In the dual-processing model, irregular past-tense forms are listed lexically but regular ones are not (except in special circumstances; see Pinker & Prince [Pinker, S., & Prince, A., 1994, Regular and irregular morphology and the psychological status of rules of grammar, in Lima, S. D., Corrigan, R., L., & Iverson, G. K., eds., “The reality of linguistic rules”, John Benjamins, Amsterdam, pp. 353-388]). The regular/irregular distinction, then, corresponds to an important structural distinction that is part of the innate architecture of the language acquisition device: this model has a rules component that is separate from the lexicon. [...] Another difference between the dual-processing model and the network model concerns the treatment of irregulars. The dual-processing model is said to handle these in associative networks, but the nature of the generalisations made about them is not specified.
- Bybee (2004), Pag. 47-48