possessive [form of a] pronoun

[Robert Whiting]

On Thu, 2 Mar 2000 ECOLING at aol.com wrote:

>Bob Whiting has widened our set of examples. He thereby brought
>up yet another wholly new line of argument, which was not covered
>by preceding messages.

>I still refer back to the message "possessive [form of a]
>pronoun" for the parallelism which it *did* contain, and which is
>strongly supported by more detailed considerations, as shown
>below.

First of all, I was surprised to see that message posted since
the moderator pulled the plug on this discussion on 3 October
1999.  And I wouldn't have replied to it except for the fact that
the poster promised not to respond (but I should have known
better).

The moderator was quite correct to stop the discussion because it
was another terminological discussion that could have been
resolved by simply consulting an up-to-date reference book.  The
discussion centered around the fact that possessive pronominal
determiners are not possessive pronouns and that therefore it is
improper to call possessive pronominal determiners possessive
pronouns because syntactically determiners are not pronouns
(although there is often a close semantic relationship and many
determiners can also function as pronouns).

Larry Trask has pointed out that general dictionaries are often
not up-to-date on grammatical terminology and therefore not to be
trusted as sources of definitions.  However, I find that, despite
some inconsistencies and overgeneralizations, 'my' dictionary
(Webster's Encyclopedic Unabridged, 1994) is quite strong in this
area so I provide the following definitions from that source
which will be needed in the following discussion:

     determiner 2. Gram. a member of a subclass of English
     adjectival words that limits the nouns it modifies in a
     special way and that usually is placed before descriptive
     adjectives, as _a_, _an_, _the_, _your_, _their_.

     pronoun Gram. any number of a small class of words found in
     many languages that are used as replacements or substitutes
     for a wide variety of nouns and noun phrases, and that have
     very general reference, as _I_, _you_, _he_, _this_ _who_,
     _what_.  Pronouns are sometimes formally distinguished from
     nouns, as in English by the existence of special objective
     forms, as _him_ for _he_, or _me_ for _I_, and by
     nonoccurrence with an article or adjective.

     possessive 4. Gram. indicating possession, ownership,
     origin, etc.  _His_ in _his book_ is a possessive adjective.
     _His_ in _The book is his_ is a possessive pronoun.

     genitive _Gram. --adj. 1. (in certain inflected languages)
     noting a case used primarily to indicate that a noun is a
     modifier of another noun, often to express possession,
     measure, origin, characteristic, etc., as _John's hat_,
     _man's fate_, _week's vacation_, _duty's call_.  2. noting
     an affix or other element characteristic of this case, or a
     word containing such an element.  3. similar to such a case
     form in function or meaning.

     adjective 1. Gram. any member of a class of words that in
     many languages are distinguished in form, as partly in
     English by having comparative and superlative endings, or by
     functioning as modifiers of nouns, as _good_, _wise_,
     _perfect_.

Probably not even Larry T. would find much to dispute in these
defintions (although I wait to be corrected), except for the more
or less interchangeable use of adjective and determiner.  Larry
has argued that determiners are not adjectives because they can't
fill the same slots.  Specifically, he pointed out that the slot
'the ______ book' cannot be filled by a determiner but can be
filled by an adjective.  This, however, is the only restriction
on use, and for the rest of their ranges determiners and
adjectives are coextensive although there are other distinctions
between determiners and adjectives (e.g., determiners do not have
comparative and superlative forms).  I would prefer to categorize
determiners as a subclass of adjectives, rather than saying that
determiners are not adjectives.

The restriction in use comes from the fact that determiners of
various classes are mutually exclusive within their classes.
While one can have as many coordinate adjectives in an NP as one
wants, an NP can have only one determiner of a particular class
at any particular level.  Thus a slot like 'I like ______ books'
can be filled by either a determiner ('I like the books', 'I like
his books', 'I like some books') or (any number of) adjectives
('I like long, interesting, inexpensive books') or even a
determiner and any number of adjectives ('I like some long,
interesting, inexpensive books') while a slot like 'the ______
book' already contains a determiner in the NP and thus cannot
have another of the same class.  If a determiner and adjectives
are used together, the determiner normally precedes all
adjectives.

>First, I am not knowingly trying to mislead anyone,
>despite Bob Whiting's claim:

Then perhaps you are at your best (or worst) when you aren't
trying.

>>You have simply tried to slide one past us by using a form
>>where the determiner 'his' and the possessive pronoun 'his' are
>>isomorphic.

>Not so, though I should not have to defend myself against
>such a groundless ad-hominem assertion.

You are still responsible for what you say and do, regardless of
whether you did it knowingly or not.

