Sunday, April 5, 2009

A zillion questions about linguistics

I sometimes wonder what linguistics is about. "It" is about various and sundry things, to be sure. But does anything tie them together? Earnest-minded readers may here already suspect the baleful influence of Luhmann in the background - and they would be right.

This is how the foreword to Social Systems begins:
Sociology is stuck in a crisis of theory. Empirical research, quite successful on the whole, has increased our stock of knowledge, but has not led to the construction of a theory that would unify the various subdisciplines. Being an empirical science, sociology cannot abandon its ambition to validate its pronouncements against data gathered from reality, however old or new the bottles may be into which the pressings are poured. Yet sociologists have been unable to apply precisely this principle to demonstrate that sociology has its own special area of study, and is a unified scientific discipline. Resignation about this state of affairs is so profound that no one even tries anymore.
One might reasonably expect linguistics to have some kind of general, unifying theory. But what are the common ideas of phonology and grammatology, say? Chomsky didn't think actual languages were much worth bothering with, as I remember - UG is embedded in a reductionist program. At a pinch, would one say phonology and grammatology are "just very different aspects of the same thing, language"? What is this "language"? Is it different from "languages"?

Isn't it a bit strange that this kind of unanswered, perhaps even unproductively formulated, question seems to be completely irrelevant to the everyday work of linguists? But does "we're all just laboring in the vineyards of the Lord, so go away" say all that could usefully be said in answer to questions that are seldom if ever asked?

Is there a more satisfactory account of "what is linguistics about" than there is of "what is sociology about"? Surely physicists should have something useful to say about physics (not just about physicists), which after all is part of the physical world? Physicists were forced to start doing precisely this about 100 years ago, upon the advent of quantum mechanics to explain how physics is non-trivially intricated in physics. Surely linguistics is non-trivially intricated with language? If Sapir/Whorf had made their claims with regard to the terminology of linguists themselves, not Eskimo Joe's words for snow, the ensuing ruckus might have been more productive.

On the internet, discussion threads among linguists often seem to me to be squabbles about taxonomy: "is this an adverb or an adjective?", "did /x/ mutate into /y/ or the other way around?", "is [a] an etymological descendant of [y], or is [a] on a different branch?". It reminds me vividly of the enormously productive classification activities in botany and zoology up into the mid-19th century. Darwin drew heavily on the results, and contributed his own. Then he took a step further, and blew many of his colleagues out of the water.

Are languages just "there, to be studied", like species used to be? Do languages evolve from other languages on the analogy of organic species? If phonology is not a subfield of biology, or musicology or acoustics, what is it? Is everybody sure it has nothing to do with semantics? On what principles has this been demonstrated? Or is this an axiom, making linguistics more like a field of mathematics? As presently visible in discussion threads, at any rate, many parts of linguistics seem to me more like cross-dressing cousins of ontology.

15 comments:

marie-lucie said...

Darwin was actually inspired by the model of the evolution of languages (the main preoccupation of 19th century linguistics) in his ideas about the evolution of species.

It is true that linguistic discussion threads on the internet are mostly about small details, but that is because languages are so complex. And it is true that current linguistics is rather fragmented (and so are some of the sub-fields), but it does not mean that no one is thinking more comprehensively. The image that just came to my mind is of a spiderweb anchored in different points - language touches almost everything connected to human life but is distinct from it.

At least one person has tried for a comprehensive view of language:
Kenneth Pike 1967: Language in Relation to a Unified Theory of the Structure of Human Behaviour (The Hague: Mouton)
This is a huge book, probably very dated theoretically by now (I opened it once but was scared off by the sheer bulk of the book).

Stuart said...

Oh no, not another huge book! I should look at it, though, based on the title. I still hope someone will give me a longish, stroppy answer to my likewise post.

Luhmann wrote yards of stuff, but I'm only at first down so I don't worry. With reference to your remark chez Hat, I don't find him at all hard to understand, in German at least . You know how I complain about English translations of German contemporary philosophers. It would be so much more efficient for people to have learned German if they want to read some of these guys.

The lucidity of L'Être et le Néant! The horribleness of the English version!

marie-lucie said...

I find L'Être et le Néant pretty horrible in the original too!

marie-lucie said...

The image of language as a web seems to be quite common, but what my image was focusing on was the connectedness of language to other features of life, from the physical (phonetics) to the social, the emotional, etc, which are outside of language but anchor it in reality both concrete and abstract.

Stuart said...

