Institutional Partnership
The infrastructure for a documented reference corpus for visual-forensic archaeological cataloging is built. What the pilot requires is 150–300 well-documented objects from institutional collections — and a partner with the collection depth to co-build and co-steward the resulting instrument.
The tool is a working prototype, not a proposal. It has been run on real objects:
Three-pass analysis
Blind observation → documentation comparison and epistemic labeling → RAP Protocol and provenance assessment. Every claim is anchored to a physical observation or a named primary source.
27-principle fingerprint
Each object is scored against 27 perceptual principles drawn from Brody, Shafer, Hegmon, and Schaafsma. Scores are stored as a vector in canonical order, built for similarity search when the corpus reaches density.
Provenance integrity assessment
Eleven named structural gap patterns applied to every documentation record — no institutional anchor, no excavation record, private collection terminus, acquisition chain anomalies, and others. Severity-coded HIGH / MEDIUM / LOW.
Contradiction detection
Documentation claims are cross-referenced against blind observations. When what the catalog says about construction, condition, or iconography conflicts with what the object shows, the tension is flagged explicitly.
Reference corpus infrastructure
Database schema, distribution API, catalog management UI, and corpus access page are complete and live. The corpus itself needs populating — authenticated, well-documented objects from institutional collections. That is what the pilot builds.
Epistemic labeling
Every claim in the analysis is marked [OBS], [INTERP], [CONS], or [THEORY] at the sentence level. Interpretive claims require two independent observations. The RAP Protocol flags claims with insufficient anchors — it is structural and cannot be turned off.
The reference corpus is a structured data-collection effort, not a raw donation intake. It targets objects that can anchor statistically meaningful comparison across traditions.
Scale
150–300 objects across three traditions — Mimbres, Ancestral Puebloan, Hohokam. Thirty objects per tradition is the minimum for meaningful fingerprint distribution analysis. The pilot targets fifty per tradition.
Provenience
Objects with known provenience: excavation record, institutional anchor, and an unambiguous chain of custody. The reference corpus draws only from documented contexts — Swarts Ruin, NAN Ranch, Pueblo Bonito, and comparable assemblages are the target sources.
Documentation
Existing catalog record with tradition attribution supported by primary sources. Physical analysis data (XRF, petrography, temper) is a strong plus but not a prerequisite for corpus inclusion.
Custody
Objects already in institutional custody. The pilot does not intake undocumented donated material. If the chain of custody is unclear, the object is ineligible for the reference corpus.
The instrument is designed for use as decision support in human review workflows — not as a replacement for human expertise, legal process, or tribal authority.
Any project working with Southwest archaeological collections operates within obligations that are legal, ethical, and relational — to the tribal communities whose cultural heritage these objects represent.
Tribal consultation
The project design builds tribal consultation in as a structural requirement. For collections subject to NAGPRA review or culturally sensitive imagery protocols, tribal representatives review methodology before analysis begins on those objects.
Advisory review authority
The project advisory board includes representatives from tribal communities with cultural connections to the collections being analyzed. Advisory review authority over methodology is substantive.
Restricted records handling
Analysis records for culturally sensitive objects are handled under the same access controls as the underlying catalog records. Public display of analysis output for sensitive objects requires explicit institutional authorization.
Imagery protocols
Objects subject to culturally sensitive imagery restrictions are analyzed with those restrictions in force. Analysis of ceremonial objects or objects under tribal protocol is not performed without explicit tribal authorization.
The design intention is explicit: build the instrument, validate it against a documented corpus, and transfer stewardship to a museum, tribal organization, or research consortium with long-term care capacity. Funders own the result.
On project completion, the full instrument transfers to the anchor institution or a named research consortium: source code, prompt architecture, schema, eval harness, and the reference corpus itself. No ongoing licensing fees. No subscription model. No dependency on the original developer.
Long-term stewardship candidates include a museum with existing Southwest collections, a university-based research center in archaeology or materials science, or a tribal organization with collections management capacity. The partner institution and the development team identify the steward jointly during the pilot phase.
Three established federal programs align directly with this project's scope:
IMLS — Collections Stewardship
Institute of Museum and Library Services Collections Stewardship grants fund exactly this: infrastructure to document, authenticate, and care for institutional collections. A shared reference corpus and fingerprint instrument for Southwest ceramics is squarely within scope.
NEH — Preservation and Access
National Endowment for the Humanities Preservation and Access grants cover documentation methodology and digital infrastructure for collections with humanities significance. The analytical framework — grounded in Brody, Shafer, Hegmon, and Schaafsma — situates this within humanistic scholarship on material culture.
NAGPRA documentation grants
NAGPRA documentation funding specifically supports projects that build the evidentiary record for repatriation review. This tool produces NAGPRA-relevant documentation as a direct output: provenance gap assessments, contradiction flags between catalog attribution and physical evidence, and fingerprint alignment with tradition assemblages. A project that surfaces repatriation candidates systematically is directly fundable under this program.
If you are a collections director, program officer, or institutional researcher evaluating whether this project is worth a conversation, the right next step is a direct email.
contact@thegrammarofthings.com