Abstract
This position manuscript examines the limitations of treating human language as both the interface to and the internal substrate of machine reasoning. It argues that future AI systems should distinguish communicative representations from the structures used for computation and inference. The complete argument and formal citation metadata will be published only when a public manuscript is available.
Public record
This page records the disclosed manuscript without implying acceptance. A public PDF, complete citation metadata, code, and artifacts will be added only when they are legitimately available.