which rule predicts that asterisk?

[Arnold M. Zwicky]

more on the Possessive Antecedent Proscription, this time in response
to recent mail from a non-linguist colleague (who would like to point
out that the message to me "was prompted by a less-than-careful
reading" of huddleston & pullum, The Cambridge Grammar of the English
Language):

> From: Arnold M. Zwicky <zwicky at csli.stanford.edu>
> Date: Mon Oct 6, 2003  8:34:10 AM US/Pacific
> To: ...
>
> On Thursday, October 2, 2003, at 03:54 PM, you wrote:
>
>> I write responses to questions of English grammar and usage on a
>> website... Here's  a portion of what I wrote last spring in answer to
>> a query about the controversial
>> PSAT sentence.
>>
>> ...The Cambridge Grammar of the English Language does state that the
>> subjective form of the pronoun may not be used to refer to a
>> possessive
>> noun:
>
> no.  they do not state this.  they state a restriction that rules out
> instances of (retrospectively) anaphoric pronouns, in a very specific
> structural configuration.  it happens that the antecedent in this
> configuration can be either possessive or not, so that some examples
> of pronouns with possessive antecedents happen to be ruled out.
> however, the restriction cited in CGEL -- it's not original with them
> -- wouldn't rule out *any* of the examples [i've cited recently],
> including the toni morrison example from the PSAT.
>
>> ?Without the support of Ann's mother, she would not have survived (not
>> correct)
>
> CGEL gives this one, which i'll label (i), with an asterisk, not a
> question mark.
>
>> The same source, however, presents the following utterance as a
>> (correct)
>> example of reference using the objective form of the pronoun:
>>
>> Without the support of Ann's mother, I wouldn't have been able to
>> persuade
>> her to seek medical help (Section 2.4.1, p. 1478)...
>
> here's the offending configuration:
>
> 1.  retrospective (not anticipatory) anaphora, i.e., the antecedent
> precedes the pronoun; the effect is not found for anticipatory
> anaphora, as CGEL shows.
>
> 2.  the pronoun is the subject of the sentence; the effect is not
> found for pronouns in any other function, as CGEL notes.
>
> 3.  the antecedent is inside a sentence-initial PP (adverbial)
> modifier.  other types of sentence-initial modifiers are fine: cf. (i)
> with
>   (ii) If Ann's mother hadn't given her support, she wouldn't have
> survived.
> (CGEL doesn't say this, but it's implicit in their reference to PP
> specifically.)
>
> 4.  this modifier must be *preposed*, not merely sentence-initial.
> that is, there must be some sense in which the modifier "belongs" in
> the VP.  ordinary sentence-modifying PPs are fine (so long as there's
> no problem with foregrounding/topicality/etc.); CGEL gives
>   (iii) In view of Paul's special circumstances, he was given extra
> time.
> to which i can add things like
>   (iv) According to Paul's view of the universe, he deserved extra
> time.
>   (v)  In Toni Morrison's latest book, she attacks her critics.
>
> CGEL carefully uses the word "preposed", but without laying out why
> "sentence-initial" would not have sufficed.  this is a subtlety that
> most readers probably won't appreciate.  if you don't catch that, then
> you'd expect (v) to be bad, and (v) is a type of example that *does*
> figure in many handbook discussions of the PAP.  but even if you miss
> the subtlety, the CGEL restriction doesn't rule out
>   (vi) Toni Morrison's genius enables her...
>
> note that possessives aren't specifically mentioned in any of this.
> the antecedent in question can be either possessive or not; CGEL gives
> two examples of each.
>
> but... wait!  a reasonable person might look at the four clauses in
> the proscription above and throw up the hands in horror.  why on earth
> should *just this* assortment of conditions produce ungrammaticality?
>
> there is a very nice answer to this, due (if i remember correctly) to
> george lakoff, from a long time ago:  sentences with truly preposed
> initial PP modifiers show the same conditions on anaphor-antecedent
> linkages as sentences with the PP modifiers where they "belong"; (i)
> is bad because (vii) -- a classic violation of the precede-command
> constraint on such linkages (in my opinion, one of the very few
> genuinely structural conditions on them..) -- is bad:
>   (vii) She would not have survived without the support of Ann's
> mother.
> (out on the reading in which "she" refers to Ann).
>
> so... there are some wonderfully fascinating phenomena here, but they
> have nothing in particular to do with the PAP.
>
> arnold
>