February 2009 Archives

The Burdens of the English Teacher


Guest columnist Evanthia O. Rosati at the Irrascible Professor has a funny column on the perils of being an English teacher out in society. She and I both agree - if you are so unfortunate as to be at a party with dip hanging from your mouth, there are more important issues then remembering if it's "who" or "whom."

As a linguist, I appreciate an English teacher who can remember that written grammar has it's place, and it may not be at the mall (even if she is compelled to correct mall flyers). I also have to guiltily confess that linguists have an unfair advantage over the English literature professions.

Depending on our moods linguists can play EITHER the descriptive card or the prescriptive card. For instance, I play the descriptive card when I write "one criteria" instead of "one criterion", but I can easily play the prescriptive card if an etymology is involved (no really, the word "administrate" is only a recent neologism...use the older "administer" instead).

If you are in a dispute with me, you will get it coming in going. It really isn't fair - but I refuse to give up the advantage.

Modelling Language Survival?


I was recently asked evaluate a mathematical model of language death as language competition from an Ask-A-Linguist panel. The paper referred to was Modelling the Dynamics of Language Death (Nature, 2003). It's an interesting idea...in theory, but does it work and does it tell anything new?

This particular paper (actually a "Brief Communication") proposes an equation to model decline in speaker population for Scottish Gaelic, Welsh and Quechua. I will presume that the equation they derived closely matches observed data found in their census figures.

What I'm not entirely certain is how certain variables, such as "Status" were derived. As they note, status is an important factor apart from sheer number of speakers. Without understanding how a status index is determined it's hard to evaluate how this model would predict decline in the future.

Which Factors?

In fact I would say that "Status" can be tricky to define. Sometimes status can be equated to political power, but not always. There are cases where a language of the political elite ends up dying in a region because the rulers switch to the language of the populace. English survived Norman French (several centuries worth) and Greek survived the Latin speakers of the Roman Empire although it succumbed to Ottoman Turkish (except in Greece).

The authors mention another factor of "mixing", but it does not appear to be factored in as an official parameter that I could determine. One factor which allowed Welsh to survive under a non-Welsh government was those most people in Wales were monolingual, meaning that relatively fewer people felt the need to be bilingual in English and Welsh. Today urbanization and mass media in a relatively small number of languages changes that. If you want to watch a televised soap opera, your language options are still limited.

Similarly, speakers or voluntarily emigrate to new communities may often have children who speak languages, although sometimes new community nuclei (e.g. Chinatowns) can emerge, at least for a few generations. Because this model only studies the effects of populations in situ, it can't take the effects of a literal population move (or disapora) into account and determine if they are relevant factors.

Another very important factor not mentioned is whether there is an ongoing language elimination campaign. It is possible for minority communities in the same location to remain bilingual...if they are tolerated by the local government. For instance, Pennsylvania German, the language of the Amish is not only living, but expanding in the U.S. (Page, personal communication). But we know from North America, Australia and the U.K. that a lot of languages lost speakers when the children were forced to be educated in English only (or French only in the Breton). Similarly, outlawing publications or official communication in a minority language will cause the "status" of the dominant language to artificially rise. Forced deportations and separations have similar effects.

Other Cases

Finally, I would say that there are mysteries of language life and death that any model of language death and survival must answer. One is why Norman French was eliminated in favor of the less prestigious (but more numerous) English. Compare with Peru where Quechua has not yet overtaken Spanish in any realistic way, although there are still many speakers remaining. Another is predicting the spread of Arabic through Egypt (dominating Coptic), Iraq and North Africa (wiping out Latin), but NOT further east. What combination factors are important? Is religious belief a factor after all? Is intermarriage a factor?

I do think that it's possible to mathematically model this, but I think it's important that the linguistics be taken as seriously as the math. I applaud the efforts of the authors, two fine applied mathemeticians, but I don't believe that census figures and a simple "status" index alone can tell the entire story. I would want to know more about the specific factors.

A Final Warning

I would end this discussion by pointing out that no language can really be considered "permanently" safe. Egyptian was the language of a power for several millennia before they were ultimately conquered by the Romans. Even then, late Egyptian survived long enough to become the basis of the Coptic church, but eventually Arabic came to dominate in Egypt. Coptic now survives as mostly a liturgical language, but Egyptian was spoken a lot longer (3000+ years) than it has been "dead" (ca 1000-500) years.

According to this Web history, Coptic too was a subject of a deliberate language elimination campaign...just like Welsh and Breton.

"Q" to the World


I'm listening to Lily Allen's new album, but I have to confess that I chose the "Clean" version. One of the songs that was obviously edited was the political protest song "F**k You" (let's just say I'm not expecting Allen to vote Republican soon).

I've heard a lot of edited songs throughout the years, but this may be the best. Normally the offending word is changed or just replaced with a silly sound, but here there a silly sound, but only partial deletion which leads to the trippy effect of Allen singing "Q very much. Please don't stay in in touch....Q...Q...Q."

Phonetically it's a very simply explanation because the final /k/ and /yu/ of "you" combine to form /kyu/ which is how we pronounce the letter name "Q" (i.e. /fʌk yu/ ➔ /kyu/) - it's confirmation that there really is a /y/ there, even if it's not spelled. It's also an good case of how phonological processes can cross word boundaries, but it's also interesting that no one has taken advantage of this before. Maybe it's something that got noticed in the final sound edits.

How Real are Proto Languages?


Last week I attended a very interesting session from guest speaker J.P. Mallory, an archaeologist who has branched into "Indo-European". Fortunately, I have appreciated Mallory's ability to bring a skeptical eye to the field in terms of figuring out how archaeology and language connect.

A question that arises in historical linguistics is when we construct a proto-language, how "real" is it? Some believe that what we reconstruct was literally spoken in the past and others believe that any reconstruction is completely artificial. As usual, I'm somewhere in the middle.

An interesting point Mallory made is that after two centuries of work, we can really reconstruct only a percentage of Indo-European roots. Pokorny's comprehensive dictionary lists between 2,000-3,000 roots, but most real languages many more roots (even the "low tech" languages). Word counts are hard to lock down because it can be hard to distinguish roots from words, but a conservative estimate would be in the tens of thousands. If we do a very conservative estimate of 10,000 roots, I suspect that we only have reconstructed 20% of them.

Beyond that, there are other issues which make you wonder how well you can reconstruct a proto-language. One is that almost all "languages" have dialects of one sort or another. In fact, when English "arrived" in Britain, it probably came in a series of dialects which are well attested in Old English. Similarly, English emigrated to the U.S. in as 4 major dialect classes. You could never really reconstruct "Proto-English" as spoken in Britain.

In case you're wondering if non-power languages are immune to breaking up into dialects, the answer is that it may be worse, especially if a language has been in the region for a long time. For instance, the Mayan language has survived, but it survives as a fairly large language family.

But is a proto-language completely fake? I think that one aspect that's fairly realistic is the sound. Although we don't have a perfect representation of the Indo-European sound system, I think we be fairly confident on a lot of forms. Similarly, I think one can be reasonably confident on some basic grammartical forms. The one part that continues to elude scholars is sentence structure, and I think it will continue to do so.

So is it foolish to reconstruct proto-languages? Definitely not. I think we can learn a lot about how language changes. For instance, the fact that reconstructed Proto-Romance is so different from attested Classical Latin shows how different street Latin was from written Latin. Reconstruction is also a good tool for determining if a root was borrowed recently, a few centuries ago or has been there all along.

Reconstruction is a good tool, but as with other tools, there are still limits which we need to determine.