implied/implicit/*imagined* (was Re: PSAT Glitch)

[Arnold Zwicky]

a small addition to the vocabulary found in usage handbooks,
from
  Theodore M. Bernstein, The Careful Writer: A Modern Guide
    to English Usage.  NY: The Free Press, 1965.  [I'm quoting
    from the 1998 paperback edition.]

p. 349:
  IMAGINED, BUT NOT ACTUAL, REFERENCE WORD.  "The broad-spectrum
  antibiotics have so simplified pneumonia therapy that it is
  now frequently treated at home."  When the writer set down _it_
  he had "pneumonia" on his mind, but "pneumonia" is serving here
  as an adjective; the only noun is "therapy," which, of course,
  is not what the _it_ refers to.

[this is the entire subsection on imagined reference words.  it
comes close to possessive antecedent territory, but doesn't actually
enter.]

so the writer merely *imagines* that he [the pronoun is bernstein's]
has referred to pneumonia by using the word "pneumonia".  it's that
old "serving as an adjective thing".

it's hard indeed to defend the idea that "pneumonia" here is an
Adjective in category, or even that it's an Adjectival in function.
"pneumonia therapy" is a garden-variety N+N compound, and "pneumonia"
shows no syntactic or morphological properties of an Adjective.  in
fact, it shows no properties of an Adjectival, either.  to see this,
note that there are at least a few english words apparently belonging
to the category N that can serve in the Adjectival function - for
instance, "monster", as in "a monster rally" 'a huge rally'.  these
words have secondary accent (the main accent is on the head N), and
they can be conjoined with other Adjectivals: "monster RALLY"; "a
monster, endless rally" 'a rally that's both endless and huge'.
contrast this with the N+N compound "monster rally" 'rally of
monsters', which has primary accent on the first element ("MONSTER
rally") and doesn't allow clear Adjectivals to conjoin with this first
element: "monster, endless rally" can't refer to a rally that's both
endless and involves monsters.

bernstein's example does have a point of interest.  it certainly has a
N+N compound "pneumonia therapy" in it, and usually the first N in
such a compound is unavailable as an antecedent for a pronoun, because
of its semantics (as i mentioned last week in this thread): the first
N refers not to an individuals or a group, but to a type, sort, or
genus.  there are well-known exceptions, however; if the first N is a
proper name or a mass noun, then its semantics would not prevent it
from serving as the antecedent of a pronoun.

*but*... the referents of non-head elements are, ceteris paribus, less
salient than the referents of heads (there's a kind of iconicity at
work here), so that non-head elements are less easily available as
antecedents of pronouns than head elements.  but various other factors
can affect the salience of one or another referent.  one such factor
is preceding discourse context.

the pneumonia therapy sentence is much improved if we supply discourse
that makes diseases (rather than treatments) especially salient, as
in:
  A number of diseases have become much more treatable in recent
  years.  For example, the broad-spectrum antibiotics have so
  simplified pneumonia therapy that it is now frequently treated
  at home.

once again, the moral is that if we're talking about effective
writing, we shouldn't judge sentences in isolation and we shouldn't be
training our students to.

yes, i understand that multiple-choice exams are a whole different
world from writing well, or from teaching effective writing.  i'll get
to that eventually, in another posting.

arnold (zwicky at csli.stanford.edu)