Why Study Foreign Languages?

My ability to speak multiple languages is a large part of who I am.* Admittedly, the more I languages I learn, the less mastery I have over each of the languages I speak. But I decided a while back I was ok with trading depth for breadth because I adore the process of starting from scratch, of gradually bringing once dormant characters to life, of working with my own insecurities and stubbornness as people respond in English to what must sound like pidgin German or Italian or Chinese, of hearing how the tone of my voice changes in French or Spanish, absorbing the Fanonian shock when a foreign friend raises his** eyebrows upon first hearing me speak English, surprised that my real, mother-tongue personality is far more harsh and masculine than the softer me embodied in metaphors of my not-quite-accurate French.***

You have to be comfortable with alienation to love learning foreign languages. Or perhaps so aware of how hard it is to communicate accurately in your mother tongue that it feels like a difference of degree rather than kind to express yourself in a language that’s not your own. Ferdinand Céline captures this feeling well in Mort à Crédit (one of the few books whose translated English title, Death on the Installment Plan, may be superior to the original!), when, as an exchange student in England, he narrates the gap between his internal dialogue and the self he expresses in his broken English to strangers at a dinner table. As a ruthless self critic, I’ve taken great solace in being able to hide behind a lack of precision: I wanted to write my undergraduate BA thesis (which argued that Proust was decidedly not Platonic) in French because the foreign language was a mask for the inevitable imperfection of my own thinking. Exposing myself, my vulnerabilities, my imperfections, my stupidity, was too much for me to handle. I felt protected by the veil of another tongue, like Samuel Beckett or Nabokov**** deliberately choosing to write in a language other than their own to both escape their past and adequately capture the spirit of their present.

But there’s more than just a desire to take refuge in the sanctuary of the other. There’s also the gratitude of connection. The delight the champagne producer in a small town outside Reims experiences upon learning that you, an American, have made the effort to understand her culture. The curiosity the Bavarian scholar experiences when he notices that your German accent is more hessisch than bayerisch (or, in Bavarian, bairisch, as one reader pointed out), his joy at teaching you how to gently roll your r’s and sound more like a southerner when you visit Neuschwanstein and marvel at the sublime decadence of Ludwig II. The involuntary smile that illuminates the face of the Chinese machine learning engineer on his or her screening interview when you tell him or her about your struggles to master Chinese characters. Underlying this is the joy we all experience when someone makes an effort to understand us for who we are, to crack open the crevices that permit deeper connections, to further our spirituality and love.

In short, learning a new language is wonderful. And the tower of Babel separating one culture from another adds immense richness to our world.

To date, linguae francae have been the result of colonial power and force: the world spoke Greek because the Greeks had power; the world spoke French because the French had power; the world speaks English because the Americans have had power (time will tell if that’s true in 20 years…). Efforts to synthesize a common language, like Esperanto or even Leibniz’s Universal Characteristic, have failed. But Futurists claim we’re reaching a point where technology will free us from our colonial shackles. Neural networks, they claim, will be able to apply their powers of composition and sequentiality to become the trading floor or central exchange for all the world’s languages, a no man’s land of abstraction general enough to represent all the nuances of local communication. I’m curious to know how many actual technologists believe this is the case. Certainly, there have been some really rad breakthroughs of late, as Gideon Lewis-Kraus eloquently captured in his profile of the Google Brain team and as the Economist describes in a tempered article about tasks automated translators currently perform well. My friend Gideon Mann and I are currently working on a fun project where we send daily emails filtered through the many available languages on Google Translate, which leads to some cute but generally comprehensible results (the best part is just seeing Nepali or Zulu show up in my inbox). On the flip side, NLP practitioners like Yoav Goldberg find these claims arrogant and inflated: the Israeli scientist just wrote a very strong Medium post critiquing a recent arXiv paper by folks at MILA that claims to generate high-quality prose using generative adversarial networks.*****

Let’s assume, for the sake of the exercise, that the tools will reach high enough quality performance that we no longer need to learn another language to communicate with others. Will language learning still be a valuable skill, or will it be outsourced to computers like multiplication?

I think there’s value in learning foreign languages even if computers can speak them better than we can. Here are some other things I value about language learning:

  • Foreign languages train your mind in abstraction. You start to see grammatical patterns in how languages are constructed and can apply these patterns to rapidly acquire new languages once you’ve learned one or two.
  • Foreign languages help you appreciate how our experiences are shaped by language. For example, in English we fall in love with someone, in French we fall in love of someone, in German we fall in love in someone. Does that directionality impact our experience of connection?
  • Foreign languages force you to read things more slowly, thereby increasing your retention of material and interpretative rigor.
  • Foreign languages encourage empathy and civic discourse, because you realize the relativity of your own ideas and opinions.
  • Foreign languages open new neural pathways, increasing your creativity.
  • Foreign languages are fun and it’s gratifying to connect with people in their mother tongue!
  • Speaking in a foreign language adds another level of mental difficulty to any task, making even the most boring thing (or conversation) more interesting.

