One type of descriptive approach is called structural analysis and its main concern is to investigate the distribution of forms in a language. The method involves the use of “test-frames” that can be sentences with empty slots in them. For example:

There are a lot of forms that can fit into these slots to produce good grammatical sentences of English (e.g. car, child, donkey, dog, radio). As a result, we can propose that because all these forms fit in the same test-frame, they are likely to be examples of the same grammatical category. The label we give to this grammatical category is, of course, “noun.”

However, there are many forms that do not fit those test-frames. Examples would be Cathy, someone, the dog, a car, and many others. (That is, we wouldn’t say *The Cathy … or *The the dog … here.) For these forms, we require different test-frames, which could look like this:

Among the other forms that comfortably fit these test-frames are it, the big dog, an old car, Ani Difranco, the professor with the Scottish accent, and many more. Once again, we can suggest that these forms are likely to be examples of the same grammatical category. The common label for this category is “noun phrase.”

Observing that it fits in this second set of test-frames, and not in the first set (*The it makes a lot of noise), allows us to improve on the older, Latin-influenced, analysis of pronouns in English. In the older analysis, pronouns were described as “words used in place of nouns.” We can now see that it is more accurate to say that pronouns are used in place of noun phrases (not just nouns). By developing a set of test-frames of this type and discovering which forms fit the slots in the test-frames, we can produce a description of (at least some) aspects of the sentence structures of a language.