Research

Ohio State Dissertations in Linguistics (OSDL)

Gayle Ayers (1996)

Nuclear Accent Types and Prominence: Some Psycholinguistic Experiments

Advisor: Mary E. Beckman

Download .pdf file [PDF]

Abstract:

This dissertation explores the prominence of words in simple English sentences produced with declarative intonation patterns. Four components of prominence are addressed: acoustic, structural, perceptual, and informational prominence. The prominence of three accent statuses (nuclear accented, prenuclear accented, and unaccented) and three types of nuclear accent (regular, downstepped, and expanded pitch range) are investigated. Compared to regular nuclear accents, downstepped accents have lower fundamental frequency peaks, and expanded pitch range accents have higher peaks. Sentences had nuclear accent placement in early, medial, or late sentence position ("normal sentence stress"). Three experimental tasks are used to test whether the three nuclear accent types have different degrees of prominence: phoneme monitoring, question-answering, and cross-modal naming.

The accent statuses and nuclear accent types had the predicted differences in acoustic prominence. Nuclear accented was the most prominent, followed by prenuclear accented and then unaccented. Within nuclear accented, expanded pitch range was the most prominent, followed by regular and then downstepped. Accent status had a significant effect on phoneme monitoring speeds in the direction predicted by the acoustic prominence relationships; nuclear accented targets were responded to more quickly than non-nuclear accented targets. However, reaction times to prenuclear accented and postnuclear unaccented targets were indistinguishable, suggesting that the primary perceptual split is between nuclear accented and not nuclear accented. Reaction times to regular and expanded pitch range nuclear accents were indistinguishable, but downstepped nuclear accents were responded to least quickly, suggesting that they are less perceptually prominent than non-downstepped accents. A question-answering experiment investigated the informational prominence of the accent statuses and accent types. Subjects answered faster when the relevant part of the sentence in the sentence- question pair was most informationally prominent and slower when another part of the sentence was most prominent. The differences between nuclear accent types were not significant. Accent status had little effect on the speed of lexical access of target words in the cross-modal naming experiments. The experiments showed that nuclear accent placement and accent type influence sentence processing, and that nuclear accent is not a uniform category in terms of prominence.