Newsgroups: comp.ai.philosophy
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!bloom-beacon.mit.edu!gatech!howland.reston.ans.net!vixen.cso.uiuc.edu!sdd.hp.com!hplabs!hplntx!curry
From: curry@hpl.hp.com (Bo Curry)
Subject: Re: What's innate? (Was Re: Artificial Neural Networks and Cognition
Sender: news@hpl.hp.com (HPLabs Usenet Login)
Message-ID: <D3swMp.Luz@hpl.hp.com>
Date: Fri, 10 Feb 1995 20:23:13 GMT
References: <3gtu3i$rf3@mp.cs.niu.edu> <D3GBKw.F5D@spss.com> <3guoku$bci@mp.cs.niu.edu> <D3LG9D.G18@hpl.hp.com> <3h69pv$en3@mp.cs.niu.edu> <D3nB2v.H6q@hpl.hp.com> <3h8tei$mab@agate.berkeley.edu> <D3p5D7.F5A@hpl.hp.com> <3hbgfe$eks@mp.cs.niu.edu> <D3qzBq.45r@hpl.hp.com> <3he4k1$8nb@mp.cs.niu.edu>
Nntp-Posting-Host: saiph.hpl.hp.com
Organization: Hewlett-Packard Laboratories, Palo Alto, CA
X-Newsreader: TIN [version 1.2 PL2]
Lines: 111

: >: >: curry@hpl.hp.com (Bo Curry) wrote:
: >But do you have a consistent way to define "the meanings people
: >desire to communicate" which is independent of their expression
: >in language?

Neil Rickert (rickert@cs.niu.edu) wrote:
: You are asking the impossible.  The meaning of "define" already
: requires that an explanation be given using language.  But meaning
: itself does not require language.  There is good evidence to suggest
: that dogs can have meanings, even though they do not have language.

You may simply "describe" the non-linguistic meanings of the
dogs, etc, if they can't be "defined".

Knowledge of various kinds is organized and represented
in some way in our brains. Can we agree on some terminology?
I'd like to talk about "semantic knowledge", by which I mean
knowledge of the sort which can be expressed in language
(as opposed to, say, knowledge of how strawberries taste,
or of how to ride a bicycle), and about "linquistic knowledge",
by which I mean knowledge of the structures, customs,
and vocabularies of the language itself.
I think we can leave phonetics out of both categories,
since the work on deaf signers strongly suggests that
it isn't relevant to the sort of things we want to talk about.

So dogs, presumably, have neither "semantic" nor "linguistic"
knowledge, though they have knowledge of other sorts.

I'll be happy to modify my terminology, if you have a better
suggestion, but not my categories.

: The assumption that everything cognitively important is based
: on language is, in my opinion, a fallacy.

It isn't, however, a fallacy to which I subscribe.

: >: >3. The large majority of SOV languages put question words at the
: >: >end of sentences, and form postpositional phrases. The large
: >: >majority of SVO languages put question words at the beginning,
: >: >and form prepositional phrases. It is not uncommon for
: >: >a language to shift from SOV to SVO (consider, in English,
: >: >archaic and poetic phrases like "til death do us part").
: >: >When a language does so, however, the question words and
: >: >prepositions shift in synchrony.

: >: Again, these regularities could be semantic in origin.

: >Can you explain how, in more detail? This one is not obvious to me.

: I was deliberately vague.  You started talking about questions, then
: you threw in the "til death do us part" which is not a question.
: Therefore I had trouble fully comprehending your point, so gave an
: correspondingly imprecise response.

"til death do us part" was an example of an archaic English SOV
word order. It was intended to illustrate that, even in our own
language, such SOV <=> SVO reversals are not unheard of. It had
nothing directly to do with questions. A good editor would
probably have asked me to insert a paragraph break before "It is not
uncommon ...".

: >What do you mean by "semantic structures"? Doesn't the existence
: >of a "structure" imply some sort of syntax? What is syntax, in
: >fact, if it is not the structures we impose on our knowledge?

: Why should structure imply syntax.  If I jump into the river for a
: swim I will get wet.  Evidently there is some relationship between
: the concepts "swim", "river", "beach" and "jump".  This relationship
: is probably quite complex, but I don't see that it is syntactic.
: Yet, although the structure is not syntactic it would be surprising
: if our sentences did not, in some way, reflect this semantic
: structure.

But if you express some idea involving rivers, beaches, etc, using
language, you have to impose some structure on your utterance.
That structure *is* the syntax, or grammar, of the language.
It may reflect deeper structures in the underlying knowledge
representation, or it may not. This is an empirical question.

But the structure of the *utterance* is not *identical* to the
structure of the underlying knowledge representation, even in
the unlikely case that they are isomorphic. It has much lower
bandwidth, at the very least. It makes perfect sense to talk
about these things (the syntactical structure of utterances,
and the semantic structure of the brain's organization of
knowledge) separately, and to give them different names,
and to enquire about how they might be related.

I don't deny that they *are* related, nor, I believe, would Chomsky.

: Since you presented the analogy "Who did you see John with?", I can
: guess the meaning you want me to assign to the sentence.  But if
: I heard that sentence without being presented with such hints, I
: would have no idea what it was supposed to mean.

: If I try to reorganize the sentence, I might get "Did you see John
: and whom?" It strikes me that the result is ambiguous.  It could be
: asking for a yes or no answer as to whether I saw, or it could be
: asking for the person represented by "whom".  The original sentence
: using "with", in the form that you presented it, has no corresponding
: ambiguity.

But those are all grammatical ambiguities, which I don't deny.
The point is that the declarative forms "I saw John and Mary" and
"I saw John with Mary" have, in many contexts, precisely the
same meaning. They are interchangeable. The fact that I can
invert one, and not the other, is a fact about the syntactical
roles of "with" and "and", not about their semantic roles.

Bo