>(The Postscript notes that I normally do just the reverse of that!)

>I *did* choose the word "his" deliberately because unlike some of
>the other genitive forms ("determiners", as many call them), it
>does look more like an apostrophe-s form.

I don't believe this!  You are telling me (and the whole list)
that, after having badgered Larry Trask for almost a year with
claims that his selection criteria for Basque words *could* bias
his conclusions, you have deliberately [your word] selected your
data specifically for its bias towards your conclusions?  Shame on
you.  As far as doing scientific work is concerned, I am afraid that
we are simply on two different planets.  I regard knowing how data
selection can bias results as a way of avoiding it myself or
recognizing it when others do it.  You seem to see it as a way of
deciding which method to use when you want to demonstrate that
your conclusions are correct.  Constantly questioning the motives
and methods of others while steadfastly refusing to examine your
own is simply the height of arrogance.

>But I did not choose it because of homonymy between
>determiner (my) and pronominal (mine) forms: (his, his).
>I was not at all concerned with the pronominal forms
>(his, hers, mine, yours, ours, theirs).
>Bob Whiting's examples involve such pronominal forms,
>but my examples did not involve them.
>So there was nothing to "slide past" anyone.

The fact that the fault that you were accused of did not happen
to be the actual fault that you practiced does not exonerate
you.  It is rather like being charged with driving 60 mph in a 25
mph zone and claiming in your defense that you weren't going a
bit over 50.

I claimed that the choice of 'his' biased the data sample in
favor of your conclusion.  I am willing to accept your claim that
the reason I gave for why this was true (the determiner and the
pronoun are isomorphic) never occurred to you.  But this does not
change the fact that the selection of 'his' as the data sample
does bias the conclusion.  Since you admit that you chose 'his'
for this very purpose, the specific reason why it biases the
conclusion is immaterial.

But if that was your reasoning then there are a number of
compelling reasons why you made a poor choice.  Given this line
of reasoning, the obvious choice for an example is 'its' because:

     (1) 'its' looks much more like an apostrophe-s form than
     'his' does (I cannot remember ever seeing anyone write
     <he's> for <his>, but I constantly see <it's> for the
     pronominal determiner <its>).

     (2) 'its' is used only as a pronominal determiner in
     English, not as a possessive pronoun.  Therefore the
     possibility of confusing the determiner with the possessive
     pronoun does not exist.

     (3) 'his' is the *only* personal pronoun where the
     determiner and the possessive pronoun are isomorphic.
     *All* of the other determiner:possessive pronoun pairs use
     different forms for the different functions (my:mine,
     you:yours, her:hers, its:--, our:ours, their:theirs).

In short, your example only works if you use 'his'.  This is the
only way that you can get 'the man we met yesterday's' to always
map to 'his' and thus justify your equation.  If you use 'the
woman we met yesterday's' it sometimes maps to 'her' and
sometimes to 'hers'.

>If you add to the following examples I used the specification
>that *we are talking about the determiner use of "his, my, etc"*,
>which was certainly the context in which I was writing
>(and please note that Larry Trask had also not been discussing
>"mine" etc., and said so in response to Pat Ryan)...

Nobody knows what you are thinking unless you say what you think.
One can only determine what you mean from what you write.
Personally, I think it is rather fanciful to expect listmembers
to remember the context of a discussion that was killed by the
moderator six months ago.  And please note that possessive
pronouns like 'mine', etc. were not being discussed because
everyone agrees that they are possessive pronouns.  What was
being discussed was the relationship between the possessive
pronominal determiners and the possessive pronouns and the
question was "if the possessive pronouns are pronouns, why aren't
the pronominal determiners pronouns too?"  In this discussion,
the possessive pronouns are taken as given; but the question of
the relationship of the pronominal determiners to them is not
likely to be resolved by eliminating them from the discussion.

>Then there is no way of adding Bob Whiting's examples,
>because they do not fit that context.  Here were my examples,
>now extended only by adding a first-person example as well.

>> The relation:
>> he :: his
>[> or
>> I :: my   ]

>> is from syntactic and semantic points of view
>> essentially the same as the relation

>> the man we met yesterday :: the man we met yesterday's.

>This statement remains valid, as I think all linguists know.

