lay audience by ranting about people they don't know and arguments they are unfamiliar with. 

A detail complaint: He can be disingenuous with citations. On page 110 he uses the wording "the most recent version of Universal Grammar", but turn to the footnote on 264 and he cites publications from 1981 and 1993. In a book published in 2015, citations from 81 and 93 hardly count as recent. See also page 116 where he cites "a more recent study" that was actually published in 2004 (and probably conducted in 2002).

I don't want to be critical of a book that argues a position I align with, but I must be honest. This book just doesn't cut it. 


Nuts and Bolts of Applying Deep Learning (Andrew Ng)

I recently watched Andrew Ng's excellent lecture from 2016 Nuts and Bolts of Applying Deep Learning and took notes. I post them as a he...