In linguistics, productivity is the degree to which native speakers use a particular grammatical process, especially in word formation. Since use to produce novel (new, non-established) structures is the clearest proof of usage of a grammatical process, the evidence most often appealed to as establishing productivity is the appearance of novel forms of the type the process leads one to expect, and many people would limit the definition offered above to exclude use of a grammatical process that does not result in a novel structure. Thus in practice, and, for many, in theory, productivity is the degree to which native speakers use a particular grammatical process for the formation of novel structures. A productive grammatical process defines an open class, one which admits new words or forms. Non-productive grammatical processes may be seen as operative within closed classes, but only previously formed and learned structures show the results of those processes.

In standard English, the formation of preterite and past participle forms of verbs by means of ablaut (for example, singsangsung) is no longer considered productive. Newly coined verbs in English overwhelmingly use the ending -ed for the past tense and past participle (for example, spammed, e-mailed). There are more recent ablaut forms, however—e.g. snuck as the preterite and participial form of sneak, or dove as the preterite of dive. Such cases are standardly said to have been created by analogy instead of by productive application of ablaut rules, though this may to some extent be begging the question. Similarly, the only clearly productive plural ending is -(e)s; it is found on the vast majority of English count nouns and is the almost exclusive means used to form the plurals of neologisms, such as FAQs and Muggles. The ending -en, on the other hand, is (at least relatively speaking) no longer productive, being found only in oxen, children, and the now-rare brethren. In the hacker sociolect, however, the plural -en became at least ephemerally productive for words ending with /-ks/ (on the analogy of ox:oxen), as illustrated by the plurals VAXen, unixen and emacsen, for example.

Productivity is, as stated above and implied in the examples already discussed, a matter of degree, and there are a number of areas in which that may be shown to be true. As the example of -en becoming productive shows, what has apparently been non-productive for many decades or even centuries may suddenly come to some degree of productive life, and it may do so in certain dialects or sociolects while not in others, or in certain parts of the vocabulary but not others. Some patterns are only very rarely productive, others may be used by a typical native speaker several times a year or month, whereas others (especially syntactic processes) may be used productively dozens or hundreds of times in a typical day. It is not untypical for more than one pattern with similar functions to be comparably productive, to the point that a speaker can be in a quandary as to which form to use —e.g., would it be better to say that a taste or color like that of raisins is raisinish, raisiny, raisinlike, or even raisinly? (All four can be found on the Internet.)