City of Buffalo, Estados Unidos
Historical linguists have long been divided in their views about the mechanisms behind paradigm leveling, with many invoking a special mechanism related to a universal preference for paradigm uniformity while others attribute leveling to the same mechanism responsible for other types of analogical change. I argue that although ‘proportional’ analogical innovation plays a major role in paradigm leveling, it cannot account for all cases, and that something akin to the ‘interference’ mechanisms commonly associated with contamination and folk etymology account well for the non-proportional instances. I further show that all of the mechanisms involved in paradigm leveling are also implicated in other types of analogical change, and I argue against the need to posit any universal bias against (stem) allomorphy.