CHAT: Wierzbicka

Gerald,

I wish you better luck and more patience than I had with Wierzbicka.
As I said, I really think she's on to something in terms of practical
technique, but not in terms of knowing how things really work.

Let me clarify my previous position with the Magic of Graphics:

The Cognitive Linguistics school (Langacker, Lakoff, and several
others) presume something like the following:

A relates via B to C, where:

A = words and constructions of the language,
B = mechanisms relating linguistic to nonlinguistic human experience
C = nonlinguistic human experience.

Wierzbicka on the contrary would say that:

A relates via D to B and C, where:

A = nonprimitive words and constructions,
D = configurations of semantic primitives,

And B and C are the same.

For Langacker and Co, to understand language, we must try and
investigate A, B, and C.

For Wierzbicka, B and C are off limits, because they are
nonlinguistic and therefore you can't "define" them in the most strict
sense.  We can try and figure out what D are (while leaving them
"undefinable" and therefore avoiding B and C), and we must investigate
A only in terms of D.

This might be a reasonable statement of a methodology, but
Wierzbicka's claim is that it is not just a methodology, it is
*actually true* that there exists a small, discrete set of primitives,
D, and that, quite seriously, when we understand any words, the
process includes our minds reducing them to the level of primitives
(thousands, possibly, for every single word, since the explication of
the simple word "mouse" requires several pages of text and is still
very incomplete!).  It is only when, say, the five, ten, or twenty
words of a sentence have been interpreted as a vast cloud of
paragraphs upon paragraphs of primitives, that they are interpreted in
terms of nonlinguistic reality.

Yes, she says in so many words, fairly late in the book, that she
sees this as the real mechanism of linguistic interpretation.  She
rightly says that we should not be amazed that the brain should be
capable of such vast amounts of computation.  I don't think we should
be amazed by the scope of it, but I think we should be amazed by the
apparent needlessness of it.

B -- the interface between linguistic and nonlinguistic reality --
has to exist, whether or not Wierzbicka considers it amenable to
investigation by linguists.  If B can relate D to C, then there's no
reason it couldn't relate A to C without the intervention of D.
Indeed, the existence of the intervening "vast cloud of primitives"
layer seems amazingly convenient for lexicographers like Wierzbicka,
and amazingly inconvenient for, oh, minds and brains.

Wierzbicka does differentiate between using her primitives as a
methodology and a real theory.  She makes it clear that she is making
the stronger claim -- that the human mind really does happen to
process language in a way that is amazingly convenient for
lexicographers -- but allows that people who do not believe it may
still be interested in her work for its practical applications.

Interestingly, her primitives, in the state to which they have
expanded, allow her to include many of the key ideas of the
semanticists she pooh-poohs, such as meanings which are structured via
image schema which are defined in relation to the human environment,
or meanings which are structured via metaphoric mappings (see the
primitive LIKE), and so on.

For all this, she *is* worth reading.  Her theory *is* worth looking
into -- it's a brilliant theory of translation, given its
methodological choices.  Good luck with it.  (I got about up to the
"evidentials" part before it just became too much for me.)

A few specific replies to your comments:

Gerald Koenig wrote:

> ... have just begun reading Wierzbicka, but I haven't seen
> any denial of prelinguistic thought process.
Not denial; just labeling of it as "off limits" for investigation by
semanticists, because it is "indefinable."  She clearly is no fool,
and it would take a fool to deny the existence of the world of
nonlinguistic thought and experience.

> What Wierzbicka says to me is that these
> concepts are primitives in the sense that everyone knows them.
And in the sense that combinations of them fully determine the
meanings of non-primitive terms, a much stronger claim.

> Well some concepts are priviledged, if we are going to model the
> world with symbols. Maybe some kind of occam's razor limits our
> directional primitives to 6 Cartesians.  Maybe evolution can't
> manage an n dimensional manifold. We would have very complex ear
> canals indeed to orient ourselves. It's hard to imagine a number
> system that is simpler than binary, and that's what we are using
> here as basis. Some things are just, well fundamental, like
> axioms or the "given". I don't know if matter itself can be
> infinitly subdivided. It seems that  infinity could
> be one directional. Paradoxes are avoided in logic by limiting
> the size of certain sets.
Whoo, that's a lot of stuff, and I'm not quite sure how it all
relates, but let me say -- some of the people Wierzbicka criticises
would strongly agree that there are human-universal structures of
meaning and constraints on meaning.  (Lakoff says exactly as much
in _Women, Fire, and Dangerous Things_.)  Wierzbicka's peculiar claim
is that, for all linguistic purposes, those universal structures
are summed up in a set of somewhat less than 100 universal *words*
and the structures that connect them.  In other words, we need to look
for them in layer D.  Others would find them in layers B and C, but
Wierzbicka is not willing to look at B and C because they are not
linguistic.


> I will look for that in my reading. So far it seems to me that she
> is just fighting for the priviledged status of the primitive set,
> now up around 89. If she is claiming that all meaning is an exact
> combination of primitives, it is tantamount to denying the reality
> of the real number system.
She does claim in so many words that all linguistic meaning is
combinations of primitives.


> She is working on a universal grammar, also presumably biologically
> and hence physically based. She's a serious finder of fact, and will
> probably destroy quite a few theories by that alone.
Yes.  She's very good at what she does.  Very very good.

> Ed, I respect your take on Wierzbicka, and will be looking for your
> caveats while reading the book.
Thanks. :)

Ed