Take a string of letters with apostrophes, substitute the regular expression [a-z]+ for each apostrophe. If there is a unique English word that matches that pattern, then let the contracted form with apostrophes be a valid contraction of that word. Currently existing contractions are grandfathered in, despite violating this rule. Maybe we need an alternate character that is not an apostrophe.
(Tangentially, there seem to be very few current cases where leaving out an apostrophe leaves a valid sentence or fragment with a different meaning. Maybe apostrophe-s versus s-apostrophe versus s-alone for possessive singular versus possessive plural versus adjective (maybe name) ending with an s. It is ironic that apostrophes are what people are sticklers about; they are almost purely decorative -- vestigial.)
(Another tangent: there is ambiguity between apostrophe-s meaning "is" or possession: Bob's back.)
Perhaps forbid apostrophes in the first or last position.
What contractions will exist? List them in order of word frequency (and probably stop before the words become too obscure: trying to figure out an obscure word with missing letters is too hard.)
Finding these contractions seems computationally difficult.
No comments :
Post a Comment