Print

Print


Hallo conlangers!

On Monday 19 November 2012 01:09:58 Logan Kearsley wrote:

> On 18 November 2012 10:18, And Rosta <[log in to unmask]> wrote:
> > Logan Kearsley, On 15/11/2012 16:50:
> >> Human languages *don't do* arbitrary variable bindings
> >> (except maybe for that spatial pronoun thing in sign languages, which
> >> I'm probably doing a horrible job of referencing accurately).
> > 
> > What are arbitrary variable bindings, such that human languages don't do
> > them?
> 
> Picking a strictly arbitrary symbol to stand in for a referent
> as-needed. Predicate calculus variables have no inherent meaning; as
> many as you need are made up as needed and their names can be
> interchanged with no consequence for the interpretation of a
> proposition. Human languages don't allow the creation of new new
> variables at will, nor do they allow arbitrary assignments of variable
> symbols to referents; human pronouns/anaphors always have restrictions
> that drive the choice of which name to use for which referent. Thus, I
> know of no language that has, e.g., two different third-person
> pronouns that are completely interchangeable; maybe one is proximate
> and the other is obviative, or maybe there's some other distinction,
> but changing which one you use for which referent will change the
> interpretation of the sentence, unlike in predicate calculus where if
> I choose to refer to one thing as x and another thing as y, it makes
> no difference if I switch them around or pick completely different
> names.

Yep.  Human languages, as far as I know them, don't do this kind
of thing.  (You can never say "no natlang does it" - the next yet
undocumented indigenous language of whichever remote corner of
our world may turn out to show the relevant feature, unless, of
course, we are talking about something that would render the
language human-unusable.)  Variables in predicate calculus can
be likened to freely assignable anaphoric pronouns; one can say
that programming languages such as C have an open class of
assignable pronouns.  Where several deictic or anaphoric pronouns
occur in a natlang, they are usually not freely assignable, but
their assignment is governed by rules regarding gender, deixis,
obviation or other syntactic or semantic criteria.  (Though many
languages have constructions such as "a man - let's name him Joe"
which are indeed quite similar to the freely assignable variables
of programming languages; and a loglang may grammaticalize such
constructions.)

> On 18 November 2012 13:40, Jörg Rhiemeier <[log in to unmask]> wrote:
> > Hallo conlangers!
> > [...]
> > Yes.  I feel that Logan strives for a *naturalistic* loglang,
> > which is of course a contradiction in terms (if I have
> > misunderstood you, Logan, I apologize).  Loglangs and
> > naturalistic conlangs are two different kettles of fish, and
> > attempts at reconciling both in a single conlang tend to fall
> > in between and end up being neither.  A loglang indeed gets
> > quite close to a "spoken predicate calculus", otherwise it is
> > not really a loglang.
> 
> "Naturalistic loglang" is, perhaps, an accurate explanation, but a
> misleading one. And explained it well, but I'll try to put the idea in
> my own terms: natural languages do not unambiguously encode logic,
> thus a logical language will not be naturalistic.

Indeed not!  That was my point.

>       But some means of
> encoding logic make use of structures / strategies that are found in
> natural languages, and others make use of structures and strategies
> that are *not* used in natural languages, and thus making a log*lang*
> as opposed to a logic notation
> is a matter of picking those structures and strategies that do
> actually occur in natural languages, even if no natural language uses
> them all at once or in quite the same way. That is an interesting
> pursuit, whereas simply trying to represent logic with phonemes with
> no other restrictions and no regard for the ability of humans to
> process it is a trivial exercise.

I now see what you mean.  It is indeed quite easy to wrap a formal
calculus in human-pronounceable phonemes; it is much less trivial
to represent a formal calculus in something that works essentially
like a human language.  Certainly, such languages as Lojban occupy
a middle ground between natural languages and formal calculi.
They have features of one and features of the other, and are thus
different from either.

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