On 9/13/07, <b class="gmail_sendername">Yorick Wilks</b> <<a href="mailto:Yorick@dcs.shef.ac.uk" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">Yorick@dcs.shef.ac.uk</a>> wrote:<div><span class="gmail_quote">
</span><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<div>...there's lots of stuff out there on this way of doing things (as John Sowa keeps saying, if I read him right) and were not really adding much here by just chucking more examples at each other.</div></blockquote>
<div><br>By "this way of doing things" do you mean relating meaning and grammar?<br><br>I agree there is a lot of work which tries to relate meaning and grammar. However, I don't think there is a lot of work which takes my position that we can model semantics by making ad-hoc syntactic generalizations over corpora.
<br><br>I've seen nothing which suggests the analysis of syntax must be ad-hoc, because grammatical generalizations over text are necessarily incomplete, for instance. While this is essential to my treatment, because you can't get the detail of semantics without it.
<br><br>In particular John is suggesting somewhat the reverse of my position, by insisting that we need a(n unknown) semantic representation, separate from syntax, to make up for the shortcomings of syntax.<br></div><br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<div><div>Rob--I dont get your last bit:</div><span><blockquote type="cite">Failure to treat the syntax of "supported" on an ad-hoc basis in this way means you have no way of capturing the information that, in a grammar of English, "supported" = "accompanied" but also "supported" != "accompanied". That means you will be unable to capture detailed syntactic restrictions which prevent you from saying "slightly odd" things like "Tom accompanied his tomato plant to the garden (where he planted it.)"
</blockquote></span>Surely, thats not the contrast you want--it's rather that "I accompanied the man with a stick" has exactly the syntactic pattern of the first (support-tomato) example l, but that means nothing at all --their joint class membership there leads to NOTHING---because of the quite different "semantic grammars" of the concepts.
</div></blockquote><div><br>No that's the contrast I want. I'm showing "supported" and "accompanied" have the same syntax in some cases, but not others. John would argue to make that distinction you need to go to semantics. I'm showing it can be done purely on (ad-hoc) syntactic grounds.
<br><br>Your contrast demonstrates something else. It demonstrates that the two words "supported" and "accompanied" can be seen as selecting (ad-hoc) a different class for the same _context_.<br><br>It is really my point again (that we can handle these "semantic" distinctions using ad-hoc syntax.) It is just you are looking at things the other way around.
<br><br>What happens is the choice of word "supported" or "accompanied" selects a class of contexts which have the same syntactic properties as "the man with a stick" (a different class in each case), and it is these which give you the impression that "the man with a stick" has one or other meaning. For instance, if the word used selects a set of contexts which includes the context "tomato plant" we will see one meaning ("supported" will do this), but if it selects a class which does not include "tomato plant", we will see another ("accompanied" will do this.)
<br><br>Note: you need an ad-hoc treatment of syntax for this to work. Otherwise the classes ("the man with a stick" = "tomato plant" or "the man with a stick" != "tomato plant") will be conflated, and "the man with a stick" will always be the same.
<br><br>-Rob<br></div></div>