Externalizing Tongue Articulation Through Forearm Touch: A Wizard-of-Oz Study of Tactile Articulatory Encoding for Arabic Letters

Speech articulation involves tongue movements that are largely hidden from direct perception, which makes them difficult to explain, observe, and guide during pronunciation instruction. This paper asks whether articulatory information can be externalized through touch and decoded by a listener. We introduce a tactile encoding scheme that maps regions of the tongue onto spatial locations on the forearm, combined with four tactile patterns (Tap, Hold, Trill, and Push) representing articulatory manner. To evaluate whether this encoding is interpretable, we conducted a Wizard-of-Oz study with six native Arabic speakers, in which cues were delivered manually by an experimenter rather than by a haptic device, so that the mapping could be examined before committing to a particular hardware implementation. Seven Arabic letters were used as targets, drawn from the full inventory of twenty-eight; participants were not informed that the target set was restricted, so responses were made against an open set. Following instruction on the general tactile–articulatory encoding principle, but before target-specific corrective feedback, participants identified the intended letter on 29.76% of trials, exceeding the nominal uniform-response reference levels defined for the open response set. After a brief exposure phase, accuracy reached 75.00% and response time fell from 14.48 s to 4.40 s. Critically, errors were not randomly distributed: they clustered on letters that are articulatorily adjacent to the target, and in several cases on letters that were never presented as targets (e.g., dāl, lām and nūn), a pattern that is difficult to explain by rote association alone. Some confusions involved targets encoded at neighboring forearm sites, suggesting possible spatial-discrimination limitations. We report these findings as preliminary evidence that tongue articulation may be represented in a decodable form through spatial touch. We make no claims regarding pronunciation learning, retention, or pedagogical efficacy, which this design cannot address.