LINGUIST List 5.1090

Fri 07 Oct 1994

Disc: French clitics

Editor for this issue: <>

Directory

1. Richard Hudson UCL, French clitic pronouns
2. John E. Koontz, Re: 5.1088 French clitics

The discussion so far doesn't seem to have mentioned Y and EN. Are these
inflectional affixes too? What about examples like (1)?

(1) Jean en mange beaucoup. `John of-it eats much' = `J eats a lot of it'

The point of this example is that EN seems to be linked to BEAUCOUP rather
than to MANGE.

Dick Hudson
Dept of Phonetics and Linguistics,
University College London,
Gower Street,
London WC1E 6BT
uclyrahucl.ac.uk

Message 2: Re: 5.1088 French clitics

In response to the joint response by Julie Auger and Philip Miller to the
recent discussion on LINGUIST concerning the status of clitic pronouns in
French, ...

I thought that it might be appropriate to comment on some of the material
they present in light of some exposure to North American languages which
have always been analyzed as being of the type that French and/or Italian
may be becoming. In particular, regarding the objections they cite from
Larrivee:

> Larrivee (LINGUIST 5.1074) suggests that this analysis is problematic
> on the basis of various phenomena, concerning (i) the possible
> morphological complexity of clitic pronouns;

This seems to be no objection in itself to a form being regarded as affixal.

In Siouan languages there are often nearly systematic vowel differences
between the otherwise similar agent and patient pronominal forms for a
particular person. These are presumably relics of what amounts to case
marking, but they are not usually systematic enough to have led anyone into
treating them as such in a synchronic context.

Looking further, in several Siouan languages the inclusive pronominal affix
consists of two separate parts, separated from each other by another
morpheme of a different function, e.g., the Omaha-Ponca inclusive patient
pronominal has variants wa ~ wa ... a ~ awa. The Winnebago inclusive
patient pronominal has variants wa~a~g ~ wa~a~g ... a. [V~ is a nasal
vowel; x ~ y are variants! JEK] Although there is no functional load to
the two parts individually, clearly here we have pronominals with very clear
morphological structure, and even, I might say, morphosyntax.

In another example, in the formation of dative forms from non-dative forms
it is often the case that the vocalism of the pronominals is shifted. In
particular, in Omaha-Ponca, several pronominals, and a locative prefix, if
present, may be affected with the set affected being determined by the
pattern of the underlying stem (what locatives, if any, are present, etc.).

In Omaha-Ponca I prefer to think of this as sort of superfix or
``automorpheme'' applied to the inflected form as a whole, but, in effect,
especially in less exuberant situations like that in Dakotan, all of the
affected pronominals and locatives have separate dative formations, and so
some morphological structure.

Probably not relevant here is the circumstance that the plural category is
generally marked separately from person in Siouan: the pronominal is
prefixed or infixed; the plural follows and is often what is called an
enclitic in Siouan circles, though I would not care to vouch for its
qualifications under a particular theory.

> (ii) the mobility of clitic pronouns;

The point of insertion of the pronominal in a typical Siouan language is
determined by the stem form, and is constant for a particular stem, barring
some personal variant in identifying stems' forms. Some forms prefix all
pronominals, some prefix the inclusive and infix the others, some infix all.

In Dhegiha languages like Omaha-Ponca, pronominals of the form V (all first
persons or inclusives) jump to the front of any derivational or inflectional
/wa/ prefix on the same stem (sometimes a verb form may include several
coordinate or subordinate/superordinate stems, so we have to be clear that
the wa must be on the same stem).

In typical Algonquian languages the position of pronominals varies from verb
order to order for the same verb. Some orders prefix; some suffix. Orders
correspond generally to moods in a somewhat extended sense, e.g.,
interrogative, dependent, etc. Some languages develop one pattern at the
expense of others; most have traces of both. This variation from order to
order seems exactly comparable to the circumstances that govern variation in
placement of pronominals in French: indicative, interrogative, imperative.

> (iii) the fact that subject clitics are restricted to subject function.

Siouan languages provide at least two series of pronominals, one for
"agents" (in a nontheoretical use of the term), and one for "patients"
(idem). As is fairly universally known, some intransitive verbs use agent
affixes/concords for subjects, and some patient pronominals, but if a verb
takes both agent and patient pronominals, the agents are usually the
subjects. Bearing in mind that third person is mostly unmarked, il, or,
better je, is subject if it is there to exercise the option.

The unusual cases involve small classes that take two patient pronominals
(e.g., in Dakotan) or that have an additional 3rd person argument. Third
person arguments are generally unmarked, so these additional arguments are
not categorized (case marked) by a pronominal. Examples would be verbs like
Omaha-Ponca gi'uda~ `for X to be pleasurable to one; for one to enjoy X',
where it seems that X can't be in the first or second or inclusive person,
though one can be.

I think that is is safe to say that languages which affix multiple
pronominals filling multiple case roles in lieu of putting case marking on
the NP have about the same sort of anomalies and regularities in case
marking as those that do. Usually subject markers mark what we'd like to
call a subject on other criteria, but not always.

As Auger and Miller say:

> Finally, we fail to understand the relevance of the third argument ("it
> ['il'] is restricted to subject function")

Yeah! Moi aussi.

> [There are] ... numerous types of data arguing in favor of lexically
> attached inflectional status. ...

All points adduced by Auger and Miller well describe the behavior of affixal
pronominals in Siouan and other language families of (at least) eastern
North America. It is true that the patient pronominals generally seem more
easily divisible and less intricately entwined with everything else. This
is generally assumed to be because they are newer as affixes, but perhaps
French teaches us that this isn't the only possibility.

John Koontz