I'm not so sure about anchoring in a common reality. Take the fact that we disagree so strongly whether EN is horribly or lucidly written. It appears that we can then only signal to each other "about EN", but I, at least, can't understand how you could come to emit "horrible" signals about this particular book. It might make me wonder if we are talking about the same aspects of the book. But suppose I don't think or talk about the matter thereafter.

This is quite different from the situation in which we agree about some aspect of EN. I would then be experiencing no difference between our points of view: and yet it may turn out later that you meant by "lucid" something unexpectedly different from what I meant by it. And yet, precisely because we appear to be in agreement, I have no particular motive to think or talk about the matter thereafter.

In other words, only an ongoing conversation gives you things to think about. Agreement is a bus-stop - it's where you get off and stop thinking. Disagreement is what happens between bus-stops. So we have to keep talking, and in particular disagreeing, if we want to get anywhere. I view language less as an anchor, than as a sail.

I don't have EN at hand. Here's the first extract I found on the net. It's about the philosopher at a keyhole. He comes to realize that it's all in his head - the point being that it is there, not elsewhere. "Other people" (autrui-sujets) are not just "out there" trying to find me, while I am "in here" trying to hide from them. "Ce qui est douteux, ce n'est pas autrui lui-même, c'est l'être-là d'autrui". The problem (if it is one) is not how to deal with other people, but how not to deal with them. In a melodramatic moment of Huis Clos, Sartre says "L'enfer, c'est les autres".

... ce ne sont jamais des yeux qui nous regardent : c'est autrui comme sujet. Reste pourtant, dira-t-on, que je puis découvrir que je me suis trompé : me voilà courbé sur le trou de la serrure ; tout à coup j'entends des pas. Je suis parcouru par un frisson de honte : quelqu'un m'a vu. Je me redresse, je parcours des yeux le corridor désert : c'était une fausse alerte. Je respire. N'y a-t-il pas eu là une expérience qui s'est détruite d'elle-même ?

Regardons-y mieux. Est-ce que ce qui s'est révélé comme erreur c'est mon être-objectif pour autrui ? En aucune façon. L'existence d'autrui est si loin d'être mise en doute que cette fausse alerte peut très bien avoir pour conséquence de me faire renoncer à mon entreprise. Si je persévère au contraire, je sentirai mon coeur battre et j'épierai le moindre bruit, le moindre craquement des marches de l'escalier. Loin qu'autrui ait disparu avec ma première alerte, il est partout à présent, en dessous de moi, au-dessus de moi, dans les chambres voisines et je continue à sentir profondément mon être-pour-autrui ; il se peut même que ma honte ne disparaisse pas : c'est le rouge au front, à présent, que je me penche vers la serrure, je ne cesse plus d'éprouver mon être-pour-autrui ; mes possibilités ne cessent pas de "mourir", ni les distances de se déplier vers moi à partir de l'escalier où quelqu'un "pourrait" être, à partir de ce coin sombre où une présence humaine "pourrait" se cacher. Mieux encore, si je tressaille au moindre bruit, si chaque craquement m'annonce un regard, c'est que je suis déjà en état d'être-regardé. Qu'est-ce donc, en bref, qui est apparu mensongèrement et qui s'est détruit de soi lors de la fausse alerte ? Ce n'est pas autrui-sujet, ni sa présence à moi : c'est la facticité d'autrui, c'est-à-dire la liaison contingente d'autrui à un être-objet dans mon monde. Ainsi, ce qui est douteux, ce n'est pas autrui lui-même, c'est l'être-là d'autrui ; c'est-à-dire cet événement historique et concret que nous pensons exprimer par les mots : "Il y a quelqu'un dans cette chambre."

SnowLeopard said...