I also polled Facebook and Twitter to see what other people thought. Here’s a selection of responses:

Screen Shot 2017-06-10 at 9.20.42 AM

Screen Shot 2017-06-10 at 9.21.50 AMScreen Shot 2017-06-10 at 9.22.24 AMScreen Shot 2017-06-10 at 9.22.57 AM

Screen Shot 2017-06-10 at 10.22.12 AM

Screen Shot 2017-06-10 at 9.25.42 AMScreen Shot 2017-06-10 at 9.26.21 AMScreen Shot 2017-06-10 at 9.27.04 AMScreen Shot 2017-06-10 at 9.27.56 AMScreen Shot 2017-06-10 at 9.28.24 AMScreen Shot 2017-06-10 at 9.28.52 AMScreen Shot 2017-06-10 at 9.29.29 AM.png

The best part of this exercise was how quickly and passionately people responded. It was a wonderful testimony to open-mindedness, curiosity, courage, and thirst for learning in an age where values like these are threatened. Let’s keep up the good fight!

*Another perk of living in Canada is that I get to speak French on a regular basis! Granted, Québecois is really different than my Parisian French, but it’s still awesome. And I’m here on a francophone work permit, which was the fastest route to getting me legal working status before the fast-track tech visa program that begins today.

**Gender deliberate.

*** It really irritates me when people say French is an easy language for native English speakers to learn. It’s relatively (i.e., versus Chinese or Arabic) easy to get to proficiency in French, but extremely difficult to achieve the fluency of the language’s full expressive power, which includes ironical nuances for different concessive phrases (“although this happened…”), the elegant ability to invert subject and verb to intimate doubt or suspicion, the ability to couple together conditional phrases, resonances with literary texts, and so much more.

****A reader wrote in correcting this statement about Nabokov. Apparently Nabokov could read and write in English before Russian. Said reader entitled his email to me “Vivian Darkbloom,” a character representing Nabokov himself who makes a cameo appearance in Lolita. If it’s false to claim that Nabokov uses English as a protected veil for his psychology, it may be true that cameos in anagram are his means to cloack presence and subjectivity, as he also appears – like Hitchcock in his films – as the character Blavdak Vinomori “King, Queen, Knave.”

*****Here’s the most interesting technical insight from Goldberg’s post: “To summarize the technical contribution of the paper (and the authors are welcome to correct me in the comments if I missed something), adversarial training for discrete sequences (like RNN generators) is hard, for the following technical reason: the output of each RNN time step is a multinomial distribution over the vocabulary (a softmax), but when we want to actually generate the sequence of symbols, we have to pick a single item from this distribution (convert to a one-hot vector). And this selection is hard to back-prop the gradients through, because its non-differentiable. The proposal of this paper is to overcome this difficulty by feeding the discriminator with the softmaxes (which are differentiable) instead of the one-hot vectors.” Goldberg cites the MILA paper as a symptom of a larger problem in current academic discourse in the ML and technology community, where platforms like arxiv short circuit the traditional peer review process. This is a really important and thorny issue, as traditional publishing techniques slow research, reserve the privilege of research to a selected few, and place pay walls around access. However, it’s also true that naive readers want to trust the output of top tier research labs, and we’ll fall prey to reputation without proper quality controls. A dangerous recent example of this was the Chinese study of automatic criminality detection, masterfully debunked by some friends at Google.

The featured image comes from Vext Magazine’s edition of Jorge Luis Borges’s Library of Babel (never heard of Vext until just now but looks worth checking out!). It’s a very apt representation of the first sentence in Borges’s wonderful story: The universe (which others call the Library) is composed of an indefinite and perhaps infinite number of hexagonal galleries, with vast air shafts between, surrounded by very low railings. From any of the hexagons one can see, interminably, the upper and lower floors. Having once again moved to a new city, being once again in the state of incubation and potentiality, and yet from an older vantage point, where my sense of self and identity is different than in my 20s, I’m drawn to this sentence: Like all men of the Library, I have traveled in my youth; I have wandered in search of a book, perhaps the catalogue of catalogues…

