May 10th, 2006

ozarque figure

Linguistics; where do words come from?; afternote...

I need to do some tidying up here, to avoid confusion that I have inadvertently created. In yesterday's post I said:

wolfangel78, in the discussion of LAadan tones, asked me this question: "If you have a morpheme ma followed by a morpheme a, how does high tone get placed? IOW, how do you choose between maA and mAa?"

Because of the context in which the question was asked, I thought that the question was really about how I choose between maA and mAa within a morpheme, rather than between morphemes, and I answered that question by saying that I didn't know the answer. I shouldn't have done that; I should have checked with wolfangel78 before I tried to answer.

For clarity, then:

1. When a morpheme "ma" is followed by a morpheme "a," no high tone gets placed. Instead, an /h/ is inserted by rule (a process very common in human languages, one that linguists call "epenthesis").

2. Inside a morpheme, when two identical short vowels would occur together, one of them must have high tone added to it. That is, you can have "maA" or "mAa," but never "maa."

3. This is true for all vowels, not just /a/. The shorthand way to do it is to say that you can have "v V" or "V v" but not "v v."

4. Obviously, when I construct a LAadan word that adds a morpheme starting with a vowel to a morpheme ending with a vowel, and I insert an /h/ between those two vowels, I know exactly why I'm doing that. I'm following a rule of the sound system.

5. But when I construct a LAadan morpheme that contains two identical vowels, I have no idea how I decide whether to use "v V" or "V v."

One of the ways you convince beginning linguistics students who tell you they "don't know any grammar" that that's false is by giving them a list of hypothetical English names for a candy bar and asking them which one(s) would be possible as such a name in English. [Like: "Mngax"; "Bowda"; "Dbik"; "Shzhuua" -- from which they will (almost) invariably choose "Bowda."] And then you ask them how they know that, which -- if you're lucky -- helps them understand that because they're native speakers of English they have an array of English grammar rules in their head that they're not consciously aware of.

I would like to think that when I decide between "v V" and "V v" in LAadan I'm following some rule that I'm not consciously aware of. But I'm not a native speaker of that language; it has no native speakers. It's far more likely -- less romantic, but far more likely -- that I'm making a purely subjective choice, that I just like the way the form I've chosen sounds to me personally.

6. This is what started me thinking about the sound symbolism (and phonestheme) controversy for English. I'm very partial to the idea that one of the ways new English words are coined is by reference to as-yet-undiscovered "phonosemantic" rules that would go something like this: "English words that fit in the semantic domain of light should, whenever possible, begin with gl." And "English words that fit in the semantic domain of compression should, whenever possible, begin with skw." And so on through the set in the Lawler PDF I posted the link for, and through many other analogous sets.

7. Science-fictionally, extrapolating from what is already a controversial hypothesis (#6), the question would then be whether, in the process of constructing the LAadan language, I set up in my head -- below the level of conscious awareness -- some phonosemantic rules. Rules that I referred to when I chose the forms of words in the past, and still refer to when people write to ask me for new words, as opposed to just making words up in some wild random unregulated fashion. That would be nice. And it would be nice if it could be further extrapolated to a hypothesis that whenever someone works up a conlang, that's what they do. There is, so far as I know, no evidence for such a hypothesis.

8. Anonymous pointed out in a comment that "we know that the language was invented by the women of the Linguist Lines, and doesn't have a 'history' even in its fictional milieu...," which is one way of looking at it (although I would suggest that there's quite a bit of fictional history in the Native Tongue trilogy). On the other hand, it does have a history, since I constructed both the language and the women of the Linguist Lines.

I started LAadan by setting up an inventory of possible meaningful sounds (phonemes), and writing the simplest set of rules for their combination that I was able to devise. I made a grid of forms -- one-syllable, two-syllable, and three-syllable forms -- that were allowed by the phoneme inventory and by the rules for combining them. I took the standard Swadesh list -- the list of roughly 100 words that linguists always try to pin down first when doing fieldwork with a previously unwritten language -- and constructed all the words for that list. (Which was the hardest part of the task, by the way, and the part where I can't explain how my choices were made.) Then I constructed the "functional" chunks -- morphemes that weren't independent words, like the morphemes that indicate time and aspect and case roles and so on. And very shortly I had the 1000 words that I felt I had to have before I could write Native Tongue. Since then, words have been steadily added, mostly by other people, and I have no way of knowing how much additional vocabulary exists that has never been sent to me. Everything that does come my way gets recorded, and gets credited to its originator.