More than half the work is devoted to exploring aspects of syntax and grammar — words and phrases, reference, ellipsis, recursion, and plurals — as they are acquired by children. The approach here is driven by Chomskyian ideas of Universal Grammar, but neither assumes nor attempts to introduce any real theory. It is focused around do-it-yourself-at-home "explorations":
It's important to realize that children implicitly have a distinction between propositions and observations (where, informally, "noting a property" = an observation). We can get a sense of a child's early knowledge with this simple situation. Set an empty bowl in front of a child and say either, "Is this soup?" (asking the child to respond to a propositional claim) or "Is there soup?" (seeking a response to an observation).
Will the child answer the first question with "No, it is a bowl" and the second with "No, it is empty"? To give these answers, the child must grasp the deep difference between an identifying proposition (It's a bowl) and noting a property (It's empty).
The idea with these explorations is very much not to experiment on children, or to hothouse them, or even to benchmark them (the ages of children are given in some example conversations, but otherwise only broad hints such as "try this from age three"). The goal is to help the reader learn something about language, while perhaps providing some entertainment for the children involved.
Moving onto language diversity, Roeper touches on the role of language variation in social distinctions. Two themes are then treated in more detail. He introduces the idea that we are all bilingual to some extent, that English incorporates bits of grammar ("organizing principles") which are more typical of other languages:
one day I caught myself saying to a child under a blanket, "Can I put my feet under, too?," and the situation was so well defined that no object was required, despite the transitive preposition. The question did not feel ungrammatical, like Can I look for too? Here we see the language momentarily departing from its grammar — or, as I suggest, we see an entirely different grammatical gestalt emerge: a tiny piece of "Chinese."
Something similar can be seen with African-American English, which is moving towards an "event structure" grammar with, among other things, "a verbal system that can convey habituality and stativity".
Finally, Roeper tackles a range of more philosophical topics. He asks: "How can we be careful not to abstract too quickly, to make sure that a claim about one module of mind is not instantly applied to the whole person?" He considers in detail the problems of using linguistic evidence to decide whether or not children have a concept of false belief. And he canvasses some of the broader implications of a modular mind, attempting to find in it an argument for humanism and human dignity.
I have some disagreements with Roeper, especially in his more philosophical ventures and his seeming lack of interest in neurobiology, but it was his strange appeals to mathematics which jarred most. Apparently "emotions are infinite" because individuals may experience up to 100,000 (over a lifetime?) and their context is different. There are some odd statements about formulae: "all individuals carry unique formulas that guide their actions" and "some formula ... allows us to form instant opinions about people". (Human decision-making can be modelled by formulae and, at a different level, mathematically trained humans can understand and manipulate them, but they are not themselves parts of a mind.) And just thinking about the confusions in a sentence like "They are features of Universal Grammar that enable the child to figure out language like a complex quadratic equation" makes my head hurt.
That aside, The Prism of Grammar is a lot of fun. Though it has something even for those who know a bit of linguistics, it is really pitched at those with no such background, most likely parents or other child carers whose curiosity has been aroused by watching children.