Frequently asked questions
Clear answers about translation and language technology.
Translation memory ownership, workflow automation, machine translation, DTP, multilingual SEO, specialist translation and accessibility: clear, practical answers from a team that has managed multilingual content for more than 20 years.
Topics: Translation memory: ownership, portability and migration · Translation workflow management and automation · Machine translation and post-editing (MTPE) · Multilingual desktop publishing (DTP) · Multilingual SEO and international search · Marketing translation and transcreation · Technical, manufacturing and medical device translation · Financial and legal translation · Accessibility: PDF remediation, testing tools and subtitles
Translation memory: ownership, portability and migration
Who owns the translation memory when you leave a translation vendor or TMS?
The content in a translation memory comes from your documents, so it should belong to you. In practice, ownership depends on your contract: some vendors treat the memory as part of their service and make exports difficult. Check that your agreement states that the memory, glossaries and termbases are yours and can be exported in standard formats such as TMX and TBX at any time, without extra fees. That is how DEMA works: your translation memory, glossaries and termbases remain yours and can be exported in TMX or TBX whenever you ask, at no cost.
How do we negotiate translation memory portability before signing a TMS or vendor agreement?
Ask for three things in writing: that you own the translation memory and terminology built from your content; that you can export them at any time, not only at contract end, in standard formats (TMX for memories, TBX or Excel for terminology); and that exports carry no extra fees or delays. Also agree what happens to the data after termination. If a supplier hesitates on any of these points, treat it as lock-in.
Our provider says our translation memory is proprietary. Is that enforceable?
It depends on what your contract says, so check it with your legal team. Many agreements are silent on ownership, and the translations were paid for and created from your own source content. Even when tools or formats are proprietary, the linguistic data can usually be exported. A clear export clause in every future contract avoids the question entirely.
How do we migrate 10 years of translation memory from one TMS to another without losing segment alignment?
Export everything in TMX, including metadata such as creation dates, users and project fields, and keep the original files as a backup. Before importing, clean the data: remove duplicates and empty or misaligned segments, normalise language codes and tags, and resolve conflicting translations. Then import into a test project, run analyses against recent documents and compare match rates with the old system before switching production work. Before a migration, DEMA’s language technology team cleans, converts and tests legacy memories, so years of approved translations keep paying off in the new system.
What are the biggest risks when migrating translation memory from one TMS to another?
The most common problems are lost metadata, broken inline tags, wrong language codes, duplicated or conflicting segments, and segmentation rules that differ between tools so that old matches stop being recognised. Large memories with millions of translation units also take time to clean and test. A validation step on a sample, with match-rate comparisons, catches most issues before they affect live projects.
How can we validate migrated translation memories before using them in production?
Use a short checklist: compare unit counts before and after migration, spot-check segments across languages and years, run a tag and character-encoding check, and analyse a set of recent documents in both systems to compare leverage. Only release the memory to production once match rates and sample quality are consistent. For clients moving between systems, DEMA runs this validation as part of the migration, including match-rate comparisons on their own recent documents.
Our translation memory is fragmented across tools. How do we consolidate it?
Collect every memory you have, from all tools and suppliers, in TMX. Merge them per language pair, remove duplicates, and decide which version wins when two translations conflict, usually the most recent approved one. Align past translations that never made it into a memory. The result is one clean memory per language pair that every supplier can use. If several vendors keep working on your content, a central, vendor-independent memory keeps it consistent. DEMA consolidates and cleans fragmented memories, and when several suppliers keep working on your content, 1memory centralises translation memories across all of them.
What is the best way to manage translation memory across business units that share some terminology but not all?
Use one shared master memory and glossary for the terms and sentences everyone uses, plus separate project or business-unit memories for content that is specific to each unit. Mark terms in the glossary by domain or product line, so linguists know which variant applies. This keeps shared language consistent without forcing every unit into the same wording.
What is the difference between a translation memory and a glossary?
A translation memory stores whole sentences you have already translated and approved, and reuses them when the same or similar sentences appear again. A glossary, or termbase, controls individual terms: product names, technical terms and words that must always be translated in a certain way. You need both: the memory keeps wording consistent, the glossary keeps terminology right in every new sentence. Read more about terminology management →
How do we set up a translation memory and glossary so multilingual content stays consistent?
