Newsgroups: sci.lang
Path: cantaloupe.srv.cs.cmu.edu!rochester!rutgers!news.iag.net!usenet.eel.ufl.edu!gatech!EU.net!sun4nl!knowar!harmsen.knoware.nl!rharmsen
From: rharmsen@knoware.nl (Ruud Harmsen)
Subject: Parts of speech accurate enough?
Sender: news@knoware.nl (News Account)
Message-ID: <rharmsen.585.000D1A51@knoware.nl>
Date: Wed, 31 May 1995 18:06:04 GMT
Lines: 27
Nntp-Posting-Host: harmsen.knoware.nl
Organization: none
X-Newsreader: Trumpet for Windows [Version 1.0 Rev A]

I wonder if the traditional classification of words by part of speech 
(nouns, verbs etc.) isn't too coarse to explain the syntax rules of 
a language. What if a certain syntactic construction if only possible
with a limited number of words, but not with others, even if it makes sense
semantically?
Dutch has an example of this in "ik zit naar de sterren te kijken" 
(I sit and watch the stars). This construction is possible with
only FIVE verbs (known so far), but not with others. Viz. walk, sit, 
stand, lie and hang can be used, but not ride, run, creep, etc.

If this occurs often in a language, it will be an obstacle for computer
translation, because just a list of words and their part of speech won't 
do, you must also record "can be used with this and that construction".

Examples in English (from "A Practical English Grammar", 
Thomson & Martinet, 4th edition):

The most useful verbs which can be followed directly by the infinitive are:
agree, aim, appear ... (list of some 50 verbs, some are which are specially
marked for an extra ability to be used with a that-clause, and/or a 
"that .. should" construction. (Paragraph 241).

And paragraph 251: 
A number of nouns can be followed directly by the infinitive. Some 
of the most useful are:
ability, ambition, etc. 

