Language Futures: New Worksites for the First Digital Century

“Greater is he who prophesies than he who speaks in tongues” Paul of Tarsus, I Cor. 14.5

You can look back ten thousand years and try to understand how we got here and what kind of cycles and patterns underlie our historical development. Writers such as Arnold Toynbee or today’s Peter Turchin and his school of data-driven cliodynamics take this approach. Call it deep history.

Or you can scry forwards and imagine what it would be like for humans to communicate in ten thousand years’ time. In 1984, the eminent semiotician Thomas Sebeok was asked by the US government to speculate on the options for maintaining written (and hence languaged) warnings about buried nuclear waste 10,000 years into the future. Call that deep foresight.

In general, talk about “language” tends to evoke our deep pasts, but very rarely any serious kind of futured cogitation. Yet the convergence of digital connectivity, the need for a serious translation solution worth its name, new neuro-insights into linguistic experience, and general ecological angst is now encouraging us to look more closely and broadly at language options and their implications for the future. This merits a new narrative for “language”, and a new assembly line of work sites for the immediate future.

Language’s tragic narrative

In many cultures we are fascinated by word histories, how pronunciation, syntax and morphology have changed over the centuries, or how modern languages can be traced back to proto-tongues. We tend to idolize what the old-timers or ancient texts have to say and how they said it. Time will bake preserved language fragments into literature if you wait long enough. We regret the disappearance of certain usages, spellings, and pronunciations even in our lifetimes. And we want to preserve our own linguistic cultures from “decay” or contamination by modernity. Yes, the disappearance of entire languages year by year is rapidly reducing the pool of human articulacy (cf literacy). From this angle, language change as we move forward in time almost inevitably means cognitive, auditive or semantic loss. Remember Nietzsche’s “language is a prison-house”. It locks us into fear about an irreversibly fractured future.

This is the tragic narrative, driven by dark misunderstandings of both language and history. There is now ample evidence that science, technology, social conviction, and intellectual outreach — as well as the indirect threats of climate change and pandemics — are radically changing our understanding of the nature and possibilities of language in action. And machine learning is steadily aping human eloquence to create a powerful new digital language twin, with as yet unknown appetites.

So let’s switch directions and use our curiosity about language to usher in a new line. This time, we will need to build it around technology experimentation, a desire for an inclusive future, and even the prospect of space travel.

Language as particle and wave

The English term “language” has an over-generous scope: it embraces the entire linguistic order — both particles and waves, to borrow a loose analogy from physics. Not just the enumerable phenomena we call languages — all 7,000 or so of them characterized by linguists — but just as importantly the wavelengths that language operates across by means of symbolic action at scale. Language here is a forcefield at work in our technologies, laws, sociologies, economies, polities, ecologies and psychologies.

We can call this gestalt of energies Language Futures, with the aim of framing a more persuasive, better-grounded speculative order for both these particles (the destinies of individual languages) and their waves (technology-inspired transformations of communication). We now have some of the intellectual and physical tools, especially a starter data stack, to harness the promise of very large-scale innovation in “supporting” language activities, or inventing new ones. The next step therefore is to explore more critically what Wittgenstein famously called the “limits of my language” in a world suspended between the hugeness of the cosmos and the extraordinary learning capacity of a single brain.

Today, compute technology and the global environmental crisis are revealing both gaps in — and powerful repair methods for- the fabric of our knowledge. Yet apart from the first steps in neurobiological experimentation, a number of language revitalization programs for endangered tongues, and generalized angst about digital fakery on social media, neither science fiction nor long-range forecasting have offered any helpful parables, narratives or models to build out plausible language futures that address emerging challenges

Beyond the conceit of banana fish translators, we therefore need new ideas to inspire responses to changing communication processes. How words mean things is still a mystery. But we realize now that we can do a lot of useful exploration of the vast space of possible discourse by using machine learning models, probably combining symbolic with data methods. In the longer run some of this work may be able to help us save, support and better use many of the word’s languages, if that is what we want and need. And of course there are lots of unknowns here. But if we want to act equitably, we’ll almost certainly need a generous language data solution to address the dearth of adequate text/speech resources, and give proper attention to the prospects of over 70% of living languages.

This combination of models and data (and not just in the AI sense!) could help us reinvent our ambitions for language as a universal good for communication, translation and creativity, as well as knowledge production and personal development. This inevitably involves thinking “exponentially” as Azeem Ahzar would say about language practices, technologies and ecologies.

So I propose to open up discussion on these worksites for thought about language futures, starting with seven worksites next time. Comments, corrections and more informed contributions are naturally welcome.