Syntax All The Way Down, and All The Way Up, Too

The central claim

This page presents the central thesis of a new proposal about the internal organization of the Language Faculty.

The usual Minimalist picture places Narrow Syntax between the Conceptual-Intentional performance system and the Articulatory-Perceptual performance system. The proposal developed here keeps that general orientation, but replaces the idea of one central syntax with a sequence of five linguistic layers.

The stronger claim is that these layers are not separate boxes connected by translation alone. Each layer uses the same basic syntactic machinery: pattern matching and graph extension.

In this sense, syntax is not merely one component in the middle of the Language Faculty. Syntax is the repeated organizing operation of the whole faculty.

The five-layer sequence

The proposal divides the internal Language Faculty into five sequential layers. These layers connect the two performance systems: C-I on one side and A-P on the other.

C-I and A-P are the two performance systems. The five layers are the internal linguistic machinery that stands between them.

The names are partly familiar and partly reorganized. Semantic Component and Phonological Component are treated here not as loose interfaces, but as linguistic layers with their own structured operations. Inner Syntax, A-bar Syntax, and Distributed Morphology form the central syntactic path.

Why this matters

In many presentations, syntax appears as one component of grammar, located between meaning and sound. This proposal changes the perspective.

The central operation of syntax is not confined to one place. Pattern matching and graph extension appear again and again, at different levels of the Language Faculty.

This makes it possible to view the whole faculty as a sequence of syntactic operations, rather than as a pipeline of unrelated modules.

That is the intended meaning of the title: syntax all the way down, and all the way up, too.

Relation to Symbolic Superintelligence

Linguistic Agents approaches Symbolic Superintelligence through the internal architecture of language.

Language is not treated merely as a channel for prompts and answers. It is treated as the machinery by which agentic content is shaped, encoded, transmitted, interpreted, and recovered.

The five-layer proposal gives this program a linguistic foundation. It describes how structured content can pass from an agent toward speech or text, and how speech or text can be interpreted back toward agentic form.

The larger engineering goal is to build agents whose internal language machinery is explicit, inspectable, and symbolic.