Pinker is as much of a twit as his hair suggests: The Language Instinct is a miserable pile of unsupported and unsupportable conclusions, straw man attacks, hypocrisy leap-frogging into doublethink, shoddy reasoning, knee-jerk contrarianism, indeliberate obtusity, and gut-feeling argumentation. Pinker tries to synthesize the ideas of people smarter than he is (Chomsky, mostly), and many of these are perfectly fine the way they were originally formulated; they no longer are after Pinker is through defending them, because he understands neither the ideas nor any arguments in their favour.
As an example, do you remember the argument Dawkins once made, possibly in The Blind Watchmaker, about how very small changes can accumulate and turn into very significant results in a relatively small amount of time? He used a hypothetical population of mice that grew in average size by one percent every generation, and showed that within a few thousand generations — a mere couple of millennia! — those mice would be the size of elephants. Pinker tries to use this same story to show that small *selection pressures* can have significant results quickly:

> ``Imagine a mouse that was subject to a miniscule selection pressure for increased size—say, one percent reproductive advantage for offspring that were one percent bigger. Some arithmetic shows that the mouse's descendants would evolve to the size of an elephant in a few thousand generations, an evolutionary eyeblink.''

Which is obviously fractally incoherent, as he would have realised if he'd understood Dawkins's argument instead of just trying to repeat it to try to get a good review by him for the back cover (which he got), or even just tried to do said arithmetic. The sad part is that what he set out to argue is actually true; his bungled argument just undermined it.
This is par for the course (there are more egregious examples, but this one stuck because it's Dawkins), and Pinker repeatedly fucks over his main thesis — that language is instinctive, which was as uncontroversial in 1994, when the book was written, as it is now — in the same way. Worse than that, though, he then tries to pretend that arguments in favour of this thesis mean that language is *nothing but* instinctive and, incidentally, uniquely human, and that ``therefore'' everything from linguistic prescriptivism to animal language to Sapir-Whorf to being interested in etymology is completely and utterly wrong-headed and obviously moronic, which he tries to back up by attacking caricatures of these things or, not infrequently, the character of the people involved.

Anyway, other people have apparently done thorough jobs of taking apart The Language Instict, so I won't waste any more time on it. Not everything he says is wrong (stopped clocks and all that), and it's a lot like Penrose's The Emperor's New Mind in that some of the digressions are interesting enough, but the signal-to-noise ratio is so pathetically low that the book as a whole isn't worth your time. Read Chomsky instead; Pinker would claim he's saying the same things he is anyway, which demonstrates just how confused he is.
Maybe it's just because TLI is his first book. Elsewhere in my to-read stack is his most recent one, The Stuff of Thought; we'll see how it compares. I'm not holding my breath. Ultimately, the problem isn't that he was new to writing, because he wasn't; it's that he's a psychologist, and not a real scientist.