Disclaimer: The views expressed here are mine alone, and not necessarily those of my employer or any organization with which I am affiliated. These views are not intended to advertise or offer legal services to the reader, or to be relied on as practical advice in any respect. Apparent statement of facts may or may not have been specifically researched beforehand. Unless I expressly indicate to the contrary, the material appearing here is original work, subject to copyright protection. Any reference in the text to specific individuals or companies who are not explicitly named is unintended and purely coincidental.
Comments? You can (try to) contact me at admin (at) limitsofknowledge (dot) com. Keep in mind that I'm still learning the technical aspects of blogging, and do have a demanding job, so don't be offended if it takes me a while to respond.

Saturday, March 17, 2007 -- 6:24 pm

Since I've been throwing around the word "expressiveness" a lot, I suppose I should see if I'm able to define what I mean by the term. There's probable not so much one thing I mean by it as a cluster of overlapping or related meanings. I'm primarily concerned, I think, with grammatically-expressed meaning rather than lexical meaning. Yup'ik Eskimo has two different words meaning "to drink", depending on whether the beverage is hot or cold, and Quechua distinguishes between drinking alcoholic and non-alcoholic beverages. But any language could easily adopt vocabulary expressing these dichotomies if its speakers found that interesting or useful in some way. On the other hand, English has few if any evidential markers ("allegedly" is a very tired word) and struggles when it comes to translating those found in American languages. Take the Hopi word as, which Hopi Dictionary: A Hopi-English Dictionary of the Third Mesa Dialect defines with two columns of examples summarized as "modal of compensation: something is claimed to be so without full evidence, despite contrary factors, or how things were or will be", including expressions of desire, polite requests, relief, surprise, and doubt. I don't want to define expressiveness the way Frost defined Poetry, i.e., as what was lost in translation, because that simply isn't useful. I also don't want to dwell overmuch on syntactic or morphological idiosyncracies that may be interesting structurally, but aren't what we normally think of as expression. Arabic gets along perfectly well without a word for "of", as does Russian without "to have". Many languages dispense with "to be", a verb with the tellingly vapid name of "copula". Yucatec Mayan, seems content to hang its tense particles at the front of the sentence like a shingle without clear attachment to any verb whatsoever, while Spanish seems to have an interesting thing going with two copulas that are used in different situations, but which may not, in fact, contribute terribly much to the meaning of the sentence.

So perhaps, provisionally, I'll say that a word or morpheme is "expressive" in a sentence of a particular language when the sentence that results from its removal, whether grammatical or not, has suffered a loss in meaning. The meaning (or meanings) that is lost is the expression. It's not a satisfactory definition, I know, but perhaps I'll be able to refine it as we go along. Let's see where it takes us.

Here, I see no particular loss of meaning from the omission of the word "is", unless we're going out of our way to emphasize the fact that it's all going on right now. However, between (2) and

meaning, in the reference to past time, has been inserted.

Under this rubric, how would we go about measuring whether one language was more or less expressive than another? As an initial matter, we'd have to take care to specify what we were talking about. English has a rich complement of tenses, for example, but fails to distinguish between dual and plural numbers, and fails to distinguish between "we with you" and "we without you", like Cherokee, Quechua, and Aymara. I would therefore consider English to be less expressive than Cherokee, Quechua, and Aymara in this respect. There also seems to be a principle of economy at work here: because English obviously can express the same distinction as these other languages; it just has to expend more syllables to do it.

Friday, March 16, 2007 -- 10:05 pm

In my subway reading -- currently Hopper's Grammaticalization, I came across the term "abduction", used to describe a process of reasoning. I suppose I was already familiar with the phenomenon from my college days with Kant's Critique of Judgment, but had never heard that term used to describe it. (Briefly, whereas 'deduction' is a demonstrably sound and reliable reasoning process that applies a known law to a known object to yield an additional (but usually unexciting) fact about that object, and 'induction' attempts to generalize a law from a set of instances (if Socrates is mortal and Socrates is a man, this man is mortal, so perhaps all men are mortal), 'abduction' appears to start from a known fact, try to assume the correct law that makes that fact true, and then conclude that the other conditions required by that law have also been met. The example in the book is: Socrates is a mortal; all men are mortal; Socrates is a man. Obviously this is necessarily true only if we assume that "all men are mortal and only men are mortal", which is more information than we've been given. The authors helpfully point out that Socrates could also be mortal because he's a lizard.) In any event, it seems very clear that the process of abduction, although perilous as a form of reasoning, is one of the most important forces at work in language change, and a driving force in creativity in language precisely because it allows for unexpected results. I could see that. And I could also see abduction, like deduction and induction, being an absolutely essential tool in any reasoning apparatus.

Thursday, March 15, 2007 -- 11:42 pm

I was spending some time updating my notepads for Hopi and Georgian vocabulary, and finally nailed down the construction "I want to __" in the former, so that's been added to the slowly-evolving syntactic template page.

