PSAT Glitch

[Arnold Zwicky]

matthew gordon:
 >It's interesting to me that in the Post article the violation is
 >said to result from the pronoun('s) having an adjective as its
 >antecedent; the possessive noun "Toni Morrison's" is said to be an
 >adjective. This reminds me that in many grammar textbooks used in
 >K-12 all sorts of things are called adjectives from indefinites
 >(every, all) and numerals to possessive pronouns and
 >articles. Basically, anything that comes before a noun is an
 >adjective.

this is what happens when you don't distinguish syntactic categories
from syntactic functions. Adjective is a syntactic category;
Modifier-of-N is a syntactic function.  they are significantly
connected: most occurrences of Adjectives are functioning as
Modifier-of-N (or as Predicative), and a great many words functioning
as Modifier-of-N are Adjectives; in other words, the primary syntactic
function of Adjective words is to serve as Modifier-of-N (and as
Predicative), though such words can do other things and other things
can serve that function.  (similar remarks hold for, e.g., the Noun
(Phrase) category and the Subject (and Direct Object and Object of P)
function.)

if you don't make this distinction, then you have a conceptual
framework in which modifiers (in any sense) of N have to be classified
as Adjectives, faute de mieux.  occasionally, people classify even the
first noun of a noun-noun compound (like "Christmas cookie" or "flower
show") as an Adjective on these grounds.

a conceptual framework for syntax that is adequate for talking about
the facts of english (not to mention other languages) needs a
distinction between category and function.  it also needs distinctions
between functions that the unwary might collapse, for example, true
Modifier-of-N (prototypically, filled by Adjective) and Determiner
(prototypically, filled by Determinative), which occur in different
layers of NP structure.  i believe that the impulse to collapse
*these* distinctions arises from an identification of syntactic
function with *semantic* function: a Modifier-of-N element and a
Determiner element both serve semantically as operators on the meaning
of the N they precede.

so, in the end, you need three sets of concepts: syntactic category,
syntactic function, and semantic function.  (actually, you need a
fourth set as well - inflectional category.  so the syntactic category
Noun, the syntactic function Subject, the semantic roles Agent and
Topic, and the inflectional category Nominative Case all need to be
distinguished, even though they're non-trivially related to one
another.  all this is really just Syntax 101 or Structure of English
101 (you can see an extended scholarly presentation in huddleston and
pullum's compendious Cambridge Grammar of the English Language), but
school grammars hardly ever come close to making the required number
and kinds of distinctions and instead limp along with some
impoverished set, so that, as soon as they get past prototypical
instances, students are baffled as to how to apply the terminology
they've been taught.

note that none of this is about "formal grammar" or "theoretical
syntax" as those terms are usually applied (to formalized frameworks
for syntactic description).  none of it is especially new, either; the
ideas here were clarified by structuralist syntacticians many decades
ago, before i was even born (and i'm an old guy now).  as far as i can
see, this is pretty much what you get when you ask the question "what
concepts (and accompanying terminology) do we need for talking about
the syntactic organization of languages?" instead of the question "how
can the conceptual apparatus inherited from the latin grammarians be
used to describe the syntax of other languages?" this is not to
dishonor the traditions based on latin grammar, but only to say that
they need significant revision, extension, and complication in the
light of the facts of language (latin included, by the way).

the important point of this is that we *discover* how languages work,
and discovery is an active, ongoing enterprise, which means that the
way we talk about language changes (and improves in clarity and
coverage) over time. (not all change is a matter of scholarly
fashion.)

but how to communicate that?

arnold (zwicky at csli.stanford.edu)