The Problem With Structuralism

Begging the Question

 

I have no problem with structuralism in general - only with a specific, narrow trend within structuralism, which became briefly popular in the 20th century.  Indeed, I would consider myself a structuralist (with a lower case "s"), or at least, in my own naive way, sympathetic to structuralism, in at least two senses: first, mathematico-ontological structuralism, pioneered, in my opinion, by Henri Poincare, and developed more famously by Paul Benacerraf (and in this I am chiefly indebted to my many conversations with the contemporary mathematician, Howard Blair); and secondly, political structuralism - that is, the insistence that the best strategy in politics is not to focus on individual people, but rather to demand nothing less than to change larger, impersonal social structures.  Both of these are called "structuralism," but neither of these are the topic of the current essay.  Instead I want to narrowly discuss the brief fad called "Structuralism" - which I distinguish with the use of the capital "S" - which unfortunately is what most people are talking about when they mention structuralism.

There are many flaws in the reasoning of Structuralism, but the clearest and most undeniable one is the classical informal fallacy known as "begging the question," also known as "assuming the conclusion" (petitio principii).  This fallacy occurs when an argument starts by making an assumption, and then "derives" a conclusion that includes that assumption, as if the assumption had been proven.  Essentially, it is a form of circular reasoning.

It's not clear to me that it was the fault, or the intention, of the Swiss linguist Ferdinand de Saussure, generally recognized as the founder of Structuralism, to make this faulty argument.  I choose to read Saussure according to the principle of charity, and I think there is a way of steelmanning Saussure's work in such a way that he did not really commit this error at all, or if he did, only in a minor way that was peripheral to his central argument.  Rather, I think it was Saussure's students and soi-disant followers who misinterpreted his work and then used it to justify every kind of biased and confused pseudo-reasoning, to their hearts' content.  I think Saussure would be rolling in his grave to see what kind of garbled nonsense people were producing in the name of his terminology.  

I do not need to go very far into Saussure's work here, to get my point across - perhaps a deeper dive into Saussure could be the subject of another essay, or series of essays.  Let it suffice to say, for now, that I think that Saussure was simply assuming that the rules of any given language, including the relations between a set of signs and a set of signifiers, are fundamentally arbitrary, for the sake of simplicity in describing some very specific features of languages.  But his work was taken to be a proof that the relationship between signs and signifiers is always arbitrary, and this "principle" was taken, by the followers of the Structuralist religion, to hold not only for language, but for mythology, for cultural practices, for anthropology, for religion, for economics, for gender, indeed, for everything that is "socially constructed," indeed, up to and including science itself.  In fact, Saussure had never proven any such thing, nor had he claimed to, or intended to.  In short, this principle of arbitrariness had become a dogma.  (And the "post-structuralists," though they claimed to have broken with structuralism, actually doubled down on the principle of arbitrariness, and carried it to unforetold extremes.  Derrida's work, in particular, opens up vast new vistas for the possibilities of arbitrariness.)

When Saussure asserted the arbitrary relation of signs with signifiers, he was positing this hypothetically, for the sake of simplicity and for the sake of argument.  (Perhaps this was loosely analogous to Foucault's famous declaration, decades later, that "I am fully aware that I have never written anything other than fictions.  For all that, I would not want to say that they are outside truth.  It seems possible to make fiction work within truth...." [from his interview with Lucette Finas, quoted in "Michel Foucault: Power, Truth, Strategy," 1979]?)  To imagine that this Gedankenexperiment, as Einstein might call it, corresponds in any way with the real world is simply to misunderstand Saussure.

Doubtless, many features of many languages are somewhat arbitrary.  No question about that: there are many obvious examples.  But are all languages always absolutely arbitrary?  How could that be possible?  What would that even mean?  

If language were truly absolutely arbitrary, how could language be learned?  How would one generation acquire language from the previous generation?  If the relationship between words (or signs) and their meanings were entirely random, how could a parent teach words to their children?  Sure, there's ostension: a parent could point at a duck and say, "duck".  But how would the child know what pointing meant?  This, too, is a sign. Wouldn't that, also, be totally arbitrary?  To indicate "that thing over there," instead of pointing, couldn't a person flap their arms up and down in a 5/4 rhythm as they hop on one foot while turning counterclockwise?

The answer, of course, is that while signs might, in some sense, "start out" as completely arbitrary (or nearly completely arbitrary - more on that in a moment), they won't remain arbitrary for very long - on the contrary, very complex patterns and rules and regularities inevitably emerge.  One example is the famous "Zipf's Law": the use of any given sign is inversely proportional with its rank among common signs.  Thus, the second most common word in English, "be" should appear about half as often as the most common word, "the," and the third most common word, "and," should appear a third as much, the fourth most common word, "of," should appear a fourth as much, and so on.  Zipf's Law has been shown to hold across languages (and in other domains as well).  

