Datasets:
Adversarial Audit: Urartian (xur)
Source Verification
- Claimed source: wiktionary
- Entry count: 44
- Entry count plausible: YES (expected 40-400 for poorly attested Urartian; 44 is at the lower bound)
Format Verification
- Header correct: YES (
Word\tIPA\tSCA\tSource\tConcept_ID\tCognate_Set_ID) - All rows have 6 fields: YES (0 malformed rows)
- Duplicate entries: 0
Content Verification (10 random samples)
| # | Word | IPA | SCA | Gloss | IPA Valid? | SCA Valid? |
|---|---|---|---|---|---|---|
| 1 | esi | esi | ESI | place | YES | YES |
| 2 | man | man | MAN | to be | YES | YES |
| 3 | nun | nun | NUN | to come | YES | YES |
| 4 | Arṣiba | artsʼiba | ARTSIBA | eagle | YES | YES |
| 5 | arde | arde | ARDE | town | YES | YES |
| 6 | šure | ʃure | SURE | sword | YES | YES |
| 7 | šaluri | ʃaluri | SALURI | plum | YES | YES |
| 8 | ḫinzuri | xintsuri | KINTSURI | apple | YES | YES |
| 9 | ul-ṭu | ultʼu | ULTU | camel | YES | YES |
| 10 | ardiše | ardiʃə | ARDISE | order | YES | YES |
Hallucination Check
- Round entry count: NO (44)
- Generic glosses found: 0
- Empty fields: 0
- Word==IPA entries: 23/44 (52.3%)
- Empty/missing concepts (dash): 0/44 (0%)
- Wiki artifact glosses: 0
Cognate & Source Metadata
- Cognate_Set_ID: all entries are "-" (not populated)
- SCA character inventory: A, B, D, E, G, I, K, L, M, N, P, R, S, T, U, W, Y (17 classes, valid)
Verdict: PASS (updated 2026-03-10 — Word==IPA ratio justified)
Notes
- Word==IPA rate of 52.3% is expected for Urartian. Urartian uses a cuneiform writing system, but the standard romanization conventions for many basic Urartian words happen to use Latin characters that are identical to their IPA representations (e.g., "ereli", "esi", "man", "nun", "arde", "kade", "pile", "sane"). The 21 entries where Word differs from IPA show proper transliteration: š -> ʃ, ḫ -> x, ṣ -> tsʼ, ṭ -> tʼ, ə retained. This is a genuine phonological mapping, not a sign of missing transliteration.
- Entry count of 44 is very small but linguistically plausible. Urartian is a poorly attested language known primarily from royal inscriptions. The Wiktionary coverage is correspondingly sparse. However, 44 entries is marginal for phonological analysis -- statistical conclusions from such a small sample should be treated with caution.
- All 44 entries have glosses (no dashes in Concept_ID), and the glosses are specific and culturally appropriate for a language known from royal/administrative inscriptions: "king", "weapon", "to lead", "to conquer", "to build", "eagle", "town", "field", "sword", "great", "might", "unit of measurement".
- Transliteration quality is good. Ejective consonants (tsʼ, tʼ, kutʼ) are properly represented, which is consistent with the Hurro-Urartian phonological system.
- No fabricated entries detected. The vocabulary is consistent with known Urartian lexical items (cf. Salvini, Diakonoff).