implicit/implied (was Re: PSAT Glitch)

Arnold Zwicky zwicky at CSLI.STANFORD.EDU
Sun May 25 22:13:35 UTC 2003


this is in response to P-A-T, who brings up several important issues.
some of what i say below repeats, in a new form, things i've said
before in discussion on this topic, so some readers might be bored.
just a warning.

[in what follows i'll routinely talk about syntax as a matter of
combinations of, and relationships between, *phrases* - expressions
that may comprise more than one word, or may consist of only a single
word - rather than single words. in "Mary's deluded father thinks he
is handsome", the referring expressing that "he" is anaphoric to (on
the obvious reading) is the NP "Mary's deluded father", not the lone N
"father".]

P-A-T:

 >I look at a possessive noun in the same way I look at a verb(al).
 >While a verbal retains the properties of a verb (It can take an
 >object; it can have a subject; it retains the characteristic the
 >verb ending [-ing, -ed, -en]), it no longer functions as a verb.  It
 >functions as another part of speech--a nominal, an adjectival, or an
 >adverbial.

there are lots of different types of "verbals", calling for different
sorts of analyses.  i'm going to talk about the type that, it seems
to me, presents the gravest difficulties in analysis - nominal gerunds
like "Kim's singing the national anthem in Farsi" - and i'm doing this
because the nominal gerund situation contrasts interestingly with
the situation for possessives.

nominal gerunds are, for the most part, not problematic at all with
respect to their function (their external syntax).  they can appear in
(almost) any position that garden-variety NPs can - in particular, as
subjects, direct objects, and objects of prepositions.

the difficulties all have to do with the form (internal syntax) of
nominal gerunds, which (as is well known) have "mixed" nominal and
verbal syntax.  to compress things down to just a few observations:
  nominal property (beyond NP external syntax):
    possessive (rather than nominative) NP expressing the subject
  verbal property: the three parts of the construction - the
    possessive NP, the V-ing word, and the material following that -
    reproduce essentially *perfectly* the syntax of clauses with
    that NP as subject, the V as main verb, and the following material
    as complement to that V; in particular, direct object NPs have
    no marking by the preposition "of", as NP complements to a noun
    regularly do ("a performance of the national anthem in Farsi",
    "the singing of the national anthem in Farsi")

now, *here*s an analytic problem, one for which a variety of solutions
have been suggested.  note, though, that it won't do just to say that
"singing" in "Kim's singing the national anthem in Farsi" is both a
noun and a verb; you have to predict, or stipulate, that "singing" in
such an example is a noun *in certain specific ways* and a verb *in
other specific ways*.

on to possessives.  the problem here is much simpler.  all that's
going on is something that's as common as dirt, a mismatch between
form (internal syntax, associated with a syntactic category, like NP)
and function (external syntax, associated with a syntactic function,
like Determiner).  nominal gerunds illustrate a form-function
mismatch, too, but they involve an additional conflict entirely in the
internal syntax.  possessives, on the other hand, are unproblematic
internally: they consist of a NP, with suffix -'s on its last word,
period.

an example: AdjPs have two core syntactic functions in english,
Adnominal Adjectival and Predicative.  but all sorts of other things can
serve in the Predicative function:
  NP: I became a linguist.
  AdvP: Stock prices are way up.  Kim is there.
  PP: Kim is from California.
  VP: What I did was eat the sushi.
  Clause: The problem is that we have to go.

what we do *not* say in these cases is that the the Predicative
expressions belong to the category AdjP, in addition to the categories
listed above (and that the heads of the Predicative expressions belong
to the category Adj as well as to the category N, Adv, P, V, or V,
respectively).  "from California" can function as a Predicative, but
that doesn't make it an AdjP (as well as a PP), nor does that make
"from" an Adj (as well as a P).  similarly for the other examples.

and similarly for possessive NPs serving in the Determiner function.
(long ago i pointed out that the syntactic function that core examples
of possessives serve is not Adnominal Adjectival, but Determiner;
possessives come very close to filling the same slots as the definite
determiner "the".)  serving in this function doesn't make possessives
determinatives, or for that matter adjectives.

