open letter to andrea lunsford 2

[Arnold Zwicky]

        When the first stories about the Toni Morrison sentence
appeared, linguists as a group were baffled.  What was supposed to be
wrong with it?  Being linguists, we began canvassing native speakers
for their judgments, appealing as much as possible to the literate
elite.  Some of us used the Toni Morrison sentence as it stood, but
mostly we stripped it down to its structural essence and eliminated
content, like the presupposition that Toni Morrison exhibits genius,
that readers with a political agenda might seize on (they did); we
constructed examples like my Mary's-father sentences above.

        The results were stunning.  So far I have yet to find, or hear
of, any speaker of English who objects to the Mary's-father sentences,
or their structural equivalents, who cannot cite some explicit "rule"
that the examples violate.  Objectors are very rare, and they are all
people who remember having been taught a proscription against things
like this.  (Sometimes with dismay.  A Stanford colleague recalls
having been taught a version of this "rule" in an ESL class, and
confesses that it haunts her every time she writes a possessive.  What
a senseless waste of the attention of an intelligent mind!)  Normal
people, even very highly educated normal people who write a lot
professionally, just don't see an issue.

        To sum up this part of my letter: really good writers use
things like the Mary's-father sentences all the time, and really
smart, highly literate, people have no problem with them.  As a claim
about writing in formal standard English, the PAP is just false.

        Here is where things start to get weird.  A certain number of
hard-line sticklers for "correctness" would dismiss all of my
discussion (as they do the MWDEU and now the Huddleston & Pullum CGEL)
as utterly irrelevant.  Wrong is wrong, no matter how many people do
it, how often, or who they are.  Just because a lot of people jaywalk,
frequently, even people who are definitely of the best sort, doesn't
make it right.

        The analogy to laws is not accidental.  Many sticklers see
usage proscriptions as rules laid down explicitly by authorities on
acceptable language, in roughly the same way that legislative bodies
and courts act as authorities on acceptable behavior by, respectively,
laying down and interpreting laws.  (There are other sorts of
sticklers.  Some share with the legalistic sticklers the folk belief
that a language exists exterior to its speakers, but locate the
language not in the courts of the authorities but in a sort of
Platonic heaven, in which case we learn about correct language through
revelations to its priests, the usageists.)

        Pursuing the analogy to laws, we should expect that
authorities on usage would provide some justification for their
proscriptions.  Jaywalking is prohibited because it is, in many
circumstances, dangerous.  You have to wait for the red light to turn
green before proceeding across the intersection because going ahead
anyway is also, in many circumstances, dangerous.  (Not every stickler
sees the need for justifications.  There are some who draw the analogy
not to laws but to social customs, as in dress, for instance.  Such
sticklers tend to *embrace* arbitrary, unmotivated requirements - the
arcana of English spelling provide paradigm examples for people like
this - because your willingness to conform to arbitrary strictures
like dress codes shows that you'll make sacrifices to gain membership
in a social group, with all of the rights and responsibilities
appertaining thereto.)

        But (starting not only a sentence, but even better, a
paragraph with "but") to return to proscriptions and their
enforcement: prohibitive laws tend towards context-independence; they
apply even when their reasons for being are absent.  If you walk
across an empty street in the middle of a block in the middle of a
night, you are still jaywalking.  If you drive across a prairie
intersection while the light is red, even if you can see that there
are no cars within miles in any direction, you're still jumping a red
light.  And in either case you can be fined, even arrested; this
actually happens to people.

        The intention here seems to be that people should automatize
their behavior, so that there's no chance of their getting themselves
in trouble, even if that means the occasional pointless
inconvenience.  The principle is: If It's Sometimes Inappropriate,
It's Always Unacceptable.  Or, for short: Just Say No.

        Usage proscriptions can arise in this way.  If some split
infinitives are clunky and awkward, then keep the student from
committing such a gaffe by prohibiting *all* split infinitives.  If
some sentences beginning with "but" or "and" show a want of explicit
connection to the preceding discourse, then the student should avoid
any such sentence.

