Tuesday, November 15, 2011

The Plantation of the Atlantic, XII: Evidence, Experts and the Source of the Susquehanna of th

Historically, by 1650, and probably by about 1600, the Eastern seaboard of North America was linguistically organised in a way that no-one found remarkable at the time, but one that I want to single out as an important historical fact. Coastwise from Nova Scotia and perhaps Newfoundland all the way to the Carolinas, native populations spoke a very closely related group of languages designated today, and since the early Nineteenth Century at least, as Algonquin. Inland, along a continuous water transport corridor extending from the lower Saint Lawrence down the Richelieu River-Hudson-Mohawk-North Branch of the Susquehanna-Potomac-Shenandoah, could be found Iroquian languages, including Huron, Erie, the languages of the Five Nations, Susquehannock and perhaps others, Tuscarora and Cherokee.

I've said before that I don't think that it is at all an accident that Cooperstown is located at one of the most important waystations along this inland route. And, I would add now, Washington, D.C. (and Gettysburg) at others. That being said, there remains the question of how languages come to be aligned with geography.

This is a question that, in the first instance, inclines me to talk about a much smaller, walkable, professional geography. It is the case that some humanities departments at the University of British Columbia have their offices in one of the wings of the great spaghetti maze that is the Buchanan complex. So when the rain is pounding down on Vancouver, as has been known to happen, you might find yourself cutting through a long corridor of offices to get to your next class, rather than braving the courtyard outside. If you have a moment to linger, you can get some idea of what a department is about by reading the cartoons, funny clippings, and newspaper stories that people like to post on doors and notice boards. As an introduction to a field of study, it sure beats trying to read yourself into a textbook!

Now, one might ask why an expert in another field would want to do something like that? We have experts for a reason. No-one goes to an historian for an opinion about string theory, and the default presumption is that a physicist who wants to talk about history has gone a little dotty. Yet some of these departments bring together specialists from many fields. Take a trip down the hall through Classical Studies, though, and you'll  see the point. The bulletin boards joke about archaeology, literature, and history, because Classical Studies contains (Classical) archaeologists, (classical) literature specialists, and (classical) historians. First come archaeological sites, then Greek literature, beginning perhaps 472BC by the absolutely most savagely skeptical, reductive approach I can imagine taking, then, sometime in the 430s, Herodotus of Halicarnassus writes his Histories under the patronage of the Alcmaeonidae. History is handed down from door to open door down the Classics Studies hall from archaeologist to literary scholar to historian, and the Classicists get to do all of us a favour by figuring out how history is going to relate to archaeology. Clearly there will be important lessons here for historians working on other areas at a similar transition point, such as the Eastenr Seaboad in the peri-Contact phase.

To follow the baton a little further, I'm going to take a walk through a campus that only exists anymore as a nostalgic memory, exiting out the north doors of the Buchanan complex to walk down back alleys and through a grove and across a field in the dripping rain under lowering skies* to the Vancouver School of Theology. Though even if we did that, we'd have to take the time machine of nostalgia back a long time further yet to the point where bible scholars intruded on the archaeologists's fields of toil to tell tales of pharaohs, Phoenicians, and Hittites.Not to mention Pythagoras studying hermetic secrets with the Jewish  monks of Mount Carmel. Crazy? Perhaps. But mixed with uncomfortable facts.

So the Classicists sought allies amongst the ranks of the philologists, as I have discussed before. Allies with nigh unbeatable scientific talk swept away uncomfortable facts, bringing Greeks into Greece from the north at whatever time science currently leaves us (1200BC? 1500BC? 1900BC? 2500BC? 7000BC?) before hooking over the Aegean and colonising the Anatolian shore about 900BC or whatnot, just straight up exterminating the Orientals.

Which last? It seems invented --implausible, even. As though a story about Nineteenth (and Seventeenth) century colonialism is being resituated in the past so that its heuristic can have a longer lever arm with which to move the present. Stupid analogy? Let's be clear. It's not what the sources say. How do you get there with vehement certainty? In part, it's the jargon, but I prefer to visualise a sputtering academic, of a particularly embarrassing and embarrassed kind, babbling about Achaeans Indo-Europeans and Mediterranean Semitics and miscegenation.

On the one hand, I'm conjuring with a particularly repellent straw man here, and my only defence is that they existed in their myriads. On the other, I'm invoking a beguiling heuristic. If races are things given in themselves, separate and equal, then languages and cultures come in boxes that we can shuffle around the maps of our ancient mind's eye. Forget archaeology! There is no science for understanding the history that comes before history like historical linguistics! We've already applied this "science" to the Greeks, and there's no reason not to apply it to the Indians. I'll bet that you can't wait to find out where the Algonquins come from. (Hint.)

