Every brand has words that must always be translated the same way: product names, feature names, legal terms, words that must never be used. A termbase is where those decisions live. When it works well, translators, reviewers and AI engines all speak with one voice. When it is missing, the same button gets three different names in three different screens.
What Is a Termbase?
A termbase (short for terminology database) is a structured collection of approved terms and their translations, usually with extra information such as definitions, part of speech, usage notes, examples and status. It is more than a simple two-column glossary. A good termbase also records which terms are forbidden and why.
In a CAT tool or TMS, the termbase is connected to the editor. As translators work, matching terms are highlighted with the approved translation, and QA checks flag segments where the wrong term was used.
Termbase vs Glossary vs Translation Memory
| Glossary | Termbase | Translation memory | |
|---|---|---|---|
| Unit | Single terms | Single terms with metadata | Whole sentences or segments |
| Format | Often a spreadsheet | Database, exchanged as TBX, CSV or XLSX | Database, exchanged as TMX |
| Purpose | Reference list | Enforce consistent terminology | Reuse previous translations |
| Used by AI | Rarely | Yes, to guide engines | Yes, applied before AI |
What Is a TBX File?
TBX (TermBase eXchange) is the international standard format for exchanging terminology between tools. It is defined in ISO 30042. A TBX file is XML-based and organizes data in concept entries: one concept, with terms in several languages, each with its own notes and attributes.
In practice, many teams exchange terminology as CSV or XLSX because spreadsheets are easier to edit. TBX is most useful when you move a large, structured termbase between professional tools and want to keep all the metadata.
What to Put in a Termbase Entry
- Source term and its approved translations.
- Definition or short explanation, so translators understand the concept.
- Part of speech, which matters when a word can be a noun or a verb.
- Usage note, for example "use in UI only" or "capitalize".
- Status: preferred, admitted or forbidden.
- Context or example sentence that shows the term in use.
- Domain or product, if the same word means different things in different products.
How to Build a Termbase Step by Step
- Extract candidate terms from your UI, documentation, marketing pages and existing translations.
- Prioritize. Start with product names, features, legal terms and the 100 to 200 terms that appear most often.
- Define and approve each term with product owners and in-market reviewers.
- Translate and validate terms in each language before large projects start.
- Connect it to your tools so translators see terms while they work and QA checks enforce them.
- Maintain it. Add new terms with every release and retire outdated ones.
For a deeper look at governance and workflows, read our guide to terminology management and translation glossaries.
Why Termbases Matter Even More With AI
Machine translation and LLMs produce fluent text, but they do not know your product vocabulary. Without guidance, they may translate a feature name differently in every file. Feeding your termbase to the engine, and checking the output against it, is one of the most effective ways to make AI translation safe for customer-facing content.
Common Termbase Mistakes
- Too many entries with no priority, so translators stop checking.
- Terms without definitions or context.
- No forbidden terms, so old product names keep coming back.
- A termbase that lives in a spreadsheet no one opens during translation.
- No owner, so the list is never updated.
Termbases in WordBeam
WordBeam includes a term base on every plan. Terms are highlighted in the CAT editor, used to guide AI engines such as DeepL, GPT and Claude, and enforced by QA checks before delivery. You can import terminology from spreadsheets such as CSV or XLSX and keep one source of truth for every language.
Frequently Asked Questions
What is the difference between a termbase and a glossary?
A glossary is usually a simple list of terms and translations. A termbase adds structure and metadata, and is connected to translation tools so terms can be suggested and checked automatically.
How many terms should a termbase have?
Quality matters more than size. Start with a few hundred high-impact terms and grow it with each project.
Can I import a TBX file into any CAT tool?
Most professional tools support TBX or a spreadsheet import. Check which fields are mapped, because some metadata may not transfer.