Machine Learning in Translation introduces machine learning (ML)
theories and technologies that are most relevant to translation
processes, approaching the topic from a human perspective and
emphasizing that ML and ML-driven technologies are tools for humans.
Providing an exploration of the common ground between human and machine
learning and of the nature of translation that leverages this new
dimension, this book helps linguists, translators, and localizers better
find their added value in a ML-driven translation environment. Part One
explores how humans and machines approach the problem of translation in
their own particular ways, in terms of word embeddings, chunking of
larger meaning units, and prediction in translation based upon the
broader context. Part Two introduces key tasks, including machine
translation, translation quality assessment and quality estimation, and
other Natural Language Processing (NLP) tasks in translation. Part Three
focuses on the role of data in both human and machine learning
processes. It proposes that a translator's unique value lies in the
capability to create, manage, and leverage language data in different ML
tasks in the translation process. It outlines new knowledge and skills
that need to be incorporated into traditional translation education in
the machine learning era. The book concludes with a discussion of
human-centered machine learning in translation, stressing the need to
empower translators with ML knowledge, through communication with ML
users, developers, and programmers, and with opportunities for
continuous learning.
This accessible guide is designed for current and future users of ML
technologies in localization workflows, including students on courses in
translation and localization, language technology, and related areas. It
supports the professional development of translation practitioners, so
that they can fully utilize ML technologies and design their own
human-centered ML-driven translation workflows and NLP tasks.
