"pronoun" is semantic or distributional?

[Larry Trask]

Lloyd Anderson writes:

> Another way of seeing why our long discussions
> have been at cross purposes is the following.

> Pat Ryan, in attempting to conform to the terminology
> being used by Larry Trask, wrote the following:

>> [PR]
>> Perhaps the discussion could be foreshortened. 'My' could perhaps be termed
>> a "pronominal determiner".

> This is a perfectly reasonable position, *GIVEN* that Pat is using
> the word "pronoun", here in the form "pronominal",
> as a semantic-functional term (referring to persons etc.).
> I believe that is an accurate statement, even if Pat's explanations
> have not said so explicitly.

> Trask is however using "pronoun" as a distributional class.
> For Trask, "pronominal" and "determiner" are a contradiction
> because both refer to distributional classes, distinct classes,
> one standing for a full noun phrase, the other as a modifying
> element part of a noun phrase (loosely put).

Yes, except that I prefer 'specifying' to 'modifying'.

But I don't agree that 'pronoun' can or should be used as a label for a
semantico-functional class, since the term is already in use for a syntactic
class.  To apply the same label to two entirely different things is merely to
invite unnecessary confusion.

> Back to Pat Ryan's terminology:
> "possessive pronoun" is perfectly reasonable when both terms
> are taken in their semantic-functional senses.
> But Trask does not use "pronoun" that way.
> (Nor do I, when I am dealing with distributional classes.)

[snip]

> On the other hand, Pat Ryan could also point out that in
> "her book", the "her" stands also (in Trask's analysis I think also)
> for the entire noun phrase modifying book in
> (the woman who came yesterday's book),
> and thus may be regarded as the genitive or possessive form
> of the pronoun "she", in the contrast "she" vs. "her book",
> just as "the woman who came yesterday's" is the genitive
> form of the noun phrase "the woman who came yesterday".

> That is all consistent in the standard analysis, I believe?

Not necessarily.  Syntactic analysis does not, in general, work in terms of
which things can "stand for" other things, whatever that might mean.

Anyway, the determiner 'her' does not intrinsically mean 'the woman who came
yesterday'.  It might be so interpreted in a given case, but so what?  This is
still no argument that possessive 'her' is a pronoun -- now is it?

Consider a parallel case:

John: "I'm looking for a book with illustrations by Aubrey Beardsley."

Mary: "I have such a book."

Now, in this context, 'such' is clearly to be interpreted as meaning 'with
illustrations by Aubrey Beardsley'.  But does this observation make 'such'
a prepositional phrase?  No?  Then how can possessive 'her' be regarded as a
pronoun, on the basis of a parallel argument?

> My point, as it has been for some time, is that
> despite whatever Pat Ryan may or may not understand
> of the type of distributional analysis represented by many of us,
> his terminology was quite consistent and sensible,
> and used an older tradition in the meaning of "pronoun",
> a semantic-functional one.
> He should not be beaten upon for that.

Well, this is debatable -- the "consistent and sensible", I mean, not the
beating-up part.

> Even Trask's distributional usage might be criticized by a purist,
> in that for him a "pronoun" does not stand for a "noun"
> but rather for a noun phrase.  So we should all give up the term
> "pronoun"?  I certainly don't advocate that, despite how
> misleading it may be to some.

We all realize that the label 'pronoun' is a little unfortunate, and that
'pro-NP' would be preferable in principle.  But the term is traditional and
established, and all linguists know what it means, so there is no good reason
to change it.  We are not so foolish as to commit the etymological fallacy.

Mathematicians apply the label 'imaginary numbers' to a certain class of
numbers.  This name too is a little unfortunate, since it misleads outsiders
into believing that the "imaginary" numbers are somehow less real than the
"real" numbers -- which they are not.  But again the term is traditional and
established, and all mathematicians know what it means, and so they have seen
no reason to change it. If outsiders misunderstand it, that's just their tough
luck -- and it's the same with us.

> Larry Trask was kind enough to take the time to distinguish
> semantic-functional senses from distributional senses,
> but I think had not acknowledged that much of the discussion
> was motivated not by a lack of knowledge on Pat Ryan's part,
> but rather by Pat Ryan using "pronoun" in a semantic-functional
> sense.  With different definitions, the discussion was bound
> to be unfruitful.

No doubt.  But I notice that no semantico-functional definition of 'pronoun'
has been advanced, either by Pat Ryan or by anyone else.  Anybody like to try?

> So can we please stop trying to prove each other wrong,
> and get back to discovering interesting things about the real world?

I'd love to.  But I honestly do not believe that 'pronoun' should, or even can,
be usefully applied to a semantico-functional category -- especially to a
hypothetical such category which nobody has tried to characterize.

> The following is a refinement, dealing with a more difficult
> "edge" case.

> Paul Postal (I think it was) many years ago argued that in
> expressions like "we linguists" the "we" was the head of
> the construction and the "linguists" was something like
> an appositive (I don't remember the details just now).

This analysis makes some sense, though it's not so easy to defend in English,
with its lack of agreement.  In languages with more agreement, the comparable
construction usually shows first-person agreement, not third-person, confirming
that the pronoun is best taken as the head.

> I don't think this kind of construction is usefully laid
> up against "those linguists", arguing the reverse of Postal's
> position that "we" can be a determiner, because it is
> understood as "we, who are linguists" (non-restrictive),
> more than as a restrictive "those linguists who are we"
> in the manner of "the house which is here" ~=~ "this house".

I query this analysis.  When I say 'we linguists', I normally mean '*all*
linguists (including me)', and *not* 'a few people (including me) who happen to
be linguists'.  In context, I might mean something different.  For example,
within my university, I might conceivably say 'we linguists' to mean 'all the
members of the Linguistics Department', but then I still couldn't use it to
mean only 'some of the people in the LD, including me'.

Anyway, note that, in English, this odd construction is confined to 'we' and to
plural 'you': nobody permits *'I linguist' or *'she linguist' or *'they
linguists'.  The construction is therefore somewhat marginal, and it should not
be invoked too freely in drawing conclusions about English syntax.

Larry Trask
COGS
University of Sussex
Brighton BN1 9QH
UK

larryt at cogs.susx.ac.uk