The effects of linguistic factors on analysis of relative fundamental frequency in typical speakers
Date
2013
DOI
Authors
Gattuccio, Caitlin Irene
Version
OA Version
Citation
Abstract
Research has shown that the relative fundamental frequency (RFF) surrounding voiceless obstruents may be used as an acoustic correlate for laryngeal tension. This suggests that RFF could potentially be used as a measurement of vocal hyperfunction, a voice pathology characterized by overexertion and increased tension ofthe laryngeal muscles. Despite this potential relationship between laryngeal tension and RFF, there has been little research into what factors contribute to creating stimuli that effectively produce reliable and valid RFF data for subsequent analyses. This study sought to investigate the impacts that linguistic factors have on RFF elicitation in healthy speakers. Two hypotheses were tested: Whether there is a significant difference across voiceless obstruents (i.e. /f/, /s/, /∫/, /k/, /t/, /p/) in terms of eliciting consistent RFF values, and whether sentences containing just one voiceless obstruent or many voiceless obstruents (i.e. "unmixed" vs. "mixed" sentences) are more effective in eliciting stable RFF. Twenty-eight sentences were developed, each containing 3-6 instances of RFF; there were 18 "unmixed" sentences, sub-grouped by obstruent, and 10 "mixed" sentences containing a variety of voiceless obstruents in each sentence. Twelve healthy adults were recorded producing this corpus of sentences and these samples were analyzed using acoustic analysis software. Results of this analysis showed that there were statistically significant differences across the voiceless obstruents in terms of producing consistent, stable instances of RFF, and that the differences appeared to be connected to the manner of articulation of the obstruent (i.e. fricatives were more stable than stops). No significant difference was noted between the types of sentences (i.e. mixed vs. unmixed) but qualitative differences were noted between the two groups (e.g., increased data loss in the mixed sentence group due to increased frequency of vocal irregularities). These results provide some insight into the factors that contribute to creating effective stimuli for eliciting reliable and valid RFF, and the information gathered should be taken into consideration for future studies.
Description
Thesis (M.S.)--Boston University