By way of explanation: I use various snippets of transit time to study vocabulary: Aymara (formerly Yucatec Mayan) is home to gym, Quechua (formerly Nahuatl) is gym to home, Hopi is home to subway, Turkish is subway to office, Georgian is office to subway, Swahili is subway to home. I moved away from Mayan and Nahuatl simply because I had learned the (comparatively meager) vocabulary in my textbooks, and now need to work on making grammatical sentences with them. In all my free time.

Thursday, March 15, 2007 -- 9:01 pm

It seems to be an evening for short posts. I wonder why it is that streetlights have a way of turning off when I walk under them. This has happened to me on a regular basis my entire life, but in recent weeks it's been happening every day. I imagine there's a rational explanation for this somewhere.

Thursday, March 15, 2007 -- 8:51 pm

As I continue to think about Bybee's book and to read further in the clumsily-named field of grammaticalization, I'm starting to lose track of what my original objections were. I think that my objections had a lot more to do with their methodology and mode of presentation, which seemed rather scattershot, than with many of the conclusions. I can't seriously argue, for example, with the conclusion that particular expressions in particular languages have undergone certain recurrent, analogous evolutions of meaning, such as obligation/intention/future, and that there seem to be recurrent patterns in language evolution. But I can argue that I found the evidence presented in support of the conclusions, and sometimes the reasoning itself, a little impoverished.

Thursday, March 15, 2007 -- 8:40 pm

It appears from today's New York Times that earthworms are posing an ecological disaster for forests in the Northeast, because they're not native to North America (news to me), make the soil unsuitable (among other things, too alkaline) for a lot of native trees like oaks, and poison a lot of plants by making the soil too rich in nutrients. I can certainly appreciate that last consideration, since that's the gravest threat my indoor cacao trees have had to cope with. To Learn: Soil Science

Thursday, March 15, 2007 -- 8:36 pm

To amend yesterday's post slightly, I suppose the current state of quantum computation theory would require that, in order to model and imitate human mental processes through quantum computation techniques, our neurological activity or other quasi-reasoning processes would have to be expressible in terms of a factoring problem. I haven't yet given enough thought to the problem to have an opinion on whether human thought processes could be reduced to a typographical process of Goedel-numbering. I guess that's what I get for making things up as I go along.

Wednesday, March 14, 2007 -- 9:06 pm

According to tomorrow's issue of Nature, a company is offering what purports to be a 16-bit quantum computer for sale. There seems to be no consensus about whether the thing works, or how to program it, or even what practical applications it might have. I have to confess that, despite considerable (albeit largely uninformed) enthusiasm for the topic, I'm skeptical on two counts: I just have trouble believing, based on my prior readings in various science journals over the years, that the company has successfully entangled as many as 16 qubits without environmental noise triggering spontaneous decoherence, when I don't recall ever seeing a peer-reviewed study claim to successfully entangle more the five. I also have trouble imagining the user's manual. I've dabbled in assembly language -- how do you program the thing? Does it come with an operating system? And at 16 bits, how does it measure up against the Commodore Vic-20 I played with a quarter-century ago? Perhaps I'll be forgiven for assuming this is a hoax, and therefore just a publicity stunt.

The fact that someone would try it, though, suggests that quantum computation may be the next great psychic bubble in a fascinating continuum starting with the messianic cults of the early years of the Christian Era, followed by what I like to think of as the Golden Age of Heresies, and including such notable highlights as the tulip craze, the Red Scare, and all the hubbub about nanotech. As I may have commented earlier, and this may simply be a side-effect of my general dimness or lack of training, I have yet to come across a satisfactorily intelligible explanation of quantum computing, what it purports to accomplish, and how. Nor have I heard of any actual applications for quantum computing other than the ability to search a very large database quickly for certain criteria, and to factor large numbers, which sound to me like different ways of saying the same thing. So why the enthusiasm?

I suspect that much of the excitement, aside from applications to cryptography, has to do with the prevailing cult of technology. People want to believe that science and technology will be able to solve all our problems, one way or another, and the mystique surrounding quantum computation suggests that people treat quantum computing as a realizable form of divination. But suppose we invented a quantum computer and figured out how to program it. What would we program it to do that we can't already program digital computers to do? Discover the cure for cancer? Reverse the trend of global warming? Achieve world peace? It's not as if we had already had the data to solve these problems, and were simply having trouble crunching the numbers. But I suspect that a lot of people, without giving too much thought to the matter, assume that quantum computation will somehow permit us to transcend our very human limitations of ignorance and reason.

