The Predicate Domain: A Unified Architecture for NonTensed Verbal Forms Verbal Classifier Hypothesis, Head–Tail Manifold Geometry, and Amplitude-Based Selection

We advance an architectural account of the Predicate Domain as a head–tail manifold 
centered on a generative verb-slot Θ. The central claim is that non-tensed verbal forms –
bare stems, infinitives, participles, gerunds, aspectual complexes, voice forms, and particlebound forms – are not a heterogeneous residue of syntax but the structured outputs of a 
single generative locus. The architecture rests on three commitments: (i) the Predicate 
Domain is a five-region geometry ℙ = C_out ⊕ S_out ⊕ Θ ⊕ S_in ⊕ C_in in which Θ is 
the unique tensed head, subordinate regions host classifier material, and coordinate regions 
host relational operators; (ii) the non-tensed inventory is a closed set of eight subtypes 
selected by a deterministic amplitude function σ: Amp(Θ) → 𝕋 and ordered by seven 
closure laws; (iii) the geometry and the classifier logic are universal invariants of clauseinternal predicate organization, as attested by English, Japanese, Mandarin, Hindi, Yoruba, 
Arabic, and Turkish. We develop the formal apparatus (operators, placement functions, 
refinement composition), prove eight theorems, report an executed corpus study of chain 
integrity in COCA (n = 300; 100% contiguity, 0% interruption), and adjudicate the 
framework against Distributed Morphology, cartographic syntax, first-phase syntax, 
restructuring theory, and Voice-based accounts. Finiteness, we conclude, is positional rather 
than lexical; classifier chains are generated rather than assembled; and tense, aspect, voice, 
causation, and event-type refinement emerge from one operator system