The Opposite of Babel | Etan Nechin

The Opposite of Babel: Power of Language and Algorithmic Ghosts

Because the messenger’s mouth was heavy and he couldn’t repeat (the message), the Lord of Kulaba patted some clay and put words on it, like a tablet. Until then, there had been no putting words on clay. 

— Sumerian epic poem, Enmerkar and the Lord of Aratta


In the story of the Tower of Babel, language functions in two ways: first, as an unstoppable, unifying force, and second, as the ultimate punishment. Babel was a tower of language: its bulk and breadth came from the bricks of a singular language. The tower reached towards the heavens because the language was understood to king and fool, a language with no divisions. This created a not only a theological crisis, but a linguistic one:  centralized, ultimate power cannot be dyarchical; there is only space for one language of authority. To curb Man’s ambitions, God created chaos, not by destroying the tower or drowning the people, as he did in the flood. Paradoxically, he created chaos by establishing a new order: Language, which was the ultimate unifier, became a dividing agent, the ultimate barrier between people. This gave birth to new hierarchies to be formed on earth. Linguistic order, therefore, became political order. 

From Babylonia to Medieval Europe to Colonial times, language has always been used as a vehicle for power, creating hierarchical divisions to tower over the governed subjects. If that is the case, how was it used, and who helped shape it into a tool of control?

Ironically, the first examples of what we call linguistics, namely, developing a written language as a system, originated in and around Babylonia. This systemizing of words came from a need to clarify bureaucratic discourse. Members of the Akkadian elite classes etched series of nouns on cuneiforms to enshrine their meaning. Cuneiform script is one of the earliest forms of writing. The terms weren’t written in Akkadian, the Mesopotamian vernacular. Rather, these nameless scribes inscribed them in Sumerian, an extinct language preserved for liturgical and scholarly purposes. Sumerian was a language isolate, meaning it has no known genealogical roots or relationship with other languages; its power came from its separateness. Through this language of exclusion, the ruling class created a legal and religious code. One famous example is the Code of Hammurabi, a set of laws regulating Mesopotamian society. This code, engraved on a stele, was fixed: not only inscribed on rock but enshrined in legal speak, which is resistant to the flow of natural speech. This code was immovable because it belonged to a caste who used the fixed code to secure their status. The language offered the ruling classes sole access to truth. Thus hierarchical social ordering was done through making lingual differences visible and  incomprehensible to the lower classes.

Linguistic power was dependent on proximity as well as literacy. Those in the temple and palace held linguistic keys that enabled them to come and leave at their gates. Their power was granted from the divine up above and the etymological roots of language. The farther one was from the lingual center of power, the less influence over the official language they had, and the more vulgar—from the Latin vulgaris, meaning common or low—they were considered.

Where the poet or priest spoke and wrote in the lingua franca—a static, written language—the vulgar person spoke in the vernacular: a fluid, ever-changing language, containing dialects, slang, idioms, other vernaculars from different languages. Vernaculars continually developed and evolved because they weren’t fixed to any official legal or religious code. The term vernacular comes from the Latin vernaculus, meaning indigenous, which in turn derived from verna, a house-born a slave in the Roman Empire. Vernacular was the language of the domesticated subjects tilling the fields and building mansions—the language of the powerless inside the system of power.

One of the primary sources of power of Lingua Franca was this interconnectedness with vernacular. Latin was the language of truth, and members of the elite classes worked tirelessly to maintain its status. But this interconnectedness was its main source of weakness: it was tied in a Gordian Knot to the vernacular. Its fixedness made it a static stele: an unchanging rigid set of rules—and grammar—trying to resist the swaying of society and time.

The first to capture the essence of vernacular power was Dante Alighieri, the Italian poet who wrote The Divine Comedy—the magnum opus of vernacular. Upon his exile, driven out of the Florentine sphere of influence into the Italian countryside with its farmers and artisans, those domesticated subjects, Dante realized the inherent power of “natural language.” In his essay De vulgar eloquentia, Dante writes that while Latin is immobile, vernacular contained an animating quality. Those powerless—outside of the bureaucratic class, outside the lingua franca—could utter their concerns and demands as defined by their alternative point of view, using their alternative grammar. Because they were invisible to power, they could create new structures that resisted this centralized force. Therefore, instead of being kept in order by lingua franca, people could resist by creating a new linguistic order which reflected their values, ethics, and politics.

