PSAT Glitch

Arnold Zwicky zwicky at CSLI.STANFORD.EDU
Tue May 20 18:20:05 UTC 2003


many thanks to P-A-T (P2052 at aol.com) for this compendium of advice
from recent handbooks.  i have a wide range of responses, which
i'll provide in no particular order.

1.  students certainly *are* being warned against possessive
antecedents for pronouns, though exactly (a) which situations are
banned varies from handbook to handbook, (b) the formulation of the
proscription differs from handbook to handbook, (c) the handbooks
differ as to whether they give a rationale for the proscription, and
(d) they differ as to what that rationale is.

about (a): faigley gives the INCORRECT example
  Because Susan had been reading Don Delillo's novel,
    she went to his book signing at the local bookstore.

as several people have commented on here, proponents of the
proscription usually permit *possessive* pronouns with possessive
antecedents; it's just nominative and accusative/objective pronouns
that are barred.  so other handbooks would permit faigley's
example.

a similar problem arises with possessive *pronoun* antecedents.
here the problem is how to apply the technical vocabulary in the
handbooks.  words like "his" and "my" are labeled as possessive
pronouns, but also as adjectives, and in addition it's unclear whether
"he" and "I" are 'implicit in' - i'll get to this later - "his" and
"my".  so maybe things like
  My faith sustains me.    His faith sustains him.
fall under particular proscriptions.  the second of these examples
presents another issue: are sentences to be judged as correct or not
on their own, or in context?  if it's judged on its own, then it
violates at least some of the versions of the proscription.  but
if it's judged in context, then it might be ok; the reasoning would
be that preceding context would supply an antecedent (the same one)
for both "his" and "he", so that "he" doesn't *really* have "his"
as its antecedent.

taken literally, the Stegner/Sauer/Hach formulation - "Avoid using a
pronoun to refer to a word in the possessive case" - would bar both
"Don DeLillo's book increased his fame" and "My faith sustains me".

about (b): some versions (faigley's and hult/huckin's, for instance)
require "specifically named antecedents" or "specific antecedents" and
bar "implied antecedents".  others (DiYanni/Hoy's and Gorrell's, for
example) say that only nouns can serve as antecedents of pronouns and
bar "adjectives and possessives" (and sometimes other things) as
antecedents.  fowler/aaron list what count as implied nouns ("a noun
may be implied in a modifier, a possessive, another noun, or some
other word or phrase") *and* say that such implied nouns cannot serve
as antecedents of a pronoun.

baker merely advises against ambiguity of antecedent (with the
wonderful tiger's eye example), and conlin/herman merely advise
keeping certain pronouns (subordinating ones) close to their
antecedents (giving an example of a relative clause extraposed from a
possessive NP).  these are rules of thumb, designed "for clearer
communication" (conlin/herman), and it would be hard to argue with
them as useful pieces of advice, when applied judiciously.  the other
formulations are, however, general proscriptions.

about (c) and (d): like baker, the opdycke/benedict discussion i cited
before refers explicitly to ambiguity avoidance.  indeed, three of the
bad writing examples contain a potential ambiguity:
  the tiger's eye... it (baker)
  the table's finish... it (reinking/hart/hart)
  Mary's mother... she (stegner/sauer/hach, also opdycke/benedict)

conlin/herman refer explicitly to avoiding difficulties (not
necessarily ambiguities) in antecedent finding.  i think they're on
to something here (see below), but something that's more general than
problems with relative pronouns.

otherwise, the handbooks just stipulate a "rule".

2.  meanwhile, the fact remains that the various proscriptions are
routinely violated in the writing of the elites.  i've started
collecting an example or two each day from the New York Times, which
i skim through every morning.  recent catches:

 5/19/03, from the editorial Judicial Nominees and Gay Rights:
"Mr. Pryor's supporters say he was just doing his job..."

 5/20/03, from bennett ramberg's op-ed piece, Safety or Secrecy:
"I argued against the commission's presumptions - especially the idea
that there would be a timely warning before any attacks.  It did
finally concede that the risk of truck bombs should be taken more
seriously.

 5/20/03, from an obit for david ives:
"Mr. Ives's career with WGBH spanned 40 years, during which he was
responsible for bringing programs like..."

the "the commission's presumptions" example is especially interesting,
because it involves a 'derived nominal', "presumptions".  both derived
nominals and nominal gerunds are interpreted rather differently from
more central examples of nouns with possessive determiners; for these
special nominals, such determiners are "argument possessives",
supplying a syntactic argument (for nominal gerunds, always a subject;
for derived nominals, either a subject or a direct object) for the
verb base of the nominal, as in the following:
  derived nominal:
    Kim's realization that she was in love with Sandy
    Kim's realization that Sandy was in love with her
    the city's destruction only a few years after it was founded
  nominal gerund:
    Kim's realizing that she was in love with Sandy
    Kim's realizing that Sandy was in love with her
    Sue's contradicting herself

within the complements of these special nominals, there's absolutely
no problem with pronouns that refer back to the argument possessive,
as in most of the examples above.  in fact, it's hard to see how to
recast sentences involving such nominals, without major alterations in
their structure.  certainly, just messing with pronouns vs. nouns
won't do, since then you get something that is grossly divergent from
the intended meaning:
    Kim's realizing that she was in love with Sandy
    Kim's realizing that Kim was in love with Sandy
    her realizing that Kim was in love with Sandy

3.  what bothers me about the handbooks, beyond the differences in
their advice and the fact that their proscriptions don't conform to
actual language use, is that it's so hard to figure out how to apply
them.  i put myself in the position of a student who's trying to
following the "rule" in my textbook, and i despair.