Yes, so long as you restrict the context to the use of
determiners.  But any linguist will also tell you that "syntactic
and semantic" equivalence is the same as "functional"
equivalence, not the same as "formal" equivalence.  And
membership of words in classes is determined by function, not by
form (particularly in English with its general lack of
inflectional elements; perhaps less so in areas like Semitic
languages where form is very closely tied to function).  The
analogy shows that both NPs in the genitive and pronominal
determiners function as determiners, not that NPs in the genitive
are formally identical to possessive pronouns (despite your, as
you admit, deliberately choosing 'his' to suggest that they are).
Now pronominal determiners function as pronouns in that they can
be substituted for NPs.  But they can only be substituted for NPs
when the NP is being used as a determiner/adjective, not when the
NP is complete in itself.  But since they are then functioning as
determiners, they are pronominal determiners and not pronouns.
Is there anyone else who can't see the difference?

>A "syntactic and semantic" "relation" is not the same as a
>slot-filler word class.

If the "relation" is equivalence, then it is precisely the same.
Syntax is the rules by which forms can be combined to create
(grammatically correct) utterances.  Semantics is meaning, pure
and simple.  Thus "syntactic and semantic" equivalence of forms
means that the forms can replace one another in the same
combination of forms and still produce a grammatical utterance
without changing the meaning concept (but not necessarily the
specific meaning) of the utterance.  So all you are doing by
setting up analogies to demonstrate "syntactic and semantic"
equivalence and then not giving examples of the constructions is
creating the slots and then refusing to fill them.

>In the cases of "he", "I", and "the man we met yesterday",
>dominant uses, and the use I exemplified,
>are as full noun phrases.
>In the cases of "his", "my", and "the man we met yesterday's ",
>dominant uses, and the use I exemplified,
>are as determiners of full noun phrases
>(referring by "determiner" here to function and positions
>of some occurrences, not to slot-filler word class).

By Jove, you've got it (except for the slot bit).  Determiner is
a functional class.  Determiners are not full NPs but modify NPs.
An NP in the genitive case can be used as a determiner.  A
pronoun can be substituted for an NP.  A determiner can be
substituted for (fill the same slot as) another determiner.  Hold
that thought.

>The latter are not full noun phrases in most usages, as we all agree

Precisely.  And since you state that Larry Trask has stated that
pronouns replace full noun phrases, what replaces these "not full
noun phrases" are not pronouns.  It's that simple.

>(though the third of these can be used also as a full noun
>phrase, in the more complex structures discussed below).

But in that case it can't be replaced by a pronominal determiner,
only by a possessive pronoun.

>End of main line of argument (again, and just as previously stated).

Except that this time your own explanation has shown how faulty
the conclusions based on this argument were and how the choice
of the data was made to support the conclusion.

>***

>Now as to the tangent, elaboration, or extension of range
>of the discussion to pronominal forms like "mine, yours, ...":

>Bob Whiting alleges that the lack of a complete parallel between
>forms shows that an equivalence of syntactic and semantic
>relations does not exist.

>This is not the case, as when we talk about word classes,
>determined by slot-filler criteria, we often find only partial
>equivalences. We even find that the surface form "this" can
>function both as a determiner and as a pronoun (as I think the
>participants in these discussions wish to call them).  So a
>relation "this :: this" functions syntactically and semantically
>equivalently to "my :: mine".  Merely one such asymmetry.

Yes, merely one.  Practically any deictic or quantifying
determiner can function also as a pronoun (this, that, some, any,
each, every, many, etc.).  And it is easy to tell how it is being
used.  If it is part of an NP it is a determiner and if it
replaces a full NP it is a pronoun. (Henceforth [d] and [p] will
be used to mark isomorphic determiners and pronouns
respectively.)

     Some[d] books are red, some[p] are blue, some[p] are black.

     Q:  Are all books red?
     A:  Not all of them are; some[p] are, some[p] aren't.
         This[d] one is red, those[p] are blue, and those[d]
         others are black.

And so on ad nauseum.  But since the usage can always be
determined from the environment, this distinction is not a
random, sporadic use of "this :: this", but rather it is a
predictable "this[d] :: this[p]" which functions syntactically
and semantically equivalently to (= fills the same slots as) "my
:: mine".  No asymmetry left.

>***

>Whiting's parallels were these:

>>This is the man we met yesterday's book.
>>This is the book of the man we met yesterday.

>>This is his book.
>>This is the book of he.*

>I find the fourth of these ungrammatical, of course.

Of course.  That's what the asterisk after it means.

>I also find the second one odd, *either* in possessive sense
>*or* in another sense which it might conceivably express,
>a book about the man we met yesterday.

There is also another sense, which apparently hasn't occurred to
you yet:  a book written by the man we met yesterday.

>For the possessive sense, the sense in the examples I had given,
>(note again that I had *not* given examples with pronominal
>usages),

Actually, you hadn't given any examples at all.  All you had
given was analogies of "syntactic and semantic" equivalence
without any examples of usage.  You created the slots but didn't
fill them.  Using 'his' in your analogies without giving usages
does not make it clear whether you are using the determiner or
the pronoun.  And saying that everyone should have known what you
were talking about from a six-month old context is taking a lot
for granted.

