September 15, 2007

Family Values in Biology and Linguistics

In my post on The Arachnid Threat, I compared the cooperation of the twelve families of spiders involved to cooperation between human beings and Lar Gibbons. The basis for this comparison was the fact that human beings and Lar Gibbons, like the spiders, belong to different families within the same order. Without knowing exactly which species of spiders were involved (not reported in the news item), distance on the traditional family tree was the only measure available to me.

Over at Evolgen, RPM points out that this is not a good distance measure because mammalian taxa are much less diverse than arthropod taxa. Using estimates of the time depth of separation of the various taxa involved, he concludes that a more apt comparison would be cooperation between human beings and marsupials, such as kangaroos, koalas, opossums, and Tasmanian devils.

This improved estimate makes the arachnid threat even greater, but it also raises an interesting point about linguistic and biological taxonomy. In the traditional Linnean classification system, the nodes are assigned categories according to the following hierarchy:

     Kingdom
          Phylum
               Class
                    Order
                         Family
                              Genus
                                   Species
                                        Variety

(The mnemonic that I learned for this many years ago is: Kangaroo Pouches Can Offer Fuzzy Gorillas Stomach Vibrations. There are many others.)

The intention was that the members of one taxon at a given level would be about as diverse as the members of another taxon at the same level. Given two species, it was possible to argue about whether they were two species of the same genus, or of different genera within the same family, or of different families within the same order, and so forth. The point is, a Linnean classification consists not only of a tree, which by itself contains only information about the sequence of divergences, but also node labels that indicate the degree of separation, which, in evolutionary terms, is naturally interpreted as time depth.

Until very recently, biologists had no reliable way to measure time depth. Now that they can, by comparing DNA sequences, they know that some taxa have much greater time depth than others of the same category. Nonetheless, they continue to use what we now know to be a flawed taxonomic system. The "taxonomic bias" to which RPM says I fell victim is actually a fault of biologists, who continue to use the meaningless Linnean categories. Why did the biologist who studied the Lake Tawakoni spiders report that they belonged to twelve different families rather than something more meaningful, such as "species separated by as much as so many millions of years"?

This is one respect in which linguistic classification is superior to biological classification. Linguistic classifications, like biological classifications, contain trees, but there are no categories like kingdom and phylum. (Linguistic classifications are actually forests, since it is not established that all languages are related.) The only category distinction that you will find in mainstream genetic classifications of languages is the distinction between language and dialect, roughly comparable to the biological distinction between species and variety, and although we use it for some purposes, the great majority of linguists will tell you that it has no real meaning, or that if it does have any meaning, it is social, not truly linguistic.

Hierarchies of categories like that used in biology have been proposed in linguistics, but they aren't in general use. Indeed, use of terms from these proposals, such as phylum and stock, is a good indicator that you are dealing with crank work. Generally speaking, a family is a group of languages whose relationship is solidly established, whereas a phylum or a stock is a group of languages for whose relationship solid evidence is lacking.

I find it ironic that biologists, who now have much better means than linguists for subgrouping and determining time depth, continue to make use of a flawed taxonomic system which linguists have been smart enough never to adopt.

[Update: I have been pointed to this site, which is written by the scientists involved and contains details of the species as well as photographs.]

Posted by Bill Poser at September 15, 2007 02:51 AM