LINGUIST List 5.1204

Sun 30 Oct 1994

Disc: French clitics

Editor for this issue: <>

Directory

1. Lambrecht Knut, Re: French clitics

Dick Hudson asks:

>If `object pronouns' are just inflections, why is it that they are treated
>like separate object NPs by the rule that makes a past participle agree
>with its object provided the latter precedes it?
>
> (1) Je l'ai e'crite (la lettre).
> I it have written (the letter)
> (2) La lettre que j'ai e'crite ...
> the letter that I have written (fem sing)
>
> (3) J'ai e'crit_ la lettre.
> I have written the letter
>
>I agree with Auger and Miller that genuine inflectional morphology should be
>invisible to syntax, contrary to much GB practice, but clitics do seem to
>have a genuine intermediate status as units which are part of syntactic
>structure but which are also treated like parts of other words (e.g. for
>what I imagine is an uncontroversial example, take "'re" in "We're ready").

As I see it, there is no reason to say that the object-agreement rule Dick
mentions entails that bound object pronouns are treated as separate object
NPs. (I know, "bound object pronoun" is a cumbersome expression, but since
I fully agree with Auger and Miller's analysis - much of which, incidentally,
was anticipated in Lambrecht 1981 (Topic, antitopic, and verb-agreement in
non-standard French. Amsterdam: Benjamins) - I think it is necessary to change
our terminological habits.) Depends what we mean by "agreement". If agreement
is seen as an obligatory formal dependency relation between two **sentence
constituents** (e.g. an NP and a verb) the observation is relevant. But (i)
there is no obligation to define agreement that way (we could say, e.g.,that
an inflection marker agrees with a referent, rather than a constituent -- this
is what I did in my 81 monograph), or (ii), if we want to keep the traditional
definition of agreement, there is no obligation to refer to Dick's data as
cases of agreement.

 I would analyse the object (agreement) marker on "e'crite" in (1) and (2)
in exactly the same way as the subject (agreement) marker in

(4) Je suis assis-e 'I'm sitting' (spoken by a female)

Not only is it established beyond reasonable doubt that 'je' is *not* an
NP (very nicely demonstrated already in Kayne's "French Syntax", 1975 - whose
analysis was foreshadowed by Benveniste's - except that Kayne, of course,
makes it into an NP in deep structure), but the subject 'je' is never marked
for gender at any "level" in French.
 The only way we can explain the subject agreement on 'assis-e' is by
allowing inflectional morphology to refer to pragmatics, which does not seem
like a revolutionary claim to me. The past participle (I could also have
taken an adjective like 'grand-e') "agrees" (or whatever term we want to use)
with the gender (or is it sex, I don't know anymore) of the person who happens
to be talking. (Needless to say that I do not believe that in so-called pro-
drop languages the subject marker on the verb agrees with a dropped pro, as
e.g. in Spanish "cre-o" 'I believe', where according to some the suffix -o
is triggered by an invisible subject NP "yo" etc; but of course, this is a
theoretical matter that can be decided only on theoretical grounds. - As far
as I remember, Julie Auger and Philippe Miller expressed views similar to
mine with respect to the "cre-o" issue, but I'm not quite sure anymore.)
 The fact that in (2) the NP 'livre' precedes the past participle, as the
head of a relative clause, is irrelevant, I think. What counts is that in
the *discourse* the referent be known to be feminine by the time the participle
is uttered. I don't think it's primarily the presence of the bound object
pronoun *before* the participle that triggers the agreement but the fact
that a pronoun *can be used* in the first place, i.e. the fact that the
referent can be assumed to be taken for granted by the addressee at utterance
time. There are some interesting facts about agreement in relative clauses,
by the way, which I think confirm this view: while in the restrictive relative
in (5) we get the expected agreement,

(5) La lettre que j'ai e'crite e'tait addressee au Pres'sident Clinton.
 the letter that I have written(FEM) was addressed to Pres. Clinton

in the c'est-cleft relative clause in (6) many speakers don't get it:

(6) C'est une LETTRE que j'ai e'crit, pas tout un livre.
 it is a letter that I have written, not a whole book

This is so, I believe, because of the presuppositional structure of the
cleft construction. The proposition that's known at the time the sentence
is uttered is 'speaker wrote something', the unknown part being that it was
a letter. The letter is the focus. So even though 'lettre' precedes the
past participle, there is no object agreement because the referent is not
pragmatically established as part of the proposition of which the participle
denotatum is part.
 The moral of the story: pragmatics can determine inflection (or whatever
we want to call it). By the way, the French facts mentioned by Dick are well
known to Africanists (see Benji Wald and Keith Allen on Swahili (I think),
Bresnan & Mchombo on Chichewa, etc.). There is object agreement in those
languages whenever the object is a topic, i.e. taken for granted as something
talked about in the discourse. When the object is a focus, no agreement.

Knud Lambrecht
UT Austin
(lambrecuts.cc.utexas.edu)