Hallo conlangers!

On Saturday 24 September 2011 19:19:35, And Rosta wrote:

> R A Brown, On 24/09/2011 08:33:
> > On 24/09/2011 04:57, Alex Fink wrote:
> >> On Fri, 23 Sep 2011 21:17:19 +0100, And
> >> 
> >> Rosta wrote:
> >>> Fithians could speak human languages far better than
> >>> humans (i.e. produce and comprehend grammatical
> >>> sentences of far greater complexity than humans can
> >>> cope with in real time).
> >> 
> >> That also seems not to strike at the main point, which I
> >> take to be human comprehensibility of full-blown Fith.
> > 
> > AMEN!
> > 
> > And's statement above is, of course, scientifically
> > untestable as Fithians are a fictitious race.
> 
> However, it is trivially simple, with a mere second or two's reflection, to
> translate my statement into something substantively equivalent that is
> scientifically testable: A machine with the processing abilities
> attributed to Fithians could speak human languages far better than humans
> (i.e. produce and comprehend grammatical sentences of far greater
> complexity than humans can cope with in real time.

Sure can such a machine be built (or rather programmed, as this
seems to be a matter of software).  But is that meaningful from
a *linguistic* standpoint?  Linguistics deals with languages
that are *actually spoken by human beings*, not with abstractions
from such languages that could be processed by computers but not
by human beings.  Formal grammars are only *abstractions*, used
to treat particular aspects of human language, not the languages
themselves.

A test with a computer programmed to emulate the Fithian language
capabilities as described by Jeffrey may be meaningful with Fith,
but not with a human language.
 
> >> I dropped the ball on the last conversation on Fith, and
> >> so this will be a retread, for which I apologise.
> > 
> > No apology needed. The points you make are IMO very
> > relevant. It was not Jeffrey Henning's intention that Fith
> > be human-usable and AFAIK it is _generally_ accepted that
> > full blown Fith is not. I really need to see firm evidence
> > to the contrary and, so far, I have not seen it.
> 
> It can also be demonstrated that full-blown English is not human-usable
> either. The lack of human-usability of full-blown Fith does not in itself
> prove that Fith is "a language with some kind of
> mathematically/grammatically implausible syntax structure".

As I have said, that is meaningless.  A Chomskyan specification
of the grammar of English may allow for "grammatically correct"
utterances no human beings could ever cope with, but that only
means that such a formal specification is only an imperfect
model of the actual language as it is spoken by its community.

Of course, with a conlang, we start with a specification of
the language (not necessarily a *formal* specification, though
- most of us describe their conlangs in plain wordage, not in
Backus-Naur forms or anything like that) that exists before
the language is spoken, if ever, by anyone.  With a natlang,
any formal specification that could be used to program a
computer to parse the language, is a *secondary abstraction*
from the observed linguistic behaviour of the speakers.
 
> > On 23/09/2011 21:17, And Rosta wrote:
> > [snip]
> > 
> >> I have claimed that there are three different and
> >> discrepant Lojbans: (1) the formal definition, (2) the
> >> langue internalized by the ablest speakers, (3) the
> >> langue inherent in actual mainstream parole. (1) is
> >> unspeakable, but (2) is speakable and is functionally
> >> equivalent to (1); if (1) had been created by people
> >> with some kind of reasonable understanding of human
> >> language, it could have been like (2) and hence
> >> speakable.
> > 
> > This IMHO is hair-splitting. By Lojban I mean and understand
> > the language developed by the Logical Language Group and
> > used by the Lojban community.
> 
> You can fudge these together in your mind so that to you they appear to be
> one thing, but in that case there's no point in discussing the
> human-usability of Lojban. Only by splitting the hairs you disdain to
> split, and drawing the distinctions I drew, can one judge the
> human-usability of Lojban (-- i.e. how close are the three types of
> Lojban).

Of course, any evaluation of the human usability of Lojban must
start with the observed discrepancies between the specification
of the language released by the LLG and the actual usage of the
language by the Lojban community.  If one finds that the Lojbanists
consistently use only a subset of the language, that points at
difficulties with the elements not used, and it remains to be
explored *why* those elements are not used.
 
> >> I find it rather annoying when people cite the existence
> >> of (3) as being of linguistic, rather than strictly
> >> social, significance: it is as if one were to study the
> >> French speech of english schoolchildren trying to speak
> >> French to each other,
> > 
> > The analogy is simply false. French is the language of _la
> > Francophonie_. English schoolchildren are attempting to
> > speak that language. Lojban, as I understand it, is the
> > language of the Lojban community.
> 
> Some Lojbanists think as you do, probably a minority of them nowadays. Most
> think of (1) as the true Lojban. Certainly it's (1) that has given Lojban
> most of its kudos, as the most explicitly defined of all languages.

Sure.  Lojban is a *conlang*, and that means that we have a
specification of the language that *predates* any usage of the
language.  We have no such specification for *any* natlang.
 
> > Your comparison above could just as well be expressed as:
> > "I have claimed that there are three different and
> > discrepant Frenches: (1) the formal definition of L'Académie
> > française, (2) the langue internalized by the ablest
> > speakers, (3) the langue inherent in actual mainstream parole."
> > 
> > Sorry - I know I am only an amateur linguist, but to me (3) is what I
> > understand as the main manifestation of French, with (1) and (2)
> > being artificial dialects.
> 
> For your own purposes it's fine for you to think that (3) is the true
> Lojban, and to apply the name Lojban to it. And I can kind of see why you
> said "people do actually both write and speak Lojban" in response to the
> question of whether Lojban exemplifies "a language with some kind of
> mathematically/grammatically implausible syntax structure" -- because if,
> for you, Lojban is whatever people do when they claim they are speaking
> Lojban (i.e. Lojban Type 3), then of course it is human-usable. But I
> would have hoped that you'd have realized both that I too knew that Lojban
> Type 3 is spoken, and that it was the distinction among the three types of
> Lojban that made pertinent the question of whether it exemplifies "a
> language with some kind of mathematically/grammatically implausible syntax
> structure".

Certainly, "Lojban Type 1" (the language as laid down in the
specification of the Logical Language Group) and "Lojban Type 3"
(the language as used in practice by actual Lojban speakers) are
not the same thing, and any observed discrepancy between the two
may shed light on the limits of human language processability.

--
... brought to you by the Weeping Elf
http://www.joerg-rhiemeier.de/Conlang/index.html
"Bêsel asa Êm, a Êm atha cvanthal a cvanth atha Êmel." - SiM 1:1