Elan Barenholtz on Language, Music and Art

How did language develop its rich, complex properties and what led to your epiphany about language’s nature? Elan framed a recent realization that language may possess an internal, self-generative structure revealed by large language models. He argued that modern language models work by learning statistical relations in sequences rather than by possessing external world knowledge or reference. Elan proposed that human speech may be language “running on its own,” with the speaker acting mainly as a prompt to an internal informational system. He reported conducting computational and empirical work over the past ~18 months to test and develop this intuition about language. Elan suggested language likely evolved from preexisting biological sequence-generating mechanisms rather than appearing de novo in humans. He proposed music as a possibly purer autogenerative domain that may have preceded or informed early language forms. EMIC analysis (emic) is used to derive internal musical rules from large samples of transcribed songs to enable generative composition that insiders can validate. Generative analysis aims to produce compositions that are stylistically authentic based on derived internal structures. Historical precedents exist for algorithmic composition; the Arca Musarithmica mechanically generated musical settings for Latin texts in the 17th century. Was the 17th‑century Arca Musarithmica an actual physical machine? Are all musical forms capturable by emic/generative techniques and do musical universals exist analogous to linguistic universals? Elan raised two core research questions: whether all musical forms are capturable by this technique and whether cross‑cultural musical universals exist. Ethnomusicologists are generally skeptical of declaring broad universals and emphasize cultural distinctiveness. Drew argued that the act of singing is a likely universal behavior with implications for intonation, audiation, scales, harmony, and rhythm. Does disciplinary bias (postmodernism) explain ethnomusicologists’ reluctance to assert universals? Quranic recitation exemplifies a vocal tradition that resembles music to outsiders but is treated as elevated intonation internally, and reciters often receive formal singing training. Serialist and some 20th-century art music intentionally separate musical sound from social function. Most music functions socially to connect people and express shared feelings beyond mere acoustics. Postmodern art and music challenged conventional notions of beauty and meaning by redefining what counts as art. Peirce’s semiotic categories (icon, index, symbol) map cleanly onto musical modes of signification. How is music conveying meaning when tonal pieces evoke emotions without explicit narrative? Indexical musical features, like minor chords, commonly signal sadness through shared or embodied associations. Cultural familiarity determines whether listeners register symbolic musical cues and extract meaning. Language and music both operate as relational systems that create expectations and structure meaning without direct reference. Music functions as a program that generates emotional responses rather than transmitting explicit narratives. Popular-song phrases can reliably evoke emotion across broad audiences due to stable structural hooks. How do you define and operationalize “meaning” in language versus music within this autogenerative framework, and what concrete predictions does this yield for experiments? In an autogenerative framework, both language and music are treated as systems that generate expectations over time, but they differ in what those expectations are anchored to and how “meaning” stabilizes. How “meaning” is defined