Etym — The Comparative Indo-Europeanist

The Comparative Indo-Europeanist — laryngeals, reflex chains, the long descent of words.

The reconstruction shelves: the kaikki proto-extracts, IE-CoR cognacy, LIV, the Leiden Latin dictionary, StarLing’s bases, and the curated loan edges. Non-IE lanes of the same shelves ride their own axes too — dual-tagging, never folding.

New here? The Quickstart sets up the library in minutes.

The shelves

A source wears every desk it serves — these eleven answer this desk. Holdings are read live from the catalog and dated; a shelf with nothing synced yet says so.

Private research materials under personal grants are not listed.

Languages on this desk (live doc-or-entry counts as of 19 September 2026): rus 18,239 · ine-pro 9,117 · gem-pro 7,711 · sga 6,564 · gmw-pro 5,551 · sla-pro 5,431 · ine 4,981 · ote 2,558 · ron 2,270 · itc-pro 2,211 … and 52 more (nabu axis etym lists all).

The desk’s instruments

• The reconstruction shelves: the kaikki proto-extracts (wiktionary-recon), IE-CoR cognacy (iecor), LIV’s PIE verbal etymons (liv), de Vaan’s Latin (edl), the five StarLing bases including Vasmer’s Russian (starling), and the curated loan edges (sabellic-loans).
• The walk: nabu etym closes the reflex graph hop by hop, flagging (loan) edges along the way; nabu define *headword scopes to the reconstruction shelves; nabu cognates crosses the crosswalk with the alignment hub.
• Non-IE lanes of the same shelves ride their own axes too — dual-tagging, never folding.

Working the etym desk

The generic axis surfaces — every desk answers to these, in working order (enable once, sync, then query):

nabu enable etym               # first time: put this desk's shelves in this box's profile
nabu sync etym                 # fetch/refresh the desk's enabled members
nabu list --axis etym          # the shelf census, this desk only
nabu axis etym                 # the desk card: members, holdings, gold coverage
nabu search WORD --axis etym   # a query scoped to this desk's shelves

This desk’s own surfaces:

nabu etym богъ --lang chu             # a multi-hop proto-to-proto walk with per-edge loan flags
nabu define *bʰer-                    # a leading * scopes to the reconstruction shelves; --long expands
nabu etym rufus                       # the Sabellic-to-Latin loan etymon
nabu cognates nt --langs got,chu      # same-root verses across the aligned witnesses
nabu language zle-ort                 # decode an etymology cognate-list language code

Ask your model

With the MCP server connected, this desk answers conversational research questions. Each example ran live against this library:

• “Walk прьстъ back as far as it goes.” → nabu_etym прьстъ (lang: chu) — The full chain *pьrstъ → Proto-Balto-Slavic *pírštan → PIE *per- “before, in front”, each shelf with glosses, licenses, and corpus-attested cognate counts (Sanskrit pára ×9,769, Latin prīmus ×1,345, Hittite peran, Greek πρίν with gold and silver counts listed separately).

Terminal setup

• Mostly Latin and IPA reconstruction text. Cyrillic reflexes render under Noto Sans Mono; Old Italic reflex headwords (sabellic-loans) need font-noto-sans-old-italic. No RTL or CJK concerns.

The full guidance, per script, is on the display page.

One of the twenty-five research desks; the flat shelf map is The Library and the reasoning is docs/axes.md.