|
Recent News
Featured Articles
»
Ten Things I Learned the Hard Way About Medical Translation
|
In this article, I argue that terminology is actually the easiest part of medical translation. What determines whether a translation is merely acceptable or genuinely reliable is a set of disciplined working habits. I discuss ten practices that I consider essential, from identifying the target audience and handling abbreviations carefully to verifying numbers and units, following a clear source hierarchy, writing effective queries, and carrying out focused quality checks. Above all, I emphasize the importance of resisting assumptions and recognizing the limits of one's expertise.
To illustrate these principles, I draw on a hospital medical record project that presented several practical challenges. The file was heavily table-based, which stripped away context in the CAT tool, and required constant judgment calls about layout, readability, untranslated identifiers, laboratory values, clinical shorthand, and interface strings with strict space constraints. The experience reinforced a point that runs throughout the article: the translation itself is often only a small part of the job. In medical translation, accuracy, consistency, and professional discipline matter far more than terminology alone.
|
Recent Articles
»
Poza tłumaczeniem: co może ujawnić analiza wielojęzyczna i OSINT
By Bartosz Pelka
| Published 09/7/2026
| Getting Established , Marketing Your Language Services , Miscellaneous , Technology , Translator Education , Other Specialties , Translator/Client Matching , Other , Polish , News , Opportunities | Not yet recommended
|
Większość pracy wywiadowczej i w OSINT kończy się na angielskim. To, co naprawdę ciekawe, zaczyna się tam, gdzie tłumaczenie w ogóle nie powstaje — na przykładzie jednego cyberataku i pięciu języków pokazuję, ile ta luka faktycznie kosztuje i dlaczego na ten sam problem trafia się w due diligence, wywiadzie konkurencyjnym i ocenie ryzyka.
|
»
Beyond Translation: What Multilingual OSINT Can Reveal
|
|
Most OSINT and intelligence work stops at English. The interesting part starts where translation never happens at all — using one cyberattack and five languages to show what that gap actually costs, and why due diligence, competitive intelligence and country risk all run into the same problem.
|
»
Seven Hundred Languages and a Font Problem
|
|
Indonesia has more than 700 living languages, and keeping them alive requires two different things that people tend to conflate: a record and a speaker. The record side runs into concrete obstacles — scripts that reached Unicode late or, like Pegon, not at all; Javanese material trapped in legacy fonts that no search engine can read; thousands of digitized *lontar* sitting as images because nobody has transliterated them; recordings labelled too vaguely to be useful. On the technology side, roughly ten of those 700 languages have usable corpora, and the parallel text that does exist is mostly scripture, which teaches machine translation a sermon register unfit for a tenancy agreement. That gap is where translators have a stake: orthography decisions, terminology for modern domains, legal and medical register, evaluation standards, and control over who gets to train on an elder's voice. But none of it makes speakers. Digitization buys time; only a household passing the language to a child keeps it alive.
|
»
OmegaT 6.1: How I Fix Word Export Errors Every Time
|
|
I learned that OmegaT-to-Word export problems are usually not because OmegaT destroys a document, but because Word files are complex .docx archives made of XML that OmegaT rebuilds using inline tags to preserve formatting. From my experience, the most important lesson is that those tags are not clutter; they are what keep the Word file structurally intact, so deleting, moving, or misplacing them can corrupt the exported file or scramble the formatting. I also see OmegaT 6.1 as a worthwhile release because it brings Java 21+ support, native Apple Silicon and ARM Linux builds, improved spell checking, LanguageTool 6.5, better packaging, and important bug fixes, but I know those upgrades do not eliminate the common causes of Word export failure. When an export fails, I first work out whether the file is truly broken or whether Word is simply blocking it, usually by opening it in LibreOffice; if LibreOffice opens it, I treat the issue as a Word trust, location, filename, or Protected View problem, but if it fails there too, I go back to OmegaT and check the tags. My routine is to clean messy Word source files before translating, keep filenames simple, preserve every tag exactly, avoid the Remove Tags option for Word files, run OmegaT’s tag and issue checks before export, export to a plain local folder, close Word before generating the target file, and always keep the original source file so I can regenerate the document without re-translating anything.
|
»
Proofreading Legal Translation: The Rules, the Limits, and the Brief You Should Write Yourself
|
|
Most arguments about legal proofreading start with the word "proofreading" itself. This article sets out eight rules for revising legal and judicial texts, explains where a linguist's job stops and a lawyer's starts, and makes the case for writing your own specification instead of asking the client for one.
|
No articles found.
|
printer version
