Terminology Management in Translation: A Practical Guide
Terminology management is the practice of defining, approving, storing and maintaining the terms your organization uses, so that every writer and translator uses the same word for the same concept in every language. In translation, it usually means a glossary or termbase that is built before the work starts, linked to the translator’s CAT tool and updated as the product changes. This guide shows how that works, what a good glossary entry contains, and how to apply it to technical, automotive, financial, medical, legal and game content.
TL;DR
Terminology management means approving one term per concept and making it available to everyone who writes or translates your content.
Poorly managed terms raise translation costs, create documentation inconsistencies and, in regulated industries, can create safety or compliance risks.
The workflow has five stages: identify terms, research and define them, approve them, distribute them, then maintain them.
Automated extraction gives you a first draft of a glossary, but a person must clean and validate it.
Start before translation begins. Fixing terms after a 100,000-word project is far more expensive than fixing them before.
Table of Contents
What terminology management means in translation and localization
Every company has its own vocabulary: product names, part names, legal concepts, interface labels, internal abbreviations. In one language, colleagues absorb this vocabulary without thinking. In translation, each term needs an approved equivalent, and without one, each translator picks their own.
Terminology management is the discipline that prevents that. It covers collecting terms, defining them, deciding which variant is approved, recording the decision and keeping it current. The international standard ISO 704 sets out the fundamental principles and methods for terminology work, and the field is wider than translation alone: the same approved terms serve writers, support teams, marketing and engineering.
For a translation buyer, the practical output is one of three things:
A glossary: a simple two-column or multi-column list of source terms and approved translations.
A termbase: a structured database in which each term has a definition, status, context and other metadata.
A style and terminology package: the termbase plus rules on tone, capitalization, units and forms of address.
People search for “translation glossary creation”, “terminology database translation” or “terminology management services” and mean roughly the same need: a controlled vocabulary that the translation process respects.
Why it matters: consistency, cost and risk
The business case rests on three effects.
Consistency. Microsoft’s localization guidance says controlled, well-defined terms improve consistency across the product, documentation and marketing, help establish brand voice and reduce later localization time and cost. A user who sees three different Hungarian words for one button, or a technician who reads two names for one part, loses trust in the product.
Cost. Searching for the right term eats time. Research by Austermühl (2001) reported that term searching can take up to 75% of a translator’s time. A Canadian professional starter guide puts experienced translators at 20–25% of their time on terminology research, against 40–60% for inexperienced ones, and notes that systematic terminology management reduces time spent on translation and revision, and therefore costs. When approved terms are already in the file, that research time shrinks, and so does the review cycle.
Risk. According to the Enosema Foundation, poorly managed terminology leads to higher translation costs, documentation inconsistencies and potential safety or compliance risks in regulated industries. A dosage instruction, a torque specification or a contract clause cannot afford two interpretations.
There is a search benefit too. Consistent term usage across organizational content supports product development, communication and search engine optimization. If customers search for one Hungarian phrase and your site uses another, you lose the click.
How the process works, step by step
The Enosema Foundation describes five core stages in formal terminology workflows. Here is how each looks in a real translation project.
1. Identify and extract terms
Gather the source material: manuals, UI strings, previous translations, product databases, legal templates. Automated term extraction can pull out frequent candidates. Microsoft recommends stabilizing source terms early, then extracting candidates and cleaning and pruning the list.
Expect noise. Extraction tools flag common phrases, fragments and one-off names. Remove them. A useful termbase is short and precise rather than long and unreliable.
2. Research and define
For each real term, write a short definition in the source language and find how the target-language market actually says it. Check standards, regulations, competitor-neutral industry publications and the client’s own earlier documents. If two Hungarian variants circulate, record both and mark one as preferred and the other as deprecated.
3. Approve and validate
Terms should be defined and approved together with members of the development team, not by the translator alone. Engineers know what a part is; the translator knows what the local market calls it. The approver should be someone in the client’s company who can say yes or no, ideally a native speaker of the target language who knows the product.
