Discriminant Features and Temporal Structure of Nonmanuals in American Sign Language
We propose a computational approach to determine which features of the linguistic model are more informative of the grammatical rules of facial nonmanuals in American Sign Language. The proposed approach can be used in five types of sentences.
- Hypothetical Conditionals
- Yes/No Questions
- Wh-Questions Postsposed
Summary of the Project
We have developed a software in MATLAB and Java to analyze the temporal relations in American Sign Language. Our software can upload a set of tab delimited files created from ELAN and automatically find the most discriminant structures. The whole explanation of our method can be found here in our paper
Current Version Functionality
Leave one sentence out crossvalidation
It finds the most discriminant features leaving one sentence out for testing and using the rest of them for training.
N fold crossvalidation
It divides the samples in N different subsets and one of them is used for training and the rest of them for testing.
Full data analysis
The method can be tested using all the available data for training and the same samples for testing.
Adjustable thresholds (see paper)
Adjustable regularization parameter
- ELAN integration
The software can be downloaded here and a brief tutorial on how to use the program. We also have created a modified version of ELAN, the original software and terms and conditions can be found here.
COPYRIGHT @ 2014 C. Fabian Benitez-Quiroz, Aleix M. Martinez
Department of Electrical and Computer Engineering
The Ohio State University
This code is an implementation of the algorithms described in: Benitez-Quiroz CF, Gökgöz K, Wilbur RB, Martinez AM (2014) Discriminant Features and Temporal Structure of Nonmanuals in American Sign Language. PLoS ONE 9(2): e86268. doi:10.1371/journal.pone.0086268
This software is distributed WITHOUT ANY WARRANTY. Use of this software is granted for research conducted at research instituations only. Commerical use of this software is not allowed. Corporations interested in the use of this software should contact the authors. If you use this code for a scientific publication, please reference the paper above.