On the other hand, what I see as the most interesting potential application is in artificial intelligence. This is another field about which I'm admittedly too ignorant to have any right to comment, but it seems to me that one of the major obstacles in attempting to mimic the behavior of the human brain is that one little processor can't possibly imitate the collaboration of 10 billion neurons in real time. There's simply too much to do, even if we had figured out the relevant algorithms. Programming would be a chore, assuming it could be done at all. But wouldn't a quantum computer, properly programmed, be able to speed up that process dramatically? Just a thought.

Tuesday, March 13, 2007 -- 9:04 pm

Now that I've started another book on grammaticization, it's starting to look like Bybee, et al.'s arguments are standard fare in discussions of language evolution. I suppose it's possible that my resistance may not be based so much on a rational objection as it's a matter of getting used to the idea. But I find a couple aspects of what I'm learning about grammaticization theory surprising. One is that it suggests that people are constantly struggling against the expressive constraints of their language, and chipping away at its grammatical and lexical regime to spare themselves the trouble of concise expression and to communicate meanings that are not otherwise convenient. The other is that speakers of a language cannot choose to preserve and perpetuate useful modes of expression from decay. I suppose the latter is a commonplace, actually. Only the fossils argue for the preservation of archaic forms, while the whippersnappers find their slouching grammar perfectly suitable for present meanings. I've read enough accounts of elder speakers of endangered languages lamenting the corrupted grammar of their children and grandchildren to recognize that.

What I find much more surprising is the suggestion inherent in what I've read so far that people find their language inadequate for the meaning they wish to express, and so end up stretching their grammar out of shape in their struggle to get their point across. My naive assumption about language change had been based loosely on the cursory overview I received in my one and only linguistics class something like 15 years ago; I recognized that phonetic change over time was an established fact, and that new phonetic rules could cause once-regular inflections to become irregular, and that the development of pidgins, creole, and less drastic social upheavals could cause a general erosion of established grammatical forms in favor of syntactic or periphrastic expressions, which could over time themselve erode and be re-analyzed as inflections of one sort or another. As I said, naive. I always assumed, I suppose, that barring some sort of dramatic linguistic upheaval, a community's language eventually settled into some kind of expressive equilibrium, or conceivably added nuances and idioms to express more subtle relationships as society grew in complexity. I've heard, for example, that emerging languages like Nicaraguan sign language (I have no direct sources on this) have become increasingly complex as they are learned and molded by successive classes of school children. Contemporary theory, however, seems to suggest that there is no expressive equilibrium or progression of increasing expressiveness -- just trends by which different lexical items become grammaticized, and then gradually drift and generalize in meaning along probable pathways until they have eroded into disuse, both phonetically and semantically, and are supplanted by other forms.

Now, I haven't been able to assimilate enough material at this point to reach my own conclusions about whether Bybee's evidence, or the evidence in my own library, happens to support this theory. Bybee seems to have avoided many of the languages -- Greek, Coptic, Persian, Mandarin, Sanskrit, Dravidian languages -- for which, in one form or another, we have millennia worth of documentary attestation. And I would think that Bybee's sample of languages -- many of which are not well documented and do not seem to enjoy long literary traditions or long periods of social dominance -- would tend to be biased against hypotheses of expressive stability or increasingly sophisticated modes of expression, simply because there may not be as solid a social framework for exploring the expressive bounds of one's language and passing along the progress made. On the other hand, it's clear that you don't need much in the way of material culture to develop an interestingly subtle and complicated language, as the Inuit, Navajo, Aymara, and Tariana know well. Ultimately, I suppose the only way to test my suspicions is to come up with some kind of objective measure of "expressiveness" (which is not terribly different from Bybee, et al.'s, method of counting "grams") and use it to assess, at prescribed stages in their evolution, an interestingly large group of languages that are reasonably documented over a thousand years or so. The operative questions would be whether the overall expressiveness of each language changed over time, and whether languages gained expressibility, lost it, or stayed the same over time with respect to individual modes of expression, such as evidential markers.

But this doesn't even touch on one of my more general concerns, which is the notion that people consistently struggle to find words for their thoughts and end up pushing and warping the grammar of their language in the ordinary process of using it. This suggests to me that the language isn't given them the tools to express what they're trying to say in the first place, and so seems to undermine my assertion that anything we might wish to say is ultimately expressible in some way, even if it's long-winded or cumbersome. Now, I could understand people choosing to replace long-winded or cumbersome expressions with more concise ones, but not opting to do without those meanings altogether. Although the history of our species may contradict me, it seems unlikely to me I would like to say that people who had once been able to express some meaning or distinction to their satisfaction would be reluctant to abandon that ability. (Of course, it's not the preservers of those archaic forms who tend to abandon them, but their children, who never learned them in the first place. But still.)

On the other hand, perhaps I just don't have enough information to be making these sweeping generalizations about language change at this point. I can only speak from my own experience. And my own experience has always been that at no point in my life have I struggled to find English words suitable for my meaning. So I find it difficult to suppose that English-speakers would go out of their way to force clumsy expressions to take on new meaning because they found existing grammar confining and ill-equipped to address their grammatical needs.