Differences between vernacular and the Lingua Franca were explicit, from the Latin explicitus meaning unobstructed.  When power can be seen unobstructed and when the margin between languages is clear, the language of power can be resisted and reshaped. 

 It would take centuries, but vernacular would increasingly replace Latin as the language of the bureaucratic sphere, which would lead to the chipping away of the Church’s hegemonic rule. As Benedict Anderson observed in Imagined Community, one of the critical elements that led to the waning influence of Church was “the revolutionary vernacularizing thrust of capitalism.” Writing and later printing en masse in vernacular (the first “bestseller” was Martin Luther’s German translation of the Bible) created a counterforce to the Church because it was inclusive—anyone could print, and anyone who could read could understand. Language became increasingly local: if Latin belonged to the Almighty Church that reigned everywhere and on everyone, then French or German belonged to a specific place and a particular group of people. Vernacular increasingly became national: printed matter was written in French or German or Spanish to serve the population local to it. Thus, the margin grew not only between vernacular and lingua franca but also between the Church and the nation. It is no wonder then that a democratization process began as Italian, French, or Spanish—spoken, common languages—became the lingua franca. It’s no wonder then, that after the French Revolution, Henri “Abbé” Grégoire, the revolutionary Catholic priest, advocated in his treatise Rapport sur la Nécessité et les Moyens d’anéantir les Patois et d’universaliser l’Usage de la Langue française (report on the necessity and means to annihilate the patois and to universalize the use of the french language) eradicating regional dialects and universalizing the use of French. Language cross the class divide; peasant and ruler spoke the same tongue. 

But when the bureaucratic lingua franca and everyday vernacular began to become one, the line between those in positions of power and those feeling its effects began to blur. However, traditional hierarchical orders were preserved because national and state apparatuses adopted complicated modes of expression and communication, their bureaucratic “tongue” was heavy. 

But what happens when the language of power manipulates local, common language, and masquerades as our daily vernacular? In our digital age, where people aren’t subjects but citizens, where the law isn’t divine but common, and issues are addressed in the public sphere, our language is reshaped by sophisticated algorithms that mediate our conversations while rendering power relations invisible. 



“Does he understand his sentence?” the Explorer asks…

“No,” the Officer answers. “He’ll learn it on his body.”


      Franz Kafka, The Penal Colony


Kafka was the clearest articulator of modern pre-algorithmic bureaucratic society: A faceless machine operating with no consideration to the individual. It was the individual’s task to interpret these mercurial organizations. Misinterpretation came with immediate and present consequences to the person navigating the bureaucratic maze. Although increasingly faceless, the blow from the system to the body was real. The bureaucratic machine, although incomprehensible, showed its results in the public sphere; the system, and its output, were corporeal.

On a societal level, individuals could resist this power by becoming a multitude or a crowd “carrying the closeness to uncomfortable physical contact.” The way to resist faceless power was showing the multitude of faces and bodies—to push against it. The multitudes could do so because power, now held by the state, could be resisted through the public sphere was still centralized; its structure still vertical.

Today, we are living in a decentralized world. There is no center of power, but different institutions—corporate, governmental, and supranational organizations such as The World Bank or the WTO—simultaneously working together and against each other. Moreover, we are living in a world where technology is the primary medium for most of our communication. We are becoming less involved in spontaneous, immediate, and direct interactions. Thus, language is increasingly becoming mediated by external forces. 

Traditional language mediation was done by TV, radio, or print newspapers, which were work as a one-way street, from producer to the viewer. On an interpersonal level, there was mail. But correspondence, apart from referring to the exchange of letters, also means close similarity. We always engage with someone or something knowable to us. Moreover, language isn’t mediated by external forces. There’s a clear marker separating the private from the public: the envelope and the seal. The contents of the letter are private while the envelope or seal engage in the bureaucratic sphere. It is no wonder, then, that in the US, opening someone else’s mail is a federal offense.

This all changed when in the 1980s when Internet Service Providers (ISPs) began connecting people across the world, and email “hosting” sites began to pop up. In 1993, email came into the public lexicon. In the late 1990s internet use became ubiquitous all around the world, from 55 million people in 1997 to 400 million in 1999 to an estimated  of more 4 billion internet users today. 

