For the purposes of the present paper, connectionist models and the network model make essentially the same claim: type frequency is a major determinant of productivity. [...] In the network model, it is the frequency of the derived word itself that is significant and is represented as lexical strenght, not a mapping between two forms. In fact, the higher the frequency of the derived form, the weaker the mapping between it and the basic form.
- Bybee (2004), a pag.48

The model of Bybee [Bybee, J. L., 1985, "Morphology: A study of the relation between meaning and form", John Benjamins, Philadelphia, PA; 1988, Morphology as lexical organization, in Hammond, M. & Noonan, M., eds., “Theoretical Morphology”, Academic Press, San Diego, CA] was developed to account for crosslinguistic, diachronic and acquisition patterns in complex morphological systems. The basic proposal is that morphological properties of words, paradigms and morphological patterns once described as rules emerge from associations made among related words in lexical representation. A major difference between this model, which I will call the "network" model, and structuralist models containing rules, is that actual usage in terms of both type and token frequency plays an important role in establishing and mantaining representation.
- Bybee (2004), a pag.44