Overgeneration is a concept related to generative grammar and computational linguistics. These branches of linguistics attempt to set up the grammar of a language based on a series of rules. Linguists then generate words and sentences based on the rules they have set for a particular language. Ideally, the words and sentences created will be perfectly grammatical, but that is not always the case.

Sometimes the grammatical rules overgenerate grammar. This means that words and sentences are created that seem grammatical according to the rules but are not recognized as such by a native speaker of the language. Let's look at a couple of examples.

Let's say that the rules of a language allow for a noun to have a determiner (like an indefinite article or a definite article). In English, this gives us noun phrases like “the book” or “the books” or “a book.” However, we would not say “a books.” Notice the indefinite article combined with the plural noun. The noun phrase fits the rule that nouns are allowed to have determiners, but native speakers know that “a books” is not grammatical. This is a case of overgeneration.

The rules of a language might also allow a prefix to negate a verb. For example, we do something, but we can also undo something. The prefix “un-” negates the verb “do.” But if we said something like “I'm going to untip over that lamp,” we know right away that we have left the realm of conventional grammar. We mean for “un-” to negate “tip,” and according to the prefix rule, that should work. Yet it doesn't. It is another case of overgeneration.