PLA Project - English to Turkish

I need support on an ongoing initiative that measures paragraph-level translation quality. A new linguistic annotation initiative designed to evaluate and improve paragraph-level translation quality across multiple languages. This project offers linguists and annotators the opportunity to contribute to the advancement of high-quality translation systems.

Project Scope

Each assignment (HIT – Human Intelligence Task) consists of a source text paired with its translation. Annotators will be responsible for:

- Reviewing the full HIT to understand context, tone, domain, and terminology
- Identifying and flagging source text issues (e.g., incorrect language, garbled text, or punctuation-only content)
- Assessing the acceptability of translations on a 0–100 scale
- If no translation is present, marking as “Fully Missing Translation”
- Highlighting corresponding spans between the source and target texts
- Flagging target-side errors, including hallucinations (translated content not present in the source)

Note: Sentence boundary errors are anticipated and should not affect scoring.

Workload and Availability

- Up to 4,000 HITs per language available for annotation.

Expected Pace

- Average of 4.5 minutes per HIT (approx. 389 words per HIT)