Machine translation and post-editing analysis

Some clients compensate projects based on a post-editing analysis. You can read the full details on this page in Phrase's own knowledge base, but in short it means that after the translation is completed, the final segments are analysed against their “best possible suggestion” to calculate the edit distance between the two. This “best possible suggestion” will be a TM match if one was available for a given segment, but it can also be an MT suggestion.

The important thing to understand here is that the post-analysis will calculate the edit distance between the “best possible suggestion” and the final translation even if the linguist doesn't actively utilise the suggestion, by for example wiping the MT suggestion in the text or not inserting leverage from the TM or MT engines. Therefore, to make sure that the post-editing analysis feels fair to you, you should make full use of the TM and MT suggestions that you see while translating.

Even if a project is compensated based on post-editing analysis, there will be a pre-analysis available which is used to estimate the amount of work and for sending the preliminary POs. However, this pre-analysis is only an indication of the final compensation, and the Sandberg PM will be updating the task when the post-editing analysis is available, before they release the task for invoicing. It's worth noting that the pre-analysis TM leverage does not necessarily reflect the post-analysis - a 75% match in the pre-analysis can be a 100% match in the post-analysis (if the translator makes no edits to the suggestion) or a 0% match (if the translator changes every single word in the suggestion).

Your PM should always clearly state when a project is paid using post-editing analysis when offering a task to you and when sending the preliminary PO, but if you are ever unsure, you can double-check this with them before agreeing to a project.