        Now, I think this is a really rotten way to teach writing.
There's no attention to context, the nature of the audience, the
writer's rhetorical purposes here.  Just abstract "correctness", which
you, Andrea, have repeatedly, in print, disavowed, in favor of
attention to what might be *effective* in writing:
        ... we present grammar and mechanics as tools to use for a
         writing purpose, not simply to use "correctly".  (L&C, p. vii)

(Just one passage from among many I could have quoted.)  You're
explicitly not a stickler; you mostly provide suggestions, not hard
rules; you look for context; and you want to provide justifications in
terms of effectiveness.  Good for you.  I'm on this train with you.

        But somewhere along the line the train occasionally gets
shunted off onto the stickler tracks.  Look at the L&C version of the
PAP.  I suspect you just inherited this from your own teachers, other
handbooks, and the like, but there it is.  It's a blanket
condemnation, of the same sort as a prohibition against "double
negation".  That is, it's framed not as advice about effective
writing, but as a statement about what counts as standard English.  In
line with that, L&C has no mention of context at all.  The implication
is that decontextualized examples of what's bad and what's good are
sufficient, because the bad examples are bad as they stand; no amount
of context could possibly help.

        It turns out that the L&C version of the PAP represents one of
the two main approaches to the possessive antecedent issue in
relatively recent handbooks: most of the recent manuals (and, I
believe, all of the scholarly grammars) say nothing whatsoever about
any problem with possessive antecedents (beyond the general advice
that the reference of pronouns should be clear in context), but a few,
like L&C, issue absolute proscriptions.

        How could this have come about?  It wasn't always so, as a
glance at the MWDEU entry for "pronoun with possessive antecedent"
(p. 779) will make clear.  This entry builds on Theodore Bernstein's
1971 discussion in _Miss Thistlebottom's Hobgoblins_ (NY: Farrar,
Straus and Giroux), which is nuanced and sensible:
        What, then, shall be concluded about this rule?  Just this:
        The rule shall be considered valid whenever it functions to
        preclude ambiguity.  That would make it apply to a sentence
        such as this: "John's roommate said he had a headache."  But
        if there is no possibility of ambiguity and observance of the
        rule would serve only to gratify the strict grammarian's sense
        of fitness, forget it.  (p. 115)

Bernstein looks at three earlier authorities, all of them tending
towards the soft side, though coming at the PAP from three different
directions: no explicit justification ("Pence", i.e. R.W. Pence's _A
Grammar of Present-Day English_ (NY: Macmillan, 1947) [I think
Bernstein was citing a book written by R.W. Pence and D.W. Emery, not
by Pence alone; I'll quote from the 2nd edition of Pence & Emery,
1963]), justification in terms of ambiguity ("Harper's", i.e. John
B. Opdycke's _Harper's English Grammar_, rev. by Stewart Benedict (NY:
Harper & Row, 1966 [my ed. is 1965])), and justification in terms of
antecedent finding ("Follett", i.e. Wilson Follett's _Modern American
Usage_ (NY: Hill and Wang, 1966)).

        I'll get to Pence in a while.  As for Harper's, it points out
that sentences like
        Mary's mother thinks that she is very intelligent.
(I've altered the example in insignificant ways) present a real
problem in understanding: *who* does Mary's mother think is very
intelligent, herself or Mary?  To avoid the problem, avoid personal
pronouns with possessive antecedents.

        This is silly, as Bernstein observed.  It's like banning steak
knives because some people use them ineptly and cut themselves by
accident (not to mention the fact that you could use one as a
weapon).  In the case at hand, context can settle things perfectly:
        Mary's parents have a very high opinion of themselves.
          Mary's father thinks he is absurdly attractive, and
          Mary's mother thinks she is very intelligent.
        Mary's parents have a very high opinion of their daughter.
          Mary's father thinks she is absurdly attractive, and
          Mary's mother thinks she is very intelligent.

