Jörg Rhiemeier, On 07/06/2011 20:39:
> Hallo conlangers!
> I wanted to reply to this, but wasn't sure, then forgot about it,
> but here is my reply.
> On Thursday 02 June 2011 22:36:34, And Rosta wrote:
>> I meant: Why does a fictional natlang loglang's implausibility make
>> fictional natlang loglangs a bad idea?
>> I don't see why implausibility entails badness; and I can think of various
>> virtues that implausible fictions can have (for example, in this instance,
>> by contemplating the virtues of a natlang loglang (within an implausible
>> fiction) we might gain insights into the consequences of nonfictional
>> natlangs' lack of loglanginess; fantasy fiction can often illuminate our
>> understanding of the real world).
> Of course, what makes a good or a bad story depends on whom you
> ask, but most people would say that an implausible story was a
> bad one.  It is true that fantasy fiction (or, for that matter,
> science fiction) can often illuminate our understanding of the
> real world, but even there, most people care about plausibility
> (in the confines of an alternative reality where *some* things
> work differently than in the real world, of course).

I think it's truer to say that implausibility is generally felt to be bad when it has no compensatory virtues. Dickensian plots are implausible but much delighted in. Likewise Rom Com happy endings. But anyway, you yourself acknowledge here that people do embrace alternative realities where some things work differently than in the real world. So I imagine your original logic was something like "(1) Natlang loglangs in fiction are implausible; (2) implausibility in fiction is a bad thing unless there are compensatory virtues; (3) I dislike loglangs so see no possible compensatory virtues in that implausible fiction; therefore (4) natlang loglangs in fiction are a bad thing". I think Step 3 is necessary to get the logic to work.

>> My definition of "loglang" is a language that unambiguously encodes
>> predicate--argument structure (which I take to subsume
>> quantifier--variable structure), without having to leave much of
>> predicate--argument unencoded.
> What exactly do you mean?  A language that clearly states which
> semantic role which argument plays, avoiding such ambiguities as
> the SAE notion of a "subject" which can actually play just about
> any semantic role imaginable?

No, a language that unambiguously shows, as predicate logic notation does, what is an argument of what.

>>        I agree that, on the basis of our current
>> state of knowledge about loglang design space, a full-blown loglang would
>> probably be unacquirable as a natlang.
> At least, it would probably be "unevolvable", if there is such
> a word.

Unevolvable, but not uninventable and perhaps not unspeakable. But (I think, based on my current understanding of the loglang problem and the human language faculty) unacquirable.
>>        But a slightly diluted loglang that
>> encodes predicate--argument structure unambiguously but leaves a lot
>> unencoded (-- which is what ordinary Lojban usage is like) doesn't strike
>> me as problematic to acquire.
> I cannot evaluate to which degree a language like Lojban is
> acquirable or not.  There is a community of Lojban users, which
> indicates that the language is not completely unusable, but some
> people say that those Lojbanists actually speak a less than
> perfect version of that language.

It used to be that most of what was encoded in Lojban bore little resemblance to what the speakers intended to communicate, and neither speakers nor hearers realized this. I suspect this remains the case today. However, there has been at least one highly competent speaker. Lojban is probably easier than other lgs to speak well, because it comes with a much more explicit definition of how it works. It's also the case that some aspects of the explicit definition are unusable and must tacitly be replaced by usuable alternatives.

>To me, at least, Lojban
> grammar looks very difficult, though I don't see that it would
> be unlearnable, but then I never seriously attempted to learn
> that language.

Most of Lojban is unnecessarily complicated, but even so, there's not much that strikes me as difficult, other than the allomorphy -- and even that is easy to grasp; it's just hard to memorize and use. Because Lojban's documentation is also comparatively explicit about Lojban's structure and meaning, there's also a sense in which it is difficult for learners who find it difficult to think explicitly about syntax and semantics; it's not uncommon to find people who are very good at picking up languages intuitively, but hopeless at thinking about them analytically, and such people would likely describe Lojban as difficult, for the very same reasons as I describe it as easy.
> And I have seen too little of Livagian to judge whether it would
> make a plausible natlang or not ;)

Everybody but me has seen too little of it. It's debatable whether Livagian is a natlang: it's not learnt natively, and is not spoken in real-time (tho it can be heard in real-time), but it is much-used and is the main medium of some sorts of communication, especially written. This strikes me as plausible. What's implausible is how the language ever came to be created in the first place -- it could only have been a Panini-type figure, able to invent a wholly new sort of syntax (possibly blending this with morphological material from a preexisting natlang) and able to prevail upon others to use it.