Alignment is the process of matching corresponding source and target segments from previously translated documents, such as approved PDFs, Word files, or legacy manuals, to create or update a translation memory. It links sentences or sub-sentence units across languages so that CAT tools can later retrieve them as potential reuse candidates in subsequent new projects and reduce duplicate translation effort.
Alignment is used when a client provides previously approved translated documents without a translation memory, or when merging legacy content into an existing TM. A project manager or language technologist runs alignment in a CAT tool, then reviews and cleans the resulting pairs before storing them for reuse in future jobs.
Automatic alignment often mismatches segments when source and target texts have merged sentences, deleted passages, or different paragraph counts, producing translation units with wrong translations paired together. If these errors enter the TM, future projects may receive incorrect fuzzy matches, undermining consistency, triggering QA failures, and requiring expensive manual cleanup.
For creative and consumer content the failure mode is flatness, not inaccuracy — and flatness passes an accuracy check. Terms here also carry technical constraints (reading speed, character budgets) that are part of the brief.
Angel Translation locks terms like this into a project terminology base before expert review begins, so the same source term resolves to the same target term across every document in a submission — and stays consistent in the next revision. See the 8-step AI + expert workflow.
Send us the files. Acknowledged in 4 hours, quoted in 24.
Request a Quote →Content Globalization solutions →