Why I love the Web

Shakespeare in Basque!

Udarako egun batekin al banizu konpara?
Zu, ederrago ta legunago izanik[.]

Ergelaren ipuña dirudi: zarata ta bizitasunez betea, baña mamibagea.

Ots, monja-etxera!

(Hmm, will anyone but lhn have the tenacity to source all three quotes? Let's see...)

Dialect vs. language: Loose ends, 1

I'm not sure how much interest remains in my further ramblings of the insupportability of distinguishing "languages" from "dialects".[*] Much of what I was going to post in entries actually came out in replies during the ensuing discussion, but there were some bits I felt that I never adequate responded to. Chief among them is this question from lhn:

Though does the lack of existence of a bright line boundary make it entirely political? I understand that whether Catalan and Castilian are languages or dialects of Spanish or whatever may not be a scientific question. But would it be meaningfully possible to claim that English was a dialect of Mandarin or vice versa?

At the risk of sounding glib, I'll ask in return, "Meaningful to whom?"

I think one of the fundamental problems underlying the whole debate is a widespread lack of understanding of how linguists go about determining what varieties are related to other varieties and how. It's easy to mock those who go about claiming that "English is a Romance language" due to its extensive borrowings from French and Latin or that "Chinese is related to Japanese" because they share a writing system, but how many people who can correctly assert that "English is a Germanic language" have an idea what that really means rather than simply relying on the authority of the EB or their high-school English teacher?

The traditional classification is built on a phylogenetic metaphor of "language evolution": Languages give birth to "daughters" who together constitute "branches" in a larger "family" (which may in turn be grouped with others into a larger "superfamily" or "phylum"). Despite the fact that they are transmitted to "offspring" in a most Lamarckian fashion, borrowed vocabulary and other "acquired traits" can't change the fundamental "genetic affiliation" of these linguistic varieties. This is called the "Stammbaum model", from the German word for "family tree".

The trouble is, as you can see, languages don't actually have genes. They are human artefacts, much like clothing, and can be altered and discarded in the same way (although not nearly as casually). It was once thought that certain core features were so resistant to borrowing that they could be used as reliable indicators of ancestry, but broader study of the many tens of thousands of spoken varieties in the world has demonstrated that there is nothing which cannot be transmitted from one to another. The "wave model" is one attempt to account for these cross-cutting influences in a graphic, typological manner, but it's still applied primarily to varieties within the same "family", and thus represents more of an additional dimension to the Stammbaum model rather than a replacement for it.

Furthermore, linguists like R. M. W. Dixon, who has done extensive fieldwork among the Aboriginal languages of Australia, argue that Stammbaum model is the product of conditions prevalent in Europe and other areas which have seen the sudden invasion and diversification of a single "genetic stock". As such, it's inadequate for accounting for the types of relationships seen in areas of relative ethnolinguistic stability (like Australia) where changes diffuse constantly between neighbouring languages.

A classic example of the problems this creates for historical linguists is the relationship between Quechua and Aymara. The similarities are certainly striking--perhaps even too much so. As my friend zompist often points out, they're more what you'd expect from massive borrowing than from genetic relationship. But is the transmission from Quechua to Aymara or vice versa? In the absence of written records, it's impossible to say.

Furthermore, there's the possibility of there being both a deep genetic relationship and one based on a more recent exchange; after all, this is what is argued for the members of the proposed Altaic family, most of which have been in close contact for thousands of years. As a result, not just vocabulary has been borrowed, but even morphology, word order, and vowel harmony--features once thought so "stable" that they were definite evidence of genetic affiliation! The greater the time-depth, the harder it becomes to distinguish mutual influence from common descent.

In light of all this, outrageous claims about the affinity of superficially divergent varieties don't seem so obviously misguided after all. Who among us would be able, on our own, to demonstrate beyond a reasonable doubt that, say, Norwegian and Romani are related but not Ottoman Turkish and Persian?

[*] For those who can't get enough of my mouthing off on the subject, be sure to see my contributions to a one-sided debate on the subject in the context of Ibero-Romance--surely a place where politics is very much at the fore.