Theiling Online    Sitemap    Conlang Mailing List HQ   

what is a loglang?

From:Mark P. Line <mark@...>
Date:Wednesday, May 5, 2004, 13:35
Some things that popped up in a few recent posts made me squirm uneasily
and wonder what it is exactly that we all think a loglang *is*.

For my part, I always distinguish between _logic_ and _reasoning_,
analogous to the differences between Artificial Intelligence and Cognitive
Science and between analytical (formal) linguistics and cognitive
linguistics. I believe that a logic ('logic' without the indefinite
article is the name of a discipline whose subject matter is logics) is
something we design to support (some subset of) reasoning.

Although it would be possible, I guess, to envision a conlang which
somehow embodied or expressed a (formal, previously defined) logic which
would then in turn be used to support reasoning, I generally think of a
loglang as a language that is intended to support reasoning directly: the
loglang *is* the logic, as it were, even though it uses linguistic forms
rather than mathematical or other symbolic forms.

If that's true, then the only requirements for a loglang (other than those
placed on any conlang in terms of how far you're willing to let it diverge
from natlang evolutionary space) would have to do with the kind(s) of
reasoning it should support and with the degree to which it should do so.

Finally, I have found that the best way to support reasoning of the
_conceptual_ kind is to define an ontology (or several, as long as they
work together). That's the approach in my upcoming Waldzell Conlang:
define an upper ontology (i.e. a consistent set of definitions that form
the "highest", most abstract layer at the top of a potentially unlimited
hierarchy of lower ontologies) and map it directly onto the conlang.
Construction of lower ontologies then requires nothing more than

Yes? No?

For extra credit: Most of us generally think of _conceptual_ reasoning
when we think of logic and loglangs, I reckon. But are there other kinds
of reasoning that a loglang might support (assuming that's what you think
a loglang is)?

-- Mark


And Rosta <a.rosta@...>