You seem to be asking several questions here, none of which I'm qualified to answer. But I'd suggest that physics is probably unique among the sciences in being able to offer a Standard Model or Grand Unified Theory, however flawed. If you consider other fields, mainstream journals seem to be exploring issues today that would likely have been considered heresy a few decades ago -- epigenetic inheritance, for example. Another example would be mycology, where important or fundamental traits of most fungi, such as toxicity, are simply unknown because no one's yet collected the data, and even classifying an examplar to the genus level can be an enormous challenge. I think of linguistics as another field where we still know too little to generalize with confidence. Sure, you could try to assemble a grand unified theory of language without collecting more data, but it would most likely be wrong -- and so wrong as to perhaps not have been worth the effort. With regard to syntax, as I understand it, the sentence structure OVS was declared "non-existing" in the 60's, before the discovery of Urarina in Peru, and Udi in the Caucasus has also created a stir for reasons that I'm less clear on. A discussion of phonology without an examination of Khoisan clicks or the world's sign languages, or theorizing about meaning without considering, say, Amazonian Tariana's obligate markings for evidentiality, or Yup'ik Eskimo's complex but highly informative palate of demonstrative pronouns, seems equally premature. In other words there's simply a staggering amount of data left to assimilate, much of it still unknown to science or, at best, gathering dust in some university's forgotten archive of unpublished manuscripts, before the field can venture to formulate something meaningful. As to whether there are deeper, unifying themes connecting phonology, syntax, and other subfields within linguistics, I suspect that there are, especially from the standpoint of language acquisition. I seem to recall a basic tenet being that children simply were not exposed to enough linguistic material to be able to deduce the phonology, syntax, and other structural rules of their native language without being endowed with an innate framework or template to help them fill in the logical blanks, and that these subfields were all trying, in part, to deduce the nature of those templates. But marie-lucie and other experts can opine more knowledgeably as to whether that sort of project is still in fashion. I imagine this wasn't the enlightening comment you were hoping for, but I hope it motivates some better qualified reader to help point you in the right direction.

Stuart said...

Thanks, SnowLeopard! Of course what you say is enlightening, precisely because we are not quite talking about the same thing, as it appears (see my post above that I apparently squeaked through just before yours). I think all that "agreeing to disagree" stuff marks a pretty tepid frame of mind. That doesn't mean I think people should be at loggerheads with each other all the time. Rather, there's no honest way to agree to disagree. Dropping a subject, or not bringing it up, is a different thing entirely.

SnowLeopard said...

Ah. In that case I seem to have misunderstood the point of your post entirely-- sorry about that. I can only assure you that, going forward, I will undertake to misunderstand you differently.

I suspect that when most people "agree to disagree", they don't literally mean what they say; that would be an admission that the issue was irrelevant and they'd been wasting their time. (The Luhmann quote does suggest, however, that some such apathetic people do walk among us.) The way I mean the phrase, and I suspect others do as well, is to agree to reserve final judgment until we have more data, or else to await the arrival of some third party who can serve as tie-breaker. In other words, it's an agreement that neither side has abandoned its principles in the short term, while both still hope for eventual resolution. Such diplomacy need not be either dishonest or tepid, and under appropriate circumstances may serve as a respectful olive branch by someone otherwise perceived as heavy-handed or condescending, and which in fact does clear the road for an eventual political solution.

But in all candor, I don't see what this has to do with your original post, and so I fear that I'm just cluttering up your tidy comment section.

Stuart said...

SnowLeopard: You say you think of linguistics as another field where we still know too little to generalize with confidence. But there is no lack of data about phonology, word origins, grammar and so on. And general principles about these things are in daily use - "a phoneme modifies into another phoneme", "the meaning of words changes over time", "grammar is analyzed with O and V and S".

As to generalizations altogether, I never understood the "induction" that is supposed to name a process by which encompassing statements are obtained from disparate data. That subject has pretty much dropped by the wayside. The common practice in all kinds of science nowadays, I would hazard to say, is not to worry about how generalizations arise, but instead to see whether, in different contexts, this set of data and this set of statements are in accord with each other. When they are, this set of statements can be called a generalization, but that is not so important as the accordance.

This is a very static and misleading description of what is goes on in, say, molecular biology,and even in mathematics - but I won't try to go into that here. One eye-opening book I've read in this connection is Rheinberger, Experimentalsysteme und epistemische Dinge [Experimental systems and epistemic things]. Rheinberger is a molecular biologist and historian of science. Currently he is head of the Max Planck Institut for the History of Science, in Berlin.

It was only inasmuch as you seemed to be writing about "grand unified theory" as a kind of generalization about generalizations, that I answered that we seem to be not quite talking about the same thing. Farther on in your post, you write (coyly omitting to mention the name of Chomsky)

As to whether there are deeper, unifying themes connecting phonology, syntax, and other subfields within linguistics, I suspect that there are, especially from the standpoint of language acquisition.

Now that would be more like a theory of the kind I was talking about. The only trouble is, none of the linguistical discussions I see on the internet invoke such a theory. Chomsky himself seems to have changed his mind several times, but I know nothing more about that.

Even O and V and S were around before the idea of Universal Grammar (which only applies to grammar, not to phonology). The Great Vowel Shifts are not understood in the context of a theory which applies to grammar as well. But it's not that I am hell-bent on having linguists find such a theory. There seem to be several kinds of theory already in practical use, but unanalyzed and unnoticed as such. They work as "that's the way it is" ideas. What they all have in common is a vague notion something like this: "there is a species/genus which is language, whose varieties/species are called languages. Languages have characterics which can be classfied. The characteristics are subject to variation over time. Languages cross-fertilize each other". In other words, 18th-century taxonomy of varieties.

