LINGUIST List 5.1132

Mon 17 Oct 1994

Sum: French Clitics

Editor for this issue: <>


  1. , French clitics

Message 1: French clitics

Date: Fri, 14 Oct 94 11:44:38 EDFrench clitics
From: <>
Subject: French clitics

There's been a nice debate following my posting a while ago on the treatment of
French clitics. I'll try here to summarize the messages I have received, and to
disamnbiguate my owm position.

It seems that treating French clitics as "lexically-attached" elements, as has
been proposed (this position has been argued by Miller and Auger on LINGUIST,
vol. 5.1088) derives from a universalist perspective: French sentences with
clitics and dislocations apparently ressemble closely "exotic" languages Ss
with verbal case morphemes and real NPs. John Kootz intervention (LINGUIST, vol
 5.1090) went in that direction; so did Jacques Guy, quoting data from Swahili.
I have received a similar comment from Stephen P Spackman. I don't have the com
petence to discuss the data from any of those exotic languages, and while those
comparisons might be enlightening, I do think we should look deeply into a lang
uage's data and formulate an hypothesis before jumping to other languages.
I otherwise fail to understand the motivations for proposing the "lexically at-
ched" treatment for French clitics. I also missed the distinction between agree
ment markers and other lexically attached elements in Quebec French; the crite-
ria for this distinction are not obvious to me. I further see several problems
with this analysis. (So does Benoit Robichaud at Montreal U, who is writing a
thesis showing that the existing arguments for an inflexional treatment
of French clitics "cannot be seriously used to promoted such a treatment".)
One syntactic problem that occured to me is multiple dislocations: If clitics
are morphemes and the dislocated NP is case endowed, what happens in multiple
dislocation contexts such as "Miranda, la coiffeuse, elle, elle pritend que
rifirendum ne passera pas, l'imbicile!"?? There would be further problems with
these Ss:

 1. Le docteur, il lui a parli.
 2. Le docteur, il a parli avec lui.
 3. Le docteur, il a parli avec ce sale imbicile toute la journie.

As far as I can see, they have identical structures. Now, if we assume that the
dislocation is case endowed in (1), I would suppose the same happens in (2) and
(3); but then, as other constituents are already case endowed in (2) (avec lui)
and in (3) (avec ce sale imbicile), this would lead to two attributions of the
same case, violating case requirements (or something along those lines). I have
further already mentionned that this treament would have to suppose that "il
lui a parli" has no case endowed NPs, since clitiks don't qualify in this treat
ment for case (I suppose); and yet, the sentence is perfectly grammatical. So
French would have to be pro-drop. That doesn't seem to be desirable for French.
NPs cannot just be dropped in any circumstances; in general, either a full NP
or a clitik has to be there for the sentence to be acceptable.
A second set of problems are mophological in nature; are clitics words or mor-
phemes? If they are words - as Miller and Auger themselves recognize by calling
them pronouns - then treating them through "lexical attachement" amounts to say
that the lexicon is deriving syntactic units, which mixes unwelcomely levels of
analysis. I have tried to indicate on my first posting that clitics are words:
they have a part of speech, in some case a grammatical morphology, they are
mobile - although not free-ordered, words can be inserted between them and
their verbal head ("Ne pas dire", where "ne" is the clitic), and they can have
a function.

Most of Miller and Auger's illustrations to defend their standpoint revolves
around the morphological, positionnal and syntactic complexity of clitics; this
is however not an excuse to treat them via the lexicon; many syntactic phenome-
non are extremly complex and anyway, the lexicon isn't a grammatica cloaca.
I hope I have made my points sufficiently clear and that the proponents of this
treatment will be able to make it clear to me and to all other LINGUIST fellows
what the general rational behind their hypothesis.
Cordialement, Pierre.

Pierre Larrivee
Departement de langues et linguistique, Universite Laval, Quebec, Canada,
G1K 7P4
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue