How to Maintain an AI Translation Glossary Over Time
You built a translation glossary six months ago. Product names, technical terms, brand language — all documented and loaded into your AI translation tool. But since then, your company launched two new products, renamed a feature, and expanded into a new market. Is your glossary still accurate?
Most teams set up a glossary once and forget about it. That works for a while, then the glossary becomes a source of errors instead of a guardrail against them. This article covers how to maintain your AI translation glossary as a living resource that keeps up with your business.
Why Glossary Maintenance Matters
A translation glossary tells the AI engine how to handle terms that it would otherwise translate generically. "CloudSync" should stay "CloudSync," not become "cloud synchronization." "First responder" in a safety document should be translated differently than in a press release. Without a glossary, the AI makes its best guess, and that guess changes from document to document.
But a glossary only works when it reflects current reality. An outdated glossary is worse than no glossary at all because it introduces errors you would not otherwise have. If your glossary says to translate "Project Atlas" a certain way, but the product was renamed to "Atlas Platform" three months ago, every translation will carry that error until someone updates the entry.
Setting Up a Maintainable Glossary
Start With the Right Structure
Before thinking about maintenance, make sure your glossary has a structure that supports it. At minimum, each entry should include:
- Source term: The word or phrase in the original language
- Target translation: The approved translation or a "do not translate" flag
- Part of speech: Noun, verb, adjective — some tools use this to apply the term correctly in context
- Context note: When and where this translation applies (for example, "in safety documentation" or "marketing context only")
- Date added: When the entry was created
- Last reviewed: When someone last verified the entry is still correct
- Status: Active, deprecated, or under review
This structure gives you the information you need to maintain the glossary efficiently. When you see that an entry was last reviewed eight months ago and your company has since rebranded, you know to check it.
Choose the Right Tool
A shared spreadsheet works for small glossaries with one or two contributors. For larger teams or glossaries with hundreds of entries, consider a dedicated tool or database. Some translation management systems include glossary features with version history, approval workflows, and integration with translation engines.
The key requirement is that the glossary needs to be easy to update. If updating an entry requires finding a file, downloading it, making the change, uploading it, and reloading the tool, people will skip updates. Friction kills maintenance.
Maintenance Triggers
Do not wait for a scheduled review to update your glossary. Set up triggers that prompt updates as part of your normal workflow.
Product and Feature Changes
Whenever your company launches, renames, or retires a product or feature, update the glossary the same day. Add this step to your product launch checklist. If marketing announces "DataVault" and the glossary still says "SecureStore," every translation produced between the announcement and the glossary update will be wrong.
Assign glossary updates to the same person who handles product documentation updates. They already know about changes and can make glossary updates alongside doc updates.
New Document Types and Audiences
When you start translating a new type of document or targeting a new audience, review the glossary for relevance. A glossary built for technical documentation may not cover the marketing terms needed for a new campaign. A glossary built for employee communications may not address customer-facing terminology.
Before starting a new translation project, scan the glossary for gaps. Are there terms in the source document that the glossary does not cover? Add them before translating, not after reviewing.
Translation Review Feedback
Every time a reviewer corrects a term during post-editing, that correction should feed back into the glossary. This is the most powerful maintenance trigger because it is based on real errors found in real translations.
Make it easy for reviewers to flag glossary updates. Add a "glossary update needed" column to your review tracking, or set up a shared document where reviewers can log terms that need attention. Review and process these updates weekly.
Market and Language Expansion
When you start translating into a new target language, the glossary needs a new column. Some terms that work in one language need different handling in another. "Do not translate" rules may need exceptions. Cultural considerations may require different approaches for different markets.
Plan a glossary expansion sprint when entering a new market. Have a native speaker review existing glossary entries for cultural appropriateness and add market-specific terms.
Scheduled Maintenance
In addition to event-driven updates, schedule regular glossary reviews. The frequency depends on how fast your terminology changes.
Monthly Review (Fast-Moving Teams)
If your company frequently launches products, updates features, or operates in a rapidly changing industry, review the glossary monthly. Focus on:
- Entries added in the past month: Are they correct and complete?
- Entries flagged during review: Were the corrections processed?
- Recent translation errors: Are there patterns that suggest missing glossary entries?
Quarterly Review (Stable Teams)
If your terminology changes slowly, a quarterly review is sufficient. Use this time to:
- Check all entries added or modified since the last review
- Verify that product names and feature names are current
- Remove deprecated entries that are no longer relevant
- Update context notes based on new document types or audiences
Annual Audit
Once a year, do a full glossary audit. Read every entry, verify it against current product documentation and brand guidelines, and remove anything that is no longer accurate. This is also a good time to assess whether your glossary tool is still adequate for your needs.
