"confuses X for Y"

[Arnold M. Zwicky]

On Aug 26, 2007, at 11:46 AM, i wrote, about "confuse X for Y":

> ... i took it to be an extension of the pattern
>   mistake X1 for X2
> to the verb "confuse", which overlaps semantically with "mistake".
> that would put this case in the same ballpark as the first stage of
> the "substitute" story, in which "substitute" appears in the pattern
>   replace OLD by/with NEW
> in the place of the semantically related "replace".
>
> in the story that david denison tells about the second stage, there
> is blending, of the two patterns
>   substitute NEW for OLD [original "substitute"]
> and
>   substitute OLD by/with NEW [encroached "substitute"]
> to give
>   substitute OLD for NEW [reversed "substitute"].
> if this story is right, the blend must have spread pretty fast,
> since reversed "substitute" is fairly common in the u.k. and now
> spreading in the u.s.]
>
> if the spreading-blend story is plausible here, then it can't be
> immediately excluded in the "confuse" case, which could in
> principle have started with a blend of
>   confuse X1 with/and X2
> and
>   mistake X1 for X2
> that spread rapidly.
>
> but i need to think more about this case.

i've now thought some more about it, and i now think that reversed
"substitute" isn't a blend, that is, it's not a type-2 combo, and
that "confuse for" isn't either.  the first is certainly a combo, and
the second might be, and both could have on occasion have occurred as
production errors, but a much more likely story is just that of
pattern extension (in which speakers work creatively, though
unconsciously, with the materials of their language).  pattern
extensions happen all the time, along natural paths, and they can be
devised by many different people on many different occasions.  theyll
be likely to spread because they seem intuitively reasonable to other
speakers.

that might be a sufficient story for "confuse for" (which sounds
better and better to me with every passing hour) and for the first
step in the history of "substitute" (the development of encroached
"substitute").  encroached "substitute" has the advantage of putting
the arguments in an order that mirrors the temporal sequence of
entities (old before new), so it's no surprise that it's spread so
successfully.

the second step (the development of reversed "substitute") does, i
think, involve a combo, but not of type 2.  let me discuss a case
that i think is analogous.

for some period, my daughter elizabeth produced past participles like
"wroten" (instead of "written").  when you hear just this much,
you're likely to say: oh, a blend, a slip in which the competing
forms "written" and "wrote" are combined in the heat of speech
production.  now, such inadvertent errors might occasionally happen
(for people of any age), but that wasn't what was going on for
elizabeth.

first, the PAST-N participles occurred for virtually all possible
verbs: if, in the adult standard language, a verb has a /(@)n/
participle with a stem distinct from the stem of the past, then
during this period elizabeth produced a PAST-N participle for that
verb.  for every such verb, and, as far as we could tell, every time
a past participle was called for -- in both perfects and passives.
(the only exceptions, as i recall, were the past participles of the
"grammatical" verbs BE ("been") and DO ("done"), which conformed to
the adult standard.)

second, she gave no evidence whatsoever that these forms were slips.
during this period, she never corrected herself or gave any sign of
unease about her productions, absolutely resisted the modeling of the
"correct" forms by her parents and others, and reliably produced the
PAST-N versions on "wug"-type tasks.  that is, she was producing
exactly what she intended to produce.

the account i gave in my 1970 article on these forms ("A double
regularity in the acquisition of English verb morphology", Papers in
Linguistics 3.411-8) was that elizabeth had (correctly) learned two
things about english:

1.  certain verbs (call them N-verbs) have a past participle in /(@)n/.

2.  with a few exceptions, the past and past participle forms of a
verb are identical.

she then did her best to make both of these pieces of knowledge as
general as possible, by building the past participle of N-verbs on
their past tense forms: "wroten" and the like.  the PAST-N forms are
combos of some sort -- the suffix of "wroten" coming from fact 1, the
stem from fact 2 -- but not blend slips.

i want to say something similar about reversed "substitute" -- that
those who first produced it got the assignment of OLD and NEW to the
two arguments of "substitute" (the direct object and a P-marked
oblique object) from temporal iconicity (OLD before NEW), as in
encroached "substitute", and the choice of the P "for" from knowledge
about P selection for original "substitute".  (this is a paraphrase
of denison's analysis.)  the result is a construction with features
from two different sources, but it doesn't posit a production slip as
the source of the combination.  instead, speakers are creatively,
though unconsciously, manipulating the materials of their language.

reversed "substitute" then spreads, because it makes sense to other
speakers.

arnold

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