The classification of words and other elements must be based on structural, i.e. grammatical, features. Meaning-based classification is not very helpful for two reasons: a) definitions of parts of speech based on meaning tend to be ambiguous; b) such definitions do not conform to the grammatical features of a language (Gleason: 115ff.; Fries: 65-86, 87f., 202ff.). If classification is derived from structure, however, there are two sets of criteria which may be used, and these do not always coincide: 1) One set of criteria may be derived from the words themselves, i.e. from morphological criteria; 2) Another may be derived from the ways a word may be used in a sentence. The ideal would be to achieve a classification that would integrate these two sets as fully as possible.