Pronouns, nouns and other syntactic myths

[Bill Croft]

David's discussion of Mickey's examples illustrates the
shortcoming of talking about "pronouns" or "nouns" or "NPs"
as categories of particular languages. In fact, all such debates
about the status of grammatical categories are really debates about
the constructions that are being used to define those categories.

    For example, David writes about Mickey's first example:

>Consider first the hide-and-seek example, eg.
>
>(1) John's it.
>
>Admittedly, this is parallel to the article-less construction
>
>(2) John's goalkeeper
>
>but this usage of "it" seems to be much more syntactically restricted:
>
>(3) John's the ??it / goalkeeper today
>(4) John's a really good ??it / goalkeeper
>(5) They're playing with two ??its / goalkeepers
>
>It seems to me that "BE it" is some kind of an idiom in English.

Well, not necessarily. 3-5 are not examples of the construction in 1-2;
3-4 have articles, and 5 is not an equative or predicative construction
at all. 1-2 are examples of a [Sbj be Role] construction. Since "it" defines
a role in hide-and-seek, it is OK in this construction. Same with "goalkeeper".
Other so-called genuine nouns are not OK in this construction:

(6) *Fido's dog.
(7) *That's chair.

So does this mean that "dog" and "chair" are syntactically restricted? Of
course. Does this mean that "dog" and "chair" are less nouns than "it" and
"goalkeeper"? I believe this question simply doesn't make sense. Anybody
who does detailed distributional analysis of a language---which you have to
do, if you want to have an empirically accurate description of the language
---will discover that there is no "noun" or "pronoun" category; there are
many constructions, some extremely common, others less so, that define
categories of words/phrases that can fit in them. Those categories criss-
cross one another. There are patterns of course, in fact of the same type
as the patterns found in cross-linguistic universals like implicational
hierarchies and grammaticalization paths. But the patterns don't all line up
to define some small finite set of syntactic categories; and syntacticians
shouldn't be looking for them. Instead, syntacticians should be doing detailed
and careful studies of the constructions found in languages, the distribution
classes defined by them, the functions of the constructions---which will
explain a lot of the distribution---and the patterns of distribution across
constructions.

     Another way in which it doesn't make sense to talk about "pronouns" as
a syntactic class is that the phoneme strings called "pronouns" are in fact
not
the same from context to context. Consider the following, again mostly from
David & Mickey:

(8) It's under the sink.
(9) Jane's it.
(10) That dog is an it.

Examples 8-10 each contains the phoneme string "it". But in 8, "it"
is a referring expression; in 9, "it" defines a role; in 10, "it"
defines a class (neutered animal). This is not unlike the phoneme
strings "ship" in 11 and in 12:

(11) The ship sailed into the harbor.
(12) I shipped my belongings to England.

The phoneme string "ship" meaning VESSEL in 11 should not be identified
with "ship" meaning SEND in 12. Regressing temporarily to traditional
terminology, [ship/VESSEL] is a Noun and [ship/SEND] is a
Verb, or at least, they are different lexemes occuring in
different constructions. That's no different from the three "it"s
in 8-10, except that the differences are subtler and the 9-10
examples are often overlooked.

    One could say we should totally ignore meaning in doing syntax.
"It" is "it" is "it", and "ship" is "ship", period.
(I don't think this is the view of anyone who has taken part in the
discussion; but I don't think I'm constructing a straw man here.)
So "ship" is a noun-verb and "it" is a pronoun-noun. But it's
obvious that the reason for the so-called formal distribution of phoneme
string equivalence classes is because those phoneme string equivalence
classes are polysemous, and their various meanings allow them to
occur in the constructions that they do, in turn because of the meanings
that the constructions have. Ignoring meaning means depriving oneself
of the explanation for the syntactic distribution patterns.

Bill Croft


(For fascinating discussion of the articled and article-less predicate
nominal constructions in English, see:)

Bolinger, Dwight. 1980. Syntactic diffusion and the definite article.
Bloomington: Indiana University Linguistics Club.