The Spelling of Cannot

[Laurence Horn]

At 4:28 PM +0000 1/29/01, Lynne Murphy wrote:
>--On Monday, January 29, 2001 10:20 am -0600 Natalie Maynor
><maynor at CS.MSSTATE.EDU> wrote:
>
>>A student recently asked one of my colleagues why "cannot" is
>>usually spelled as one word while the other negatives aren't.
>>Is there an answer?
>>    --Natalie Maynor (maynor at ra.msstate.edu)
>
>
>
>My hypothesis for why it's one word:  it means something different if it's
>two words.
>
>I cannot sing.  = I am not able to sing.
>I can not sing (if you like).  = I am able to not sing.
>
>No one seemed to do spell it as one word in South Africa, and a lot of
>people don't in the UK either.  NODE says that both forms are acceptable,
>but 'cannot' is more common.  Whereas in US English, I'd say that a
>sentence like "I can not fit into this dress anymore" is wrong.
>
>Lynne
>
>
My 1972 UCLA dissertation is, as far as I know, the first place in
which this observation is made, or at least "explained", but only
insofar as the impossibility of "cannot" (and of course "can't")
allowing wide scope for the negative.  For me, "can not" allows both
scopes in principle (and I think you'll find that a text count will
turn up more occurrences with the same NOT CAN meaning that "cannot"
or "can't" must have), but in practice writers may avoid the two-word
spelling for this meaning since the one-word spelling is available.
This is an instance of the general Elsewhere Principle and in this
case not something fixed by the lexical entry itself.  (Note that the
standard dictionary entry for "cannot" simply glosses it as 'can
not', which is wrong for the reason Lynne cites.)

What I was trying to argue for in my thesis was a general tendency to
avoid lexicalization of modal-negation complexes with the semantics
of [possible/permitted [not]] or [not [necessary/obligatory]] as
opposed to the more readily lexicalized [not [possible/permitted]]
and [obligatory [not]], an asymmetry that partakes in a much more
general conspiracy reflected in e.g. the fact that no natural
language lexicalizes "nall" [ = 'not all'] while many lexicalize 'not
some' or 'all not' (Eng. "no", "none").  I propose a
Gricean/pragmatic explanation for this asymmetry that I'll spare you
here.  Within this general framework, I characterized "cannot" as an
orthographic lexicalization that obeys the same semantic constraints
as the morphologically lexicalized "can't".  But to respond to
Natalie's question, there's no simple answer to her students's query,
since exactly the same asymmetry is found with "couldn't" as with
"can't"--the former too is possible only when the negation takes
scope over the modal, with the meaning 'not possible/permitted/able',
never with the meaning [could [not]]--yet no "orthographic
lexicalization" of the form *couldnot occurs.  That is, even though
we can predict that "couldnot" could only mean what "couldn't"
means--[not [could]]--we cannot predict that it does not, in fact,
occur.  I suspect the difference between "cannot" and *"couldnot" is
a matter of frequency and phonology, but I won't try to speculate
how, or why--as she notes--"cannot" is the only orthographic-only
contraction we have.

larry