This study was conducted to classify tongue motions of elderly people from surface electromyography (EMG) signals of the suprahyoid muscles detected at the underside of the jaw. The EMG signals were measured via 22 surface electrodes mounted onto a special flexible boomerang-shaped base. Root-mean-square features and cepstrum coefficients features of the EMG signals were extracted as features. An algorithm based on a support vector machine for pattern recognition was used to classify tongue motions. Results showed that six tongue motions were classifiable with accuracy of 86.1±9.6%.