Start with the content you already have: align past translations into a memory and extract your key terms into a glossary, then have your experts approve the terms. From then on, every project uses both, every approved translation is added back to the memory, and the glossary is updated when products change. This is also how TUBE, DEMA’s translation solution for technical documentation, starts: our language technology team builds both from your existing content.
How does translation memory management improve workflow efficiency?
Reused sentences are not translated again, so you pay less for repeated content and receive translations faster. Linguists work with approved wording instead of researching it again, reviewers correct fewer inconsistencies, and every update builds on the previous one. The effect grows over time: the more you translate with a well-maintained memory, the more you reuse.
Does our team need translation memory software or a translation management system (TMS)?
If you mainly need consistent, cheaper translations and work with a language partner, you do not need to buy software: your partner works with translation memory in its CAT tools and gives you the memory. A TMS makes sense when you manage many suppliers, internal reviewers and integrations yourself. Either way, insist that the memory is yours and exportable.
How do translation vendors price translation memory leverage, and are the discounts negotiable?
Most providers analyse each file against the memory and charge different rates for new words, fuzzy matches (similar sentences in bands such as 75 to 99 percent), exact matches and repetitions. The bands and discounts vary between suppliers and are usually negotiable, especially for recurring programmes. Ask for the analysis behind each quote and for regular leverage reporting, so you can see whether reuse is growing. DEMA reports reuse transparently, release after release, so you can see exactly where the savings come from.
What is a realistic translation memory leverage rate?
It depends on how repetitive your content is and how long you have been building the memory. A first project usually has little reuse; regular updates of manuals, software strings or product documentation often reuse a large share of previous content. The most reliable way to estimate it is to analyse a sample of your files against your existing memory. If you want a realistic estimate before committing, DEMA can analyse a sample of your files against your existing memory.
Translation workflow management and automation
How can we automate translation workflows efficiently?
Automate the repetitive steps: file preparation, translation memory and terminology lookup, machine translation where it fits, automated quality checks and file delivery. Keep people where judgement matters, in post-editing, review and approval. Start with one content type that repeats often, measure turnaround and cost, then extend. With TUBE, DEMA brings your glossary, your translation memory and machine translation controlled by both into one workflow, managed with our linguists.
How do we implement translation management without disrupting our current workflow?
Start where translation already happens: keep your file formats, authoring tools and approval steps, and add translation memory, terminology and quality checks around them. Begin with a pilot on one product line or document type, compare the results with your current process, and roll out step by step. A good partner adapts to your workflow, not the other way round.
Can we integrate translation management into our development workflow and sync translation memory with our pipeline?
Yes. Software strings and documentation can be exchanged in standard formats such as XLIFF, JSON, YAML or properties files, so new and changed strings are sent for translation automatically and returned in the same format. Because the memory recognises unchanged strings, only new content is translated. Agree on the file format, the trigger for each translation round and how translated files are merged back before you start. DEMA proposes the exchange format and set-up that fit your repository and release cycle, so translation follows your development process instead of slowing it down.
How do we reduce translation turnaround times and costs through automation?
The biggest savings come from not translating the same content twice (translation memory), not fixing the same terminology repeatedly (glossaries), using machine translation where the content allows it, and catching errors automatically before review. Clean source files also save time. Measure cost per project and turnaround before and after, so you can see what each change delivers.
What is the fastest way to get multilingual content live in a new market without an existing translation memory?
Start by aligning any translations you already have, even old ones, and agree a short glossary of key terms before translation starts. Prioritise the content customers need first, use machine translation with post-editing where it fits, and build the memory from day one so every later update is faster. When clients enter a new market, DEMA helps plan the first release: aligning existing translations, building the glossary and choosing where machine translation can safely speed things up.
What are best practices for managing translation workflows in global marketing teams?
Keep a central glossary and brand style guide, give each market a reviewer who approves terminology early, brief translators on the campaign goal rather than only the text, and use transcreation for headlines and claims. Plan local review into the schedule and feed every approved change back into the translation memory, so the next campaign starts from approved wording.
How do scalable translation services help global companies maintain brand consistency?