in fact, possessives can serve in other functions as well, in various
special constructions:
  Predicative: This pencil is Sandy's.
  Object of P: a friend of Sandy's
  Object of P or V: We went to Sandy's.  We visited Sandy's. ['place']
  Subject or Object of V or P: Sandy's outran Kim's. Sandy's ran in
    front of Kim's  [need context, but that's easy]

the way to make sense out of these simple form/function misfits is to
provide separate vocabularies for syntactic categories and syntactic
functions; to stipulate default associations between categories and
functions; and to stipulate non-default associations in particular
constructions.

 >Except in some nonstandard dialects, one cannot say, "She sewing."
 >But, "Sewing is her favorite hobby" is acceptable.  We even have
 >special names for these "tricky" phenomena (gerunds, participles,
 >infinitives).  When a word has both the form and function of a verb,
 >it is finite and there should be no confusion.  But when it has the
 >form of a verb and the function of other parts of speech, do we
 >still dare teach it as a verb?  Or do we teach it as what it is
 >functioning as?  Many textbooks discuss the verbal only briefly at
 >the end of the chapter on verbs.

see above.  inventing ad hoc labels for an endless array of putatively
mixed categories doesn't provide an actual analysis of the facts of
the language, only names for problems.  instead, we need to state
generalizations about how syntactic categories and syntactic functions
are associated with one another, and about how these associations are
connected to semantics and to information structure and discourse
organization.

 >There is a similar reaction to the possessive noun (Maybe it should
 >be called a "nounal" to capture that inconsistency between form and
 >function.)  While it still obviously has characteristics of a noun
 >(One can easily recognize that it began as a noun), it is no longer
 >functioning as a noun.  It now functions adjectivally.  Like the
 >verbal, which looks like a verb but functions as a nominal,
 >adjectival or adverbial, the possessive noun is no longer considered
                                                            ^^^^^^^^^^
?considered by whom?

 >to be functioning as a nominal.  Thus, just as the verbal retains
 >characteristics of the verb itself, but loses the verb's key
 >defining quality--finiteness--, the possessive noun retains the
 >recognizable (surface) features of a noun; however, it loses the
 >essence of a noun.  That is, it can not be pluralized; it can not
 >serve in a nominal position (as subject or object), nor,
 >consequently, can it be the antecedent of a pronoun).  Of course, if
 >you drop the apostrophe s ('s), the word regains its "nounness," but
 >then you are back to square one.

you seem to have a serious misunderstanding here about how case
inflection works in syntax.  inflecting a NP for a particular case
limits the external syntax of that NP; case morphology is *for*
indicating syntactic relationship.  in a reasonably case-rich
language, a nominative NP is restricted in syntax; it can't serve as a
Direct Object, Indirect Object, or Object of P.  an accusative NP is
similarly restricted; it can't serve as a Subject.  ablative and
instrumental NPs are more restricted still; they can't serve as
Subjects *or* Direct Objects, but only as certain subtypes of
Adverbial and as Object of certain Ps.  possessive/genitive NPs are
limited (for the most part, but see above) to function as a
Determiner.  (indeed, we *call* a case "possessive" or "genitive" just
*because* this is the default syntactic function of NPs with that
case.)  this is as true of english as of latin or thousands of other
languages.

there is no conversion of NP (or, as you choose to put it, N) to
some other category.  your belief that there is such a conversion is
just an artefact of your working with an impoverished vocabulary for
talking about syntactic categories and syntactic functions.

 >I think these authors use the word implicit to capture the
 >discrepancy between form and function--that is, between how it looks
 >(form) and how it behaves (function).

somewhere along the line here you should have gotten worried about
the fact that your way of talking about the syntax of possessives
*predicts* that they should not be available as antecedents for
pronouns, when in fact there is massive evidence that they are
routinely and unproblematically available for just this purpose.
that's a great big flag that there was something wrong with your
assumptions.

arnold (zwicky at csli.stanford.edu)



More information about the Ads-l mailing list