The Fall and Rise of Universals on Relativization

Abstract

This article discusses linguistic universals concerning relative clause constructions, which are relatively well-studied, both by typologists and theoreticians. It turns out that several universal statements formulated in the paste .g., in Downing's (1978) seminal work-must be weakened to tendencies or less on the basis of present knowledge. Following Odden (2003), statistical universals are rejected for the reason that cross-linguistic statistics is inherently unreliable, and may have nonlinguistic causes. However, some absolute universals and universal implications concerning relativization still stand; moreover, some interesting new ones can be formulated. If these universals can be maintained, they constitute (indirect) hypotheses concerning the human language faculty, which need to find an explanation within a specific linguistic model.