There was a recent flurry of linguistic excitement when Dr Gerard Cheshire, from the University of Bristol, claimed a breakthrough in deciphering the Voynich manuscript. The 15th century book, rich with illustrations, is written in an unknown language and script. Since it was “discovered” by book dealer Wilfrid Voynich in 1912, the mysterious manuscript has been a plot element in books and movies and studied by cryptographers and linguists.


Dr Cheshire believes it is written in proto-Romance, an ancestor of today’s Romance languages that include Portuguese, Spanish, French, and Italian. According to Dr Cheshire, “The language used was ubiquitous in the Mediterranean during the Medieval period, but it was seldom written in official or important documents because Latin was the language of royalty, church and government. As a result, proto-Romance was lost from the record, until now.”


It is not the first time that a breakthrough has been made in translating lost languages. The most famous breakthrough came with the discovery of the Rosetta Stone in 1799. This enabled the deciphering of Egyptian hieroglyphs through cross reference to more well known languages also on the tablet.


There are many endangered languages in the modern world. In fact some have predicted that 90% of the current world languages will be extinct by 2050. How many languages have already come and gone is unknown. The Voynich manuscript is a rare exception. Because unlike when animals become extinct, there are rarely any fossils left when a language dies.


Many endangered languages are just going to die. It seems inevitable. But can’t we save some of them? Of particular significance in the quest to preserve endangered languages are the languages of indigenous people whose cultures have been swamped, but whose population remains steady. Examples are the Sioux tribe of North America, the Maori of New Zealand and the Ainu people of Japan.


We now have tools and knowledge to aid language teachers and learners that have never been available. In the 20th century an enormous amount of research and investment went into techniques and strategies for  language learning and teaching. The internet and particularly learning management systems, such as Moodle, allow individuals and groups to create learning materials and deliver them at low cost. Modern techniques such as shadowing, extensive reading and the Leitner flashcard system are generally applicable and can be used within an LMS to great effect. The use of video can magnify the ability of a teacher with scarce language skills, to reach a large number of learners. Video can also preserve authentic language resources for use in the future.


The dawn of AI brings with it even greater potential for preserving endangered languages. Because of evolving technology and patterns already discovered in natural language processing to date, it now requires fewer samples of speaker audio to build models for speech recognition and text-to-speech systems, The presence of speech models for endangered languages, provides a foundation for building higher level educational activities such as speech recognition powered flashcards and text-to-speech passage readers. These are engaging, fun and can be used in the absence of a human teacher. At the time of writing the number  of languages supported by Google Cloud Speech API is 120, and growing.


Poodll, a maker of language teacher tools for the Moodle platform, has produced a passage reader, speech recognition flashcards, and oral reading fluency evaluation systems that are powered by AWS machine learning services. The extension of these tools to different languages is limited only by the underlying language support of the AI platform.


In summary, organisations and governments seeking to preserve endangered languages now have access to modern techniques and technology, that are known to be effective.  These have already been developed, documented and expert advice is available. Combining these with the power of a modern LMS such as Moodle, would be a solid investment in securing the language. The development of AI speech models for the endangered language would be a true sign of commitment to the language’s survival.




About the author

Justin Hunt is the founder of, a company focussed on developing tools for online language teaching and learning.