Akswara: A Writing System Based on Human Speech Perception for Cross-Linguistic Use

Writing systems are typically developed within the phonological and historical boundaries of individual languages, while phonetic transcription systems prioritize fine-grained articulatory precision that exceeds the requirements of everyday communication. This creates a gap between how speech is produced, how it is perceived, and how it is commonly written. Human listeners reliably interpret speech despite substantial variation in articulation, suggesting that perceptual organization, rather than exact phonetic specification, is sufficient for functional communication. Akswara proposes a sound-based writing system grounded in this perceptual reality. Conceived as aksara suara—a script of sounds—it encodes speech according to how sounds are perceived rather than how they are exhaustively articulated. Consonants are organized through a region × manner framework that models broad articulatory regions as perceptual categories, while vowels are treated as regions within a continuous vowel space anchored by points of maximal contrast. The system maintains a compact and stable core inventory of 26 consonant categories and ten vowel regions. Implemented as an abugida, Akswara treats consonants as primary units carrying an inherent low vowel, with diacritics used to encode non-default vowel qualities, syllable structure, and secondary articulations. This layered encoding mirrors phonological hierarchy while supporting cross-linguistic adaptability. The paper presents the theoretical motivation, phonological organization, and graphic encoding principles of Akswara, and illustrates its application across multiple languages as a perceptually grounded, language-independent writing system.