Quality frameworks
Talk to an expert
Service description
You want to know what you are paying for. Rate your vendors. Select the best MT engine or translation workflow for your content and target languages. Define the scope of post-editing. Monitor quality against agreed service levels.
Various techniques are available for MT quality estimation, ranging from automated methods to manual linguistic QA, though their results aren’t always comparable. A quality framework based on error typology helps to reduce subjectivity and eliminate bias in evaluations. The resulting scores can be combined with edit distance and time for comparable aggregate metrics.
Why choose espell?
espell uses a comprehensive framework adapted to different use cases. Error categories remain consistent, but their assigned weights reflect specific criteria for any given content type.
Our experience with assessing and improving the quality of transcreated content is distilled into a QA matrix. This addresses aspects like creative adaptation, tone of voice, fluency, culture, etc.
For MT post-editing, different service levels are defined based on the standard error categories. This helps to scope and limit the MT PE effort to the key quality criteria.
Our pricing is always custom-tailored to your needs.
"Our customer is one of the top streaming content providers, with tens of millions of subscribers worldwide. When launching their services for 10+ new markets, they needed a multilingual communications partner to design a localization process from scratch that would cover diverse needs across different platforms. The espell team helped in setting up an integrated and automated authoring, content management and translation management environment, while also building a selected team of linguists to provide a seamless copywriting and localization service."
Márta Snopek
lead project manager