Translation Memory (TM)
A database that stores previously translated segments for reuse and consistency.
Definition
Translation Memory (TM) is a database that stores source text segments paired with their translations. When similar or identical text appears in new content, the TM suggests existing translations, saving time and ensuring consistency. TM systems typically show match percentages: 100% (exact match), 101% (context match), or fuzzy matches (75-99% similar). TM is a core feature of professional translation workflows and TMS platforms.
Examples
- →Exact match: 'Save changes' appears again → TM suggests the same translation
- →Fuzzy match: 'Save all changes' matches 'Save changes' at 85%
- →Context match: Same segment in same file location → higher confidence
Frequently Asked Questions
How does translation memory save money?
TM reduces costs by reusing existing translations. If you've already translated 'Login' in 50 places, updating your app won't require re-translating that word. Enterprise TMs can have millions of segments, dramatically reducing new translation volume.
What is the difference between TM and MT?
Translation Memory (TM) stores human translations for reuse. Machine Translation (MT) generates new translations using AI. They're complementary: use TM for known content, MT for new content, then human review for quality.