Reinforcement in inflection classes: Two cues may be better than one

Abstract

The paper presents examples of inflection class reinforcement, where an inflection class becomes more different from its neighbours than it was. This is a manifestation of a diachronic tendency for more overt marking, similar to Kuryłowicz’ first law for analogy. It is also a manifestation of redundancy, which is characteristic of inflection. If inflection classes do not serve any purpose (as claimed by, for example, Wurzel 1986 ), it would seem strange that they should be strengthened. So, classes are not mere junk; reinforcement of class distinctions (polarisation) testifies to the autonomy of morphology (cf., for example, Aronoff 1994 , Carstairs-McCarthy 2010 , Maiden 2011 ). A change in Swedish conjugation is argued to be reinforcement. We consider a case where declension wins over gender in Norwegian, the opposite of what has been claimed to be the norm for that language. That case is also reinforcement. I suggest that reinforcement may be thought of as a kind of simplification. That also fits with the geographical distribution of a specific change. The paper also presents some evidence that a fairly ‘concrete’ definition of inflection classes can sometimes be useful. Inflection classes often latch on to extra-morphological properties in the cases examined, as claimed by Natural Morphology ( Wurzel 1984 ). Yet it is problematic for Wurzel that inflection classes (mere ‘ballast’, in his view, Wurzel 1986 : 76) should be strengthened. The finding is, however, expected if ‘[…] autonomously morphological structure […] can be a dynamic, self-reinforcing factor in morphological change’ ( Maiden 2005 : 168).