The Big-Bag-O'-Words View of Language, and Some Related Errors
Geoffrey K. Pullum

The prevailing view of what a natural language is, tacitly maintained
not just by the proverbial layperson but also by most writers on
language and many academic researchers in fields from literature to
psychology, boils down to the claim that a natural language is a big
bag of words.  Natural languages are, in effect, equated with their
dictionaries.  The evidence of this is everywhere, in journalism,
popular writing, and much academic discussion of language. My review
of this evidence will be mercifully brief, even cursory.  But I will
argue that it is worth thinking a little about the big-bag-o'-words
view carefully, first because it is instructive to attempt to
reformulate the reasons why it is (I believe) utterly mistaken, and
second because there are recent movements within theoretical syntax
to reinstate it in disguise, and if I am right those efforts are at
least to some degree misguided.  

Related reading:
This talk is really a commentary and elaboration on an essay about
language by Barbara Scholz and myself that has just appeared in the
`Concepts' series in Nature (27 September 2001).