The Guardians of Cultural Heritage are Robots 

From March 9 to March 30, the number of new users on Duolingo jumped by 148% in the US alone. It seemed language learning was the new “pandemic hobby,” right up there next to baking bread. But unlike in sourdough school, you don’t need a person to teach you a language – although that is preferable – you just need an app.  

Millions of people around the globe are moving online to learn anything from French to Navajo. But with all this newfound time to commit to new year’s resolutions past, we have to consider the cultural value attached to such endeavors. Language learning, along with other pandemic hobbies, begs exploration beneath the façade of social media sensation. What we might find upon deeper reflection is that amongst a rapidly changing and uncertain world, people are turning to their traditions and cultures to get by.

Take bread-making. Social media sensation? Yes. But even more so it is a centuries long tradition with deep cultural ties. Almost every country has a variation of the beloved flour, water, salt trifecta. Language is likewise endemic to different cultural formations. The difference is that the resurgence of language learning is occurring against the backdrop of rapid decline in spoken languages around the world.

The UN declared 2019 the International Year of Indigenous Languages in an effort to draw attention to the unprecedented number of languages in danger of disappearing worldwide – 40% of the world’s estimated 6,700 languages to be exact. But this linguistic predicament isn’t new.

We’ve been in a cultural crisis of sorts with regard to language for centuries. As globalization spread around the world, with it came a need for common understanding – a lingua franca. French was the ‘language of diplomacy’ from the 17th century until recently, when the digital age marked the eclipsing of French by English as the dominant language. It is even argued that from the very infancy of computing technology, despite being 1s and 0s, the structural logic of code is more intuitive to English speakers. So, the language of the computer, and thus the global infrastructure, was designed to favor native-English speakers. 

A massive controversy exists in the literature of globalization over the tradeoffs between bridge languages for international business and the safeguarding of non-dominant, cultural-linguistic forms. So much so that in 1954 the UN recognized its first constructed international auxiliary language, Esperanto. Developed in 1887 by Ludovic L Zamenhof, the language – meaning “he who hopes” – was born out of the thought that common understanding might lead to peace. Striving for cultural neutrality, Esperanto has only 16 basic rules and no exceptions. But it is undeniably Eurocentric; three-fourths of the root words are borrowed from Romance, Germanic/Slavic, and Greek languages. Even so, leaders within Europe vehemently opposed Esperanto. Stalin referred to it as “that dangerous language.” Hitler described it as a tool for Jewish world domination, and France banned the language from schools. Even the US called it the “aggressor language” in their Cold War era army training exercises. The language was feared for its potential organizing power of the working class during a time when communist uprisings remained a dominant fear. As time moved on, these associations were hard to shake and the world was reluctant to adopt a culturally void language.  

The failure of Esperanto teaches us that languages don’t grow out of thin air. Language is embedded in culture, and vice versa. It is precisely this relationship between language and culture that makes a constructed international language non-feasible. Esperanto accrued its own cultural products in the form of cinema and literature, but no obvious culture. The language had nowhere to call home. Its neutrality was both its appeal and its demise.  

Some might ask when faced with the dilemma of diminishing linguistic forms, is saving languages truly necessary? Especially when new ones are being created that more or less ensure their demise. Is it not inevitable, much like the extinction of animals?  

Call it a cultural crutch, but language is arguably the defining feature of a culture. At its core, language is the cultural DNA which provides the building blocks for identities, through which people preserve their communal histories, customs, traditions, and memories. More than simply grounding a culture, it informs cultural values and modes of thought. The French language, for example, treats emotions as subjective, whereas the Italian language considers emotions as fact. Turning away from romance languages, an Aboriginal Australian language expresses in terms of directionality, given the strong connection to the land. Cultural differences are quite literally baked into language - no bread pun intended. 

---

So now, when we look to the increase in Duolingo users and the company’s endangered language course offerings, we might consider the role of technology in language preservation. 

What if we could assure the preservation of cultures with a computer? Artificial intelligence is increasingly being explored for its potential role in preserving languages. The appeal of this technology in an ethnolinguistic application is that AI-powered language codification ensures to some extent the preservation of the culture to which it is tied.  

Duolingo has been using AI to break down the dominant language model, both in their learning algorithm as well as in their efforts to make endangered languages accessible to the masses. But so long as the language of the computer is English, the realm of technology will prove culturally exclusive. Linguistic researches point to a need for AI-powered technology to be culturally sensitive. In short, when languages are codified into language processing algorithms (learning models), they need to be able to respond to inputs in different ways depending on the cultural context or mode of thought. Until Duolingo moves closer toward 100% contextual accuracy, the merit in digitizing endangered or extinct language declines. For without contextual accuracy, is it really the same language at all? It’s certainly not the same culture.

Despite its current downfalls, codification brings hope for the safeguarding of languages from the inevitable tide of globalization. In the past few years, Duolingo has added learning modules for endangered languages including Navaho, Welch, Gaelic, and Pidgin. But codifying these languages is only one part of preservation. Access is the second part. Fort without access, these languages are essentially digitized fossil accounts.  

Futurist Thomas Frey touches on this when he envisions a Global Language Archive, a living museum – “the Louvre of Languages” - where both endangered and extinct languages can be learned. He outlines the minimum requirement for language archival as sufficient video, audio, and written documents (including facial expressions, intonations, gestures, symbols, body language, values, etc.) for an AI “Language Recreation Engine” to reassemble a functional language that can be taught. Such advanced AI (100% contextual accuracy) does not yet exist but is already being developed.

What is particularly indicative of the cultural trend that kicked off this conversation is the people behind the numbers. Who are they, and what are they learning for? Duolingo found that most often, it is native members of the endangered cultures who engage with, or demand the release of, endangered language modules. The intent, presumably, is to connect with their cultural heritage and identity. But if the demand is coming from native speakers, is there a role to be played by linguistic outsiders? Codification needs to be a combined effort between technology, native, and non-native speakers. Native speakers must take part to ensure contextual accuracy in codification, and non-native speakers to validate the effort to codify in the first place, as they are technology’s cultural gatekeepers.  

So, when we ask what’s the point of trying to save languages, we might consider, who are we to efface history? Is it not wildly ignorant to disregard the value - inherent and functional - these languages have in shaping if not our world, our world view? To let languages wither is not far from engaging in revisionist history. We might all be better off if we acknowledge far-reaching cultural-linguistic flows as integral to our own identities.