4. Distribute and implement
Approved terms are only useful if they are in front of people at the moment they work. For translation, that means loading the termbase into the CAT tool so that approved terms are highlighted and flagged when the translator deviates. Microsoft’s guidance adds that dedicated terminology solutions integrated with CAT tools can support the standardized TBX format and share termbases with stakeholders.
5. Maintain and update
Products change, regulations change, and so do preferred expressions. Assign an owner, set a review rhythm, and record every change with a date and a reason. Notify everyone who uses the terms when something changes. A termbase that nobody updates becomes a source of errors within a year or two.
How to create a translation glossary
You do not need special software to start. A spreadsheet works for a first version. What matters is the structure and the discipline.
Microsoft’s guidance suggests that an initial termbase can come from automated extraction, after which a terminology expert works with the product team to clean the list and define and approve selected terms. The same page lists the kinds of information an entry can carry: definition, source or context, taxonomy, grammatical usage, date, author, status, examples, exclusions, related terms and product applicability.
For most business glossaries, a smaller set is enough:
Field | What it contains | Why it helps |
Source term | The English term in its base form | Lets the tool match it in new text |
Approved translation | The Hungarian (or German, French) equivalent | The decision you want enforced |
Definition | One or two sentences in plain language | Separates homonyms such as “driver” (software vs. person) |
Context example | A real sentence from your content | Shows usage, case endings and plural forms |
Part of speech and grammar | Noun, verb, gender or other notes | Hungarian suffixes and German gender depend on it |
Status | Approved, proposed, deprecated | Tells the translator what not to use |
Do-not-translate flag | Yes or no | Protects product names and trademarks |
Date and owner | When it was set and by whom | Makes maintenance possible |
Two habits make a glossary work in practice:
One concept, one entry. If “rear axle” and “back axle” both appear in source documents, the source should be fixed first. Translating inconsistent source terms multiplies the problem.
Short definitions. Nobody reads a paragraph while translating at speed.
Glossary, termbase and translation memory: what each does
These three get confused, and the confusion causes real project problems.
A termbase is the central database in which researched terms and their information are recorded so that translators, editors and writers can reuse approved decisions instead of repeating research. A translation memory stores previously translated sentences. The difference matters: a termbase stores information at the term level, while translation memory systems typically work at the phrase level and are not specifically designed to verify terminology consistency.
In practice: a translation memory will remind the translator how you translated a whole sentence last year. A termbase will warn when a single word in a new sentence departs from your approved choice. You want both, and the termbase should be built first, because a memory full of inconsistent terms will faithfully reproduce them.
Tools and standards worth knowing
You do not need to master standards to buy translation, but three names help when you speak with vendors or IT:
ISO 704 covers principles and methods of terminology work.
ISO 30042 (TBX) provides the framework for representing terminological data in TermBase eXchange format. If your glossary is in TBX, you can move it between tools without retyping.
ISO 26162:2012 specifies criteria for designing, implementing and maintaining terminology management systems, meaning the software used to collect, maintain and access terminological data.
The practical advice: keep your terminology in an exchangeable format, store it where more than one person can reach it, and avoid locking your only copy inside one vendor’s private system.
Applying it by industry
The five stages stay the same. What changes is where the risk sits and who must approve.
Technical and product documentation
Technical terminology translation fails most often on near-synonyms: “fitting”, “connector” and “coupling” may be three different parts or one part named three ways. Build the glossary from the bill of materials and the UI strings, and tie each term to a part number or screen where you can. Microsoft’s advice to translate and lock key terms before the rest of localization begins prevents each translator from inventing a different equivalent.
Automotive
Automotive content mixes OEM-specific names, supplier jargon and regulatory language. Keep separate columns for the OEM’s official term, the common workshop term and the abbreviation. Mark which one appears in customer-facing manuals and which in service documentation. Variants used in the workshop but not in the manual should be recorded as acceptable in one context and deprecated in the other.
