Part of me loves the idea of building new languages for fun, and I've often considered learning one myself. If nothing else, I don't imagine there's a great amount of competition in the teaching space for sci-fi languages like Klingon and Avatar's Na'vi.
But every time I get close to giving it a go, I back off. Why?
We can start with the "big one": Esperanto. I've mostly specialised in Romance languages, a choice which is in part laziness. But only in part. The more I learn of the Romance languages, the more I'm fascinated by the way the languages form a continuum. Corsican, for example, has the unique feature of having two major dialect groups that have a phonology much like Italian (in the south) and far more "Iberian" in nature (in the north). In the south, double vowels are "geminated" (lengthened) as in Italian. In the north, they aren't, and a single vowel may be "lenited" (softened, weakened, lightened). So while in the south, the island name "Corsica" is said much like an Italian would expect, the pronunciation in the north softens the second C by voicing it -- making it sound identical to a G. Spanish people call the island "Córsega". Past participle endings in Italian almost all have a /t/ sound: -ato, -ito, -uto; in Spanish they have a /d/: -ado, -ido. And in the south of Corsica it's a T sound and in the north a D (both represented in the written form by the same single letter T). But of course the "D" of -ado in Spanish in some accents is weakened to /ð/ ("th" of English "the")... and if you put a single D between two vowels in Northern Corsican, you get the same sound.
When I first looked at Esperanto, I saw that it took Romance roots and Germanic roots, and it modified them in ways that were not possible within the two language families themselves. As I learn more Romance languages, I find I'm more able to deal with variations, so I can almost understand languages like Portuguese and Occitan, even though I haven't learnt them. My fear with Esperanto, then, was feeding "false data" into the language function -- polluting the natural spectrum that I am acquiring with points that would mislead my brain and reduce my ability to understand one language from another.
It's an unprovable assertion, and no-one's going to be able to prove otherwise with enough certainty to make me risk it.
My latest temptation was the idea of using a simple language such as Toki Pona as a test case for a language learning application/framework that I've been trying to develop. I figured that its minimalist featureset and its completely formalised, regularised grammar would make it a simple and quick language to program and test.
But what scared me off this time was V.S. Ramachandran's notion of "synkinesthesia" -- the idea that language is about shape at some level. When thinking about pronouns and demonstratives, I always remember watching Ramachandran on TV demonstrating the "pointing" that a speaker will do with their lips when using many of these words. I was helping someone learn Gaelic over the summer, and she could never remember her "here,there,yonder" distinction, so I pointed out to her that "seo" feels close (the tongue stays in the back of the mouth), "sin" points forward a bit (palatalised N) and "siud" is far away (dental T, tongue nearly leaves the mouth). It seemed to help her (at the time, at least).
But that's something that no conlang I know of covers. Perhaps there are conlangers incorporating Ramachandran's ideas -- it's an idea I've toyed with myself in the past -- but then again, it's still just theory.
Any conlang can only encode what is known in theory, and will miss many of the subtleties of real languages that have evolved through natural usage and change. It may even miss one of the really core ideas of real language.
It's the same, then, as my original concern about Esperanto, but the system isn't one of sound changes and a few superficial syntactic differences: no, what I'm worried about now is that I train myself out of recognising the underlying principles of natural language by teaching my brain an artifical language that doesn't exhibit them.
Subscribe to:
Post Comments (Atom)
3 comments:
Esperanto may not be perfect, but I've used it successfully in Africa, South America and Europe, and it does the job.
This suggests to me that the distinction between natural and artificial language is more apparent than real. My own experience after many decades of using Esperanto is that a planned language or conlang can be "internalised" as well as any mother tongue. For example, Esperanto uses “jam” more readily than “already” is used in English. Esperanto’s word order is potentially freer than that of English. I learned neither of these things from textbooks, but from talking to competent speakers of the language.
I’m curious about your fear that “polluting the natural spectrum that I am acquiring with points that would mislead my brain and reduce my ability to understand one language from another.” I think it’s a false fear. I’m a linguist of sorts, and I haven’t come across any such pollution. I don’t expect Dutch to sound like Esperanto or do be able to create words in the way Esperanto does.
As for V.S. Ramachandran's notion of "synkinesthesia", I see elements of that in Esperanto. “Panjo” for “Mum”, and Manjo, an affectionate form of Mary / Maria have some softness and gentleness to them. It wasn’t planned, of course.
Don't back off. You might even enjoy it!
I'm not saying that Esperanto is difficult to learn because of its unnaturalness -- in fact, the opposite is obviously true: it was designed to be easy to learn.
The fact that it can be and is used successfully the world over doesn't change the very real distinction between natural and constructed languages: natural languages evolve according to rules that we do not yet fully understand; constructed languages are built either A) in accordance with a partial understanding of natural rules or B) in ignorance of those rules.
It is often said that the brain is just a big statistical inference engine, and that language is just a set of interrelated probabilities.
If this is true, then feeding unnatural datapoints into the inference engine will bias the results. I do not want my brain to unconsciously infer that some of the (very) unnatural derivations in Esperanto are on the spectrum of natural change.
Besides, I refer you to my previous comment:
It's an unprovable assertion, and no-one's going to be able to prove otherwise with enough certainty to make me risk it.
For those who want Klingon, rather than Esperanto, as the international language see http://www.youtube.com/watch?feature=player_embedded&v=8TQGVh025E4
Post a Comment