Security is spoken before it is secured
Early warnings, negotiations, testimony and public-health messages succeed or fail in the language people actually speak.
Comparative Heritage Project
Language Research Portal
A research environment for studying the world's languages as instruments of national and global security: how they carry power, how meaning survives or fails in translation, and how human and machine systems can be held to account for it.
Leadership
Every conflict is argued, every peace is negotiated and every warning is heard in a human language. To defend a people is to understand how they speak, and to measure how faithfully their words are carried by others and by machines.
Under the direction of Dr. David Olali, Founding Director of the Comparative Heritage Project, this portal brings linguists, translators, security analysts and technologists together around one conviction: that human languages are strategic heritage. Grounded in his Comparativity Complex Theory, which treats heritage as a source of power that adapts or dies, and in the Project's Heritage Security Intelligence framework, the research asks where language protects, where it endangers, and who decides. Dr. Olali's leadership sets the standard: rigorous human review, open method, and research that serves the communities whose languages it studies.
Language research portals
Choose a region to begin. All regions share one reviewer portal.
African Languages
Languages of the continent and its diasporas, from Yoruba and Swahili to Amharic and Zulu.
Enter portal →European Languages
The languages of Europe, from the widely spoken to the regional and minority languages that share its borders.
Enter portal →Nordic Languages
Danish, Faroese, Finnish, Icelandic, Norwegian, Swedish and the Sámi languages of the north.
Enter portal →Slavic and Eastern European Languages
Russian, Ukrainian, Polish and the wider Slavic, Baltic and Eastern European family.
Enter portal →Middle Eastern and North African Languages
Arabic in its varieties, Hebrew, Persian, Turkish, Kurdish and the Amazigh languages.
Enter portal →Asian Languages
The languages of South, East, Central and Southeast Asia, where much of humanity speaks and translates daily.
Enter portal →Indigenous Languages of the Americas
Quechua, Nahuatl, Guaraní, Navajo and the many Indigenous languages of North, Central and South America.
Enter portal →Pacific and Oceanic Languages
Māori, Hawaiian, Samoan, Tok Pisin and the languages of the Pacific islands and Australia.
Enter portal →Why language is a security question
Mistranslation is a vulnerability
A single misrendered term can misdirect aid, inflame a dispute or distort intelligence. Translation quality is a security question.
Machines now mediate meaning
AI translation is deployed faster than it is tested, least of all for the languages most of humanity speaks. Evidence has to come first.
Heritage is power
Languages carry memory, authority and identity. Who controls how a language is represented shapes who is heard, believed and protected.
Research programmes
The portal grows programme by programme. African Languages is open to reviewers now; the others are in development.
ActiveIn development In development In development In development In development
African Languages
Human linguistic review of machine translation across African languages: adequacy, fluency, terminology and the preservation of meaning.
Open the African Languages portal →Languages of the Diaspora
How languages carried by migration keep, lose or remake meaning, and what that means for belonging, representation and trust.
Sacred Texts and Translation
Scripture, translation and power: who renders sacred language, for whom, and with what consequences for identity and conflict.
Language and Security Intelligence
The linguistic dimension of Heritage Security Intelligence: early warning, influence and misinformation as they travel between languages.
Endangered and Minority Languages
Documenting languages at risk, and measuring who is left unheard when public systems and AI serve only dominant languages.
Machine Translation and AI Across World Languages
Reproducible, human-reviewed evaluation of the models now translating between the world's languages, with model, language and version on record.
How the research works
- Human judgement first. Trained reviewers who speak the language evaluate every translation.
- Reproducible evidence. Each evaluation records the model, language, version and reviewer, so results can be checked and repeated.
- Comparative by design. Findings are read across languages and traditions, not one at a time, to show where power and meaning shift.
- Accountable to communities. The people whose languages are studied are partners in the research, not only its subjects.