>I believe such a usage more normally occurs with a relative
>clause following, than in the simple form given by Whiting:

Whether a relative clause is to be expected after the NP or not
is determined by whether the head of the NP is definite or not so
the rest of this paragraph is just argumentum ad infinite
digression.

>This is the book of his   which we were talking about.
>This is the book of mine   which we were talking about.
>This is the book of John's   whih we were talking about.
>This is the book of the man we met yesterday's   which we were talking about.
>     [as opposed to the books belonging to the man we met yesterday
>     which we were not talking about]

>The "of..." phrase in each case has at least sometimes been treated
>as a transformation of something with a bit more concrete content:

>This is the book from among mine which we were talking about.
>This is the book from among John's which we were talking about.

>or even
>This is the book [which is mine] which we were talking about.
>This is the book [which is John's] which we were talking about.

[end of digression]

>We also have the following form, which to me is more colloquially normal
>with the apostrophe-s.  I believe it shows that, like "his",
>the form "the man we met yesterday's" *can* function
>either as modifier (parallel to "my")
>or *sometimes* as a full noun phrase (parallel to "mine").

Spot on again.  And the fact that we can determine which form has
to be used in a given environment shows that it is not a vague
"*sometimes*" but "under certain conditions".  But you have the
basic idea correctly stated: 'my' is a modifier, 'mine' is a
pronoun ("functions as a full noun phrase").  And this is the key
to the entire thing.  Pronouns do not act as modifiers.  If
something that looks like a pronoun acts as a modifier then it
isn't a pronoun any more.  It's a pronominal modifier.

>We then have a three-way relation,

And this is where you should have started in the first place.

>and although some may not prefer the terminology which
>immediately follows, I trust with the examples they will know
>what I am referring to:

I'm glad that you realize that the terminology is out of line.
Up to this point you have demonstated only that the second
category is a determiner, not a pronoun.  In fact, you have
rather conclusively shown that it is not a pronoun.  So labelling
it as a pronoun is what is known as petitio principii ("begging
the question").  You haven't proved that it is a pronoun, you
have begged the conclusion.

>The triple relations:

>(nominative pronoun; determiner genitive pronoun; possessive pronoun NP)

>he :: his :: his
>I :: my :: mine

>are equivalent to the triple relation:

>the man we met yesterday ::
>:: the man we met yesterday's :: the man we met yesterday's

>with some slippage that in certain contexts and styles,
>in the last triple relation,
>the first form may sometimes be used for the third form
>(but with stylistic marking and awkwardness of different forms
>for different speakers depending on additional details of the
>contexts...).
>The third of the forms in each of the relations above was *not*
>previously prominent in these discussions (except very briefly).

Which is why your bipartite scheme fell apart with a little
gentle prodding.  You claim that demonstrating that two
expressions are "syntactically and semantically" equivalent
proves that they are the same form.  I say that demonstrating
this proves that they have the same function.  Proving that two
things have the same functional relationship does not prove that
they have the same formal relationship.

You claim that since 'the man we met yesterday's' is the genitive
case of 'the man we met yesterday' then 'his' must be the
genitive case of 'he' because both have the same function (and
look: they both have <s> on the end).  This presupposes a proof
that expressions with the same function always have the same
form.  I know of no such proof.  It is simply a premise ex nihilo
bolstered by your choice of 'his' for your analogy because "it
does look more like an apostrophe-s form."

Now it is not terribly difficult to demonstrate fairly
conclusively that the pronominal determiners must be pronoun
forms (apart from the obvious formal similarity) and that they
must be genitive forms, but it requires the tripartite relational
system.

To simplify it and expand it to cover all the personal pronouns:

   NP   :: NP's[d] :: NP's
   I    :: me      :: mine
   you  :: your    :: yours
   he   :: his[d]  :: his[p]
   she  :: her     :: hers
   it   :: its     :: --
   we   :: our     :: ours
   they :: their   :: theirs

This scheme accounts for all the genitive uses of NPs.  Since
'NP's' is genitive and can be replaced by any form in the third
column the other members of the column must be pronouns (forms
that can be substituted for NPs).  Since pronouns can have cases,
the evidence all converges on the supposition that these members
of column three are the genitive forms of the personal pronouns,
especially since all the forms with the exception of the first
person singular end in '-s' which is characteristic of the
English genitive.  Note that this is different from going
directly to 'his' in column two (or even 'its') and claiming that
it must be the genitive because it has an '-s' ending and is
congruent with one (and only one) use of the genitive 'NP's'.

