"cannot" vs. "can not" (from Zuckerberg to Zimmer)

[Laurence Horn]

Ben's excellent "Gray Matter" piece in today's Times "Sunday Review" (yes, that's what the old Week in Review and older News of the Week in Review has been rechristened), the "former On Language columnist for the New York Times Magazine" (and whose fault is *that*, o Gray Lady?) explores the 'multibillion dollar lawsuit that might hinge on whether [Mark Zuckerberg] capitalized the word "Internet" or whether he spelled "cannot" as one word or two'.

See http://www.nytimes.com/2011/07/24/opinion/sunday/24gray.html for the full article.

I don't have anything to add to Ben's account, which delves into the research protocols of certain forensic linguists with whom some of us may be acquainted, except to note that contrary to the standard if not ubiquitous practice of lexicographers to gloss CANNOT simply as 'can not' or, in the language of the OED, 'the ordinary modern way of writing can not', this is not really an adequate definition.  In fact, "cannot" can only be used (more or less interchangeably) for 'can not' if the latter is understood as 'not possible/permitted/able (to)', never if it means 'possible/permitted/able not (to)'.  In other words, "cannot" is exactly like "can't" semantically, and not (always) like "can not", as seen in paradigms like the following

(1)  A priest can not marry.    (true on either Catholic or Episcopalian reading)
(2)  A priest cannot marry.     (true only on Catholic reading)
(3)  A priest can't marry.        (true only on Catholic reading)

What I'm calling the Episcopalian reading can be brought out more clearly by intervening material:  "A priest can, if he wishes, not marry" or "A priest can always not marry".  
(Some history: Zwicky and Pullum's superb 1983 paper in Language on negative auxiliaries points out that "can't" isn't simply a negative cliticized onto "can", and I discuss the pattern in (1)-(3) in my 1972 dissertation.)

So if Zuckerberg's e-mail horde included something like (4),

(4) You can not invite the Winklevoss twins to my next surprise party.  :-(

it would be useful for the forensic linguistics team to know whether the context favored the "can't" [= 'not possible'] interpretation or the "can always not" [= 'possible not'] meaning.  In the latter case, no spelling-as-one-word option would have been possible, and this would affect the relevant authorship determination statistics.

LH

------------------------------------------------------------
The American Dialect Society - http://www.americandialect.org