Q: Can you please dwell in some detail on why “can not” is now usually written as “cannot”? Is there a linguistic reason for this uncontracted form? Or is it just one of those irregularities that cannot be accounted for?
A: When the usage showed up in Old English, the language of the Anglo-Saxons, it was two words.
One of the oldest examples in the Oxford English Dictionary is from the epic poem Beowulf, perhaps written as early as the 700s: “men ne cunnon” (“men can not”).
And here’s an expanded version that offers context as well as a sense of the Anglo-Saxon poetry:
“ac se æglæca etende wæs, / deorc deaþscua duguþe ond geogoþe, / seomade ond syrede; sinnihte heold / mistige moras; men ne cunnon / hwyder helrunan hwyrftum scriþað” (“all were in peril; warriors young and old were hunted down by that dark shadow of death that lurked night after night on the misty moors; men on their watches can not know where these fiends from hell will walk”).
The combined form “cannot” showed up in the Middle English period (1150 to 1450), along with various other spellings: cannat, cannatte, cannouȝt, connat, connott, conot, conott, cannote, connot, and cannott.
The earliest OED example with the modern spelling is from Cursor Mundi, an anonymous Middle English poem that the dictionary dates at sometime before 1325: “And þou þat he deed fore cannot sorus be” (“And thou that he [Jesus] died for cannot be sorrowful”).
In contemporary English, both “cannot” and “can not” are acceptable, though they’re generally used in different ways. The combined form, as you point out, is more common (Lexico, formerly Oxford Dictionaries Online, says it’s three times as common in the Oxford English Corpus).
Here’s an excerpt from the new, fourth edition of Woe Is I, Pat’s grammar and usage book, on how the two terms, as well as the contraction “can’t,” are generally used today:
CAN NOT / CANNOT / CAN’T. Usually, you can’t go wrong with a one-word version—can’t in speech or casual writing, cannot in formal writing. The two-word version, can not, is for when you want to be emphatic (Maybe you can hit high C, but I certainly can not), or when not is part of another expression, like “not only . . . but also” (I can not only hit high C, but also break a glass while doing it). Then there’s can’t not, as in The diva’s husband can’t not go to the opera.
Getting back to your question, why is “cannot” more popular than “can not”? We believe the compound is more common because the two-word phrase may be ambiguous.
Consider this sentence: “You can not go to the party.” It could mean either “You’re unable to go” or “You don’t have to go.” However, the sentence has only the first meaning if you replace “can not” with “cannot” (or the contraction “can’t”).
In The Cambridge Grammar of the English Language (2002), Rodney Huddleston and Geoffrey K. Pullum say that “You can’t/cannot answer their letters” means “It is not possible or permitted for you to answer their letters,” while “You can not answer their letters” means “You are permitted not to answer their letters.”
In speech, Huddleston and Pullum write, any ambiguity is cleared up by emphasis and rhythm: “In this use, the not will characteristically be stressed and prosodically associated with answer rather than with can by means of a very slight break separating it from the unstressed can.” The authors add that “this construction is fairly rare, and sounds somewhat contrived.”