Stuart said...

In quoting from your post I should have included the subsequent sentence as well, to make the Chomsky connection:

I seem to recall a basic tenet being that children simply were not exposed to enough linguistic material to be able to deduce the phonology, syntax, and other structural rules of their native language without being endowed with an innate framework or template to help them fill in the logical blanks, and that these subfields were all trying, in part, to deduce the nature of those templates.

Are you sure that deduction is what is going on when people speak? Language acquisition requires practice, just as learning to play the piano does. When you play the piano, are you performing acts of logical deduction? Let's imagine that some kind of deduction was involved in the acquisition of the ability, but was not needed later. Isn't it odd that the deduction is seen as preceding the ability? I would have thought command of the practice was a prequisite for making deductions about it - since you can deduce only from what you know.

Suppose we replace "deduce" with "being guided by templates and patterns". What is the advantage of positing "deep" patterns? Can't we regard the process of acquiring language as the process of practicing patterns provided by other people? As others have said, and surely better than I am doing here: universal grammar seems to be explaining something away, rather than explaining it. It replaces an unknown by another unknown with which one can do cool (at the time) mathematics, namely the Chomsky hierarchy.

That was productive only in computer science, however. The Chomsky hierarchy was shown to be equivalent to four types of automaton, and that was the end of that. Compiler theory took off from the linguistic airport, and never came back. Non-Chomskian línguists - at least all linguists not working with grammars - appear to be still playing cards, and not worrying about the next flight out, since they're content to be where they are. (Remember I'm talking about linguistic discussions on the internet, and asking whether they are characteristic of the general practice of linguistics).

Stuart said...

I somtimes get carried away with my facetiousness, and write things that sound denigrating or impertinent. When I just wrote about "linguists playing cards", it didn't immediately occur to me that that could sound like I think that what linguists do is no more important than a game of cards.

What I was thinking about was two things: playing poker (for high stakes), and just doing what one enjoys doing. Of course I do that myself at work - when I can. But the fact that I myself despise my job doesn't mean other people are feel the same way about their jobs.

Sorry, folks.

SnowLeopard said...

Well, again, I can't claim anything approaching passing familiarity with the current state of linguistics-- I'm a labor and employment attorney who took one linguistics class in college, majored in philosophy, and then went to law school because I found it discouraging that my university professors' views of the world seemed to be derived purely from talking to each other. I do collect technical materials on and study a lot of languages, though, and read the occasional book on, say, evidentiality or grammaticalization. So my theoretical credentials to debate this are probably nil, and I'll happily defer to my betters. That said, I do think that while there's a lot of data out there, it's hardly systemic, and largely incomplete, and may not be of adequate quality to perform all the tasks that may be required of it. (The only grammar of Chukchi I've been able to locate, for example, was published by an anthropologist a century ago. There may be something more current in Russian, but my Russian is in very sorry shape, as is my German, and is generally fit only for parsing museum placards.) I could try to come up with some examples of open questions if you think it's relevant. As for the quoted propositions in the first paragraph of your first follow-up comment, yes, those do appear to be working principles in the field, but lie at such a level of abstraction (compare: living organisms are comprised of cells) as to have very little practical application or predictive power. The Physics Standard Model, by contrast, has gone well beyond "matter is made up of atoms" and instead has draft equations that can be used to predict the outcomes of experiments with varying degrees of accuracy.

From my own regular reading of Nature and Science, I find that scientists these days are generally very cautious about offering bold new theories and tend to offer submissions of an extremely incremental nature -- the proposed structure of a particular protein, the discovery of a microorganism with a previously unrecognized type of metabolism, an unforeseen superconducting material. Those that offer grand new theories -- and I'm thinking of Wolfram's book, which I haven't yet read-- seem to be dismissed or even derided by those who consider themselves the mainstream. There's probably a lot to be said about the politics in that, especially the likely inability of the current system of peer review to cope with a new Grand Unified Theory of any sort. This general phenomenon may also have bearing on your complaint that linguists are only talking about taxonomies on the internet -- to a certain extent internet communications are likely to be much more topical and narrowly-focused than the sort of thing you'd find in a published monograph. If you peruse, say, arXiv.org, you'll find precious little in the way of grand unified theories even for physics, and I've found hardly anything, in print or on the internet, that can intelligibly explain quantum entanglement or precisely how quantum computation is supposed to work to my satisfaction. (On that point, Feynman's Lectures on Physics and other books, and Scott Aaronson's blog, are the closest I've been able to get, but they don't get me all the way there. Nielsen's book on quantum computation assumes far too technical a background for me to be able to handle it at this point, and anyway he seems to take the physics for granted.) That the available materials are inadequate for my purposes may be attributable to my own ignorance or bone-headedness, I readily acknowledge. I often have trouble keeping up with the program.

