Languages of Art: An Approach to a Theory of Symbols (1968) is Nelson Goodman's most influential work and the founding text of analytic aesthetics. The book applied the rigorous methods of analytic philosophy to questions about art that had been largely reserved for impressionistic criticism: What is representation? How do pictures refer? What distinguishes art from non-art? What is expression? Goodman's answers rejected Romantic assumptions wholesale. Art is not emotional self-expression—it is symbolic reference through structured systems. Pictures do not copy reality—they construct versions through conventional denotation and exemplification. The aesthetic is not a feeling—it is a cluster of formal properties (density, repleteness, exemplificational richness) that certain symbol systems exhibit. The book introduced the autographic/allographic distinction, the theory of notation, the analysis of denotation and exemplification as referential modes, and the concept of expression as metaphorical exemplification. Its 277 pages are densely argued, technically demanding, and unsentimental to a degree that alienated many readers trained in Continental or Romantic aesthetics. But the framework proved generative: it reshaped aesthetics, influenced cognitive science's understanding of symbolic representation, and now provides the most precise vocabulary available for analyzing AI's challenge to human creativity.
The book emerged from Goodman's 1960-62 Special Lectures in Philosophy at the University of London, where he first presented the notation theory to a philosophically sophisticated but aesthetically skeptical audience. Early reviews were mixed—analytic philosophers appreciated the rigor but questioned the framework's adequacy for capturing aesthetic experience; art critics found the technical vocabulary forbidding. The book's influence grew slowly through the 1970s and 1980s as its frameworks were adopted, extended, and contested across philosophy of art, cognitive science, and semiotics. By the 1990s it had become canonical—the work any serious aesthetics student had to engage, whether to build on or to reject.
What makes Languages of Art applicable to AI is its functionalism: Goodman evaluated symbol systems by what they do (how they refer, what understanding they provide) rather than by what produced them (human intention, creative genius, emotional authenticity). The functional approach opens the door to AI art—if a machine-generated image denotes, exemplifies, and expresses through the same referential structures human art employs, then it functions as art. But the same functionalism raises the bar: functioning as art requires achieving rightness of rendering, and rightness depends on purposes the rendering serves. AI can render symbols. It cannot supply the purposes that make the rendering right rather than merely plausible, because purposes require a worldmaker inhabiting the world the symbols version.
The book's Chapter IV on notation systems has become unexpectedly central to the AI discourse. The requirements Goodman specified for adequate notation—syntactic differentiation, semantic unambiguity—are the requirements that would enable a human to specify a creative intention with precision sufficient to survive machine rendering. The requirements are extraordinarily strict, perhaps impossibly strict for natural-language specification of complex creative works. What this means is that the collaboration between human and machine operates without the notational infrastructure that would guarantee the human's worldmaking intention determines the work's identity. The collaboration is post-notational: intentions are expressed in language too loose to function as scores, and the machine's contributions can alter the work's identity under the guise of merely performing it. Whether this is a liberation or a crisis is the question Goodman's framework forces into view without settling.
Goodman wrote Languages of Art in the mid-1960s while teaching at Brandeis University. The book synthesized his earlier work on philosophical logic, phenomenalism, and induction with his long-standing interest in the arts (he was a serious collector of modern art and a museum trustee). The title signaled the book's ambition: to show that the arts are languages—not metaphorically but literally, symbol systems through which versions of reality are constructed according to syntactic and semantic rules as determinate as those governing verbal languages. The claim was controversial then and remains contested now, but its influence on how philosophers, artists, and now builders think about symbolic representation has been undeniable.
Art as symbol system. Paintings, music, literature function cognitively through structured symbolic reference—denotation, exemplification, expression—not through emotional transmission or reality-copying.
Autographic vs. allographic. Arts divide by whether productive history is constitutive (autographic) or whether notational compliance suffices for work-identity (allographic)—a distinction AI makes unstable.
Notation requires precision. The five requirements for adequate notation are so strict that natural language fails them—collaboration without notation means intentions cannot be specified with identity-preserving precision.
Functionalism opens and constrains. Evaluating symbols by function rather than origin permits AI art in principle while demanding rightness in practice—a standard AI rendering often fails despite surface competence.