Way back in the world of credible scholarship, the classicists are getting better and better at founding history in archaeology. Why shouldn't we take the same approach to historical linguistics?

Actually, we should. But one more analogy here, this time with boxing. I've already linked to a brief discussion of Representative Seaborne Roddenberry's anti-miscegenation amendment to the United States Constitution, introduced  into the House in January, 1913 in direct response to Jack Johnson's July 4th, 1910 victory over James Jeffries. At least for a historian of boxing, the mood in the United States in the wake of a black boxer's achievement of the heavyweight title seems a little ...frenzied. Could there be anything more petulant than trying to shut down professional boxing over this? (In favour of college football, just in case you thought that there was a public health case to be made) So Jess Willard's highly irregular victory over Johnson in April 1915 was taken hard by some. When this white, alleged "working cowboy" stepped into the ring for another 4th of July fight in Toledo, Ohio in 1919, he might have suspected that the man glaring at him from "Kid Blackie's" corner intended to set the boxing world to rights.

As with boxing, so with historical linguistics.

So, a moment, back to the age of Representative Roddenberry, but no longer in the realm of racial anxieties and stupid laws, but rather of pure science. A young man named Edward Sapir was out in the field, making a name for himself. Having moved from Germanic Philology at Columbia to Franz Boas' emergent anthropology department, it is perhaps not surprising that he ended up in Boaz country in the Pacific Northwest, initially amongst two language groups along the southwest Oregon coast, where this trained Indo-Europeanist promptly proclaimed the existence of an almost equally diverse language family, "Penutian." Intruding on his supervisor's country, Sapir also announced the existence of "Na-Dene," incorporating the Athabaskan languages, various neighbours, and, as the historical linguists now accept, indefensibly, Haida and Tlingit.

Boas, not a big fan of language-lumping, was not impressed. But Sapir was not finished. At another point, Sapir discovered that two minor languages of the northwest California coast, Yurok and Wiyot, were "Algic" languages. That is, they were related to Algonquin, albeit distantly, and distantly to each other.

Therefore, according to the theories of historical linguistics that find languages moving with peoples from regions of high linguistic diversity to ones of low, the nearly mutually-intelligble Algonquin  languages uniformly distributed coastwise from the Newfoundland fishing grounds to the area where the the treasure fleets turned east for home was the result of a prehistoric transcontinental migration from California to the east coast.

Now that's what I call making prehistory historical! I'd also mention that it's in a well-established tradition of pulling this crap out of your ass. But, of course, this was now true crap. Science showed!

Sapir, after taking time out to prove that the Aztec language was related to others stretching up to the Oregon border, that Athabaskan (part of Na-Dene) included Navajo and Apache, came by 1925 to the belief that all of the Indian languages of North America could be reduced to only six language families. At this point, he got tenure at the University of Chicago, and, in a radical departure from normal career trajectories, from this point until his 1939 death produced nothing but new PhDs and some crazy, tossed-off ideas for them to pursue.

One of those students was Morris Swadesh, who invented the "Swadesh" list of about a hundred --or many more, if you want to be more accurate-- root words that languages never, ever borrow or change, so that if you find them in multiple languages, those languages are super-guaranteed to be "related" in the Indo-European sense. (Sometime back in the past, half the tribe went right, half went left. Right speaks A, left speaks B.) And he discovered glottochronology, the amazing rule that languages change at a constant rate, so that by comparing A and B, you can tell exactly when the split took place.

What could Swadesh's discoveries tell us about the past? A great deal, it turned out. Joseph Greenberg, a scholar not in the Sapir lineage, soon managed to show the "genetic" relationship of thousands of languages in Africa within four great families: Afro-Asiatic, Nilo-Saharan; Niger-Congo; and Khoisan. As Greenberg explained, he just lined up Swadesh lists from these languages and let his eyes rove over them. Eventually, obvious connections just jumped out at him.

People were impressed. So Greenberg announced that most all of the languages of Southeast Asia and the Pacific were linked, genetically descending from Indo-Pacific. And that all of the Indian languages of the Americas were genetically descended from a single stock, Amerind. And then we got Norstratic, and similar proposals that reconstructed a language spoken by just about everyone, perhaps 15,000 years ago.

