tainted syllable?

A few days ago a friend of mine twittered that he had just had lunch with the folks at viewzi. With lots of friends in the internet biz, it's not uncommon to see this sort of thing fall into my twitterstream, but something caught my eye. The -zi on the end of the word looked really odd. Not just odd, but ominous.

Viewzi is a new search engine. It is alleged to be different from other search engines in some fundamental and significant ways. I won't argue any of that. I do have a problem with the name though. Look at other search engine names: Google, yahoo, ask, altavista, dogpile, metacrawler. Some of them are less than appealing, but none of them conjure visions of blonde boys in brown shirts and jack boots.

It's not so much the sound of the name (I'm sure they don't pronounce it [vjutsi]), but the graphic representation of that final syllable that gives me the willies. When I searched for words ending with -zi, I found only six that are commonly used in English. Sorted by frequency, nazi was number one, with 13 appearances per million words.

Four of the remaining five words were Italian and one was from Arabic. I would suggest that when and English speaker sees -zi appended to a word that looks native (like view), the most readily available association is not paparazzi, or intermezzi, (as sinister as these can be...) but with nazi. This is only a hypothesis mind you.

I think some market testing of the name might have been a good idea. That's all I'm saying.

blog flogging

Well, finished with all the necessary course work to get my MA in linguistics. I was supposed to be writing my papers this semester, but things just didn't work out. I'm going to try to get some work done over the summer and hopefully finish in the fall. I'm doing a wee bit of contract work for a company called ordinate. They do English language assessment testing.

Aside form that I'm pursuing other interests. Hopefully I'll have more to say here as I do more research on my paper topics.

Because I'm so successful at keeping up with this blog and my personal blog, I've decided to start another one. It's called ecohacker.net. It was going to be about green DIY projects, but I think it's going to start off at least as something much more general. Lots of links to articles about green tech and environmental issues, that sort of thing.

My hope is that over time, a pattern will emerge through the chaos of aggregated data, a pattern that might show the way to the future...

If you're interested in that sort of thing, Check it out: www.ecohacker.net

Evolution of Language

I have recently been reading some articles arguing various positions on functionalist explanations of language versus generativist explanations. These articles (Joan Bybee, Martin Haspelmath, Frederick Newmeyer) have renewed my interest in language evolution in general and in one book in particular. I bought Andrew Carstairs-McCarthy's "The Origins of Complex Language" some time last year but never got around to reading it. I've picked it up again though and found a quote that I think is worth spreading around:

"In most sciences, the choice between rival hypotheses depends on their relative success in predicting the outcomes of controlled experiments; but in sciences relating to the past, the test of a hypothesis is how much of what is currently observed can be deduced from it as a logical consequence. A good hypothesis is one that is both simple and economical in the sense that a relatively large proportion of what is currently observed can be deduced from it, with relatively little reliance on chance or supplementary assumptions. A good hypothesis is also surprising, in the sense that its consequences include facts that do not seem obviously related to one another (Medawar 1967: 125). The more disparate the facts are that a simple hypothesis accurately predicts, the more unlikely it is that these facts are as they are by accident, or for reasons unconnected to the hypothesis."

Workin' on a chain shift

We have a mid-term on Tuesday in my History of English class. One of the things we need to know, is how chain shifts work. I figured I'd write about it as a way to test my recall. We'll see how it goes...

I’m sure that many of you have wondered, what the hell is a chain shift? Is it like a swing shift? Maybe time spent on a chain gang? Nope. A chain shift is a process by which sounds in a language change, leaving a gap in the phonemic inventory. In order to fill the gap, other sounds change sometimes leaving a new gap. The process continues until you have rebalanced the system.

Grimm’s law is a textbook (really) example of a chain shift. It describes the changes that took place on the Proto-Indo-European stop system in the Germanic language family. Of course, in order for Grimm’s law to function appropriately, we have to squeeze Verner’s law in the middle to help it out a bit.

Grimm’s law says that the Germanic language family took the Proto-Indo-European stop system and turned it on its ear. We took the original PIE stop inventory and juggled it up, coming out at the end with almost the same set of sounds, but in different places than the originals.

The PIE stop inventory looked something like this (somewhat simplified):

Voiced aspirated stops:		[bh]		[dh]		[gh]
Voiceless stops:		[p]		[t]		[k]
Voiced unaspirated stops:		[b]		[d]		[g]

These became (respectively):

Voiced stops:		[b]		[d]		[g]
Voiceless fricatives:		[f]		[θ]		[x]
Voiceless stops:		[p]		[t]		[k]

Again, this bit is simplified, but I lined them up this way because it shows why a chain shift is called a chain shift. Note how the sets of stops stayed aligned horizontally, but they moved vertically. Entire classes of sounds moved in ways that replaced other entire classes. This effectively pushed the other classes into changing in order to maintain differentiation within the language.

