Aesthetics and Discourse

How do musicians, listeners, and critics talk about synthesised sound? WP2 identifies and analyses the vernacular discourse of synthesis across different genre communities, moving beyond both the pejorative language of cloning and fakery that dominates popular commentary and the engineering terminology that cordons off synthesis as a specialist domain.

Top-down taxonomies of listening in electronic music are abundant, from Pierre Schaeffer's Traite des objets musicaux to Denis Smalley's spectromorphology. Yet little research has examined the vernacular terms and metaphors through which musicians connect synthetic sound to particular musical worlds. This is a conspicuous gap, given decades of musicological work attesting to the shifts that synthetic sound brought to listener categories and expectations. The growth of prompt-based generative AI systems that synthesise sound from semantic cues suggests that a listener-led discourse for timbre has been functionalised even where it has been little analysed.

Approach

WP2 employs natural language processing and critical discourse analysis to compile and analyse web-based text corpora drawn from genre- and practice-based discourse communities: listservs and forums engaging with synthesis and synthetic media across electroacoustic music, EDM, modular synthesis, hyperpop, organisations such as Afrorack and Music Production for Women, and online communities for AI-generated music. Analysing these makes it possible to assemble the lexical traits of synthesis among different groups, attending to how sounds, their control, and their perceived integrity are conceptualised.

The results will enable the creation of a richly annotated, open-access linguistic corpus whose primary purpose is to understand evolving aesthetic orientations to synthetic sound on the terms that musicians and creative technologists themselves establish. Additionally, findings will be directed toward practical settings: enhancing the cultural responsiveness of LLM training, raising new questions in timbre cognition, and improving virtual instrument design.

Methods

Natural language processing (Python/spaCy), topic modelling (TOMOTOPY), critical discourse analysis, and cognitive linguistics. Data collection via web scraping (Beautiful Soup) with cleaning and standardisation (OpenRefine).