[Lexicog] Nouns

Mike Maxwell maxwell at LDC.UPENN.EDU
Thu May 25 03:57:53 UTC 2006


Ron Moe wrote:
> Some theories of syntax confuse the situation by calling 'grits' a NP.
> Sorry, but it is not a phrase. A phrase by definition consists of more than
> one word...
 > Just because Chomsky stuck the label 'NP'
 > on a particular node in a syntactic tree doesn't suddenly
 > make 'grits' a noun phrase.

Well, I could argue that, but I won't...

> Chomsky should have searched for a better label.

He was just using a different definition, one that didn't rely on 
counting the number of words, but rather relied on the function that 
word(s) could have in higher level (clausal) syntax.  And I'm not sure 
he was the first.

As for the better label, he later invented the terms N-bar and 
N-double-bar (often written N' and N'', for typographic reasons--this 
was in the days before linguists had direct access to fonts that could 
do fancy things; the bars of N-bar and N-double-bar were supposed to be 
written on top of the N.)

 > And don't call 'hot dog' a 'noun phrase', 'compound', or
 > 'noun'. Like 'of course', it is phonologically a compound,
 > but written as two words...
 > Phonologically and syntactically they are all compounds...
 > If you call the ones written as phrases 'compounds', your parser
 > will choke. Your parser knows (even if you do not) that
 > a space separates words and that there is no such thing as
 > a 'compound phrase'.

Actually, I have written syntactic parsers that knew very well that 
there were compound phrases.  The rule was
    N --> N   N
As you say, "Phonologically and syntactically they are all compounds", 
and that's exactly correct, regardless of whether some parser can or 
cannot parse them as such.  Space characters matter to parsers, but they 
don't matter to spoken language--and for that matter, people tend to be 
inconsistent about writing such spaces, particularly in borderline cases.

 > In the meantime I would recommend that you leave the
 > part of speech field for lexical phrases
 > empty or simply assign them to 'Phrase' (abbreviation: Phr.).

That's reasonable for many cases, including idioms.  It is not good 
advice for compounds, I claim, precisely because they are not 
necessarily phrases, but they do have a specific behavior--their 
distribution is that of nouns.  There are also other sorts of multi-word 
constructions that are not phrases, such as particle verbs in English: 
"bring...to" (as in "The referee brought the boxer to") is not a phrase, 
but it should be listed in a dictionary.  In these cases, as you say, 
one might argue for leaving the POS field blank, although one might also 
invent a POS (something like "transitive particle verb" for 
"bring...to", for example).

:-)
    Mike McSwell


------------------------ Yahoo! Groups Sponsor --------------------~--> 
Get to your groups with one click. Know instantly when new email arrives
http://us.click.yahoo.com/.7bhrC/MGxNAA/yQLSAA/HKE4lB/TM
--------------------------------------------------------------------~-> 

 
Yahoo! Groups Links

<*> To visit your group on the web, go to:
    http://groups.yahoo.com/group/lexicographylist/

<*> To unsubscribe from this group, send an email to:
    lexicographylist-unsubscribe at yahoogroups.com

<*> Your use of Yahoo! Groups is subject to:
    http://docs.yahoo.com/info/terms/
 



More information about the Lexicography mailing list