Related Experiment Videos
Analysis of the three-dimensional tongue shape using a three-index factor analysis model
Yanli Zheng1, Mark Hasegawa-Johnson, Shamala Pizza
1Department of Electrical and Computer Engineering, University of Illinois at Urbana-Champaign, Urbana, Illinois 61801, USA.
The Journal of the Acoustical Society of America
|February 1, 2003
Summary
This study used 3D MRI and parallel factors (PARAFAC) modeling to analyze tongue shapes during vowel production. The analysis revealed a stable two-factor solution, explaining 70% of tongue shape variance.
Area of Science:
- Linguistics
- Speech Science
- Biomedical Engineering
Background:
- Understanding 3D tongue articulation is crucial for speech production research.
- Previous analyses often relied on 2D data or less precise methods.
- Advanced imaging and modeling techniques can provide more accurate insights.
Purpose of the Study:
- To analyze the three-dimensional tongue shape during vowel production.
- To apply the three-mode Parallel Factors (PARAFAC) model to 3D MRI data.
- To identify key modes of tongue movement and shape variation.
Main Methods:
- Acquired 3D Magnetic Resonance Imaging (MRI) data from five speakers producing nine vowels.
- Selected 65 virtual fleshpoints across interpolated tongue surfaces.
- Developed methods for MRI image alignment, fleshpoint placement, and coordinate measurement.
- Applied PARAFAC analysis to the 3D coordinate data.
Main Results:
- A stable two-factor solution was identified through PARAFAC analysis.
- This two-factor model accounted for approximately 70% of the variance in tongue shape data.
- The results indicate significant, quantifiable variations in 3D tongue configuration during vowel articulation.
Conclusions:
- PARAFAC modeling is effective for analyzing complex 3D tongue shapes.
- The identified factors likely represent fundamental dimensions of tongue articulation.
- This approach offers a robust method for quantitative analysis in speech production.