Theiling Online    Sitemap    Conlang Mailing List HQ   

Re: Language comparison

From:Sai Emrys <saizai@...>
Date:Monday, January 10, 2005, 0:46
> A previous resume: Among the features that you've proposed, there is none > that could distinguish the qualities of any existing human language. Of > course, the day of tomorrow might emerge one where they all apply, and of > course, they apply to other information systems (like programming so-called > "languages").
I am confused; how does this relate? Sure, some of them don't currently exist. That was the point, ne? To propose some features which, if they did exist, would (all other things being equal) constitute an upgrade?
> >Seems to me that the only "difference" is that it's serial, and that > >it has a known way to translate into speech. The latter is irrelevant > > No it isn't. It's what writing systems are all about.
Mind supporting this assertion? Why is it that, in your opinion, writing systems are exclusively for "writing down" speech?
> >- one can always devise one, relatively easily, for a serial code. So > >is a written form that does not yet have a manner of speech not a > >"real" form of language? > > Of course not. Why should it? It's not even a writing system (I'm not > repeating again what a writing system is).
Er... that's interesting, indeed. So, for example, if I were to take any Omniglot page and white out the places where it has phonetic equivalencies, would those no longer be writing systems? Or for that matter, if I give you a Chinese dictionary that contains none of the pronounciations but only the meanings for each character or combination of characters?
> >Hardly. The only thing evolution produces is sufficiency, not > >efficiency. I think it's a mistake to belive that (any) evolution > >results in "the best", just probably the best amongst the local > >competitors. > > (Aren't we all local competitors?) From birth, humans learn language with no > effort (with normal effort, if you will). A baby learns the language of the > people it grows with, and it doesn't depend on the language how long it takes.
... so? I would hardly say that it is "no effort", just that nobody normal seems to fail at the task, overall. If you raise a child speaking Klingon in this way, then surely they would be fluent at it. But I think it would still be more "effortful" - in an anatomical way - than some other languages. Or a better example: someone learning 1900s-era ASL. The sign for "help" used to be off to the side, and moving the elbow; it's not *hard*, but it's certainly *more* effort than the modern sign, which is at center neutral space, nearly symmetric, and quite easy to do. If it were not the case that there is an effort gradient, then why would we have natural evolutions that follow it? (E.g., phonetically going towards laxer sounds, or with signs going towards more central, small-movement, within-normal-ROM ones...)
> So am I, if somebody gives me a convincing example. You haven't, and I doubt > you ever will or anybody. But this believe of mine doesn't affect my > willingness.
Extend the above example, then. Let's say you subjected a test language - X - to phonetic drift of the normal sort. *Other than randomness and contact*, if there is some direction to it, then it would probably be towards what is "easier". So, your "evolved" X would be different than proto-X - it'd have undergone the laziness change. ... unless, of course, I'm completely misrepresenting what little I've learned of historical linguistics.
> http://www.amazon.com/exec/obidos/tg/detail/-/0521667720/
Thanks. I'll put that on ILL.
> Language has no characters, only written language has. And it has definitly > no bits.
I was quoting something I'd read recently (probably Umberto Eco - Serendipities); do you want the source? A quick Google search gives at least one article that makes a similar point: http://cse.stanford.edu/class/sophomore-college/projects-99/information-theory/entropy_of_english_9.html .
> Maybe you're just looking for a language with a large amount of grammatical > categories. Have a look at sanskrit.
Does it do that better?
> Certainly. Because our visual recognition of characters is helped a lot when > we always see them in the same orientation. We don't recognize character for > character. If you hold a book upside down, then you do. > > Okay, I admit that I don't deny the mere possibility of a characters whose > recognition doesn't depend on directionality (neither do I deny the > possibility that the earth stops from turning).
*laugh* That's a bit cynical, but I suppose that's improvement. All that it would require, at minimum, is that it be symmetric. Or written in a way that did not imply a required orientation. Wine script, as a simple example, makes directionality irrelevant; it still retains the inversion problem, but that could be solved simply by making all characters symmetric across the stem, or explicitly saying that the mirror-inversions are identical. But suppose for a moment I did devise a writing system that had this feature. Would you not consider it an improvement (holding all other qualities equal), if only a small one?