And then we got Merrit Ruhlen (of the Santa Fe Institute, of course) pronoucing the existence of "Proto-Human."

Sometimes, academic fields of study are like, I don't know;  maybe a rule for ending words in Latin or something. Or a crosstown bus; I suspect that'll be a more productive analogy than some rule about ending verbs in Latin. This crosstown bus has an unusual route. It starts out at the college, where a bunch of kids get on. They're excited. They're migrating places. Gradually, the kids get off as the bus at exciting places. If you're still aboard, you'll notice that your fellow riders are getting more eccentric. That's because the final destination of this bus is Crazytown. And by the time you get there, your fellow riders are Joseph Greenberg and Merrit Ruhlen. On the other hand, if you fell asleep early in the journey, you might be wondering when, exactly, the bus turned onto the Crazytown Aqueduct, and how far you'd have to backtrack if you got off now.

In preparing to write this posting, I skimmed two textbooks on historical linguistics; one by Lyle Campbell and one by Terry Cameron and Claire Bowern. (An Introduction to Historical Linguistics 4th ed. [Oxford, etc.: OUP, 2010]. Campbell, who presents as a fairly orthodox historical linguist in other work, is here scathing. The conventional position in historical linguistics is that Swadesh  is unsound and that only Greenberg's "early intuitions" are acceptable. To put it less politely, Khoisan and Nilo-Saharan have been demolished and Afro-Asiatic is in trouble. Wikipedia on Language Families  maintains the existence of Niger-Congo, in spite of severe doubts about Greenberg's "method." Campbell, on the other hand, drops it from his list of language families without explanation.

Given that most of the languages proposed to exist within Niger-Congo have barely been studied, that does seem like the rational approach to take. The virtue of maintaining Niger-Congo, conversely, is that if it is entertained as a valid language family, than "most" of the world's languages can be grouped into families of genetic descent. If it can't, than most of the world's languages cannot so be grouped. 

Now, Cameron and Bowern do not provide a list of language families. They do, however, present a mild defence of the Swadesh/Greenberg methods. Which seems a little crazy until you board the bus, as it were. Bowern is associated with the Pacific Linguistics Research School of Pacific and Asian Studies of the Australian National University. Unless I copied that wrong and got an extra "Pacific" in there. I'm refusing to look it up on the grounds that people shouldn't make up stupid names for their institutions and use them as part of the goddamn publishing information of their books so that you have to enter that crap into your bibliographic database in the first place.

Okay, rant aside, I link-analogised Robert M. W. Dixon's The Rise and Fall of Languages above to the beating that Jack Dempsey laid on Jess Willard for a reason, but I could also have gone the full Australian here.  Dixon, a Pommy from Gloucestershire has had a long career in the linguistics of Australia. In 1980, he published a major work on Aborigine languages. About that time, Robert M. Hale of MIT, in the course of a remarkably short period of field work in northwestern Australia, the existence of a "Pama-Nyungan" family of languages that, in its fullest form, embraced most of the languages of Australia, pointing towards "Proto-Australian."

Dixon, an expert actually on the ground, spent a great deal of time pursuing Hale's "Pama-Nyungan." The Rise and Fall of Languages is the result: a robust rejection of, in general, the very idea of language families. As is often the way on Wikipedia, turn the article here over to the discussion side to watch Claire Bowern get very upset about this, and not come off looking very good.

So, as we watch the angry Aussies fighting in the street outside and taking time out to gang up on the Pommy, it occurs to  us that we've been on the Crazytown Viaduct for a while, now. How long, exactly? More pointedly, have we been on it long enough for it to matter to the history of peri-Contact North America?

Well, back to Wiyot and Yurok. (Not this, unfortunately.) When we take Dixon's criticisms of linguists who do little field work and jump to large comparative conclusions about languages without even bothering to produce a grammar first, we have to look a little askance at Sapir's work. The breadth of his linguistic knowledge is amazing, and as a person who can barely talk or organise an essay, I'm no-one to judge the intellectual possibilities open to someone with a gift for languages. But come on, seriously. Dene; Haida; Nahuatl; Algic; Penutian....

Yurok specifically is a very small language, spoken by only a few thousand people in the Eureka Bay area at the time of statehood. Census data shows that the population had dropped as low as 800 in 1910, and while it has risen to 4000 since, it must be noted that the official definition of a Yurok Indian allows one-eighth blood quanta. It's enough that your great-grandparents were Yurok.