The next problem is to account for the forms in column two.
First, we note the transparent similarity (or even identity) of
the members of the second and third columns.  Next we note that
the heads of each of these columns are identical in form, but not
in function.  The head of column two occurs only when 'NP's' is
part of another NP (i.e., it modifies another noun or NP).  We
note that only when 'NP's' occurs in this environment can it be
replaced by another member of column two.  Members of column
three occur in all other environments where 'NP's' can occur.
Thus the members of columns two and three are in complementary
distribution since it can be predicted from the environment which
one must be used.

Any linguist will tell you that similar forms that are in
complementary distribution are most likely to be aspects of the
same form.  Thus it is highly likely that if the members of
column three are pronouns then the members of column two are also
pronoun forms.  Furthermore since both can replace only genitives
both are probably allomorphic variants of genitive pronoun forms.

This establishes that the members of column 2 are genitive
pronoun forms.  Fine, everyone says, but if they are pronoun
forms, why not just call them pronouns and be done with it?
Well, many people do.  In fact, 'my' dictionary classifies each
one of these forms as a pronoun (even though if you look at its
definition of 'determiner' you will see that it knows that they
are really determiners) and then goes on to note that they are
used as attributive adjectives.  And for the man in the street,
to whom grammar is something that he had to study in school, this
is sufficient.  If he considers these forms as pronouns used as
adjectives he will still be able to speak the language correctly.

Linguists (and some grammarians), however, are less concerned
with how to use *a* language than with how language works (please
note the difference between *a* language and language), so this
simplistic approach of "if it looks like a pronoun, it must be a
pronoun" is not sufficient.  For a linguist, words are classified
by what they do, not by how they look.  Pronouns do not act as
modifiers, so something that acts as a modifier can't be a
pronoun no matter how much it looks like a pronoun, or even if it
is transparently derived from a pronoun.

Now while these forms do have a pronominal function in that they
can replace NPs, they cannot replace just any NP.  They can only
replace NPs that are being used as modifiers.  Hence their
pronominal funcion is entirely subordinated to their function as
modifiers.  A few simple tests will show a linguist that the type
of modifier that these forms function as is a determiner so the
linguist will call them pronominal determiners.  Anyone can say
that 'her' is a pronoun to the man in the street and no one will
argue (and if someone does you can look it up in a dictionary and
"prove" that it is a pronoun).  But anyone who says that 'her'
is a pronoun to a linguist will be told "no, it is a determiner"
and if he persists in calling it a pronoun (and the linguist is
patient) the linguist will explain to him why it is a determiner.
The moral of the story is that if you want to talk to people, you
have to speak their language.

>I still plead to our list:
>Can we now stop trying to prove people are wrong in using
>traditional terminology?

And can we reject the use of arrogance as a hermenutic?

>It meant what it in fact referred to,
>and we all knew what it was intended by the author to refer to.
>Still does refer to just what it did before,
>not only for those who themselves use terms that way,
>but *also* for those who understand the terms that way
>while preferring themselves to use different terms!
>(And just as Larry Trask has pointed out that "pronoun" is an
>illogical term, because it refers to what are really "proNP"s,
>yet that we still use it successfully, so it can be the case for
>other traditional terminology.)

Terminology is important in linguistics, as it is in most
specialized disciplines.  One can use imprecise terminology as
much as one wants to in general discussions, but when one starts
talking about linguistics with linguists one should be prepared
to use linguistic terminology.

>Can we please rule out of order on this list any messages which
>go off into meta-analysis to prove someone wrong because
>of the terms they used,
>instead of dealing with the content of what we perfectly well
>know they were saying?

Not when the terms used affect the content.

>I thought that was the policy which our moderator declared
>sometime back?

And I thought that our moderator killed this discussion last
October for just that reason.  Or do we now have a new,
self-appointed moderator?

>Sincerely yours,
>Lloyd Anderson
>Ecological Linguistics

>***

>In response to Bob Whiting's ad-hominem, an incorrect one:
>>You have simply tried to slide one past us by using a form where the
>>determiner 'his' and the possessive pronoun 'his' are isomorphic.
>I may be *wrong* in some analysis I propose,
>but I don't try to "slide" things "past" anyone.
>Just the contrary, I tend to bring up counter-examples
>even to my own beliefs!
>People sometimes try to get me to shut up when we are on the same
>team, precisely because I do that, because they don't want me to
>present a more nuanced or balanced case.
>I always try to see all sides and especially do not want to overstate
>and have to retract something later!

I will simply let this stand on its own merits.

Bob Whiting
whiting at cc.helsinki.fi