In which the author sets out to refute the "nativist" view put forward by Chomsky, Pinker et al that language is an innate ability, which is biologically built in to the human brain.

Having read parts of Pinker's "The Language Instinct", and seen it and Chomsky referenced in many places, I had assumed that there was little ongoing debate on this subject. At least, this is the first book I've read I think seriously questions the nativist view. But Sampson puts forward a very persuasive set of arguments to show that Chomsky and his followers are by no means beyond criticism. As well as advancing his own view (that language is a skill which everyone must learn from scratch), he approaches the issue a bit like I think Feynman would, by checking out the evidence put forward by the nativists. He claims that most people have not actually read and checked out what Chomsky says (partly because it is difficult to read) because if they did they'd see flaws in the argument, and find that many of Chomsky's copious references (although they look impressive) do not support his position. And he does seem to be able to come up with an awful lot of examples which contradict what you'd expect to find if you accept the nativist point of view.

Sampson praises Pinker for his ability to write clearly, and finds no fault with much of what Pinker says in "The Language Instinct", but on the subject of nativism, he claims that Pinker too is arguing from a rather shaky position.

The book is very easy to read (apart from the last section where he gets into a discussion about human creativity, Popper and dualism which he says he feels he has to include to pre-empt criticism that it was something he hadn't considered) and is frequently quite funny. For example he says [p133] that Pinker describes English as having a "fixed rule" which will prohibit certain types of sentence. To check this out, he looks through a corpus for a counter-example and finds one in the second sentence of the corpus. I've just checked the Pinker book, and I don't think Pinker actually says "fixed rule" but that does seem to be the general implication of what he's saying. At any rate, the sentence Sampson quotes does seem a convincing refutation of Pinker's point.

Some other things I noted as being interesting:

• Chomsky says that "a person might go through much or all of his life without ever having been exposed..." to a particular type of grammatical construct. Sampson gives several examples of sentences that use this construct [p42]
• One nativist argument is that humans are able to distinguish between "ta" and "da" sounds (which are acoustically very similar): "why else would we hear these sounds in a fashion at odds with their physical reality, if not because linguistic considerations are built into our perception?". Sampson then quotes evidence that chinchillas have been shown to be able to make the same distinction, and no-one is claiming they have innate language ability [p58]
• sign languages have constructs that don't have analogues in spoken language. So these at least are presumably not innate [p78]
• Pinker says that consonant cluster "sr" (as in "sram") is "illegal" in English. Sampson provides two examples of names containing this cluster [p83]
• The example of "Chelsea" given by Pinker has very little detail and when Sampson looked for more he found hardly any, and questions some of Pinker's statements about the case [p89]
• The "KE" family mentioned by Pinker in reference to a putative faulty language gene had a lot more problems than the ones Pinker mentions: he picks out specific issue with language which strengthens his case, but neglects to mention other data from research on the family which weakens it [p94]
• Pinker says the plural of any compound ending in "foot" will be "~foots" (e.g. "flatfoot" -> "flatfoots"). But when Sampson searches the corpus, he found no "~foots" words but quite a few "~feet" (e.g. Blackfeet, pinkfeet). [p98]
• While nativists see the tree structure of grammar as evidence of an innate ability, Sampson says that a tree structure is statistically the most likely to appear in any complex system, and provides quite a lot of evidence to support this [p110]
• "The writings of the nativists contain airy hints about the wealth of language universals they could cite in evidence, if space permitted. If one tries to pursue those hints, though, the substance is often elusive" [p121]
• Talking of a nativist who says that the fact that different languages have identical verb arguments, Sampson concludes "What Bickerton is telling us is that we know, effortlessly and unerringly, that words which mean the same are words which mean the same" [p132]

Geoffrey Sampson has a website which looks like it has a few useful pointers to linguistics related stuff

Completed : 23-Feb-2004