In , we saw that the inconsistency between the data and the hypotheses as well as that between hypotheses can be resolved by paraconsistent tools.raises the problem of how to evaluate the paraconsistent treatment of inconsistency.will be devoted to a case study exemplifying the emergence and the usefulness of paraconsistency in generative syntax. In , we will discuss another two case studies that highlight the limits of paraconsistency. Finally, in , we will draw the conclusions from the case studies that evaluate the use of paraconsistency in linguistic theorising.

The Introduction to this chapter raises the problem of how to distinguish between tolerable and intolerable inconsistencies in linguistic theorising. In , Rescher and Brandom’s () logic will be introduced, which differentiates between paraconsistency, allowing for a special kind of inconsistency without ex contradictione quodlibet, and strong inconsistency, which is exposed to logical chaos. The workability of this distinction in theoretical linguistics will be exemplified by two case studies in Sections 4.3 and 4.4, respectively. They will illustrate, first, that tolerable inconsistency can be represented as paraconsistency, while the unavailability of a paraconsistent resolution may leave us with an intolerable strong inconsistency. Second, they will also show that in the cases examined the emergence of inconsistencies is closely related to the data handling techniques applied.summarises the solution to the problem raised in this chapter and highlights that there is a need to re-evaluate the way data are treated in theoretical linguistics; but tackling the question of how this should be done will be postponed to later chapters.