From Http://Members.Aol.Com/Lassailly/Tunuframe.Html wrote:

> > Dans un courrier dat=E9 du 14/09/99 20:12:55  , BDW a =E9crit :
> -----------
> Ed has promised me that once I delve
> deeper into cognitive grammar..., I'll get=20
> closed to knowing what exactly the semantic definitions of 'noun' and
> are - other than 'a thingie you can get, see or think of', and 'an acti=
> can perform, or get performed to you' ;-).....
> ----------
> when you're enlightened, sparkle on us please.

I'll sparkle a bit, but only dimly like a far-off firefly.

Ronald Langacker's "Foundations of Cognitive Grammar" is a large and
carefully constructed edifice, with layer upon layer of careful
definition and justification for those definitions.  L's semantic
definitions of "noun," "verb," and "adjective" are built upon more
layers of definition and justification than I care to try to duplicate
in an email message, alas.  And divorced from their context, they will
probably be no more impressive or interesting than the "thingie" and
"action" definitions which Boudewijn gave above.  However, within the
system they work remarkably well.

A THING profiles a region within a conceptual domain.  The most
prototypical nouns -- physical objects -- are primarily understood as
regions within the domain of three-dimensional space, but any
conceptual material can serve.  Regions with definite boundaries are
count nouns; regions without definite boundaries are mass nouns.

Nouns symbolize things.

An ATEMPORAL RELATION is similar in some ways to a THING; the region
that is a THING consists of a group of interconnected entities, where
an entity is absolutely anything that can be consciously or
unconsciously conceptualized.  In a THING, the entity or entities are
profiled, whereas in an ATEMPORAL RELATION the interconnections
themselves are profiled.  This is why a "group" is a THING but
"together" is an ATEMPORAL RELATION.  The same content, construed in
different ways, here becomes a THING or a RELATION.

Adjectives symbolize atemporal relations.

A PROCESS is like an atemporal relation, in that relationships are
profiled, but it is processed differently than an atemporal relation:
of Langacker's bolder proposals: he suggests that "the jumping monkey"
and "the monkey jumps" are actually processed differently by the mind;
in the former, the act of jumping is considered as a whole, while in
the latter, the act of jumping is considered in its temporal parts,
start to finish, in order.  (This does not necessarily imply that this
is conscious, or that it involves a vivid, movie-like mental picture.)
 Incidentally, Langacker distinguishes two types of process, which he
calls "perfective" and "imperfective," which are similar to but not
identical with the concept of "active" and "stative."  Perfective
processes, like count nouns, are structured, individualized, and
distinguishable, while imperfective processes, like mass nouns, are
indefinitely divisible without loss of structure.

Verbs symbolize processes.

That's the gutted, ripped-from-context,
technical-terms-left-unnoted-and-unexplained version of Langacker's
semantic characterization of parts-of-speech.  I feel rather like
someone who has claimed there exists a creature which can fly and find
its way around by sonar, and asked to provide a demonstration, has
pulled the ears and wings off a bat and presented the wrinkly bloody
little bits of them to his audience.

Ed just stopped in to see what condition his condition was in.
                   [log in to unmask]