June 7, 2022

Possibilities of Improved Terminology Adaption Features in Machine-Translation Post-Editing

By admin-emal-17 Views-No Comment

Get the full paper here: Terminology Adaption Features in MT-PE

As a freelance translator into English for several German higher education
institutions (HEI), I face the challenge of aligning terminology with client
decision-makers and authors. Furthermore, the widespread use of machine
translation (MT) by staff at higher education institutions and the integration
of terminology adaption into several MT engines has strengthened the
business case for finding a common data model and data source for
multilingual glossaries, which coincides with the aims of ISO/TC37 –
Language and terminology.

Existing terminographic resources are currently not stored and shared
consistently with all user groups, such as authors, translators and content
editors, requiring them to rely on repetitive lexicographic research. While
there are some organizational reasons, the technology is mature enough but
not yet sufficiently integrated.

For example, computer-assisted translation (CAT) tools still struggle to adapt
terminology automatically, even though several connected MT systems
already support terminology adaption. This submission aims to illustrate the
value of MT adaption to the machine translation post-editing (MT-PE)

workflow from a user perspective, hoping to convince stakeholders to
prioritize terminographic features and develop some user recommendations
for standards covered by ISO/TC37/SC 3.

Based on the argument that term validation is a significant effort, both in
human translation (HT) and in machine translation post-editing (MT-PE), and
that lack of a “single source of truth” and interoperability between computer
assisted (CAT) tools and other terminology-consuming software (such as
authoring tools) are the limiting factors, this submission provides an overview
of terminology features in selected CAT tools, describes the development of
a custom terminology management system, attempts to measure productivity
gains from term adaption in different translation tools and proposes the term
rate as the starting point for an additional indicator for productivity-relevant
MT quality.

Leave a Comment