In your second follow-up, I never meant to claim that deduction was what was going on when we learn to speak, play the piano, or otherwise -- my actual view is quite the contrary, that the mind is comprised of a large number of largely independent, specialized functions and subroutines that really have nothing whatsoever to do with logic. Wilson's Strangers to Ourselves was the first thing I've read in a while that resonated with my own intuitions on that point. I'm sorry if my writing was unclear there. Your alternate phrasing of "being guided by templates and patterns" is closer to my naive understanding of what's going on in language acquisition, precisely because, as Chomsky may have argued at some point or other, deduction is inadequate for the task. This is partly why I hoped that a qualified commenter could step in, but I remember reading about a theory in which basic phonological and syntactic features in a language were reduced to a series of binary decisons which a child could navigate through observation of the world around her, and thereby acquire the basic rules of pronunciation and word order of the language. Part of the problem may be that the binary reduction of features turned out to be too simplistic, especially for phonology. For example, I don't see how this scheme would begin to account for the radically different tone systems of Mandarin, Cantonese, Thai, Vietnamese, Tibetan, and Navajo, and the European voiced/voiceless division of consonants is completely inapplicable to Korean, which has a tripartite division in which voicing is irrelevant. I have less to say about syntax mainly because it takes so much longer to master in your study of a particular language, but I imagine that there are similarly grave impediments there as well -- the syntax of Korean and Farsi in particular seem to give me trouble, though for different reasons. Which seems to take me full circle. I, for one, despite studying from a minimum of seven languages every week, and sometimes tasting double that number if things are going well (yeah, I'm weird), may be able to entertain basic, survival-level conversation in many of them but am unable to offer general, unifying comments about their sentence structures. Things that I can speculate about -- why do so many disparate languages use the verb "to have" to form a past tense -- in turn depend on a larger library of data than I've assembled at this point. Linguists have the luxury of spending more time on such things, and better resources, but I suspect that many of them are in the same situation -- there's simply too much material to assimilate to be ready to say anything of general and meaningful significance. I remember reading an obituary in the Economist a few years back about a linguist at MIT who had held a world record by being fluent in 50 languages(he complained that Dutch was the hardest, and took him a week), and who aspired to write something systematic, but he never got around to it.

Stuart said...

I'm going to need a moment to get back to you on that...

CL said...

I know this comment thread has been quiet for a while now, but I stumbled on it and thought I'd chime in.

SnowLeopard, the theory you're thinking of is generally known as "Principles and Parameters" (Chomsky & Lasnik). The parameters are the binary decisions you mentioned and the principles are the universals that all languages obey. When this was first proposed, the identity of the parameters was taken to be an empirical question and consequently a lot of research has been done on language typology. (For example, investigating whether languages are always either head-first or head-final.)

While it does seem to be true that languages cluster (that is, if a language uses a head-first order in its prepositional phrases, it's likely to also have a head-first order in other phrases), the picture is not as clearly binary as a strict parameter treatment would have it. It's still an influential idea, but not nearly as promising a theory as it once seemed.

The Poverty of the Stimulus argument, which Principles and Parameters was developed to address, is alive and well... and the subject of much ongoing debate in the language acquisition literature. There's a pretty decent, if brief, Wikipedia article on it, if you're not familiar with it: http://en.wikipedia.org/wiki/Poverty_of_the_stimulus.

PS- I'm mostly familiar with the syntax side of the debate, but I believe that the theory was extended to morphology and phonology... and that similar clustering but not completely binary patterns have been found in those domains, as well.

marie-lucie said...

SnowLeopard, for someone who calls himself a non-linguist I think you are pretty up to date on linguistics.
At this point I don't have much more to say on the topic, but if you are looking for a grammar of Chukchi beside the one by Bogoraz there is a recent one by Michael Dunn, a dissertation done at the Australian National University a few years ago. The author speaks Russian and has spent time in the Chukotka.

The 50-language linguist was probably Kenneth Hale, who died a few years ago. He was fantastically gifted for language acquisition.