These examples are very important, because they let us see just
*what* in the context affects the interpretation of pronouns.  The key
thing is what's salient, foregrounded, or topical in the discourse
context.  (I'm not going to try to distinguish among these different
wordings.)  I would scarcely be the first person in the world to
suggest that the referent of an anaphoric pronoun needs to be high on
the salience/foregrounding/topicality scale.

        Harper's actually alludes to a "rule" for possessive
antecedents, but as if the thing were some sort of grammatical
phantom:
        There is a good grammatical rule to the effect that a pronoun
        cannot take as an antecedent a noun in the possessive case...
        But... this rule is little respected by writers and authors--
        if indeed, known...
I haven't yet checked out the first, pre-Benedict, edition of Opdycke
(1941), but I suspect it has virtually the same language.  Which means
that there is some PAP history from before 1940; I haven't yet found
discussions of it in earlier manuals, though.  MWDEU conjectures that
the origin of the rule "is likely to have been with one of those
18th-century appliers of logic to language" and suggests that it is
"only Latin grammar in disguise", but this seems unlikely, because the
MWDEU staff couldn't find an 18th or 19th century version of the PAP,
because the PAP doesn't hold for Latin, and because *no* manual I've
found appeals to Latin grammar.

        My own conjecture is that the PAP arises as a Just Say No
confection, with at least two contributing ingredients: ambiguities of
the Mary's-mother sort, and the (genuine) awkwardness of a type of
example that Follett talks about.

        Follett is, so far as I know, the first to settle on things
like "in Mary's book" as paradigm examples of PAP violations:
        ?In Mary's book, she writes about geckos.
Soon after Follett, sentences like this are *all over* the usage
handbooks, not infrequently as the sole illustrations of the power of
the PAP.  L&C (p. 216) gives just one example illustrating its version
of the PAP:
        In Welty's story, she characterizes Bowman as a man unaware
          of his own isolation.
which is "corrected" by a shift from (backwards) anaphora to cataphora:
        In her story, Welty characterizes Bowman as a man unaware
          of his own isolation.
The following exercise in L&C (13.1) supplies one further example
(#10) of a PAP violation meant to illustrate unclear pronoun
reference:
        In Tom Jobim's lyrics, he often describes the beaches of Rio.
and the exercise after that (13.2) begins with one more:
        In Paul Fussell's essay "My War," he writes about his
        experience...

        Yes, in-Mary's-book sentences are often pretty rotten, at
least out of context.  The problem is that possessives are, ceteris
paribus, backgrounded (this is a general observation about dependents
vs. heads - a kind of iconicity), and that subordinated constituents,
for instance prepositional phrases serving as adverbials, are also,
ceteris paribus, backgrounded (iconicity again), so that possessives
*inside* adverbial expressions are doubly backgrounded, and so not
great antecedents for pronouns.

        I'll omit the (now boring) demonstration that in-Mary's-book
sentences can in fact work just fine, given the appropriate
contextualization, because the real point here is that even if they
were all just totally wretched, that still wouldn't motivate the PAP,
but only some extremely restricted version of it, one that wouldn't
touch our original sentences about the admirable and intelligent Mary.

        In any case, my current speculation is that writing teachers
and editors noted sentences of the Mary's-mother sort, which are
confusingly ambiguous (out of context), and/or sentences of the
in-Mary's-book sort, in which (out of context) the intended antecedent
has a referent that's not salient enough to serve this role, and they
jumped, in Just-Say-No style, to a blanket condemnation.

        This is a mistake, on two grounds.  First, if you state a
generalization about a language, you have to be prepared to defend its
predictions, and in the case of the PAP most of the negative
predictions are just false.

        This is a point about logical discourse - about science, if
you will.

        And second, advice about writing well should focus on the real
stuff - like what establishes discourse referents, how these are
foregrounded and backgrounded, and the like - not on some rigid
short-cut "rules".  After all, we're giving advice to adults, not
little kids who can't be trusted with the steak knives.

        This is a point about good instructional practice.