 |
 |
James Chandler wrote: > > Mark Line skripted: > > > If by "artificial" you mean "divergent from known properties of > > (natural) human languages", then your conclusion may be valid and our > > views may be disjoint. > > This is really a very tricky thing to interpret. > You seem to be saying that you would not tolerate any property of an IL > that diverged from those of natural human languages. I seem to be saying that I would argue against elements in an IAL that contradicted what we know about the nature of human language. Anytime I could make a case that "human language doesn't work like that", I'd argue against such an element. It's a very tricky thing for you to interpret only because you don't know in advance against what elements I may be able to make such a case. > Does this mean that you think every property of an IL should be shared > by some natural language, Not strictly, no. It depends on what all you're calling a "property". I can easily imagine a conlang phoneme inventory that does not occur in exactly the same form in any natlang. As long as I don't know of a reason why that phoneme inventory _couldn't_ occur in a natlang, my concern is satisfied. > or all of them, No. > or some subset of them? No. > If I assume the former, how does one deal with the question of degree? > Some languages use a fairly fonetik spelling. Does that mean an IL can > have a fairly fonetik spelling also, or can we have a spelling more > fonetik than in any natural language? What does spelling have to do with the nature of human language? It has no more to do with the nature of human language than painting seascapes has to do with oceanography. > Some languages have a number of productive affixes. Does that mean we > can only tolerate a certain number of these, or can we allow all our > affixes to be productive even if this is not the case in any natural > language? What makes you say there is no natlang that allows all its affixes to be productive? If you know that, you know more than I. As I've been trying to point out, I would argue against conIAL elements which contradict what is known about the nature of human language. AFAIK, "some affixes of a natlang are always less than completely productive" is not an instance of something we know about the nature of human language. It'd be pretty hard to establish, anyway. I assume you mean productivity within a distribution class of morphemes: productivity of "noun" affixes, for instance. But that depends very much on the distribution classes that are posited for the language: what about mass nouns vs. count nouns, for instance? This boils down to the fact that you can, if you want, define distribution classes such that every affix is 100% productive for its head class. And the empirical justification of distribution classes themselves will always be very sensitive to corpus size. So I see little hope of establishing a principle as tenuous as "all natlangs have less-than-productive affixes" with any degree of justification. > It seems to me that at every turn your position will be paralysed by > problems of this sort. I've gotten pretty good at avoiding paralytic positions. You might be surprised, if you didn't jump to conclusions about what I must have "really" meant. -- Mark (Mark P. Line -- Bellevue, Washington -- <[log in to unmask]>)