Handling Glossary Conflicts
As teams grow, glossary conflicts emerge. Marketing wants "platform" translated one way; engineering wants it another. The legal team has specific requirements for compliance-related terms that differ from how the rest of the company uses them.
Resolution Process
- Document the conflict: Write down both proposed translations and the rationale for each.
- Identify the context: Is the term used differently in different document types? If so, both translations may be correct in different contexts.
- Make a decision: Assign one person (or one team) as the final authority on glossary decisions. This prevents endless debate.
- Add context notes: If both translations are valid in different contexts, document when to use each one.
- Communicate the decision: Make sure all reviewers and translators know about the resolution.
Version Control
Keep a change log for your glossary. When a term is added, changed, or deprecated, record the date, the reason, and who made the decision. This serves two purposes: it provides an audit trail, and it helps new team members understand why certain decisions were made.
Some translation management systems handle version control automatically. If you are using a spreadsheet, add a "changelog" tab or use a tool with revision history.
Handling Regional Variations
When your company operates across multiple regions that share a language but use different conventions, the glossary needs to account for this. European Spanish and Latin American Spanish use different terms for common business concepts. Canadian French and European French have distinct vocabulary. Simplified and Traditional Chinese differ in more than just character sets.
If you translate for multiple regional variants of the same language, add a column to your glossary for each variant. This prevents the common mistake of using a term appropriate for one region in a document destined for another.
Integrating the Glossary With Your Translation Workflow
AI Engine Integration
Many AI translation tools support custom glossaries or terminology lists. Load your glossary into the tool so that it applies your approved terms automatically. This reduces the number of corrections reviewers need to make.
Check your tool's documentation for glossary support. Some tools accept simple CSV uploads. Others have dedicated glossary management interfaces. The process varies, but the benefit is the same: the AI produces better first drafts.
For example, DeepL supports glossary features that allow you to specify how certain terms should be handled. Azure AI Translator and Google Cloud Translation both offer custom terminology capabilities.
Human Reviewer Access
Make the glossary easily accessible to everyone who reviews translations. If reviewers have to search for the glossary file or ask someone to share it, they will skip terminology checks. Put it in a shared location that everyone on the team can access and knows about.
Common Maintenance Mistakes
Adding Terms Without Context
"WidgetPro" -> "Do not translate" is a start, but it does not tell you when to apply this rule. What about "WidgetPro Enterprise"? What about the abbreviation "WP"? Add context notes that cover common variations and edge cases.
Ignoring Deprecated Terms
When a product is renamed, do not just add the new name. Mark the old name as deprecated and note the replacement. Otherwise, someone reviewing an old document might "correct" a valid translation by reverting to the deprecated term.
Not Reviewing After Tool Changes
When you switch AI translation tools or update to a new engine version, re-test your glossary. Different tools handle glossary terms differently. Some apply terms strictly; others treat them as suggestions. A glossary that worked perfectly with one tool may produce unexpected results with another.
Allowing the Glossary to Grow Without Cleanup
Every term added to the glossary increases the chance of conflicts and slows down lookup. Periodically remove entries that are no longer relevant. If you have not used a term in a translation for over a year and it is not a core product name, consider removing it.
Measuring Glossary Effectiveness
Track how often reviewers correct glossary terms during post-editing. If the correction rate for glossary terms is low, your glossary is working. If it is high, either the glossary entries are wrong or the AI tool is not applying them correctly.
Also track how often reviewers find terms that should be in the glossary but are not. This tells you where the gaps are. Every missing-term correction is a candidate for a new glossary entry.
Over time, a well-maintained glossary reduces review time and improves consistency across all your translated documents. The maintenance effort is modest — especially with event-driven updates — and the payoff grows with every translation your team produces.
Getting started does not require a large investment. Open a spreadsheet, list the 20 terms that matter most to your business, and add their approved translations. Load the glossary into your translation tool. After the next translation project, update the glossary with any corrections from the review process. Repeat. Within a few months, you will have a glossary that meaningfully improves translation quality and reduces review time.
For AI-assisted document translation across PDF, DOCX, PPTX, and XLSX files, tools like Jitan Translate provide the initial draft. Your glossary then guides the review process and feeds corrections back into the system for continuous improvement.
Source: https://jitantranslate.com/en/blog/pdf/translate-pdf-without-losing-formatting/
