Retour vers liste

Détail de la contribution

Auteur: Evie MALAIA

Co-Auteur(s): Joshua D. BORNEMAN, Ronnie B. WILBUR

Bioinformatic properties of sign language motion: fractal complexity of optical flow

Abstract/Résumé: A fundamental goal in sign language research is to identify the perceptual properties of the visual signal that underlie the linguistic universals in human languages. In the auditory domain, world languages and music have been described as having modulation spectra of moderate fractal complexity (1/f) (Singh & Theunissen, 2003). Perceptual research (Yu, Romero, & Lee, 2005) shows that visual (V1) neurons in the human brain are tuned to optimally respond to 1/f complexity of visual signal, suggesting a biological basis for neural sensitivity to moderate fractal complexity in visual stimuli. We hypothesized that in American Sign Language, the distinctive grammatical categories will fall within the range of moderate complexity with regard to their kinematic variability. To test this, we analyzed frequency profiles of optical flow for two types of videos: 1) natural scenes containing humans conducting everyday activities (folding laundry, assembling a Playstation), and 2) signers of American Sign Language (ASL) producing grammatically distinct signs. The frequency profile for each of the binned magnitudes of optical flow in ASL vs. human activity videos indicates more level distribution of high frequencies in sign language videos, suggesting higher fractal complexity of signing as compared to everyday human motion (Figure 1, C: lower β value in 1/fβ correspond to more fractal nature of the signal). Further research should investigate relevance of visual-kinematic properties of sign language to neurobiological mechanics of attention in the visual domain, and sign language acquisition. References Singh, N. C., & Theunissen, F. E. (2003). Modulation spectra of natural sounds and ethological theories of auditory processing. The Journal of the Acoustical Society of America, 114, 3394. Yu, Y., Romero, R., & Lee, T. S. (2005). Preference of sensory neural coding for 1/f signals. Physical review letters, 94(10), 108103.