Monday, March 12, 2007 -- 10:57 pm

I wonder if there's anyone else who regularly has problems with e-mails sent to him never arriving. It seems to be restricted to Yahoo mail and specific correspondents, who are (for me) typically eBayers or booksellers. I write and write, but only receive responses about 50% of the time. A very frustrating way to do business, but I don't want to publish the address I use for my personal correspondence.

Monday, March 12, 2007 -- 10:06 pm

I'm in the home stretch of Bybee, et al.'s Evolution of Grammar, which attempts to account for the emergence of tense, aspect, and modality in the world's languages. The book is interesting, but I can't say that I find the argument overwhelming.

I approached this book with a number of questions. Why is it that so many of the world's languages have seemingly arbitrary, and yet parallel, grammatical structures to convey abstract notions? For example, English, French, German, and Mandarin all make some use of the verb "to have" to communicate anterior events, to extents varying from ubiquitous (English: I have eaten) to applying only to transitive verbs (French: j'ai mange, German: I habe gegessen), to, as far as I can tell, confining its use to the negative anterior of transitives (Mandarin: wo meiyou qu). Are there mechanisms that trigger this, is there a natural progression of language evolution, and do languages at later stages of their evolution develop more sophisticated and subtle modes of expression, or lose it, or is there a lack of net change?

Evolution of Grammar doesn't squarely answer any of these questions. It seems to chart changes in meaning of different grammatical form over time, primarily for Western European languages, but there doesn't seem to be a clear answer as to why or how other than a collection of plausible, ad hoc hypotheses. Perhaps this has something to do with its methodology.

I see two general problems with its methodology, which are closely intertwined. One problem is that the authors purport to base their argument on data taken from 76 of the world's currently existing languages, selected more or less randomly according to criteria intended to eliminate a Western European bias. This is puzzling because their argument is about how languages change over time, but this selection process forces them to rely on often poorly-documented languages for which there is little or no data about prior historical forms. They are left to assume, much as in radiocarbon dating, that at any given moment a random selection of languages will be at different stages of development, and to try to organize them along some inferred path of development that seems to depend, in almost every case, on a generalization from the historical development of one or more better-documented languages. But how can you make a diachronic argument about language change based on synchronic data? And this brings me to the other concern, which is that the data are often so impoverished that the arguments cannot avoid being circular. For example, we are asked to accept on page 135 that "Even though the English Progressive has advanced considerably from its origins and is used in a wider range of contexts than progressives in other languages . . ., it still conveys much more than simple aspectual meaning. What it conveys seems to be directly derivable from a locative source, from a meaning 'the subject is in the midst of doing something.'" (internal citations omitted). Their general assertion here is that in many of the world's languages, one or more stative, or specifically locative, expressions, is generalized in periphrasis to convey the notion that the activity being described is going on in the background of some other event, and is further generalized thence into a progressive aspect. Perhaps this is true. But to assert that the original meaning of the English Progressive is inherently locative because it is equivalent in meaning to a locative expression ("in the midst of") seems, to me, to be assuming the expression to be inherently locative rather than proving it so. Nor, understandably, do Bybee et al. attempt to seriously tackle the question of why we, in so many of the world's languages, would make any kind of metaphorical leap from space to time, which I don't find obvious. Similar questions arise for why so many languages use "going to" to express a kind of future, and why English "keep" can mean "continue". To say that the mechanism is simple and obvious simply to beg the question, and I found a lot of the book's proffered explanations to be arbitrary and, to a certain extent, common-sensical just-so. I would rather be shown that a transition was necessary or probable based on some inherent tendency, rather than be subjected to recurring observations that people's reliance on metaphors and circumlocutions will cause the meanings of various patterns of sounds to be conferred or eroded.

Which is not to say that I fault the book entirely. It was clearly researched with great care, though I don't feel that the readers benefits from all of the fruit of that research. As can be expected, the authors picked only one or two examples from their massive database to illustrate each point. Aside from English, few languages were cited often. I was disappointed by this, in part, because I haven't yet been able to get my own copy of Fortescue's West Greenlandic (backordered for years, it seems) and was curious how different the grammatical structures of West Greenlandic were from Yup'ik, which I've spent a little time on. Also, it's clearly difficult subject matter and an ambitious undertaking to survey so many of the world's languages in one study. I do feel like I've learned a little about the subject, and the authors' hypothesis is even one I find generally plausible. I'm just not convinced that the progressions they describe are inevitable, nor have they explained to my satisfaction why they happen. The word "will", for example, may have drifted from a statement of intention to a prediction of the future in English, but it didn't in German, which co-opted "werden" for this purpose. Why the different outcome? I don't feel like we've been given a decent explanation for this.