The Utility of the Humanities in the 21st Century

I did my PhD in Comparative Literature at Stanford. There is likely no university in the US with a culture more antithetical to the humanities: Stanford embodies the libertarian, technocratic values of Silicon Valley, where disruptive innovation has crystallized into a platitude* and engineers are the new priestly caste. Stanford had massive electrical engineering and computer science graduate cohorts; there were five students in my cohort in comparative literature (all women, of diverse backgrounds, and quite large in contrast to the two- or three-student cohorts in Italian, German, and French). I had been accepted into several graduate programs across the country, but felt a responsibility to study at a university where the humanities were threatened. I didn’t want the ivory tower, the prestigious rare book collection, the ability to misuse words like isomorphism and polymorphic because they sounded scientific (I was a math undergrad), the stultified comfort that Wordsworth and Shelley were on the minds of strangers on the street. I wanted to learn what it would mean to defend a discipline undervalued by society, in an age where universities were becoming private businesses tailoring to undergraduate student consumers and the rising costs of education made it borderline irresponsible not to pursue vocational training that would land a decent job coding for a startup. Stanford’s very libertarianism also enabled me to craft an interdisciplinary methodology–crossing literature, history of science and mathematics, analytic philosophy, and classics–that more conservative departments would never entertain. This was wonderful during my coursework, and my Achilles heel when I had to write a dissertation and build a professional identity more conservative departments could recognize. I went insane, but mustered the strength and resilience required to complete my dissertation (in retrospect, I’m very grateful I did, as having a PhD has enabled me to teach as adjunct faculty alongside my primary job). After graduation, I left academia for the greener, freer pastures of the private sector.

The 2008-2009 financial crisis took place in the midst of my graduate studies. Ever tighter departmental budgets exacerbated the identity crisis the humanities were already facing. Universities had to cut costs, and French departments or film studies departments or German departments were the first to go. This shrank the already minuscule demand for humanities faculty, and exponentially increased the level of anxiety my fellow PhDs and I experienced regarding our future livelihoods. In keeping with the futurism of the Valley, Stanford (or at least a few professors at Stanford) was at the vanguard for considering alternative career paths for humanities PhDs: professors discussed shortening the time to degree, providing students with more vocational communications training so they could land jobs as social media marketers, extolling the values of academic administration as a career path equal to that of a researcher. Others resisted vehemently. There was also a wave of activity defending the utility of the humanities to cultivate empathy and other social skills. I’ve spent a good portion of my life reading fiction, but must say it was never as rich a moral training ground as actual life experience. I’ve learned more about regulating my emotions and empathizing with others’ points of view in my four years in the private sector than I had in the 28 years of life before I embraced work as a career (rather than just a job). Some people are really hard to deal with, and you have to face these challenges head on to grow.

All this is context for my opinions defending the utility of the humanities in our contemporary society and economy. To be clear, in proposing these economic arguments, I’m not abandoning claims for the importance of the humanities in individual personal and intellectual development. On the contrary, I strongly believe that a balanced, liberal arts education is critical to foster the development of personal autonomy and civic judgement, to preserve and potentially resurrect our early Republican (as political experiment, not party) goals that education cultivate critical citizens, not compliant economic agents.  I was miserable as a graduate student, but don’t regret my path for a minute. And I think there is a case to be made that humanities will be as–if not more–important than STEM to our national interests in the near future. Here’s why:

Technology and White-Collar Professions – In The Future of the ProfessionsRichard and Daniel Susskind demonstrate how technology is changing professions like medicine, law, investment management, accounting, and architecture. Their key insight is to structurally define white-collar professionals by the information asymmetry that exists between professional and client. Professionals know things it is hard for laymen to know: the tax code is complex and arcane, and it would take too much time for the Everyman (gender intentional) to understand it well enough to make judgments in her (gender intentional) favor. Same goes for diagnosing and treating an illness or managing finances of a large corporation. The internet, and, perhaps more importantly, the new machine learning technologies that enable us to use the internet to answer hard, formerly professional, questions, however, levels this information asymmetry. Suddenly, tools can do what trained professionals used to do, and at a much lower costs (contrast the billed hours of a good lawyer with the economies of scale of Google). As such, the skills and activities professionals need are changing and will continue to change. Working in machine learning, I can say from experience that we are nowhere near an age where machines are going to flat out replace people, creating a utopian world with universal basic income and bored Baudelaires assuaging ennui with opiates, sex, and poetry (laced with healthy doses of Catholic guilt). What is happening is that the day-to-day work of professionals is changing and will continue to change. Machines are ready and able to execute many of the repetitive tasks done by many professionals (think young associates reviewing documents to find relevant information for lawsuit – in 2015, the Second Circuit tried to define what it means to practice law by contrasting tasks humans can do with tasks computers can do). As machines creep ever further into work that requires thinking and judgment, critical thinking, creativity, interpretation, emotions, and reasoning will become increasingly important. STEM may just lead to its own obsoleteness (AI software is now making its own AI software), and in doing so is increasing the value of professionals trained in the humanities. This value lies in the design methodologies required to transform what were once thought processes into statistical techniques, to crystallize probabilistic outputs into intuitive features for non-technical users. It lies in creating the training data required to make a friendly chat bot. Most importantly, it lies in the empathy and problem-solving skills that will be the essence of professional work in the future.