By working from the same assets for every market: one glossary, one style guide per language and one translation memory that grows with every project. Specialist linguists, automated terminology checks and a documented review process keep the tone and the product language consistent, even when volumes and the number of languages grow.
What is a practical workflow for translation quality review in enterprise localization projects?
A proven sequence is: translation by a qualified linguist with the memory and glossary, automated checks for numbers, tags, terminology and consistency, revision by a second linguist, and in-country review by your local teams for high-visibility content. Every correction is fed back into the memory and glossary. The ISO 17100 standard describes this kind of process. It is the same sequence DEMA follows under its ISO 17100 certified quality management.
How do enterprises measure time and cost savings from automating translation workflows?
Compare a baseline with the automated process on the same content types: cost per word or per project, share of reused content, turnaround from request to delivery, number of review rounds and number of corrections. Leverage reports from the translation memory show reuse growing over time; ask your provider to share them regularly. Leverage and reuse reports are part of every DEMA programme, so the savings from automation are visible and measurable.
Machine translation and post-editing (MTPE)
How do translation agencies integrate machine translation post-editing into their services?
A professional MTPE workflow prepares the source text and linguistic assets first, generates a machine draft with the right engine, has a qualified linguist post-edit it for meaning, terminology and style, and runs quality checks before delivery. The level of post-editing depends on the use of the content. In DEMA’s workflows, machine translation is always managed by people: our linguists decide when it fits, post-edit the output and sign off every delivery. Read more about machine translation post-editing →
What are the main challenges in machine translation post-editing?
Machine output can be fluent and still wrong: omissions, added content, mistranslated terms, wrong numbers or units, and inconsistent terminology across a document. Post-editors also need clear guidelines on how far to edit. Controlling machine translation with your glossary and translation memory, and running automated checks, reduces these risks considerably.
Which machine translation solutions reduce manual process overhead for mid-market companies?
The most effective set-ups combine machine translation with your own translation memory and glossary, so the engine uses your approved wording and terms, and route the output to qualified post-editors. This removes most manual retyping and terminology research while keeping quality under control. For technical documentation, TUBE applies exactly this approach, with machine translation controlled by your own glossary and memory.
Where can enterprises find advanced machine translation and post-editing (MTPE) providers?
Look for a provider that assesses whether your content suits machine translation before using it, controls the engine with your terminology, uses trained post-editors and documented quality checks, and gives you ownership of the resulting memory. Certification to ISO 17100 for translation, or ISO 18587 for post-editing, is a useful indicator. DEMA provides MTPE under ISO 17100 certified processes for technical, financial and legal content, and the resulting memory always remains yours.
What can artificial intelligence not do in translation?
AI can produce a fluent first draft quickly, but it does not take responsibility for the result. It can miss context, legal meaning, safety implications, cultural nuance and brand tone, and it can invent or omit content. That is why qualified linguists stay in control: they decide when AI fits, check its output and sign off every delivery.
Multilingual desktop publishing (DTP)
Which translation agencies offer desktop publishing (DTP) services?
Many agencies translate only the text and leave the layout to you. Look for a provider with its own DTP specialists who prepare the source files, rebuild the layout in the original application, such as InDesign, FrameMaker or Illustrator, and check the final document in every language before delivery. DEMA’s in-house DTP team works this way, in InDesign, FrameMaker, Illustrator and other native formats.
What is a DTP file?
In translation, DTP files are documents created in desktop publishing applications such as Adobe InDesign, FrameMaker, Illustrator or QuarkXPress, as opposed to plain text or Word files. The .dtp extension itself is used by a few older publishing programs. DTP files are best translated in their native format, so the layout can be rebuilt in every language.
Can desktop publishing be automated?
Parts of it can: text extraction, translation memory, styles and templates reduce manual work considerably. But text expansion, line breaks, right-to-left scripts and images with text still need a DTP specialist to check and adjust each language version. The best results combine automation with a final visual check of every page. That is why DEMA pairs automation with a visual check of every page in every language before delivery.
What should we look for in multilingual desktop publishing solutions?
Check that the provider works in your native formats, supports all your target scripts, including right-to-left languages, uses translation memory so updates are cheaper, and checks every page visually before delivery. For teams, easy adoption matters more than features: files should come back ready to print or publish without extra work on your side.
