the toolbox.

I kneel beside my dede (“grandfather”) as he gently swings his hammer, lightly tapping a nail into the wooden cabinet through three different motions (“dācede, ā:cede, muācede,”), demonstrating as he recites each word. I’m visiting his dusty village house along the shores of the Black Sea, eagerly listening to his instructions for tackling the kitchen repair.

Beyond the physical tools he’s holding, he’s also pulling from his linguistic toolbox. My dede speaks one of Turkey’s many minority languages and is explaining how you can express the simple action of hammering a nail in various different ways. What is simply “hammering” in English becomes dācede (“hammering in front of oneself”), ā:cede (“hammering above oneself”), and muācede (“hammering from behind the nail”) in Gürcü. Learning these intricacies and concepts of relative spatiality in verbs was beyond what my mind could even comprehend; however, through my dede, I began to see new ways to communicate beyond my knowledge of just Turkish and English. With each new Gürcü word, I added to my lexicon as well as my perception.

These fascinating conversations have always been the highlight of my sporadic visits to Turkey and sparked my interest in linguistics. My dede also reveled in my curiosity, frequently teaching new words and phrases to his best student. No one else in our family spoke Gürcü. It had never been passed down to my parents. This was no choice; language suppression was a central part of the government’s aim in seeking control. Through an intense history of programs and bans that enforced one language of supremacy (Turkish), the Turkish state not only demonized the proliferation of other languages and dialects but also prevented diverse thinking and public opinion that diverged from the ruling beliefs.

This destructive dominance has precluded me from accessing the rich cultures of my heritage. There is a broad range of languages in my family, from Gürcü to Kurdish to Laz. The patterns in my family are a part of a much greater trend in the loss of the world’s linguistic diversity. Around the world, there are over 7000 more diverse languages holding their own unique knowledge and significance, yet 90% of them are predicted to disappear within the century. As dominant languages form linguistic hierarchies of social, economic, and digital power, minority languages—and their identities and knowledge—become lost in deafening silence.

The inequality of AI language technologies only exacerbates this digital and economic divide in world cultures. While AI algorithms have grown rapidly and are becoming more capable, these advancements are primarily English-centric with the disparities in data availability and the algorithms' data inefficiency. Thus, they serve different communities disproportionately. Yet it is obvious with the current growth of AI that access to these top tools will become the key to global power and opportunities. Thus minority language speakers are forced to abandon their cultures towards the utility of dominant languages or be left in the dust. With the acceleration of unequal AI, we also accelerate the death of our cultures and humanity itself. This is both a social and a technical challenge to address in AI to promote the accessibility and service of AI systems to ensure equal opportunity. 

These languages are easily forgotten. I argue that diverse language hold a functional value with 'linguistic relativism' and the creativity they enable with their unique views and knowledge embedded in the language. Yet it is a fact that a language deserves to exist simply because it is a beauty and art of humanity. AI may be unequal right now, but it also has so much potential to support diverse cultures as well, with potential for large scale preservation, documentation, and tools of empowerment. With each voice it can reach, we add creative, innovative views of the world to our linguistic toolbox.