On 31 May 2016 at 22:05, David Peterson <[log in to unmask]> wrote:

> > On May 31, 2016, at 9:23 AM, And Rosta <[log in to unmask]> wrote:
> >
> > If conlangers really were creating languages, then your characterization
> of
> > glossopoesy would be right. But I have come to think that Jeffrey
> Henning's
> > term 'model languages' (which are to languages as model railways are to
> > railways) is apter for what conlangs are. If I build a model cathedral
> with
> > plasticene, you don't laud me for having built a cathedral (much tho you
> > may laud me for having built a model cathedral).
> This is, of course, an inappropriate analogy, to put it mildly. If you
> build a model cathedral that’s a few feet high, people will look at you
> funny if you ask them to walk inside and sit at one of the pews. Even a
> badly constructed language can be used for communication, if the words are
> there. There’s nothing that can rescue this analogy: it ought to be
> permanently abandoned.

The point about the cathedral is merely that laudable tho creating a model
X might be, it does not amount to creating an X. And just because one has
invented something that can be used for communication doesn't mean that it
is a language. A model language can be used for communication too.

 > My reasons for thinking

> > that conlangs fall short of being languages, or at least never escape
> being
> > model languages, are these:
> >
> > 1. Every published conlang is a non-working model [or almost every -- see
> > (3)]. No conlang's published documentation is complete enough for the
> > conlang to be usable (as a language, mapping between thought and
> > perceptibilia). Since many conlangs are used, they are of course usable
> in
> > some sense, but what happens is that the conlang is a sketch of parts of
> a
> > language, and the users call on their tacit knowledge of natural language
> > to complete the sketch into a working model. The conlanger has created
> the
> > sketch, not the working model.
> If you take this criterion seriously, you’ll see that all natural
> languages also fail it, and humanity has, thus far, failed to produce a
> language of any kind. Assuming, though, that what you meant was languages
> are real, and since conlangs aren’t, you have to have a book that explains
> them (natlangs don’t need a book to explain them, since they already
> exist), and currently no published materials are complete enough to explain
> any conlang, then all one needs to do is produce such a book. It wouldn’t
> be challenging; it’d just take time and care. Unless your contention is
> that there is no possible book that could ever explain a conlang in
> sufficient detail, in which case I’d counter that the same is true of
> natural languages, which means that we’d end up coming back to “natlangs
> are real and conlangs aren’t”.

My contention is that nobody has produced such a book and that the task of
doing so would, at least in 2016, be immensely challenging. (My original
message did say that it can be done, but for something so simple that it
still warrants being called a model language.) Without the book, one cannot
really claim to have invented a language.

> In reality, though, what you describe above when it comes to users using a
> conlang is exactly what kids do when it comes to learning a natlang (i.e.
> take an incomplete model and flesh it out). They make tons of mistakes—most
> of which are eventually corrected—but some stick around and become a part
> of their new version of the language. These innovations aren’t taught; they
> don’t exist in books; they’re not a part of the language—that is, until
> they become a part of the language.
> The main point is this: If you accept there could be a sufficient written
> description of a natlang, then there will also be a sufficient written
> description of a conlang, even if one hasn’t been written yet.

I don't disagree with that. My message was prompted by Padraic's eloquent
description of what it means to invent a language, which seemed to
presuppose that this is what conlangers do, and my point is just that in
the history of conlanging so far, conlangs have not got beyond being model
languages. I think us conlangers should be appropriately lauded for making
model languages, but not for the far greater achievement of making a full

I also don’t see an important difference between a natlang and a conlang
> when it comes to its utility.

A fully specified working conlang is likely to lack the depth and richness
of the natlang, and richness and depth have utility. A partially specified
conlang, usable through being projected onto the users' knowledge of
natlang(s), has, I feel, about as much utility as a natlang second language
but less than a first language.