A DIACHRONIC ARGUMENT AGAINST THE SPLIT MORPHOLOGY HYPOTHESIS - ANALOGICAL UMLAUT IN GERMAN DIALECTS

This paper presents diachronic evidence from four modern German dialects against Anderson's (1982) formal dichotomy between inflection and derivation, otherwise known as the ‘Split Morphology Hypothesis’ (SMH). It examines the relative productivity of analogical umlaut rules (i.e. the analogical fronting of back vowels in morphologically complex forms leading to the creation of new ‘phonemes’) in a number of inflectional and derivational categories and demonstrates that these rules are either (i) productive in inflection and derivation, (ii) productive in inflection and diminutive nouns but sporadic in derivation or (iii) productive only in diminutive nouns. As a marker of noun plurals, analogical umlaut is lexically conditioned. On the basis of this evidence, it is argued that the productivity of analogical umlaut is determined by semiotically‐based parameters, including the semantic transparency of the relation between base and derivative and the uniformity and transparency of the formal means used to mark morphologically complex forms, all of which interact to determine the overall perceptual saliency of particular alternation types. Copyright © 1994, Wiley Blackwell. All rights reserved