part of the problem here is that the handbooks assume an impoverished
theory of syntactic categories and syntactic functions, as i've
discussed in an earlier posting.

another part of the problem is that the handbooks espouse a theory of
anaphora that is, to put it kindly, broken.  as far as i can tell,
they uniformly treat anaphoric pronouns as literal *replacements* of,
or *substitutes* for, nouns (well, really, noun phrases, but i'll
treat the handbooks gently by assuming that they identify noun phrases
with their head nouns); the function of anaphoric pronouns, on this
view, is to avoid repeating expressions.

once linguists got sophisticated about semantics, they saw anaphoric
pronouns in a very different way: these pronouns pick out, not
linguistic expressions, but *discourse referents*, and they're related
to their antecedents by the relation of *coreference*.  once you look
at things this way, many puzzles disappear.  for example, the
inability of pronouns, for the most part, to be coreferential with the
first noun in a noun-noun compound (crews: "The peanut jar was empty,
but Bob was tired of nibbling them anyway") follows from the fact that
these nouns are not referential, but generic.

*and* once you look at things this way, there's absolutely no reason
why pronouns shouldn't be coreferential with possessives, since
possessive NPs are usually referential.

4.  back to the "Mary's mother" example, which two different handbooks
cited.  the idea is that "Mary" can never be the antecedent for an
anaphoric pronoun because the head noun "mother" (well, actually, the
whole NP "Mary's mother", but let that pass) will always be available
as an antecedent as well.  bad, bad ambiguity, don't let that happen;
go for the head noun (which will also be closer to the pronoun).

but, unless you are a manic adherent to IISIIAU, there's no reason
to avoid all possessive antecedents just because some might be
ambiguous, especially out of context.  sometimes there's absolutely
no possibility of ambiguity:
  Mary's mother adores her.
   [if "mother" is the antecedent, then it would have to be:
   Mary's mother adores herself.]

even textbook nasties like "Mary's mother thinks that she is extremely
intelligent" become clear in the appropriate context: "Mary's parents
dote on her. Mary's father thinks that she will be President one day,
and Mary's mother thinks that she is extremely intelligent."  (vs.
"Mary's parents have very high opinions of themselves.  Mary's father
thinks that he will be President one day, and Mary's mother thinks
that she is extremely intelligent.")

what students need to be taught here is, in fact, something they *are*
taught: avoid wordings that are likely to present your reader/hearer
with two different plausible interpretations.  but this is exquisitely
a matter of context, both discourse context and non-linguistic
context.  in certain situations, i should avoid referring to "John
McCarthy", because my audience won't know whether i'm talking about
the linguist or the computer scientist.  if we're sitting in my
living room and can hear a dog barking, i can felicitously say, "It's
been doing that all morning", since you can fix the reference of both
"it" and "do that" from the non-linguistic context.

5.  back to pronouns with antecedents in the linguistic context.  the
task of the reader/hearer is to figure out what those antecedents are
- "competence" testing for antecedents, following follett, or better,
antecedent finding.

larry horn has pointed out here that the information structure of
sentences in discourse is relevant to antecedent finding.  in
particular, he suggested that possessives with certain kinds of
subordinate modifiers might be "too backgrounded to serve as a likely
antecedent for a pronominal".

the follett example (also the wensberg variant of it) illustrates the
point:
  On F's arrival from Virginia at LaGuardia Airport last night,
    he denied to reporters that...

so do a large number of examples from the handbooks that P-A-T cited:
  [faigley] Because Susan had enjoyed reading Don Delillo's novel,
    she went to his book signing at the local bookstore.
  [diyanni/hoy] Throughout Walt Whitman's work, he celebrates...
  [gorrell] In Nabokov's novel Lolita, he tells the story...
  [fowler/aaron] In the speaker's advice she was not concrete enough.
  [reinking/hart/hart] When I removed the table's finish, it proved...

to my ear, these range from the not too bad (the delillo example) to
the really clunky (the speaker's advice example).  the writers could
have been more helpful to a reader trying to work out the antecedents
of the pronouns.

this is subtle stuff, but real.  citing rigid rules, however, isn't
going to make students better writers.

6. most of the proscriptions refer especially to anaphoric pronouns.
but, if the accounts in terms of "implied" nouns or in terms of
ambiguity avoidance are right, they should apply equally to other
anaphoric elements.  deliciously, faigley provides an example, in
his CORRECT version:
  Because Susan had enjoyed reading Don Delillo's novel,
    she went to the author's book signing at the local bookstore.

note that "the author" is just as much an anaphoric element as "he".
why doesn't the proscription (against "implied antecedents") apply
here?

ditto for
  Every/Each girl's present made that girl happy.

and others we could invent.

7.  finally, "implicit".  i've been baffled by this technical term,
and i'd imagine students would be, too.

my problems start with the fact that "implicit" is normally opposed to
"explicit", which made me think that the contrast was between overt,
explicit, stuff and covert, "understood", implicit stuff.  but that
can't be right, because "Mary" is right there in "Mary's mother".

so my guess is that "implicit" is supposed to have its etymological
meaning 'folded in' (though how students should be expected to figure
this out, i can't imagine).  that is, an implicit element is one
that's *inside* somthing else.  if so, then the proscription looks
like some version of an anaphoric island constraint, but i've already
been down that path, and it wasn't a good trip.  so i'm still baffled.

i also have started to wonder about examples like
  My dog barks, even though I tell it not to.
this has a zero anaphor (the missing VP complement of "to"),
for which we'd supply a full VP "bark".  but "bark" is a proper
subpart of "barks".  isn't this just like the possessive anaphor
examples?

arnold (zwicky at csli.stanford.edu)



More information about the Ads-l mailing list