The Yurok language is moribund now, and spoken with a very noticeable American accent, I'd add, and probably wasn't in good shape in 1915, either. What's more, there was a Hudson's Bay post in the Bay back in the 1820s, with intermittent contact with company men since, and these would have included Algonquin-speaking Cree. At one point in the Pama-Nyungan debate, Dixon rather incautiously referred to the way that evidence proceeded from Hale's pen, he knew not how. That's an unkind and incautious thing to say about a fellow academic in an academic debate (insert Aussie joke here), but it's true  "elicitation errors" are a serious problem in interview formats. 

What's more, Sapir came under immediate, blistering criticism from Smithsonian linguist, Truman Michelson. Unlike Sapir, Michelson was an Algonquinist. I've seen Michelson's critique of Sapir criticised as methodologically unsound, but although it is very hard to track down on the web, it seems pretty devastating to me. Sapir's thesis, in contrast, is strongly defended, including by Campbell. 

I, however, suspect academic politeness. Whatever be the case, the transcontinental migration of Algonquins from California to the east coast is ridiculous. As I say, historical linguistics ought to be incorporated into our historical armament. The fact that Iroquoian and Algonquin are distributed along the major coastal and inland north-south trade routes of peri-Contact North America is important. But it can't possibly be explained in terms of this particular migration. 

Take a step back. Assume that it's not a "genetic" relationship at all. What else can create such a large linguistic unity? Well, in the peri-Contact phase, there was a single language distributed up and down the West Coast: Chinook.  This is known to have been a "trade jargon," or pidgin.  Pidgins are supposedly necessarily simple languages with common traits, but it has been proposed that they evolve into "creoles," emergent languages synthesised out of three or more original tongues with its own unique grammar and vocabulary.

Creolisation has been proposed by some as a unique process that creates languages whose main characteristics can be known a priori from presumptions of "simple" grammar" and "easy" pronounciation, however these are defined. Others question this idea of Creole exceptionalism. As in the case of the Dixon critique of language family lumpers, they appear to have the inconvenient advantage of data to back them up. 

So on this basis, can this tongue-tied scholar assert that Algonquin and Iroquian represent emergent Sprachbunds organising long-range north-south trade/exchange routes that formed along the Eastern Seaboard during the peri-Contact period, and perhaps as early as when the Greenland Norse were interacting with the Ramah Island Chert Exchange Network? (To the extent that they did?)

 I don't see why not. In fact, if we take the existence of widespread language families as something to be explained, historical linguistics have precisely given us some historic evidence that such a thing existed. Thus, the fact that the Indians encountered in Virginia in 1607 and 1632 spoke substantially the same language as the Indians encountered in Massachusetts in 1621 and in Quebec in the 1550s tells us that those Indians were in contact with each other. Again, the initial English colonies were being inserted into an existing world system.

There. Got that out of my system. Time to take a look at 

*Rain, Vancouver. I got strong reactions, thanks to Battlestar Galactica. You know how you can tell that science fiction is a guy's fantasy? Hot girls like Grace Park make the first move on guys like Tahmoh Penikett. Man. Life would be so much easier, we think to ourselves, lazy, cowardly guys that we are. Life shouldn't be easy. But they do make a nice couple.


  1. This is so far off-base it's hard to know where to start. The quick answer to "can this tongue-tied scholar assert that Algonqui[a]n and Iroquian represent emergent Sprachbunds organising long-range north-south trade/exchange routes that formed along the Eastern Seaboard during the peri-Contact period" is "absolutely not." If you want to see what a Sprachbund a millennium or two old looks like, look at the Balkans or Ethiopia or North Africa: unsurprisingly, the families/subfamilies involved are still quite easy to tell apart despite a good deal of convergence. The Indian Sprachbund is even older, and there too it's still trivially easy to distinguish Indic from Dravidian from Munda. Family spread and diversification has been observed more than once within historic times from a directly attested ancestor (Romance, Arabic, Mongolic, Neo-Aramaic, Sinitic...) whereas no one has ever observed direct evidence for any group of languages anywhere starting out unrelated and converging to the point where they could be mistaken for a linguistic family, let alone one as close-knit as Eastern Algonquian. But even if you accept Dixon's frankly rather inadequately supported model, his examples of potential convergence groups, like Pama-Nyungan and Niger-Congo, are supposed to have taken many more millennia to emerge, and their members are far less similar to one another than Eastern Algonquian languages are. More immediately, you seem to be gliding over the fact that transparently related Algonquian languages (never mind Algic) were spoken over a wide area west as well as east of Iroquoian; by "Algonquian" you seem to mean "Eastern Algonquian".