Can publishing teams use InCopy for multi-language document management?
InCopy works well for editorial workflows linked to InDesign, but multilingual version control is easier to manage at file level: one InDesign or IDML file per language, translated from the approved source and kept under clear naming and versioning. DEMA works directly with InDesign and InCopy files and returns them in the same structure, ready for your team.
Multilingual SEO and international search
What are the benefits of using a specialised multilingual SEO provider?
Translated pages do not rank automatically. A specialised provider researches the keywords people actually use in each market, translates and adapts titles, descriptions and content around them, and checks the technical set-up, such as hreflang and language structure. You get content that reads naturally and is found.
How do international SEO services handle multilingual content effectively?
Keyword research is done separately for each language and market, not translated from English. Content is then translated or transcreated around those keywords by native linguists, metadata is localised, and technical elements such as hreflang tags, language-specific URLs and internal links are checked so search engines show the right version to the right audience.
Our website exists in several languages and Search Console shows hreflang errors, duplicate content and wrong country targeting. Who can fix this?
These problems usually come from missing or inconsistent hreflang tags, pages that are not fully translated, and duplicate pages across languages. An audit identifies them page by page; the fix combines technical corrections, such as consistent hreflang and canonical tags, with properly localised content and metadata for each market.
Which translation services offer SEO-optimised content writing in multiple languages?
Look for a provider that combines native copywriters or translators with keyword research for each market, so articles are written or adapted around real search terms instead of translated word for word.
Marketing translation and transcreation
Which agencies specialise in translating marketing and advertising content?
Marketing translation needs linguists who write well in their own language, understand the brand and know the target market. Look for native marketing translators and copywriters with experience in your sector, and for transcreation when slogans and claims need to work in another culture.
Which agency does transcreation of marketing campaigns into other languages?
Transcreation recreates a message so it has the same effect in another culture, rather than translating it word for word. A good transcreation brief covers the audience, the channel and the goal of each piece, and providers often deliver alternative versions with back-translations so your team can choose.
What is multilingual marketing?
Multilingual marketing means communicating with customers in their own language and cultural context in every market: websites, campaigns, social media, product content and customer communication. It combines translation, transcreation, local SEO and a consistent brand voice across languages.
Technical, manufacturing and medical device translation
How do I find qualified specialist translators for the manufacturing industry?
Look for translators who have worked with your type of machinery and documentation, a provider certified to ISO 17100, which requires qualified translators and a second-linguist revision, and a process that uses your terminology and translation memory from the first project. DEMA selects its technical translators by subject area, from machinery and automation to CE documentation, and works under an ISO 17100 certified process with your glossary from the first project.
How do I order a specialist translation for the manufacturing industry?
Send the provider your files, ideally in their original format, with the target languages, the deadline and any existing translations or glossaries. A good provider analyses the files against your memory, quotes on that basis, assigns translators with experience in your sector and delivers the translation in the same format, ready to use. This is exactly how a manufacturing translation project runs at DEMA: an analysis against your memory, a quote based on it, specialist translators and delivery in your original format.
What is required for medical device documentation translation under the EU MDR?
The MDR requires instructions for use, labelling and other information for users to be available in the official languages of the countries where the device is sold. Translations must be accurate and consistent with the technical documentation, which means specialist medical translators, controlled terminology and a documented review process. For medical device manufacturers, DEMA translates IFUs, labelling and technical documentation with specialist medical translators, controlled terminology and second-linguist revision. Read more about medical device translation →
How do I order a specialist translation in the life sciences sector?
Share the document type, the target markets and any regulatory requirements, together with existing translations and terminology. The provider should assign translators with life sciences experience, use your approved terminology and run a second-linguist revision before delivery, as required by ISO 17100. DEMA handles life sciences projects this way, assigning translators with sector experience and keeping your approved terminology consistent across every document.
Financial and legal translation
What are the challenges of translating financial documents?
Financial texts combine precise terminology, regulatory wording and numbers that must match the source exactly. Different accounting standards and legal systems use different concepts, formats for dates and figures vary between countries, and deadlines are often tied to reporting cycles. Specialist translators, consistent terminology and automated number checks are essential. DEMA’s financial translators combine sector expertise with terminology control and automated checks on every figure before revision.