Email, like all other digital communication tools, works much differently than letters. Emails use language algorithms that are built by programmers and computational linguists. These algorithms are sets of rules based on linguistic contingencies—all the possibilities of a conversation—written in lines of code. Algorithms are units that don’t only frame the conditions for a particular conversation, i.e., a work email or Google search, but are also predictive of any permutation in that conversation.  To achieve this goal, algorithms need to “tear the envelope” in order to understand and analyze our conversations. 

But to make communication and use as intuitive and “natural” as writing a letter,  programmers and linguists need to understand how we communicate with each other—our digital vernacular—and to serve the language that feels authentic to us. The field concerned with creating these “natural” language algorithms is Natural-language processing or NLP. NLP is a sub-category of linguistics which uses computer science, informational engineering, and artificial intelligence. Computational linguists and computer programmers use NLP methods to analyze how language is being used in every interaction we make. Their focus is to understand what structures we choose to formulate our messaging, how we use language, and where we “misspeak.” Through this, they can “teach” computers to disseminate “natural human language”. For algorithms, a human linguistic error isn’t a bug, but a feature. This results in subversion of our understanding of private and public language.  

The algorithm presents us with a clear, cohesive world while hiding another. It creates this world not only by using the words we put into a search bar, but also by our geographic location (IP) and our digital history. It takes into account all the digital topographies and histories of those we communicated with in the past. The digital world is a mirror: it reflects our features to ourselves. But this mechanism operates as a one-sided mirror: we search through this knowable, personalized world, while on the other side programmers and linguists analyze our lingual behavior; they monitor our calls for “quality assurance.” They enhance the illusion of this world by listening to everything we do, even to our everyday babble.  

When technology is focused on doing this very thing – decoding human interaction – linguistics becomes a tool for appropriating human communication. When programmatic language is subsumed in our digital vernacular, it takes on a spectral quality to make our world comprehensible. The ghost isn’t in the machine, but the machine itself is a ghost.

Michel Foucault identifies this process in his book Power/Knowledge.  Power, for Foucault, both uses knowledge and reproduces it to strengthen its particular ideology. Foucault shifts attention from language to discourse. Discourse is the production of object of knowledge through language. These objects of knowledge are then disseminated as truths of our world. 

 But in the digital world, perception of truth isn’t only enhanced through manipulating language but also by subverting distance. NLP algorithms replace proximity with locality. Digital linguists in the corporate sphere, namely in the field of search, commerce, and digital communications, focus on making the language feel local to the user. Local, in this case, is a common vernacular, a language indigenous to any user’s digital sphere, this every language: German, Urdu, Suomi—is the language of origin. Therefore NLP algorithms is coded, corporate communication, masquerading as everyday digital vernacular. Digital linguists do this by tracking our most intimate interactions, analyzing them, then feeding it back to us as set truths about our world. Our movements through the digital space feel natural to us, but in fact, they’re choreographed by an unknown force.

Language is continuously absorbed into an interconnect network and reappears elsewhere irregularly: on a banner ad or a “smart reply” addon in Gmail. Therefore, the main quality of digital linguistic power can be seen as a programmatic vulgarizing of language. Bureaucratic language  of power is already embedded in our daily discourse.  Consequently our language becomes devoid of the vital quality of Dante’s vernacular.  We can’t build new linguistic structures because there is no difference between the lingua franca and our vernacular.

This is the opposite of Babel: It is a tower of language that keeps feeding on itself by our unwitting engagement. Because we are both engaged in the input and the output of this machine, we no longer separate ourselves from it, and every utterance, desire, post, we put online works to reaffirm and reinvigorate that power. It is a decentralized linguistic maze that operates without a guiding ideology nor clear grammar. And, because everything online is personalized to us, we, despite incessant interactions, are increasingly living in a world simultaneously made for us and is of our own making. In this individualized digital world, the multitude is erased; everyday language is used to keep us isolated. The ever-changing digital language works to cover up the fact that we are stagnant, that we have been “domesticated.” The opposite of Babel is not an oppressive, monolingual force building a visible edifice of control – but a disparate, multilingual divided mass, building brick by brick, a labyrinthine, invisible, singular structure of coded power that doesn’t unite and isn’t unifying; instead, it divides and obfuscates. And it keeps on growing, amassing more weight and breadth, without us realizing we are the ones building it.