At some point in middle school, my classmates and I were obsessed with a mobile game called Alchemy. The rules are simple. You start out with four “primordial” elements (water, fire, earth, and air), and combine them into new ones by placing them together on the canvas and letting them react. Newly obtained elements can then again be mixed with old ones, resulting in yet other elements. I listed a handful of such reactions below, but the original game has hundreds of them.
water + wood = boat
paper + story = book
milk + time = cheese
tool + light = flashlight
water + cold = ice
As you mix and match elements in an attempt to discover all of them, you gradually internalize the semantics of the Alchemy universe. For instance, you get to understand a book as a blend between paper and story. You learn that in Alchemy, ice is defined as a mixture of water and cold. Even if the thrill of the game is to identify elements which react with one another, the gratification of discovery is inevitably accompanied by a surprising mental model.
After picking up Chinese a couple months ago, it didn’t take long to recognize the same Alchemy-like mixing mechanic. Chinese words can be used as raw elements in the process of synthesizing new ones. Even if words are also made from smaller pieces (e.g. prefixes, suffixes, affixes, roots) in all the other languages I’ve dabbled in before, it feels radically different in Chinese. The raw elements feel so much more versatile and reactive. So much meaning lies dormant in a handful of quick strokes, it feels like they’re soaked in it, made from it. In the limit, as there are no word delimitations in day-to-day writing, whole sentences feel like rare elements obtained through long converging chains of semantic reactions. Chinese feels much closer to Alchemy than to any other language I’ve seen before:
电 (electricity) + 脑 (brain) = 电脑 (computer)
老 (old) + 师 (master) = 老师 (teacher)
家 (home) + 人 (person) = 家人 (family)
电 (electricity) + 影 (shadow) = 电影 (movie)
女 (woman) + 子 (child) = 好 (good)
After digging around a bit more, I learned that Chinese is classified as an analytical language, as opposed to an agglutinative one. These terms skillfully capture the dichotomy described above. Analysis is the study of how parts give rise to wholes, which closely matches the style of reasoning required for learning Chinese. In contrast, agglutination is the process of stitching things together, which in the linear medium of language takes the form of chained morphemes which have to be dissected sequentially.
Even if the Alchemy-like nature of analytical languages is an intriguing topic on its own, what I find even more captivating is how Chinese culture leaks through and permeates the language through the semantic reactions. Whole packs of mental models, ways of conceptualizing the world, are baked into the very fabric of the language, just as was the case in Alchemy. The very way of referring to a computer makes use of the analogy of the electric brain. A family is conceptualized in terms of the people at home. Allegedly, people of ancient China conceived of good as having a wife and a child. The language is as much an artifact as it is a carrier of a worldview, and it’s fascinating.
A detour to the expert systems of 1960s might be informative as to why this happens. If you endowed such a system with knowledge of the world (e.g. health symptoms) and with rules for reasoning about it (e.g. symptomatology of diseases), it could reach surprisingly accurate verdicts (e.g. healthy). Crucially, expert systems had to either rely on forward chaining or backward chaining. Forward chaining refers to deriving all possible facts based on the initial knowledge base (e.g. not suffering from A, B, C…), hoping that the target will be verified in the process. Backward chaining means reasoning backwards from the target, working your way back to the initial facts (e.g. What symptoms would be present if the person was to have this sickness?). Alchemy is all about forward chaining, and so appears to be Chinese on a fine-grain level. However, when reflecting on what concepts mean, what we do is closer to backward chaining, as we’re understanding them in relation to other ones. This duality highlights the link between knowing how to follow the recipe and what the end result is made of.
Let A and B be the initial facts, and let Q be the target query which has to be verified. With forward chaining, the system would derive L, M, P, and Q, in order, by applying whatever rules are available. With backward chaining, the system would see that Q depends on P, then that P depends on L and M, and so on, until all the dependencies are eventually reached and, in this case, met. (figure credits: Davide Grossi)
Expert systems aside, can analytical languages in themselves serve more practical goals? What if you could tap into the fine structure of Chinese culture by using it as the semantic backend of an Alchemy-like game? What if you could instantly swap it with another similar language? What if you could decompose concepts against a certain worldview, essentially reversing the semantic reactions? In the meantime, learning new languages appears to be one of the most reliable methods of tapping into exotic ways of conceptualizing the world, so happy learning!