Purpose
The purpose of this study is to evaluate and compare a selection of leading machine translation technologies.
What was done
We completed a series of human reviews on five of the leading machine translation (MT) technologies across seven language pairs with a focus on usability, and evaluating performance with the goal of articulating the practical business impact that MT has on translating corporate website content.
Scope of work and methodology
The resources who performed the sample reviews were professional linguists. The role of these linguists was primarily to represent the user who also is a language professional. We call these resources editors. The editors were not predisposed in any way for or against machine-translated content and were used to working with it. They have revised, edited, and commented on the target content segments where applicable. The editors were not asked to directly compare machine and human translation or to perform a service akin to LQA, but rather to evaluate the criteria of the usability of the machine translation as well as the accuracy and reliability. The main principle and perspective was the following: If it’s not broken, don’t fix it. Focus on what’s usable.
Purpose
The purpose of this study is to evaluate and compare a selection of leading machine translation technologies.
What was done
We completed a series of human reviews on five of the leading machine translation (MT) technologies across seven language pairs with a focus on usability, and evaluating performance with the goal of articulating the practical business impact that MT has on translating corporate website content.
Scope of work and methodology
The resources who performed the sample reviews were professional linguists. The role of these linguists was primarily to represent the user who also is a language professional. We call these resources editors. The editors were not predisposed in any way for or against machine-translated content and were used to working with it. They have revised, edited, and commented on the target content segments where applicable. The editors were not asked to directly compare machine and human translation or to perform a service akin to LQA, but rather to evaluate the criteria of the usability of the machine translation as well as the accuracy and reliability. The main principle and perspective was the following: If it’s not broken, don’t fix it. Focus on what’s usable.