Processing (un-)predictable word stress: ERP evidence from Turkish

Abstract

This paper investigates the way the predictability of prosodic patterns in a particular language influences the processing of stress information by native speakers of that language. We extend previous findings where speakers of languages with predictable stress had difficulties to process and represent stress information when confronted with a language with distinctive stress and investigate how the co-existence of a predictable stress pattern and exceptions to that regularity within a single language influences prosodic processing. The stress system of Turkish constitutes an instructive test case since it employs predictable stress on the final syllable of a prosodic word (e.g., mısır “corn”) and some exceptional nonfinal stress (e.g., mısır “Egypt”). Results from an event-related potential (ERP) study on stress violations in Turkish trisyllabic words showed asymmetrical ERP responses for different stress violations: Stress violations with final stress produced an N400 effect whereas violations with nonfinal stress produced a P300 effect. The application of the predictable pattern to words with lexical stress led to lexical costs and the application of exceptional stress to words with default stress to effects reflecting the evaluation of this pattern. Although final stress constitutes no alternative pattern for words with exceptional stress, participants have difficulties to judge this pattern as incorrect. In contrast, exceptional stress patterns are detected easily when applied incorrectly to words that normally receive final stress. These findings demonstrate nicely the co-existence of two phonological processing routines in Turkish speakers. Furthermore, the variability of stress patterns does not affect prosodic processing in general but instead leads to differential effects in stress perception. We conclude that stress predictability does not homogenously result in the so-called “stress deafness” effects in stress processing, but that it rather emerges only for the default stress pattern.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 42 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.