Medical MTPE and Arabic LQA by Pharmacist Post-Editors
Machine translation has reached the point where it produces readable Arabic for most general text. Medical Arabic is different. The output looks fluent, which is exactly the problem: the errors are hidden inside grammatical sentences. Lingoarabia provides medical MTPE and Arabic LQA with a pharmacist reading the output, so the faults that only a clinician notices are found before your content ships.
Where MT fails in Arabic medical text
We post-edit a large volume of machine-translated medical content, and the error pattern is consistent.
- Gender agreement. Arabic verbs, adjectives and pronouns carry gender. MT routinely addresses a female patient as male, or switches mid-paragraph, which matters in consent forms and patient instructions.
- Dosage and units. “Twice daily” becomes “two days”, mg and mL get swapped, and decimal separators drift. These are the errors a pharmacist spots in seconds and a general post-editor can read straight past.
- Negation and conditionals. “Do not take if” and “take only if” are small strings with opposite meanings, and MT engines drop or move the negation more often than anyone expects.
- Register. Regulatory Arabic has a fixed register that authorities such as the EDA and SFDA recognise. MT produces a mixture of journalistic and colloquial phrasing that reads wrong to a reviewer and may be rejected at submission.
- Terminology. Engines pick a plausible synonym rather than the term in your glossary or in the authority’s approved labelling.
A post-editor without medical training will fix fluency and leave the substance. Ours fix the substance.
Services
Full post-editing. Output brought to the quality of human translation, checked against your glossary and style guide, and revised by a second linguist where your workflow requires an ISO 17100-aligned process. Suitable for labelling, patient material, IFUs and anything that goes to a regulator or a patient.
Light post-editing. Accuracy fixes only, for internal use: literature screening, internal training, safety triage, comprehension copies. We tell you plainly when a document is not suited to light post-editing.
LQA scoring. Sampled or full evaluation of Arabic output, human or machine, against an MQM or DQF-style error typology, with severity weighting agreed with you. You receive a scored report, annotated examples and a short root-cause summary you can act on with your engine or your vendor.
In-country review. Review of Arabic produced by another vendor or by your internal team, from the perspective of a pharmacist in the region. We mark what must change, what should change and what is preference, so your project manager can close the loop quickly.
Glossary and style-guide authoring. Bilingual termbases and Arabic style guides built as standalone deliverables, then applied in the post-editing stage. Dr. Ola Siam has written Arabic style guides and glossaries for global accounts that other linguists then work to.
Tools and workflows
We work inside XTM, memoQ, Trados Studio and Wordfast Pro, and in client-side TMS and review portals, with QA runs in Xbench. If your engine sits inside your TMS, we post-edit there and your data never leaves your environment. Tracked edits, edit-distance reports and time logs are available where you need evidence of the post-editing effort.
Who does the work
Dr. Ola Siam, B.Pharm., leads the medical MTPE and LQA team. She has 17+ years of English to Arabic medical translation behind her, 5.8 million medical words, and more than one million words of UpToDate clinical reference content worked inside the client’s own platform between 2024 and 2026. Second linguists are qualified Arabic medical translators who work to the same glossary and QA checklist.
Capacity
A senior post-editor reviews 7,000 words a day of machine output at full post-editing quality, and more for light post-editing depending on the engine. For large programmes we scale with additional qualified linguists under the same lead reviewer, so terminology and register stay consistent across the batch.
Request a quote
Send a sample of the raw MT output with the source, or a batch of translated Arabic you want evaluated, to request a quote. We will tell you which level of post-editing the content needs and quote on that basis.
Frequently asked questions
Full post-editing brings MT output to human translation quality, checked against your glossary and style guide, and is right for anything a regulator or patient will read. Light post-editing corrects accuracy errors only and is suited to internal, comprehension-level use. We advise on which level a document needs.
The most dangerous MT errors in Arabic medical text are in dosage, units, negation and gender agreement, and they sit inside fluent sentences. A pharmacist recognises a wrong dose or a reversed instruction on sight. A general post-editor tends to fix fluency and leave the substance.
MQM or DQF-style typologies, with categories and severity weights agreed with you before scoring starts. You receive a scored report, annotated examples and a root-cause summary that your engine or vendor can act on.
Yes. We work in XTM, memoQ, Trados Studio, Wordfast Pro and in client-side TMS and review portals, with QA in Xbench. Where your engine runs inside your TMS, we post-edit there and your content stays in your environment.
Yes. In-country review of third-party Arabic is a regular request. We separate mandatory changes from recommended changes and preference, so your project manager can resolve the review without a second round of debate.
That is the sequence we recommend for a new programme. We build the bilingual termbase and, where useful, an Arabic style guide, then apply both during post-editing and LQA so the output is consistent from the first batch.