> The difference between writing and other methods of conveying meaning is > that writing always corresponds to speech.
Always? Odd. Most mathematical etc., forms of writing I know are rather cumbersome to translate into speech. I wouldn't call them any less "writing". Nor, for that matter, if I were to create a conlang and just skip the phonetics step.
> This is not because of writing, but because of the frames where writing is > displayed. It's not a characteristic of the alphabet.
*shakes head* Perhaps my analogy didn't come across well. Imagine, as a simple example, a written language that looked somewhat like a thoughtweb. You could even make it composed of serialized symbols on any given thread, if you want (though I think it'd be more powerful to do away with that too); as a whole, it would be markedly different from present writing systems. Capable, at least, of expressing notions in an explicitly interconnected, possibly fractal way. That it might be hard to translate into serial speech is something that doesn't bother me at all. (A literary example of this, IIRC, could be found in one of the forms of magic found in the Deathgate Cycle.)
> Well of course it does. If we were talking to each other face to face, you > would notice that I'd be getting tired and annoyed of replying at length to > your arguments.
Yes, but can you express "the car drove over the tree" very easily with just gestures? I can, with ASL. Which is pretty easy to use in parallel with speech. All I was saying in this is that the ad hoc methods of gesture are not as good at conveying meaning in a well-understood way as ones that incorporate indexing, spatial grammar, and other features found in e.g. ASL. And that thus, if you were to incorporate them, then even if you continued to use gesture only as a complement to your speech, the whole would be enriched.
> My contribution was in response to you idea that a language would be better > if it could be used in any given environment/situation. My contribution was > meant to deny this.
I don't see how it did. Elaborate, please?
> >Yes, but that wouldn't be differential. I can conceive of situations > >where this would be useful (e.g. where the message to one part of the > >audience needs to be filtered). Nevertheless, the point here is only > >that current languages are not capable of doing this (except in an ad > >hoc manner I described above, which you'll get in any movie with an > >"we've been bugged" scene)... and that it is probably possible to do > >it better. > > Tell me how (I'm willing to believe in the possibility of anything).
How to do it better? OK, some examples (which I don't propose to be the best method, just possibilities): * have a language have both signed and speech forms. At minimum, it would mean that two fluent speakers would have two modes available - which, even if they aren't able to be used in an additively parallel way, would make a much better "secondary" mode than ad hoc gesture. (Sure, you could do this by just knowing two languages - one spoken, one signed - but that's already harder.) * have gestures that are explicitly understood to add certain connotations - e.g., "I'm lying" would be one. Use these only to highlight the parts you're saying that are lies. * intersplice nouns or adjectives in sign to change your message - e.g. "I gave the card to [my brother] my customer" or "Sure, he's a [stupid] nice man". Another possibility would be to delegate some of the grammar to the signing - so that, without having access to both sources, the message would get garbled.
> All known languages are full of redundance (see the above mentioned > linguistic textbook) (well, I haven't mentioned any, but I don't any in > Engish and most are very useful). This is because, as I've explained, > language doesn't only develop into maximal denseness, but also into minimal > confusion.
Well, sure. Redundancy is useful. But if you already know that my friend is female, then how does it help to refer to this fact every time I mention anything she does or owns? It doesn't add any useful redundancy other than that I'm talking about a female - what if I were talking about my friend, and my mother? Doesn't help potentially confusing the two. Another system might. (As simple as having, e.g., "possessive-#1" "possessive-#2", etc. and instantiating them to refer to particular people. ASL does this.)
> Tell me how (I'm willing to believe in the possibility of anything). How > about - another language! :)
I don't know yet. :-P I've only had some very tentative ideas about it. And sure, with another language. ;-) That's what all of this is about. Perhaps some features could be incorporated into existing languages, but that wasn't the point I was trying to make. - Sai