Universal Grammar is made up of a set of rules that apply to most or all natural human languages. Most of these rules come in the form of "if a language has a feature X, it will also have the feature Y." Rules that are widely considered as part of UG include: If a language is head-initial (like English), it will have prepositional phrases; if and only if it is head-final (like Japanese) will it have post-positional phrases.[13] If a language has a word for purple, it will have a word for red.

Speakers proficient in a language know what expressions are acceptable in their language and what expressions are unacceptable. The key puzzle is how speakers should come to know the restrictions of their language, since expressions which violate those restrictions are not present in the input, indicated as such. This absence of negative evidence—that is, absence of evidence that an expression is part of a class of the ungrammatical sentences in one's language—is the core of the poverty of stimulus argument. For example, in English one cannot relate a question word like 'what' to a predicate within a relative clause (1): (1) *What did John meet a man who sold? Such expressions are not available to the language learners, because they are, by hypothesis, ungrammatical for speakers of the local language. Speakers of the local language do not utter such expressions and note that they are unacceptable to language learners. Universal grammar offers a solution to the poverty of the stimulus problem by making certain restrictions universal characteristics of human languages. Language learners are consequently never tempted to generalize in an illicit fashion. (83-92)