Muhakamat al-Lughatayn

I was once advised that it takes "a special kind of nut" to study Xinjiang. The comment was raised with regard to Justin Rudelson, who, in his introduction to Oasis Identities, lists the dozen or so languages he learned in preparation for his research. Like many others, I took it both as a warning as to the sorts of problems I would face in the course of my work and as a kind of challenge.

Six years later, I speak several languages passably well and read a half-dozen others. Certainly, first earning an MA in Linguistics helped tremendously: Not only did I learn a great deal about languages, how they're put together, and how to pronounce their funny sounds, I also gained an appreciation for the universality of the language faculty. (It was a very, very Chomskyist program.) That's a very reassuring thing, the idea that, really, they're all the same under the hood.

What really enabled an ongoing marathon of language-learning, however, was following links from one language to another, avenues of typological similarity or shared content. I will illustrate this with a diagram, slapped together with Netdraw in about 15 minutes:

Languages, each one a node, are labeled thus:

Ones that I have studied as "living" languages and learned to speak are circles. "Dead" languages, or at least ones that I've only learned to read, are squares.

The nodes vary by size. The larger they are, the longer I've studied them. (Mandarin Chinese: 11 years; Uyghur: 5 years; French: A bit on my own, etc.)

Then, I've labeled them by a sort of "super-family." The Romance languages, then, are all gray, as are the Germanic ones, since they belong to Western Indo-European. Persian and Russian, which form a bridge with the Asian languages, are Eastern Indo-European. Then I've done them by broad, paralinguistic area: Uyghur, Mongol, and Finnish are all under "Central Eurasian," while Chinese, Japanese, and Wenyanwen (Literary Chinese) are under "East Asian." This makes for some very clear areal distinctions in the diagram!

Finally, there are the arrows. An arrow pointing from a language to a language shows that knowing the first contributes or contributed actively to understanding and using the second. The thicker the arrow, the closer the connection.

Let's work through this. I studied 5 years of Spanish in high school. This was completely and utterly useless, but it taught me how to learn languages, more or less. This is its only contribution to Chinese, which I started in college as my third language. (Yikes, memories!) On the other hand, some years later, it was in fruitful dialogue with German: Only a good first-year German teacher got me to understand the subjunctive, which suddenly made perfect sense in Spanish, while Spanish had been my most recent European language with declinations. Finnish, which I -- looooong story -- ended up studying for a little less than a year was my first agglutinative language, and it prepared me for Uzbek, which then turned into Uyghur.

As we move left through the diagram, it becomes less historical and comes more fully to represent what goes on in my head as I learn or work with a language. Take Mongol, which I studied intensively for a summer: In my head, it was just Uyghur grammar with different words. Indirect and direct speech especially just grafted themselves onto my Uyghur grammar. What this meant, unfortunately, was that my hard-earned Uyghur started to turn into Mongol! Oh, no! Thankfully, even though Japanese also resembles Uyghur, and my intuition for Japanese derives entirely from Uyghur grammar, it seems to be just different enough that there has been no cross-contamination.

There's a similar problem, now that I'm learning Swedish. Swedish, by the way, is a surprisingly helpful language to know for Xinjiang from the 1890s through the 1930s. Anyway, Swedish started off in my head as a dialect of German, and it drew from the same resources. Now it's starting to take over. What I really need is a reason to speak German regularly, which will help me differentiate it.

On the other hand, sometimes a low level of interlinguistic differentiation is helpful. Take the big blue cluster on the left. When I approach a Chaghatay text or a Uyghur conversation, the two languages are in dialogue in my mind. In fact, I couldn't get my Uyghur speaking to improve at all for about a year until I started reading Chaghatay regularly. It's partly, I think, because modern Uyghur writing is often so stilted and dry, as well as derivative of Chinese and Russian styles, and partly because the modern stuff doesn't talk about certain things that are still central to Uyghur life. I've found it helpful to approach Uyghur as part of a big, lumpy Ottoman-Persian-Chaghatay continuum that all sounds more or less like Uyghur in my head. Indeed, to read Chaghatay or anything really interesting in Uyghur, I end up pulling my Redhouse Ottoman dictionary off the shelf or looking things up in the online Steingass Persian lexicon.

This must be tedious, so here's the lesson: It gets easier. After a few languages, some really central ones, they all start to look more or less the same. Your brain gets used to the work of absorbing vocabulary, you develop strategies for learning grammar patterns, and, eventually, you feel like you've seen it all. Now, I still think Armenian would kick my ass, and I have no illusions about learning Naxi. Yet, working my way from language to language has been a good strategy. I'm lucky that I studied Uyghur, which has such a rich vocabulary from Persian and even borrows from Russian. That made it a great starting-point.

Actually, Uyghur might just be the best language in the world, in terms of its utility. Yes, I'm biased, but I'm also getting tired of taking classes. Without learning to speak Uyghur and, over time, differentiating the sound of a Persian word as opposed to a Turkic or Arabic one, scanning through prose would be substantially more difficult. I already know half of the Persian vocabulary I learn, and about a quarter of the Arabic, as well, though I've only just begun that long journey.