CAPR: Computer Assisted Proto-language Reconstruction

CAPR is a Dockerized toolkit for comparative linguistics, combining:

• Cognate board management — track cognate sets across related languages
• FST development — build and test finite-state transducers for sound change modeling
• Web interface — Svelte UI for data exploration and FST debugging

Project Components

This repository contains two parallel research pipelines sharing common infrastructure:

Component	Description	Status
Germanic/	Proto-Germanic → Old English sound changes	Active development (62 of 1057 rows mismatched)
Burmish/	Proto-Burmish reconstruction	Maintenance mode

Both use the same web application and FST tooling but have independent data, transducers, and documentation.

Quick Start

# Start the Docker stack (Flask API + Svelte UI)
docker compose up -d

# In another terminal, proxy through Caddy
caddy run --config Caddyfile.dev

# Open http://localhost:5002

The Docker stack mounts Germanic/ by default. To switch to Burmish, edit docker-compose.yml to mount Burmish/ paths instead.

Repository Structure

capr-v3-working/
├── Germanic/               # Proto-Germanic → OE pipeline
│   ├── data/               # TSV wordlists (germanic-aligned-final.tsv)
│   ├── fsts/               # FST sources (germanic.txt)
│   ├── tools/              # Python analysis tools
│   └── docs/               # DEV_NOTES.md, debug snapshots
├── Burmish/                # Proto-Burmish pipeline
│   ├── data/               # Burmish wordlists
│   ├── fsts/               # burmish.txt
│   └── ...
├── app/                    # Shared web application
│   ├── backend/            # Flask API (compare_fst.py, refish.py, etc.)
│   └── frontend/           # Svelte cognate board UI
├── server/                 # Docker working directory (code copied here)
├── cognate-app/            # Svelte source (mirrored in app/frontend/)
├── docs/                   # Shared documentation
│   ├── references/         # Scholarly sources (PDFs, text extracts)
│   ├── runbook.md          # Operational checklist
│   └── README.md           # Documentation index
├── build/                  # Compiled .bin files (gitignored)
├── docker-compose.yml      # Development stack
└── Caddyfile.dev           # Reverse proxy config

Germanic Pipeline

The active focus is modeling Proto-Germanic → Old English sound changes via FST rules.

Current status: 62 mismatches out of 1057 lexemes (94% accuracy)

Key files:

• Germanic/fsts/germanic.txt — Main FST with all sound change rules
• Germanic/data/germanic-aligned-final.tsv — Aligned proto-forms and OE targets
• Germanic/docs/DEV_NOTES.md — Detailed research log and phonological decisions

Development workflow:

# Compile FST and check results
docker compose exec -T backend bash -c "cd /usr/app && foma -q -l fsts/germanic.txt -e quit"

# Run mismatch analysis
docker compose exec -T backend python3 tools/oe_mismatch_report.py

See Germanic/README.md for detailed documentation.

Burmish Pipeline

Proto-Burmish reconstruction using LingPy/LingRex for cognate detection.

See Burmish/README.md for documentation.

Documentation

• docs/README.md — Documentation index
• SETUP.md — Full installation guide
• USAGE.md — UI walkthrough
• docs/runbook.md — Operational checklist
• WORKFLOW.md — Development workflow

Citations

• Xun Gong & Nathan Hill (2020). Materials for an Etymological Dictionary of Burmish. Zenodo. https://doi.org/10.5281/zenodo.4311182
• List, J.-M. & R. Forkel (2022). LingRex. Zenodo.
• Hulden, M. (2009). "Foma: a finite-state compiler and library." EACL.