Functional Outcome Categorization in Human Language: A Cross-Linguistic Dataset and Benchmark for AI Alignment

Scope Note: This methodology paper documents empirical and typological constraints observed during FOCP development (v0.2). The OGL kernel specification (DOI: 10.5281/zenodo.19634102) establishes prior art for the underlying triad. FOCP, DNAHS, Personal OS, and Delta Mark rubric remain in private development under a Red Hat open-core model. Known scope limitations (named, not fixed): - Algonquian animate/inanimate grammatical classification is a known case where the organizing principle's surface does not straightforwardly reduce to agent-outcome framing. Flagged for targeted treatment in a forthcoming companion paper; not addressed in present work. - Inter-rater reliability (IRR) full study is deferred to v0.3. A single-rater pilot protocol is documented in Appendix J. - Tzeltal spatial reference is flagged as future adversarial TOC expansion target. - Cross-linguistic kinship terminology noted as untapped dataset for falsifiability criterion testing; not integrated in present work. Intended for prior-art timestamping. Pre-peer-review. v0.3 planned for arXiv submission after IRR full study.