Independent publication · natural language processing
Parallel Text explains how machines process human language: how models are built and evaluated, what makes research open and reproducible, why most of the world's languages are still left behind, and the ethical questions raised by language data. Every article cites its sources and dates its facts.
Practical guide
Listings full of abbreviations, deposits that are not what they seem, a French "caution" that is really a guarantor: what translation tools get right when you rent in another language, where research shows they fail, and how to check a lease before you sign.
Read the guideHow text becomes model input, and how to tell whether a system really works.
Methods & evaluationWhy models read text in pieces, how BPE, WordPiece and SentencePiece split words, and why the same sentence costs more tokens in some languages.
Methods & evaluationSplits, F1, leaderboards, significance tests and contamination: how to read an NLP evaluation, and how to run one you can trust.
Methods & evaluationWhat BLEU really measures, why learned metrics replaced it in many studies, and how professional MQM evaluation catches the errors scores miss.
Open models, licences, documentation and results that can be checked.
Open researchOpen weights are not open source. The parts of a model release, the OSI definition, restrictive licences and what the EU AI Act says.
Open researchCC BY, ShareAlike, NonCommercial, ODbL, LDC agreements and the EU's text and data mining exceptions, explained for people who build and share corpora.
Open researchPreprocessing, seeds, versions and silent API updates: why NLP results fail to repeat, what studies found, and a routine that makes your work checkable.
Translated text, linked lexicons, Europe's languages and the thousands of languages NLP still leaves behind.
Multilingual & low-resourceMost of the world's 7,000-plus languages have almost no digital data. What that means for NLP, why web crawls fail them, and what actually helps.
Multilingual & low-resourceFrom the Canadian Hansard and Europarl to EU translation memories and web-mined bitext: the translated text behind machine translation, and its pitfalls.
Multilingual & low-resourceWordNets, BabelNet, Wikidata and ConceptNet: how meanings are linked across languages, what NLP uses them for, and where they fall short.
Ethics & dataStereotypes in embeddings, gendered mistranslations, dialects flagged as offensive: what research has shown about bias in language technology and what helps.
Ethics & dataNames, emails and health details hide in every corpus. How the GDPR applies to NLP data and models, why text anonymisation is hard, and what to do.
Ethics & dataCommon Crawl, robots.txt, the EU's text and data mining opt-out and the AI Act: the legal and ethical questions behind training data scraped from the web.
Inputs, meanings and measurements.
Reusable, checkable science.
Beyond English.
Fairness, privacy, consent.