Newsgroups: comp.ai.philosophy
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!howland.reston.ans.net!gatech!newsfeed.pitt.edu!godot.cc.duq.edu!news.duke.edu!news.mathworks.com!udel!ssnet.com!stingray.ssnet.com!pagesat.net!internet.spss.com!markrose
From: markrose@spss.com (Mark Rosenfelder)
Subject: Re: What's innate? (Was Re: Artificial Neural Networks and Cognition
Message-ID: <D3ypu9.KKs@spss.com>
Sender: news@spss.com
Organization: SPSS Inc
References: <3gtu3i$rf3@mp.cs.niu.edu> <3hf0fa$7p@agate.berkeley.edu> <D3t179.633@spss.com> <jqbD3vrp1.K4E@netcom.com>
Date: Mon, 13 Feb 1995 23:42:08 GMT
Lines: 89

In article <jqbD3vrp1.K4E@netcom.com>, Jim Balter <jqb@netcom.com> wrote:
>In article <D3t179.633@spss.com>, Mark Rosenfelder <markrose@spss.com> wrote:
>>No, it just uses "semantics" as most linguists do.  Topic (what you're
>>talking about) and comment (what you're saying about it) is a matter of 
>>pragmatics.
>
>Whatever they are, they aren't grammar.  When people here are talking
>about other explanations than UG, they are talking about other explanations
>than grammar; if they say "semantics" when they "should" say "pragmatics",
>it doesn't change the argument.

I wouldn't be so sure that UG doesn't include any semantics.  There's a 
whole semantic side to Chomsky's theories-- stuff about Case and theta-roles.
I don't know if it's part of UG or not-- it probably is; it's described
in Cook's book (_Chomsky's Universal Grammar_).

>>The cited paragraph was part of a response to the suggestion that something
>>about the semantics of "You saw John and X", where X is an unknown person
>>whose name we wanted, prevents our asking by saying "Who did you see John
>>and?"  So far as I can see, this claim is refuted by the fact that we
>>*can* ask "You saw John and *who*?"  
>
>I think you have misconstrued the argument.  The claim was made that
>"John and Mary" was semantically equivalent to "John with Mary".
>Various arguments were given that that isn't so.  The relevance of the
>semantics of "John and X" is that, since it differs from "John with X",
>you cannot make the argument that, because "John with" is ok, "John and"
>should be ok too.

I didn't misconstrue the argument; I simply intervened when it started to
go wrong.  As it happens, I don't agree that "John and X" and "John with X"
are equivalent; even if the semantics are tolerably similar, the syntax
is quite different-- conjuncts just aren't the same as prepositional phrases
and can't be expected to behave the same.

>>As you say, "the first example fails to ask anything".  The question is,
>>why does it fail to do so?  Just asserting
>>
>>>The first one makes no sense, which is a fact about its semantics.
>>
>>is not an explanation; it's a restatement of the problem: people don't
>>say that, and have trouble understanding it.  But *why* do people
>>have trouble making sense of it?
>
>First, if we do fail to explain why it doesn't make sense, how is that support
>for UG?  

You're losing track of who's claiming what.  The claim that such examples
support UG is, I think, Bo Curry's, echoing Pinker.  All I've done in this
thread is contest claims that the observed syntactic restraints are due
to something about the semantics of "and".  I don't reject this a priori,
but the arguments I've seen have been weak.

>Second, I think there is a ready explanation in the fact that "I saw
>John and Mary", "I saw John and I left", "I saw John and touched Mary", and "I
>saw John and Mary didn't" are all valid, but only "I saw John with Mary" is.
>With "and" playing so many different possible roles, it becomes hard work to
>determine its role when the cues are removed.  

This is a promising approach, I think, though not yet rigorous enough for me.
How much "hard work" are we talking about?  More than we already do for
disambiguating many-sensed words like "run", or the semantic minefields
around quantifiers?  More than we already do with "and" in the sort of 
sentences you quoted?

BTW, I'd call this a *syntactic* argument, despite the earlier promise
of a semantic (or pragmatic) explanation.  Likewise with Marvin Minsky's
suggestion, based on the difficulty of choosing between different parsing
trees.  

>Third, despite comments by
>others here, I think that, given proper stress, following a statement like "I
>saw John and <mumble>", "*Who* did you you see John and?" does make sense and
>would be readily parsed.  

It doesn't quite work for me-- it still sounds strange.  But you're on the
right track here, trying to think up a context for a disputed utterance.
Many times people (including linguists) assert much too readily that a 
sentence is ungrammatical.  An example from the '70s: it was posited that 
"Spiro conjectured Ex-lax" was ungrammatical (because Ex-lax is not something 
that can be conjectured).  Finding an acceptable context is left as an 
exercise for the reader.  :)

>In fact, I speak that way occasionally with my geeky
>formalistic friends.  Does UG posit that the language we speak isn't a
>"natural" language?

Sure, why not?  You get UG built-in, but you can still be infected by
geeky formalistic memes.
