Possessives

[Larry Trask]

On Sun, 3 Oct 1999, Patrick C. Ryan wrote:

The moderator has -- mercifully -- pulled the plug on this thread, but
he has invited me to post a final comment.  Thanks, Rich -- I'll try to
be brief, by my standards.

> One of the principal benefits of terming this class of words
> "possessive pronouns" is that it enables us to identify comparable
> classes of words in IE, English, and Basque --- even though the
> details of their employment may differ.

Well, this reminds me of Abe Lincoln's little joke.  "How many legs does
a sheep have if you call the tail a leg?"  "Five?"  "No, four: calling
the tail a leg doesn't make it a leg."

Possessive items, in general, are "comparable" among languages only
insofar as they are translation equivalents.  Syntactically, they can
and do exhibit many different kinds of behavior, requiring them to be
assigned to various parts of speech, or sometimes to no part of speech.

The English possessives like `my' are determiners because they behave
like determiners.  They do not behave like pronouns, and so they are not
pronouns.  And they certainly don't behave like adjectives.

The Basque possessives like <nere> `my' exhibit none of the properties
of the Basque determiners and they cannot be classified as determiners.
They are also certainly not adjectives, since they behave nothing like
adjectives.  They are probably best regarded as case-inflected forms of
pronouns.

The Latin possessives like <meus> `my', on the whole, are probably best
regarded as adjectives, since they behave like adjectives in that
language.  They are certainly not determiners, and they don't look much
like pronouns, either.

Other languages use other strategies.  Some have no distinct possessive
forms at all, but simply adpose the free forms of pronouns to head
nouns.  Some use a particle to link a free pronoun to its head.  Some
use bound markers attached to head nouns, often markers bearing no
resemblance to free pronouns.  And some languages employ mixed
strategies with two or more of these in various combinations; examples
are Turkish and Jacaltec.

Classing all of these are "possessive pronouns" *tout court* is, at the
very least, unhelpful.

> The essential quality of any adjective is that it designates a
> subcategory of a catgeory of objects.

No, not at all.

This is a prototypical semantic property of adjectives in languages that
have them, but `adjective' is a syntactic category, not a semantic
category, and not all adjectives, even in English, have this property.

For example, consider `Susie is a mere child.'  Here the adjective
`mere' does not define a subclass of the class of children: children are
not divided into mere and non-mere varieties.  Or consider this one:
`Lisa is a heavy smoker'.  This time the adjective `heavy' does not
perform any subclassification of the set of smokers.  In fact, it
doesn't even apply semantically to smokers, or to Lisa: instead, it
applies to Lisa's habit.

> 'Black dog' can be represented logically as a small circle ('black')
> within a larger circle ('dog'). Possessive pronouns can be
> represented logically in exactly the same way.

For certain purposes, perhaps.  But this is not an argument that `my' is
a pronoun: rather, it appears to be a (feeble) argument that `my' is an
adjective -- which it is not.

> Of course, this is exactly the same situation as when nouns are used
> attributively: "newspaper account".

Indeed.  It begins to become clear that semantic tests are not very
useful for identifying syntactic categories.  But we linguists knew
that.

> I think your basic problem is that you need to take a serious look
> at your definition of 'adjective'.

Well, *one* of us certainly does. ;-)

[LT]

>> The noun `arrival' is not a verb because it's related to the verb
>> `arrive'.

> Perhaps not when you're playing the 'slots' but that is not the only
> game in town by a long shot!

> I sincerely feel sorry for you if you cannot see that.

You mean there's an Arkansas analysis in which `arrival' is a verb?
Most interesting.

[on the slot-and-filler approach]

> I remember the first manifestations of the 'slot system' as opposed
> to teaching grammar in the training I received as an aspirant
> foreign language instructor.

> This 'grammarless' method of teaching language was then extended to
> English instruction in the US,

I would hardly describe the slot-and-filler approach as "grammarless".
It is in fact a thoroughly grammatical approach.  Not so traditional, of
course, but certainly grammatical in nature.  Far more so than the
semantic approach which I dismissed above.

> and the terrible language skills of today's youth are
> directly attributable to this method of instruction, IMHO.

Dear me.  The linguistic shortcomings of our young people are to be laid
at the door of slot-and-filler grammar?

That's a new one on me.

Oh, by the way -- further bad news for the Arkansas school of linguistic
description.  I've recently written the entry on parts of speech for the
forthcoming encyclopedia of grammatical categories from Elsevier.  I'm
afraid my baleful influence is getting out of hand.  Doubtless the
nation's young people are in for yet another body-blow. ;-)

Larry Trask
COGS
University of Sussex
Brighton BN1 9QH
UK

larryt at cogs.susx.ac.uk