M.A. Theses in Linguisticsat the University of North Dakota (Abstracts)
Dyrud, Lars 2001
Hindi-Urdu: Stress Accent or Non-stress Accent?
The Hindi-Urdu stress system has been a problematic topic for over a century. Hayes (1995) says, "...this topic has its empirically dismaying aspects: the published descriptions almost all disagree with one another, and seldom mention the disagreement."
The literature today divides languages between 'stress accent' languages and 'non-stress accent' languages. According to Beckman (1986), 'stress accent' languages are those that use phonetic attributes other than pitch to indicate a prominent syllable, while 'non-stress accent' languages are those that use only pitch to mark a prominent syllable.
In the past, several quantitative studies of phonetic correlates of stress in Hindi-Urdu have been carried out. The question that this study seeks to answer is whether there is evidence for stress accent (in Beckman's terminology) in Hindi-Urdu; that is, is word stress in Hindi-Urdu reflected in one or more acoustic properties, independent from the pitch fluctuations that are due to intonation? Most previous studies have not directly addressed this question.
The acoustic properties of prominent and non-prominent syllables were compared, controlling for the effects of intonation (especially the presence vs. absence of prominence-lending pitch movements), by recording words in both [+focus] and [-focus] contexts. Results showed a significant effect of stress on pitch as well as on duration. However, it was also found that focus interacts with stress: for two minimal pairs in the data, stress showed a significant effect on both pitch and duration in the [+focus] condition, but on neither pitch nor duration in the [-focus] condition. This result suggests that duration does not function independently from pitch as an acoustic correlate of stress in Hindi-Urdu, and that the language is more accurately classified as having non-stress accent instead of stress accent.
In a pilot perception experiment, listeners did not perform better than chance in identifying members of minimal stress pairs spoken in a [-focus] context, while they did perform better than chance for words spoken in a [+focus] context. This result corroborates the findings of the production experiments, viz. that for the words studied, acoustic correlates of stress disappear in the [-focus] context.
The author may be contacted by email at "Lars_Dyrud@sil.org".
Download thesis from http://arts-sciences.und.edu/summer-institute-of-linguistics/theses/_files/docs/2001-dyrud-lars.pdf