Tongue machine interface (TMI) is a tongue-operated assistive technology enabling people with severe disabilities to control their environments using their tongue motion. In many disorders such as amyotrophic lateral sclerosis or stroke, people can communicate with the external world in a limited degree. However, they may be disabled, while their mind is still intact. Various tongue–machine interface techniques has been developed to support these people by providing additional communication pathway. In this study, we aimed to develop a tongue–machine interface approach by investigating pattern of glossokinetic potential (GKP) signals using neural networks via simple right/left tongue touchings to the buccal walls for 1-D control and communication, named as GKP-based TMI. As can be known in the literature, the tongue is connected to the brain via hypoglossal cranial nerve. Therefore, it generally escapes from the severe damages, in spinal cord injuries and was slowly affected than limbs of persons suffering from many neuromuscular degenerative disorders. In this work, 8 male and 2 female naive healthy subjects, aged 22 to 34 years, participated. Multilayer neural network and probabilistic neural network were employed as classification algorithms with root-mean-square and power spectral density feature extraction operations. Then the greatest success rate achieved was 97.25%. This study may serve disabled people to control assistive devices in natural, unobtrusive, speedy and reliable manner. Moreover, it is expected that GKP-based TMI could be a collaboration channel for traditional electroencephalography (EEG)-based brain computer interfaces which have significant inadequacies arisen from the EEG signals.