February 05, 2004

Inflections, genes and western Iran

As a gloss on the discussion a while back on Indo-European origins, I have just discovered the neatest thing.

Persian is odd among Indo-European languages in how low on inflection it is. In particular, it is one of the few which (like its fellow oddball English) has no grammatical gender marking. And forget middle marking, separate paradigms of endings for the past and future, etc.

Specialists tend to just assume that this is because word-final unaccented vowels tend to fall away. But this is like claiming that mammals, once they emerged, had to start eating meat. Lots of mammals eat plants -- properly, unstressed vowels MIGHT wear away. But they might not, and there was more going on in Old Persia than historical phonology. I have recently found that obscure genetic mutations might show us the light.

After all, Baltic and Slavic languages have retained an alarming amount of Indo-European inflections in all of their mumbled, word-final splendor. Germanic isn't too shabby, Icelandic being Exhibit A and German itself respectable. Really, the situation in Romance and English is rather anomalous.

But still, could Persian just be an accident? The diachronic and comparative situation suggests not.

In a nutshell, Old Persian and its close sister Avestan, what we have of "Iranian Part One" in any real way, are card-carrying early Indo-European, bristling with inflectional paradigms that barely look acquirable. Then we meet Middle Iranian and things get weird -- namely, an east-west split. In the east, languages like Sogdian look like "sons of Old Persian." Some collapses of case markers here and there, not quite as baroque overall, but still players -- "Germans," so to speak. But in the west, Middle Persian and sister Parthian are suddenly Englishes -- inflections vastly fallen away. Middle Persian is, in essence, today's Persian.

But meanwhile, not a single other Indo-Iranian language -- western or eastern -- is as naked as Persian. None are "Lithuanians" by any means, but all have some case marking left, many retain gender marking, and all have some ergativity in the past, while Persian alone lacks it.

What happened to Persian? Arabic was not the culprit, despite Persian's heavy Arabic lexical overlay. Islam came to Iran long after the lurch from Old Persian to Middle Persian, and heavy lexical borrowing does not entail structural reduction and usually occurs without it (witness Australia).

Lately I have been evaluating a hypothesis that things like this only happen to a grammar because of a spate of what Östen Dahl calls "suboptimal transmission" -- heavy non-native acquisition streamlining a language. Old Persian dropped a stitch under the reign of Darius I, and thus I have wondered whether heavy immigration at that time impacted Persian.

To suppose that Persian got a close shave in being imposed upon peoples across the Persian Empire, à la Latin in Europe, is a dead end -- the Persians did not impose the language abroad, and allowed Aramaic to be the lingua franca. So we must look to what was going on right in Persia itself.

But the historical record this far back only gives us flickers, like surviving fragments of a silent film. Indeed, Darius brought foreigners from all over to Persia to build and work. But that's all we know. The Greeks and Romans to whom we owe most of our information didn't know from "multiculturalism" and have little to say about "networks" and "ethnicity." All they really cared about was battles, court rituals and contempt, and otherwise we have little to work with but some clunky inscriptions from the king and stuff on coins.

Yet something happened to Persian. If it didn't, then there would be at least one other Indo-Iranian language as stripped down. So I started wondering: could genetic evidence perhaps shed some light on what was different in western Persia? One must leave no stone unturned, after all.

Combing the sources by geneticists tracing human migrations through reference to genetic mutations, damned if I haven't found that there is indeed a glitch in the data regarding none other than Western Iran!

A nice link can be found in a 27-author paper on Y-chromosome diversity in the Eurasian heartland by Walter Bodmer and colleagues ( available in PDF here). The mutation known as M17 is found in Europe, pours through the Caucasus bypassing Turkey and the Middle East and spouts down into India -- a nice reflection of how Indo-European would have spread (pace Renfrew). But get this -- the mutation is common in east Iran, but strangely low in west Iran!

But why couldn't people have migrated through there? Okay, a desert separates the west from the east, but coming from the north of the desert one could have taken either route. The scientists know this, and only venture that maybe west Iran harbored unusual population density (why?) or already harbored an Indo-European language (but how could this have been before M17 carriers got there?). In other words, they don't know -- and have no reason to care within their bailiwick; to them it's a mere blip in the data.

But I can't help noticing that most of the people history tells us were brought to Persia under Darius were from Turkey and the Middle East, where M17 has no juice. Could it be that west Iran is so low on M17 because of heavy admixture from migrants from these places?

I'm not done investigating this. But what a cool correlation between language and genetics this could be -- if the absence of grammatical gender and ergativity in Persian could tip us off to population mixtures that explain the odd sparseness of M17 in western Iran.

I just had to share.

Posted by John McWhorter at February 5, 2004 06:51 PM