Why would regularities emerge?  Several reasons.  For one thing, we can imagine a process something like Darwinian natural selection: mutations may be (or may not be) completely random, but a mutation that doesn't help or even hurts an organism's chance of survival and reproduction will eventually be filtered out of the gene pool, whereas helpful changes will be preserved, thus giving rise to more and more complex structures in an organism which make that organism better adapted to its environment. Similarly, although language transmission from one generation to the next will never be a perfect copy and thus languages are constantly evolving and changing, they don't necessarily change randomly or arbitrarily.  If an acquired language caused a person to make meaning of their world in a way that was perfectly arbitrary and absurd, that person would be severely hampered by this, and would not have as great chances of passing language onto others.  Language should help us to adapt to our environment, which has its own obvious regularities.  We should expect people to use the word "water" much more than they use "bitartrate."

Even more straightforwardly, there is of course simple association.  If I statistically tend to use the word "cup" more often when holding a cup in my hands, and you witness this tens of thousands of times, eventually you will associate the word "cup" with cups.  Undoubtedly, you may develop all kinds of associations due to statistical anomalies and coincidences - false positives, so to speak - detecting patterns that aren't there.  For instance, you may think the word "goodbye" means "doorknob," just because I tend to be reaching for the doorknob every time I say "goodbye".  But not all of the associations will be false positives.  Indeed, people can figure out a lot from context clues.

For that matter, language acquisition is not all "accidental," so to speak, as in the case of the person who happens to be holding a cup and saying cup.  Parents and caregivers are (often) actively, intentionally teaching their children language.  This happens in a variety of ways but for one, there is, of course, positive (and negative) reinforcement.  Sure, a child might misinterpret pointing at first.  If a parent points at a duck and says, "duck," the child might think that the parent is conveying the message, "My finger is called 'duck'."  Or they might interpret a pointing finger to mean "Mom" or "Love" or "sodium bitartrate," or any other arbitrary thing.  The relation between the sign and the signifier is indeed arbitrary - at first.  But if the parent rewards the child when they randomly stumble on the right answer, by opening their eyes wide and cooing "Good girl! Yes! You're so smart!" and either punishes or does nothing when the child randomly stumbles on the many possible wrong answers, the child will learn to home in on the "correct" meanings (however ultimately arbitrary they may be).

Chomsky, of course, thinks there is another explanation: an underlying "language module," some kind of built-in structure in human brains, which explains the many regularities among languages.  Who knows?  There might be some truth to this.  

But even if there is no such thing as a "language module," quite a surprisingly large amount of the regularity of language might be explained by simple statistical regularity.  This is, of course, how LLMs (large language models, like ChatGPT, Claude, and Gemini) have acquired language from humans.  They've been able to absorb an extremely impressive amount of the structure of language simply from "scraping" data across the internet and calculating, given a string of "tokens," (groups of characters such as syllables or words) the most statistically likely next word to follow.  It's a bit uncanny how much they've been able to glean from such a simple m.o..  (Their "training" is also a bit like classical behaviorist conditioning, as they try to maximize their "reward function".)  Maybe babies do something similar; or perhaps not.  It's too early in the research to say for sure.

I cannot let the moment pass without at least mentioning my two favorite theories for how language originated among humans or among our primate ancestors: namely, that language arose from a system of gestures, such as some kind of sign language or body language, or - even better - that language arose out of music.  Or, of course, some combination of the two, perhaps in combination with yet other phenomena.  Let's take the sign language theory for an example.  It seems to me that there's no obvious reason to assume that the sign language of our hominid ancestors was completely arbitrary.  For instance, its easy to imagine a person explaining how to cook something with a series of gestures, many of which would be anything but arbitrary - indeed, which would be inextricably pragmatically tied into a demonstration of the cooking labor process.  (By the way, Engels once said something quite similar.)

The truth is that producing anything that is perfectly arbitrary is something that human brains - especially adult human brains - are extraordinarily bad at doing.  It is almost impossible, for instance, for a human to produce a truly random series of numbers, as researchers like Figurska et al. have shown.  

In any case, it seems to me that there's still a lot to learn.  And that, in itself, is saying something.  For what the Structuralist doctrine of absolute arbitrariness ultimately says is that there's nothing to learn.  There's nothing for us to know, because it's all totally random and arbitrary.  It seems to me that this amounts to saying that you have given up trying to learn, perhaps because you don't want to know the truth.

In my opinion, some of Saussure's followers - especially many of his French followers - are committing the absolutizing fallacy in elevating arbitrariness to an absolute principle - once again, as usual, trying to smuggle theology back into their so-called theory.  Their God - the God of Infinite Arbitrariness - resembles the God of the Book of Job, who, having been petitioned by Job and his friends, who beseech Him with what amounts to an anguished "Why?" responds by not responding: "Who is this that darkeneth counsel by words without knowledge?  Gird up now thy loins like a man; for I will demand of thee, and answer thou me.  Where wast thou when I laid the foundations of the Earth?  Declare, if thou hast understanding." (Job 38)  In other words, God says, "I'll be asking the questions around here."