Catalonia Reveals Language‑Identification Bugs in AI Search
Audit AI search results for language accuracy and adjust local SEO to mitigate language‑identification bugs.
Audit AI search results for language accuracy and adjust local SEO accordingly.
Summary
AI search does not merely translate or localize results; it decides which sources and versions of reality are surfaced. Catalonia, where Catalan and Spanish share the same geography, reveals language‑identification problems in AI search. Google Translate often misidentifies Catalan as Occitan, leading to castilianized results that do not reflect the local language. The issue was publicly acknowledged by Google in 2023, and updates improved Catalan visibility in classical SERPs, but the underlying language‑identification layer remains broken. When a Catalan speaker queries AI Overviews, the system often returns Spanish answers, reflecting the same bug that propagates into AI synthesis layers. AI search inherits the assumption that the query language is unreliable, flattening sub‑national contexts in markets that appear monolingual. The problem is a symptom of broader retrieval defaults that collapse multiple languages or jurisdictions into a single statistical default. Understanding these defaults is crucial for local SEO and AI‑search strategy in multilingual regions.
Key changes
- AI search decides which sources and reality versions are surfaced
- Catalonia shows language‑identification problems; Catalan misidentified as Occitan
- Google Translate castilianizes Catalan, leading to wrong results
- Google acknowledged issue in 2023; updates improved visibility but layer remains broken
- AI Overviews often return Spanish for Catalan queries
- AI search assumes query language unreliable, flattening sub‑national contexts
- Retrieval defaults collapse multiple languages into single statistical default
- Local SEO must account for these defaults in multilingual regions