Why does every detail in financial translations have to be right?
Because financial documents are used to make decisions and are often subject to regulators, auditors and investors. A wrong figure, a mistranslated term or an inconsistent disclosure can create legal and reputational risks. That is why financial translation needs specialist translators, a second-linguist review and checks on every number. That is why every financial translation at DEMA is revised by a second linguist and checked for numbers, terminology and consistency before delivery.
When are financial translations needed?
Typically for annual and interim reports, financial statements, prospectuses, fund documents, investor presentations, audit reports, contracts and regulatory filings, whenever a company reports to investors, authorities or partners in another language or country.
How do I find qualified financial translators and order a secure financial translation?
Choose a provider whose translators specialise in finance, that works under ISO 17100 with a second-linguist revision, and that handles confidential documents securely. When ordering, share the documents, target languages, deadline and any existing terminology, and agree confidentiality requirements before the project starts. DEMA works with finance specialists under ISO 17100 and agrees terminology and confidentiality requirements with each client before the first document is translated.
How do asset managers ensure regulatory document accuracy across multiple languages?
By using one approved terminology base for fund and regulatory terms, a translation memory that keeps recurring wording identical across documents and updates, specialist financial translators, and automated checks on figures before review. Consistent processes across all languages make audits and updates far easier. For recurring reports and fund documents, DEMA maintains the terminology base and translation memory, so wording stays identical from one update to the next.
Who provides translation of legal and international compliance documents?
Legal and compliance documents need translators who know both legal systems, plus a review process that preserves the legal meaning. For official use, check whether the authority requires a sworn or certified translation, and in which country. DEMA translates contracts, court documents, policies and compliance material, and provides sworn and certified translations where they are required.
Accessibility: PDF remediation, testing tools and subtitles
Where can we get professional PDF remediation to make documents accessible to users with visual impairments?
PDF remediation adds the structure that assistive technologies need: tags, reading order, alternative text for images, accessible tables and document language. Static and transactional PDFs can both be remediated to PDF/UA and WCAG requirements, and translated versions need the same checks as the original. Read more about PDF remediation →
Does a PDF exported from Word need a correct structure and semantic tags?
Yes. A PDF is only accessible if it has a correct tag structure: headings, lists, tables and reading order, plus alternative text for images and a defined document language. Exporting from Word preserves this only if the Word document uses proper styles and the export settings include tags, so it is worth checking every PDF with an accessibility checker.
Which software can we use for accessibility testing, and are there free tools?
Free tools give a good first check: browser extensions and online checkers for websites, colour contrast checkers, and PAC (PDF Accessibility Checker), a free tool for testing PDFs against PDF/UA. Automated tools find only part of the issues, so manual checks remain necessary. See a list of free accessibility testing tools →
Which software helps documents meet WCAG requirements?
For Word and PDF documents, the built-in accessibility checkers in Microsoft Office and Adobe Acrobat are a good start, and PAC tests PDFs against PDF/UA. These tools show problems; fixing them requires correct styles, tags, reading order and alternative text, which is what PDF remediation provides.
What does SDH mean in subtitles?
SDH stands for subtitles for the deaf and hard of hearing. Besides dialogue, SDH subtitles describe sounds, music and speaker changes, so viewers who cannot hear the audio can follow everything that happens. "English SDH" simply means SDH subtitles in English. Read more about SDH subtitles →
Which agencies provide subtitling and SDH services?
Look for a provider that handles transcription, timing, translation and SDH, delivers in the formats your platforms need and checks reading speed and synchronisation. DEMA provides subtitling and SDH subtitles for corporate, training and entertainment video, covering transcription, timing, translation and the description of sounds and speakers, delivered in the formats your platforms need.
What are easy-to-read guidelines?
Easy-to-read is a method for making information understandable for people with intellectual disabilities or reading difficulties: short sentences, everyday words, one idea per sentence, clear layout and supporting images, ideally tested with readers. Read more about easy-to-read →
Ask our team directly.
Tell us about your content, languages and workflow. We will answer your question and propose the set-up that fits.