Machines now use words fluently. Do they mean anything by them?
Language is the medium in which most philosophy is done and, increasingly, the thing philosophy is about. Is meaning a matter of reference, of use, of social agreement? Confucius thought the rectification of names — calling things by their right names — was the first task of good government. Wittgenstein came to see meaning as use within a form of life. The Indian grammarians built a science of it millennia ago. Now machines produce language of uncanny fluency, and the question of what language is has become impossible to postpone.
The thinkers this collection draws together. Full profiles live under Thinkers.
Wittgenstein
Meaning as use; language as embedded in a shared form of life.
Confucius
The rectification of names as the foundation of social order.
Plato
The Cratylus: are names natural or merely conventional?
Bhartrhari
The Indian grammarian tradition on language and cognition.
Anchor texts, in translation, with the original where it matters. To be added.
Candidates include the Philosophical Investigations, the Analects (XIII.3) on names, and Plato's Cratylus.
Where the traditions agree, where they part, and why. In progress.
Reference, use, and convention across traditions, and what a purely statistical user of language reveals about each.
An LLM is a working experiment in Wittgenstein's intuition: it learns language entirely from use, representing meaning as geometry in which related senses sit close together. Whether that constitutes meaning or merely its shadow is the live dispute — the stochastic parrot on one side, richer readings on the other. The fluency is not in question; the understanding behind it is.
Terms used here, defined plainly: LLM · Token · Embedding · Stochastic Parrot
Annotated bibliography. To be added.
Selected scholarship and translations will be listed here, each with a short note on why it earns its place.