28 Oct 2007

Language waves and the satem innovation in PIE

I want to talk about the awesome, yet unappreciated, power of isogloss maps. I find that a lot of people in general misunderstand Proto-Indo-European (PIE) to have been a single, unified language when in fact this could never possibly have been the case. At any given time, PIE would have been an assemblage of dialects from start to finish. Of course, when I say that, I will probably be misunderstood to have meant that the reconstruction of Proto-IE is invalid or futile, which I stress is absolutely not my position at all.

Rather I think that by conceiving of language change in a better way, as I will explain below, we can gain a lot more insight into protolanguages than is currently the norm. To me, the confusion about language evolution stems from a popular, antiquated notion of language change as represented by the all-too-famous "family tree" diagrams. We can see the concept in this picture or in this one. As the diagram implies, language divides into various new languages as time goes on much like how a tree grows new branches as it ages. This is a horrible way to understand language change.

A better way is to recognize language, not as a single entity, but rather as a package of features (whether they be grammatical features or phonetic features) that coexist within said language. Furthermore, these features may spread like waves over a geographical area much like waves in an ocean and therefore spread into other neighbouring dialects or even completely different languages. When one language affects another, dialectologists call it areal influence.

So now, if we can truly understand that languages are nothing more than the mere intersection of spoken features within a specific geographical area and that each of these features may be represented as independently spreading waves, then we can now understand what I mean when I say that PIE was never a single language and had always been a sea of regional dialects.

Now that we have that clarified, I want to explore a crazy idea. It's untested but even if it's wrong, it should give you an example of an as-yet underappreciated process using isogloss mapping to tease out interesting details about a protolanguage's history that would otherwise go unpondered. I want to posit an idea that the satem innovation of PIE was in fact caused by areal influence from a "para-dialect", that is, a dialect lying just outside the boundaries of PIE itself. A dialect, in other words, that was almost-but-not-quite PIE. One that only shared some features with PIE because it had split away at an earlier date than Anatolian or Tocharian while innovating new features. The only way I can fully explain how this intriguing idea would work is to illustrate this in a video using my skills with Macromedia Flash.

In this rough sketch presented in my video below, the p-Satem (or para-Satem) dialect is a hypothetical paradialect that could have perhaps provoked the Satem innovation in PIE. I toy with this idea to illustrate for you how language waves and isogloss maps can help us explore new details about ancient languages.

(Hopefully this video will show up because I've never used Blogger's new video feature before. Knowing how most programmers code and being pessimistic, I'm thinking that I will encounter problems as I press "Publish Post". Let's all cross our fingers that my video will be visible.)

UPDATE
As I expected. Unskilled programmers botch up yet another "user-friendly" service. What an annoying hassle but I'm used to fixing other people's stupid programming. So I have uploaded my presentation to my eSnips account for downloading.



Sorry for the inconvenience but blame Blogger and don't forget to send them hate mail, hehe.

4 comments:

  1. Very interesting idea. Besides that you've explained the, obviously more correct, wave theory very clearly, and I'll be referring people to this website because of it in the future.

    I do have a couple of questions though: So the para-Satem language, you assume to be an extremely early 'branched off' dialect of Indo-European, that would then have said Satem Shift.

    I don't even think it's necessary to assume it was an early IE language. It could actually be a different language from a different language family. For some kind of hard to understand reason, waves never have too much trouble crossing languages.

    Similar to how both French and Dutch had the /u/ > /y/ shift, and after that /o:/ > /u:/ shift. (Maybe french 'ou' was /ow/, but it's similar enough). Although French and Dutch are distantly related, it seems difficult to assume that's the reason of influence.

    Another typical areal change is how in the Balkans the dative and genitive's uses merged.

    There's problem with this areal influence idea though. For para-satem to influence such a big group of IE, you'd expect it to leave some traces, either as living languages, or unusual words that the satem languages have in common.

    ReplyDelete
  2. Glad you like it. I find wave theory a fascinating concept and there's so much more to say on this in relation to language reconstruction.

    Phoenix: "I don't even think it's necessary to assume it was an early PIE language."

    Okay, first off, I'm only casually suggesting this to illustrate a possible scenario showing how a neighbouring para-IE dialect could affect PIE itself from the outside. I don't insist (yet) that such a p-Satem dialect really existed.

    However, in its defense (for the sake of philosophical argument), it's actually far more likely that IE was within a larger region of para-IE dialects than for IE to exist alone in a bubble. IE didn't appear out of thin air. It took thousands of years to develop and within those many centuries, other para-IE dialects surely would have emerged around it.

    You say that there is no trace of such paradialects, but I would argue that there could be traces that we can observe, for example in the way the various dialects were transformed in different regions where different paradialects would have originally been spoken before having been replaced by the growing PIE wave.

    Even if we reinterpret IE *ḱ and *k as I suggested earlier, a question still remains: Why did Satem dialects choose to push PIE's plain *k forward and palatalize it instead of the simpler option, to merge *k and *ḱ together as plain stops?

    So I suggest that the motivation makes little sense unless, dare I say, a paradialect with different features than IE had influenced it briefly. A language that, let's say, was a Uralicized IE dialect that allowed clusters like *ky perhaps. The reason why I suggest para-IE over an unknown language is because while it's safe to say there must have been para-IE dialects at one time, there's no guarantee that a completely unrelated language was present, obviously. If we're going to assume, let's reduce the number of assumptions when we can, yes?

    Phoenix: For para-satem to influence such a big group of IE, you'd expect it to leave some traces [...]"

    Don't confuse IE's "satem wave" with p-Satem in my video. While the dialect initially would directly influence the northern dialects, the satem wave then would take on a life of its own in its new host. Furthermore, keep in mind that if paradialects influenced IE this way, the traces would be very minimal because we're talking about languages that were almost IE already. Any loanwords would look too much like IE and be hard to distinguish from true IE. In other words, we're probably overlooking much of the evidence.

    ReplyDelete
  3. Great video. In about ten seconds, you'd given an insight into something I'd been thinking about (but only had fuzzy unformed ideas) and was not getting my head around for years.

    ReplyDelete
  4. It also clarified to me why there's never going to be a "Teach Yourself PIE" in the shops anytime soon! :)

    ReplyDelete