Analogie v morfologickém vývoji románských jazyků: „essai de méthode“

The article attempts to show how the traditional notion of analogy has recently been revisited and reformulated so as to explain some apparent irregularities of the morphological phenomena. The paper argues that analogical changes like the morphologization of metaphony or the distribution of stem alternants in Romance languages are to a certain extent predictable phenomena. It is claimed, however, that the predictability of such analogical changes (such as the spread of L, U and N patterns in Romance verb morphology) concerns only one aspect of the phenomenon, namely the regularity with which one particular innovation in a given cell of the paradigm affects all the other cells involved in the pattern in question. Yet another aspect, perhaps more important, is the probability that any such alternant (like the substitution of Old Spanish oyo with oigo) will come into being. It is argued that research on analogy such as conducted by Adam Albright might be considered complementary to the current diachronic research in Romance linguistics.