use(d) to

Arnold Zwicky zwicky at CSLI.STANFORD.EDU
Wed Aug 13 18:34:18 UTC 2003


first, to clarify things a bit, there are two very different items
here.  one is a predicate adjective, and the "to" is a preposition
("I'm used to miso soup but not to sushi").  the other, the one we've
been talking about, is a verb, and the "to" is the infinitive marker
("I used to eat sushi").

this verb "use(d)" has a very strange syntax.  with one exception, it
behaves just like an extremely defective main verb, a verb that has
only one, finite, form, the past.  (in particular, it lacks a base
form, and so can't occur with infinitival "to" - "*You ought to used
to eat sushi" 'It ought to be the case that you used to eat sushi' -
or modals - "*You must used to eat sushi" 'It must be the case that
you used to eat sushi'.)  the exception is that in negated clauses, in
inversion, and under emphasis, this verb occurs with one, and only
one, auxiliary, supportive "do" (in its past form "did"): "You didn't
use(d) to eat sushi", "Did you use(d) to eat sushi?", "You DID use(d)
to eat sushi"; this is extraordinary, in that supportive "do" is
universally said to require the base form of its complement, and we've
just seen that "use(d)" doesn't (otherwise) *have* a base form.

those are (most of) the facts, and they are described clearly in some
sources, in particular the Cambridge Grammar of the English Language
(p. 115).  (CGEL also notes that some speakers have negated
"use(d)n't" and inverted "used" - that is, that they treat "use(d)"
like an auxiliary, in these two respects.)

larry horn suggested here that the problem with the supportive "do"
examples *in writing* (they're fine in speech) is that neither
spelling - "use to" or "used to" - seems right.  i'm perfectly happy
with the "used" spelling, though.  CGEL has the "use" spelling, but
notes that "used" is sometimes found in these contexts.

so, i'm cool on the spelling.  but making sense of the odd
distribution of "use(d)" is quite a nice little puzzle.

arnold (zwicky at csli.stanford.edu)



More information about the Ads-l mailing list