Parsing, productivity, and word prosody: the interaction of stress assignment and morphological processing in Vedic Sanskrit

Abstract This study investigates the relationship between quantifiable morphological productivity, frequency, and the position of stress in a language with a head-dependent canonical lexical accent system, Vedic Sanskrit. Necessary background is furnished by reviewing principles of the phonological analysis of stress assignment in lexical accent systems generally, specific aspects of stress assignment in Vedic Sanskrit, and existing investigations into potential frequency effects on stress assignment and prosodic change. A corpus study on six word-formation affixes of Vedic Sanskrit based on a corpus of the oldest materials in the language reveals that three relatively more productive word-formation suffixes exhibit consistent stress patterns, determined by lexical properties of the suffix. Three less productive affixes, argued to be synchronically largely unparseable , show unpredictable stress: some lexemes have historically older stem-edge stress, others show word-initial stress, in agreement with the default stress rule of Vedic. Statistical analysis reveals that lexemes in these categories with higher token frequency are significantly more likely to exhibit the original stem-edge stress. I conclude that this data supports the claim that higher token frequency exercises an inhibitory effect on productive phonological and morphological processes, helping exceptional lexical properties to be preserved diachronically.