Autonomy and Mores in the Gig Economy – In October, 2015, I spoke at a Financial Times conference about corporate sustainability. The audience was filled with executives from organizations like the Hudson Bay Company (they started by selling beaver pelts and now own department stores like Saks Fifth Avenue) that had stayed in business over literally hundreds of years by gradually evolving and adding new business lines. The silver-haired rich men on the panel with me kept extolling the importance of “company values” as the key to keeping incumbents relevant in today’s society. And my challenge to them was to ask how modern, global organizations, in particular those with large, temporary 1099 workforces managed by impersonal algorithms, could cultivate mores and values like the small, local companies of the past. Indeed, I spent a few years helping international law firms build centralized risk and compliance operations, and in doing so came to appreciate that the Cravath model, an apprenticeship culture where skills and corporate culture and mores and passed down from generation to generation, as there is very low mobility between firms, simply does not scale to our mobile, changing, global workforce. As such, inculcating values takes a very different form and structure than it did in the past. We read a lot about how today’s careers are more like jungle gyms than ladders, where there is a need to constantly revamp and acquire new skills to keep up with changing technologies and demand, but this often overlooks the fact that companies – like clubs and societies – used to also shape our moral characters. You may say that user reviews (the five stars you can get as an Uber rider or AirBnB lodger) take the place of what was formerly subjective judgment of colleagues and peers. But these cold metrics are a far cry from the suffering and satisfaction we experience when we break from or align with a community’s mores. This merits much more commentary than the brief suggestions I’ll make here, but I believe our globalized, gig economy requires a self-reliant morality and autonomy that has no choice but to be cultivated apart from the workplace. And the seat of that cultivation would be some training in philosophy, ethics, and humanities. Otherwise corporate values will be reduced to the cold rationality of some algorithm measuring OKRs and KPIs.

Ethics and Emerging Technologies – Just this morning, Guru Banavar, IBM’s Chief Science Officer for Cognitive Computing, posted a blog admonishing technologists building AI products that they “now shoulder the added burden of ensuring these technologies are developed, deployed and adopted in responsible, ethical and enduring ways.” Banavar’s post is a very brief advertisement for the Partnership on AI IBM, Google, Microsoft, Amazon, Facebook, and Apple have created to formalize attention around the ethical implications of the technologies they are building. Elon Musk founded OpenAI with a similar mission to research AI technologies with an eye towards ethics and safety. Again, there is much to say about the different ethical issues new technologies present (I surveyed a few a year ago in a Fast Forward Labs newsletter). The point here is that ethics is moving from a niche interest of progressive technologists to a core component of large corporate technology strategy. And the ethical issues new technologies pose are not trivial. It’s very easy to fall into chicken little logic traps (where scholars like Nick Bostrom speculate on worst-case scenarios just because they are feasible for us to imagine) that grab headlines instead of sticking with the discipline required to recognize how data technologies can amplify existing social biases. As Ted Underwood recently tweeted, doing this well requires both people who are motivated by critical thinking and people who are actually interested in machine learning technologies. But the and is critical, else technologists will waste a lot of time reinventing methods philosophers and ethicists have already honed. And even if the auditing of algorithms is carried out by technologists, humanists can help voice and articulate what they find. Finally, it goes without saying that we all need to sharpen our critical reading skills to protect our democracy in the age of Trump, filter bubbles, and fake news.

This is just a start. Each of these points can be developed, and there are many more to make. My purpose here is to shift the dialogue on the value of the humanities from utility in cultivating empathy and emotional character to real economic and social impact. The humanities are worth fighting for.


*For those unaware, Clayton Christensen coined the term disruptive innovation in The Innovator’s DilemmaHe contrasted it with sustaining innovation, the gradual technical improvements companies make to a product to meet market and customer demands. Inspired by Thomas Kuhn’s Structure of Scientific Revolutions, Christensen artfully demonstrates how great companies miss out on opportunities for disruptive innovation precisely because they are well run: disruptive innovations seize upon new markets with an unserved need, and only catch up to incumbents because technology can change faster than market preferences and demand. As disruption has crystallized into ideology, people often overlook that most products are sustaining innovations, incremental improvements upon an existing product or market need. It’s admittedly much more exciting to carry out a Copernican revolution, but if we consider that Trump may well be a disruptive innovator, who identified a latent market whose needs were underserved only to topple the establishment, we might sit back, pause, and reconsider our ideological assumptions.

The image is Jacques-Louis David’s The Death of Socrates from 1787. Plato sits at the front with his head down and his legs and arms peacefully and plaintively crossed.