Steven Pinker's best-known book has some wonderful chapters, some so-so chapters, and a few that damage the credibility of the rest. Chapter 6 on how the sounds of spoken languages are formed is itself worth the price of the book. Chapter 2 on the grammatic differences between languages is fascinating. Chapters 4, 5, 7 and 8, which talk about grammar and its role in determining meaning, are well-meaning but become repetitive and obvious. When talking about Artificial Intelligence he is ill-informed and unaccountably pessimistic about future advances in the field. In Chapters 3 and 9 he proposes a "language instinct" and in chapter 10 a "grammar gene," but both hew to discredited Chompskian models and don't even try to establish any mechanism. In chapter 11 he dismisses the whole field of non-human communication in toto, citing such Christian apologists as Herbert Terrace. Instead he sets up a series of straw men, claiming that because that apes cannot master advanced grammar in human languages (undisputed), somehow this makes their mental processes unworthy of study. This contradicts his earlier claim, in chapter 3, that mental processes can exist quite independently of grammar and language. He apparently never even considers that non-human grammar may differ from ours. Worse, he doesn't even mention non-primate language research! 12 is a vitriolic dismissal of all his critics, and 13 falls into the common trap of describing evolution as "wanting to build" this or that, a common convention for which he could be excused if this were his only failing.
Throughout, Pinker maintains a breezy, readable tone full of pop-culture references - which unfortunately becomes infuriating when it's obvious he doesn't know what he's talking about.