TL;DR
This paper introduces a simplified four-cube Finite Element Method (FEM) model of the tongue to investigate the role of intrinsic tongue muscles, specifically the anterior transverse muscle (Ta), in producing the vowel /i/. By comparing FEM simulations with tagging-MRI data, the study suggests that intrinsic muscles play a crucial role in the rapid, medial compression of the tongue blade necessary for /i/ production, challenging the traditional view that extrinsic muscles are solely responsible.
What Problem Does This Paper Address?
Producing speech sounds involves complex interplay between brain signals, muscle activation, and the physical deformation of the vocal tract. The tongue, a highly muscular and flexible organ, is central to articulation. While extrinsic tongue muscles (like the genioglossus) have been extensively studied for their role in tongue movement, the contribution of intrinsic tongue muscles (which alter tongue shape without changing its position) to specific vowel sounds, particularly the high front vowel /i/, remains less understood. Previous research, including EMG studies, has primarily focused on extrinsic muscles, leaving a gap in our understanding of how intrinsic muscles contribute to the precise tongue shaping required for vowels like /i/. This paper aims to bridge this gap by modeling and simulating the tongue's internal deformation.
Methods
The authors developed a simplified Finite Element Method (FEM) model of the tongue, conceptualized as four cubic elements. This "four-cube" model divides the tongue into top-lower and front-back halves, allowing for the simulation of deformation in the mid-sagittal plane. The model incorporates the action of extrinsic muscles (genioglossus posterior - GGp, and genioglossus middle - GGm) and, crucially, the anterior transverse muscle (Ta), an intrinsic muscle.
The FEM simulation is based on a viscous elasticity continuum model for soft tissues. Muscle contraction is represented as active and passive tensile stress, with passive elasticity modeled using an isotropic exponential constitutive strain energy function and viscous stress using linear viscosity. The simulation parameters include a maximum contraction force of 0.022 N/cm². Muscle activation is linearly increased over 0-0.05 seconds, then held constant. The study compares two muscle activation scenarios: (1) GGm + GGp (representing a conventional model) and (2) GGm + GGp + Ta (incorporating the intrinsic muscle).
Datasets
The study utilizes tagging-MRI data for evaluation. Specifically, they refer to tagging-MRI measurements of tongue deformation during the /ei/ vowel sequence, including mid-sagittal imaging at 60 fps and transverse plane imaging. This data is used to validate the FEM simulation results. No new datasets were introduced.
Key Insights and Ideas
- Intrinsic Muscles Matter for /i/: The most significant insight is the strong evidence suggesting that the anterior transverse muscle (Ta) plays a critical role in producing the vowel /i/. Its contraction appears to be essential for the rapid medial compression of the tongue blade.
- Challenging the Extrinsic-Only View: This finding challenges the long-held assumption that extrinsic tongue muscles are the primary drivers of vowel articulation. The study highlights that intrinsic muscles contribute significantly to the fine-tuning of tongue shape.
- Medial Compression is Key: The simulation results, particularly with the inclusion of Ta, show a good agreement with tagging-MRI data, demonstrating a rapid and significant medial compression in the front half of the tongue. This compression is crucial for achieving the acoustic characteristics of the /i/ vowel.
- Simplified Model Efficacy: The use of a very simple four-cube FEM model proved effective in isolating and demonstrating the contribution of the anterior transverse muscle, suggesting that complex models are not always necessary to understand specific articulatory mechanisms.
Key Results
- The FEM simulation incorporating the anterior transverse muscle (GGm + GGp + Ta) showed a good agreement with tagging-MRI data regarding tongue blade displacement and lateral expansion/contraction patterns during the /e/ to /i/ transition.
- Specifically, the GGm + GGp + Ta model demonstrated a rapid transition and larger displacement of the tongue blade between 100-300 ms, along with medial compression in the front half of the tongue, mirroring observations from tagging-MRI.
- In contrast, the model without the transverse muscle (GGm + GGp) did not accurately replicate the observed tongue deformation, particularly the medial compression and the speed of tongue tip movement.
Limitations and Future Work
- Model Simplification: The authors acknowledge that the four-cube model is a simplification. It lacks the detailed anatomical complexity of the tongue and does not account for the mid-line groove formed by the genioglossus anterior (GGa) or the effects of other intrinsic and extrinsic muscles.
- Limited Muscle Representation: Only a few key muscles are modeled. The full range of intrinsic and extrinsic muscle interactions is not captured.
- Future Work: The authors suggest that future work should involve more detailed anatomical models and the inclusion of a wider range of muscles to further refine the understanding of tongue biomechanics in speech production. They also imply that this approach could be extended to other vowels and speech sounds.
Summary updated