    If you want to do historical linguistics, don't just skim Campbell's (or Trask's) textbook; do the exercises and learn to evaluate the evidence for yourself. It's not that hard, and it's highly rewarding. On the other hand, if all you want is evidence for a north-south trade network - well, the distribution of Eastern Algonquian, a close-knit branch of Algonquian, is prima facie evidence that at some point people were traipsing up and down the East Coast more readily than they were travelling inland, without bringing in any further speculation.

  2. Well, clearly I don't want to do historical linguistics. I don't feel competent to even undertake it, and the arguments within the field leave me much more pessimistic than you about my gaining the ability to evaluate the evidence for myself.

    So I'd leave it very much alone if it weren't being adduced as historic evidence. This is similar to the issue with archaeology. Historians can't not use archaeology, but methodology, hence critique, seems to be beyond us. If the community spoke with one voice about the evidence, this would be fine, but where two authors disagree? Hard.

    So what about Algonquin? Its documented spread in historic times offers a straightforward explanation. It travelled west with the fur trade through the pays d'en haut. That said, one would then have to account for significant differences in the Plains Algonquian languages in terms of contact rather than diachronic change in order to discard the west-to-east migration. Contact with fur traders would nicely account for the Algic features of the California coast languages.

    Is that possible? It seems like common sense that it is, and the heroic discoverers of vast language families who think that they can discount such possibilities, seem, by way of contrast, demonstrably full of crap to me in general.

    So, as a consumer of the work of other fields, why, exactly, should I accept the authority of Sapir over the authority of Dixon? (At least, his deconstructive authority. It's way easier to be a skeptic, of course.)

    In the case of Eastern Algonquian, given the small number of speakers and the extreme fluidity of Eastern Woodlands Indian life, the case that it is a trade pidgin rapidly coalescing into a Creole seems defensible on the model of, say, Tok Pisin. That is, defensible depending on who speaks with more authority out of the whole "Creolisation" debate within historical linguistics.

    As a hapless consumer, I find myself more persuaded by the argument for the normality of Creole languages than by people who say that they have to look "artificial" and be readily distinguished in deep historic time after their genesis. As a scholar, though, I'm all too aware of the dangers of confirmatory bias.

    In sum: it scares and confuses me when you grownups fight.

  3. I do sympathise with your predicament, having been trying to figure out enough Saharan history for my own historical-linguistic uses. But if you're going to talk about the field, more care is called for. A Sprachbund is a group of languages from different families showing common features resulting from mutual contact. A "trade pidgin rapidly coalescing into a Creole" is, by definition, not a Sprachbund: for one thing, it's a single language. And most trade languages are neither pidgins nor creoles.

    Minus the misused technical terms, I think what you're trying to suggest is that the language became Eastern Algonquian was adopted by a variety of tribes along the eastern seaboard, replacing their previous languages, due to north-south trade rather than migration. This hypothesis is compatible with Algic being real or unreal, because a trade language is almost always based mainly on a single existing non-trade languages (even Tok Pisin's vocabulary is something like 80% English), and because Eastern Algonquian is a valid subgroup of Algonquian. There's no reason to drag Sapir or Dixon into that issue; Yves Goddard would be a more useful starting point.

    However, if for some reason it's essential to your argument that Plains Algonquian is a recent result of the fur trade, then you're in bigger trouble - and you won't find support in Dixon there either. If every Algonquian element in Arapaho came from Cree within the past 400 years, then every Arapaho word ought to be either practically identical to Cree (modulo the requirements of Arapaho phonology) or completely distinct - never related through a long, decidedly un-obvious chain of regular correspondences, as they in fact are.

  4. Thanks for the critique. I'm not sure that Plains languages are important to my thesis at all, especially since I'm not sure what I'll be writing when I restart my manuscript next week. (Hittite might be another matter.)

    That being said, the towns that the Illini Confederacy built around French Missions in the 1680s are important. The Wikipedia article on Ojibway (me r scholar!) claims that it was already one of the main trade tongues of the region by 1703, but the language genesis events at Kaskaskia and Cahokia might have been awfully complicated. Ah, well. I have an expert that I can pursue now, so I should probably just look up Goddard.

    That said, I wish that I could be more confident of historical linguists' claims to be able to distinguish change in time from change due to accommodation within multilingual communities. I know that this is the standard, and frustrating demand made on specialists, but in my defence, I'm just wishing here.