Financial
Financial terminology translation is sensitive to jurisdiction. A term such as “equity”, “accrual” or “covenant” may map to a different legal concept in Hungary than in the United States. Record the jurisdiction in the definition, and where there is no exact equivalent, record the agreed explanatory rendering rather than leaving it to each translator.
Medical and legal
Here, terminology is a safety and liability question. Use the terms from the relevant regulator, pharmacopoeia or statute, record the source of each decision, and keep a clear approval trail. Our related guide on safety data sheet translation between Hungarian and English shows how fixed phrasing works in a regulated document type. Sensitive source files also need careful handling: see our notes on protecting confidential documents in translation.
Software, e-learning and games
UI labels have length limits and need consistent verbs (“Save”, “Submit”, “Apply” must not drift). Games add character names, invented words and humor, where the glossary should include notes on tone and what must stay untranslated. Courses add another layer, since the same term appears in video, slides and quizzes; our piece on localizing e-learning courses covers that.
A realistic starter plan
If you have no terminology process today, do this before your next translation order:
Pick one product line or document family, not the whole company.
Export the 100 to 300 most frequent and most critical terms from your source files.
Remove duplicates and fix inconsistent source usage.
Have one person in your company approve each term, and one native-speaking specialist approve each translation.
Hand the glossary to your translator before work begins, with a note on which terms are mandatory.
After delivery, collect the translator’s queries and add the answers as new entries.
That sixth step is where a glossary becomes a living asset. Each project teaches the next one.
Perspective
In our experience, a glossary is not paperwork. It is the shortest route to a translation that reads as if your own engineer wrote it. We work as a single translator with more than 15 years of experience in technical, medical, legal, automotive and other specialist fields. That has a practical consequence for terminology: the person who researches your terms is the person who translates your files, so no decision is lost between a terminologist, a translator and a reviewer. Every query goes to you once, and the answer applies to the whole project. We use advanced CAT tools and quality assurance checks, which means approved terms can be applied and verified across large volumes without slowing delivery. When a deadline is tight, we are often reachable after regular hours, so an open terminology question does not hold up a release. Our advice is to start with a small, approved glossary instead of waiting for a perfect one. Send us what you have, even a messy spreadsheet. We will help turn it into something the whole project can rely on. Bernadett
How we can help
We translate into Hungarian from €0.05 per word, and into English, German or French from €0.07 per word. Post-editing of machine translation is available too: full post-editing from €0.04 per word and light post-editing from €0.03 per word. Localization is quoted per project. If you have a glossary, send it with your files and we will work to it. If you do not, we can flag the recurring terms during the first batch so that you can approve them once.
To discuss a project or request a quote, go to Wordsbyme.
FAQ
What is the difference between a glossary and a termbase?
A glossary is usually a simple list of terms and approved translations. A termbase is a structured database where each entry also holds definitions, status, context and other metadata. For a small project, a glossary is enough; for several languages or many contributors, a termbase scales better.
When should terminology work start?
Before translation begins, ideally while the source content is still being written. Microsoft’s guidance recommends planning terminology from the start of a product project, because early decisions improve both the source product and its localized versions.
Can software create my glossary automatically?
Software can produce a first list of candidate terms from your documents. That list still needs manual cleaning and validation by someone who understands the subject, as Microsoft’s localization guidance states. Treat automation as the starting point, not the result.
How big should a first glossary be?
Small enough that someone will actually maintain it. For a single product line or document family, a few hundred carefully chosen terms usually does more than thousands of unreviewed ones. Add entries as real questions come up during translation.
Does terminology management matter if I use machine translation?
Yes. Machine output tends to vary term choices between segments, so approved terms matter even more when a post-editor reviews it. A glossary gives the post-editor a clear standard to correct toward, which shortens review.
Sources
Terminology management: Principles and methods (Enosema Foundation)
Managing terminology (Microsoft globalization documentation, GitHub)
Terminology Management for Translators (Fernández-Parra, 2026)
Terminology Management for Translation: A Starter Guide (COTSOES)
Terminology Management in the Localization Industry (LISA survey analysis)
Recommended
Comments