Arnold M. Zwicky
zwicky at CSLI.STANFORD.EDU
Mon Nov 7 21:33:19 UTC 2005
On Nov 4, 2005, at 8:19 PM, Wilson Gray wrote:
> "If you live _ or drive through Cranston, Rhode Island ..."
(what's the actual source, wilson?)
yes, still another kind. i have files on the following types:
1. equative/comparative -- "as good or better than"
2. superlative -- "one of the best, if not the best, student"
3. verb forms with modals -- "will and may already have finished"
4. infinitive/participle verb forms with adjectives -- "neither
interested in nor willing to"
5. verbs/adjectives/nouns taking prepositions -- "wonder but never
dare to ask"
these are standardly seen as simple "failures of parallelism", but
that way of talking about them assumes an analysis of these
coordinations that seems natural and necessary to many people, but is
in fact not the only way to think about them. (i plan to write this
up for Language Log, but i'm now about 60 postings behind there, so
maybe it won't hurt to say a bit here.)
the usual view of coordination of non-clauses is that it's some kind
of reduction of clausal coordination. on this view the source of
wilson's example would have to be something like
(If) you live Cranston or you drive through Cranston
but the first clause here is ungrammatical, so the reduced
coordination is, too. the grammatical version of the first clause
you live in Cranston
which would be a source for the (parallel) reduced coordination
(if) you live in or drive through Cranston.
here's an alternative view, prompted by a view of troublesome
disjunctive subjects, like "you or he". i've posted about agreement
with disjunctive subjects here and on LLog. the short story is that
each disjunct calls for a person/number form for the verb -- "you"
calls for "are", "he" for "is" -- and that different people respond
to a resulting conflict in different ways (some allow no resolution,
some go for the nearest determinant, some for 2nd person over the
others, some accept more than one version). the key idea here is
that in (at least some kinds of) coordination, each coordinated
element places a requirement on some third, shared, element, and that
conflicts can arise between these requirements.
that's how i'm now looking at the five sets of phenomena above. in
#3, for example, "will" calls for a base-form complement, while
perefect "have" calls for a past-participle complement; there's a
conflict, which for many people is resolved in favor of the nearer
determinant. (for others, no resolution is possible, and the example
in wilson's example, "live" calls for a complement PP with "in" (or a
few alternatives), while "drive" calls for a complement PP with
"through" or some other Ps; tthe two sets of Ps do not overlap, so
there's a conflict, which can be resolved in favor of the nearer
determinant, or not resolved.
on this way of looking at things, it would be possible for different
types of coordination to differ as to how conflicts are resolved.
that is, resolution could be construction-specific (as well as
differing from person to person), and i believe that it is. in
particular, resolution in favor of the nearer seems to be hugely more
frequent in type #3 than in type #5.
More information about the Ads-l