Monday, September 27, 2010

Classification: Lumpers more accurate than Splitters? Some reflections on what dinosaurs can teach us about potsherds


1915 Classification of Ceratopsidae by William Diller Matthew. (Wikicommons)

A recent study by paleontologist Michael Benton (University of Bristol) of the history of dinosaur taxonomy in the Journal of Vertebrate Paleontology, discussed by Scientific American, argues that lumpers (those more willing to overlook minor differences when creating classifications) are more accurate than splitters (those who prefer to spin off more classificatory types based on minor differences).

Those who named more dinosaurs had a higher rate of those species later being absorbed into other species as it became clearer that the differences once used to define a "species" were minor, with typically 50 - 70% of "species" ultimately being rejected by the field. Those who named only a few dinosaurs, however, typically had about 40% of their discoveries rejected. Benton also argues that some of the more prolific dinosaur namers may have had financial or prestige motivations.

This issue has been particularly in the news lately, with the determination that the famous genus Triceratops is simply a juvenile version of the genus that had been called Torosaurus. Dinosaur fans need not worry, as Torosaurus was named later, so its genus will be erased and lumped in with Triceratops. No need to start another Pluto guerrilla war.

Any archaeologist will bristle at the idea that we dig up dinosaurs (go ahead, ask one). And unlike in the case of dinosaurs, where a press release goes out with a new taxonomic addition, no one really cares if you've named a new type, group, or even ware of pottery. But this does bring to mind some of the issues with archaeological classification. Pots or spear points are not species, and any first year grad student (at least of Americanist archaeology) can point to the iconic Ford vs. Spaulding debate over whether types are discovered or created by the archaeologist.

Reading this article, I started to think about how chaotic archaeological typologies really are. One problem I'm quite familiar with is the issue of the word colono or colonoware. In historical archaeology, the term has a wide at and at times nearly contradictory use, incorporating ideas of both continuity and blending. It has been used to refer to pottery from sites in the Caribbean, the United States Southeast, and nearby locations that are simply vessels not created by Europeans, plates and pitchers created by indigenous potters but incorporating European design concepts, bowls and jars created by Africans and African Americans in the style and symbolism of their pre-diaspora homeland, vessels like those of a pre-diaspora homeland but reflecting ethnogenesis involving people from various African cultural traditions, and vessels not associated with potters or consumers of a particular identity but instead reflecting the dynamic changes wrought by colonialism.

Thinking of this case I can't even imagine the idea of an orderly "rejection" of a type in the manner of the paleontologists. Again, a species isn't a type, and a type may well serve a research purpose more than be an actual discoverable "thing." But types aren't often treated in this manner. They get published, or sometimes are just propagated within a regional research community, and then they start to get used by other archaeologists for possibly very different research ends than those of the classifier. A type originally designed simply to clarify a deep stratigraphic sequence might get used for examining ethnic identity, trade routes, or status differences. Neverminding unusual cases like the colonoware one, there is no real way to check the production of new types, nor how useful they are, other than through the informal process of ignoring earlier works, a process that may get entangled with more practical than intellectual matters.

Personally, I have found that I feel like "splitting" feels proper, taking note of minor differences and then pointing your index finger in the air whilst shouting "A-ha!" in your best Sherlock Holmes imitation. But then, when it comes time to actually present taxonomy to the world, "lumping" instincts kick in out of caution. At least within my research materials. Lumping in with someone else's typology, that opens up whole new issues (is this really the same thing? Is there a regional variation?). And the cycle begins anew.

1 comment:

Mix Tu'ux said...

I have bumped up against these problems myself. It seems to me that not only do archaeological types never get rejected, but we are flatly not allowed to ever reject them. Since they are artificial constructions from specific research questions they do have a reason for their creation, but then they are wedded into the discourse as concrete realities.

The two types Joventud and Sierra were defined based on a stratigraphic difference at Uaxactun, not on any inherent difference between them. Yet now every Maya ceramicist is crossing their eyes trying to see a difference between Joventud and Sierra. Smith's original distinction was entirely valid, but he never claimed that there was a physical difference between these two types. Yet when I suggest we should merge them together as a single stylistic type with a broad chronological distribution people get upset.

Argh. Classification!