We present the results of an experimental study that we designed in order to assess whether subjects are able to identify the vowels on the only basis of the tongue somatosensory feedback. This experiment involves a novel tongue positioning task in which the participants are visually guided to reach different tongue postures in the range of the [e-Ɛ-a] articulations without using speech motor control strategies. The target tongue postures and actual tongue postures were measured using Electromagnetic Articulography. Once they reached the tongue postures the participants, which auditory feedback was masked with a white noise, were asked to whisper and then to identify which of the [e-Ɛ-a] vowel was associated to the reached tongue posture. Whispered sounds were evaluated after the session by independent listeners. Eight subjects performed the task.

We evaluated the proprioceptive responses using qualitative and quantitative approaches. These analyses indicate two main results. First, there is a strong inter-subject variability in the adequacy of the proprioceptive answers with respect to the answers that we expected, knowing where each reached tongue posture is located in the [e-Ɛ-a] articulation range: four subjects performed highly consistently and in agreement with the expected answers, while the others four subjects performed poorly, with a less regular pattern of answers and a small agreement with expected answers. Second, despite these important inter-subject differences, the overall pattern of proprioceptive and auditory answers were similar in each subject: for the four subjects who performed well in the proprioceptive categorization task their whispered sounds were also consistently categorized by the independent listeners, and for those who performed bad in the proprioceptive identification task, their whispered sounds were not well-categorized.

Overall, our results suggest that (1) vowel categorization is possible on the only basis of the somatosensory feedback and (2) its accuracy is similar to the accuracy of the auditory perception of whispered sounds. We discuss the origin of the differences observed across our eight subjects and the possible mechanisms involved in the proprioceptive categorization: directly from the somatosensory feedback via a comparison with stored somatosensory patterns associated with each vowel category or indirectly by projecting the somatosensory feedback into the auditory domain via internal models?