A segment is a defined unit of source text, usually a sentence, heading, list item, or UI string, that a CAT tool isolates for translation. In localization files and translation memory systems, segmentation rules determine how text is divided, stored, and matched, which directly affects leverage, consistency, context visibility, and the ability to reuse previous translations across projects and language pairs.
Segments are the default view in CAT tool editing panes, translation memory matches, QA checks, and word-count reports. Project managers inspect segmentation in file-preparation steps to adjust rules for source documents such as legal contracts, UI strings, or subtitles so that text units are not too long or too short.
Changing segmentation mid-project can alter context, split sentences in ways that mislead translators, and reduce translation memory matches, creating inconsistent translations across entire files. Translators may also translate isolated short segments too literally when surrounding source context is not shown, especially for pronouns, lists, UI strings, headings, or product names.
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 →