Here’s how Grimm’s and Verner’s really worked together to make this happen:

Step 1: All PIE voiced aspirated stops become fricatives.

bh > β

dh > ð

gh > ɣ

Step 2ː All PIE voiceless stops become fricatives.

p > f

t > θ

k > x (h)

Step 3ː Voiceless fricatives became voiced when surrounded by voiced sounds and if the preceding syllable was unstressed (Verner’s law).

f > β

θ > ð

x > ɣ

s > z

Step 4ː All PIE voiced stops become voiceless.

b > p

d > t

g > k

Step 5ː Voiced fricatives sometimes become stopped and z rhoticized.

β > b

ð > d

ɣ > g

z > r

Each of these steps had to happen in this order, and had to be completed before the next step began, or we would not see the reflexes that we do in the Germanic languages.

Chain shifts can be tricky to notice, because they have a tendency to wipe away their own evidence. For example, according to Grimm’s law, PIE [p], [t], and [k] disappeared in early Germanic, but were soon replaced when [b], [d], and [g] became voiceless. The result is that you see a language full of voiceless stops. This obscures the fact that these voiceless stops went missing. Where you notice is when you look at cognates from other Indo-European language families. For example compare Latin pater to Modern English father. Latin retained the [p] that was rendered [f] by Grimm’s law in the Germanic languages. This example also shows a reflex of Verner’s law. The medial voiceless alveolar stop [t] became the voiced interdental fricative [ð](th sound) through a combined application of Grimm’s and Verner’s laws. Note that in Latin, the stress is on the second syllable. This means that the [t] became the voiceless [θ] then, because it was between two voiced sounds and following an unstressed syllable, it became voiced: [ð].

Please understand that this representation of Grimm’s and Verner’s laws is simplified. I know that there were PIE stops with secondary rounding, I know that step 5 has a lot of caveats, etc. Don’t hassle me about that stuff. On the other hand, feel free to hassle me if there is a glaring error. I’d rather know it than think I know it. Know what I mean?

Of course, wikipedia has a pretty good explanation of both of these things already:
Grimm's law
Verner's Law
Chain Shift

Almost a year

OK. I know it's been a long time. What can I say. I've been busy.

In any case, the Mexico trip was great. Teotitlan del Valle is a beautiful place, with wonderful people, and Zapotec is a really interesting language. It was a great opportunity to do real field work, and I'm going back again this summer to continue with the project. There are pictures here.

I'm taking a basic syntax class and a history of English class this semester. Both of them are more or less review for me, but I'm happy to have it. It's been a long time since I've done either, and next semester is going to be a killer.

I'm going to try to start writing again. Hopefully it will be interesting (for you that is).

Aymara resources

For those interested in Aymara, here are some resources:

textbook (scroll down and click on the "go" labels.)

complement agreement


Aymara home page

Ethnologue

rosetta project

textbook index

quechua contact article

Aymara Sessions 3 and 4

Things are going very well with our Aymara speaker. The elicitation is moving along faster, and is getting more and more targetted. there are times during the elicitation when people get little flashes of intuition/inspiration. You can see it on their faces as they raise their hands eagerly hoping to ask the next question.

This has led me to rediscover the importance of memory. Everybody says that it's important to come prepared for elicitation, that you need to analyze your data after every session so that you know where to work from in the next session, etc. I've noticed though that memory for your data is absolutely necessary if you are to have those little flashes of insight. The insight comes from recognizing a pattern. To see a pattern, you have to have more than one example. If you are to see a pattern "in the moment," that is to say during the elicitation, you have to be able to recall the data that you've acquired from previous sessions. If you can do this, it allows you to redirect the elicitation on the fly, and to run down any avenue that presents itself.

I've noticed two important thing about language in general so far. The most obvious is the clutter. I'm used to performing morphological analysis on data that has been substantially cleaned up. Natural language it turns out, is really really messy. I have said that we are very lucky to have a consultant who is well educated and knows a bit about linguistics. Yes and no. Yes because it helps him to understand what we're trying to get at, but no because he is aware of a lot of variation in his own language and will give us multiple forms of the same thing, sometimes prefacing with "Well, I think it's really supposed to be x but some people say it y and I say z." What the hell am I supposed to transcribe for that? I want to transcribe one dialect, not three.

This experience has also really brought home to me the notion that human languages just aren't that different from one another. I've studied a few Indo-European languages and Japanese, and I don't see much in Aymara that differs too radically from anything I've experienced before. It's comforting in a way to see that a language that should be about as foreign as you can get (to me) still has derivational and inflectional endings, a pretty normal counting system, and verb constructions similar to Japanese and German.

I'm loving every minute of it.