LINGUIST List 5.444

Tue 19 Apr 1994

Disc: Generics

Editor for this issue: <>


  1. Knud Lambrecht, Re: 5.436 Generics

Message 1: Re: 5.436 Generics

Date: Mon, 18 Apr 1994 13:20:35 Re: 5.436 Generics
From: Knud Lambrecht <>
Subject: Re: 5.436 Generics

Like Julie Auger, I find the discussion on generic c,a fun and worth
pursuing. Julie and I have discussed the matter of the status of bound
or atonic subject pronouns for a while (with each other and in publications).
I think Julie has demonstrated convincignly that Quebec French (QF) has gone
a step further than European French (EF) if that is the right term) in that
the subject pronoun has become obligatory in more contexts in QF than in EF.
The quantifier issue is especially interesting. I can't see any French
speaker saying

(1) Quelques linguistes ils ont beaucoup parle' des pronoms en franc,ais.
 some l. they have much spoken about pronouns in French

instead of, either, in written French

(2) Quelques linguistes ont beaucoup parle'....

or, in spoken French (pretty much obligatorily instead of (2))

(3) Y a quelques linguistes qui ont beaucoup parle' ...

while apparently (1) is just fine in QF. I know (thanks to the GARS
team in Aix-en-Provence) of a few attested examples like the following

(4) Un monsieur il a dit que...
 a man he has said that

where the quantifier is the indefinite article. These examples all seem
to be from children (sentences like (4) sound distinctly childlike to me),
which is of course water on Julie's mill. Maybe in another hundred to
five hundred years everyone will say (1) without blushing.

Everyone seems to agree, at least since H. Paul, O. Jespersen, Vendrye`s,
that French atonic subject pronouns have become more and more tightly
connected with the verb over the centuries and are used more and more
obligatorily in cooccurrence with a full lexical NP; there also seems to
be general agreement that the sequence pro+V (where pro = atonic pronoun)
is a constituent of type V (plus any bars or boxes you might want to add)
rather than a sequence of type [NP + V].

(There is also the issue of whether we should continue speaking of "clitic"
pronouns in French, rather than of atonic/bound pronouns or (subject)
prefixes. I have always thought (and argued in writing) that the term
clitic is a misnomer for "je, tu, il" etc. and I was pleased to hear
Julie Auger and Rich Janda at the last LSA meeting arguing just that. I
think (but have no proof) that the use of the term clitic for French bound
pronouns got established because Dave Perlmutter, who was working on Slavic
clitics, decided to apply the term to French in his seminal early book on
deep and surface structure constraints (I forgot the the exact title).
The term "clitic" was convenient at the time because it was theoretically
compatible with the transformational view: a clitic is a word that sticks
to another word but it doesn't really belong there because by its deeper
nature it really is an independent constituent that fits a syntactic
position in a tree structure. We now have a pretty good understanding
of what clitics are (thanks to Zwicky and others) and it is easy to show
that the French pronouns don't pass the tests. We now also have
(fortunately) plenty of theories in which grammatical relations are
not necessarily expressed configurationally. Let's get rid of the term
"clitic" in the analysis of French!)

As I see it, the real theoretical issue is not whether modern French
subject pronouns should be called pronouns or prefixes and whether

(5) Les Romains ils+sont fous!
 the Romans they are crazy

is an instance of left-dislocation or of the canonical NP VP sequence.
To the extent that we are dealing with a squishy phenomenon I don't
think we have gained much by opting for one term rather than another.
There are lots of tests for distinguishing free from bound morphemes,
and French atonic pronouns clearly are of the latter type. But to my
knowledge noone has a solid enough formal definition of the category
"pronoun" that would allow us to decide that in "ils+sont fous" ILS is
no longer a pronoun.

The really interesting question is, in my opinion, WHY linguists are so
keen on drawing the line between French pronouns and prefixes and between
canonical and dislocated sentence patterns and what they do once they
have decided to put "je" etc. into either category. I have a strong
suspicion that the main reason for discussing the issue, in generative
grammar at least, is because of the fundamental assumption, which
generative grammar has directly taken over from "traditional" grammar,
which has it from Aristotle, who probably has it from people before him,
that a real good sentence has the form "Socrates currit" (well, "Sokrates
trechei" for Aristotle) "The farmer kills the duckling" or "The man hit the
ball", with a full lexical subject NP and a predicate phrase after it (or
before it, as the case may be). Ancient grammarians called this sentence
type the "oratio perfecta", the type expressing a "complete thought". In
the case of Aristotle and other philosophers and logicians this assumption
does no harm to the extent that these people weren't interested in sentence
structure but in sentence meaning and human thought. In the case of
generative grammar, the assumption has had deep consequences and has
influenced syntactic theory without anyone being clearly aware of it (or
at least without stating it explicitly).

The generative grammarian's desire has been to reduce as much as possible
all sentences to the NP-VP type (with or without AUX or INFL or AGR or
whatever), where the NP has the function of expressing the subject argument
of the verb. Deviations from this type have typically been addressed in such
terms as "clitic doubling" (you say the same thing twice) or "pro drop"
(you drop an essential element from the surface, but deeper down it's really
there). The reason for this, I think, is the preconceived idea of what a
sentence is. In the case of a large number of modern syntacticians this
idea is crucial, because grammatical relations are defined structurally
and because the same grammatical relation is assumed to be instantiable
only once for a given predicate. So, for example, to say that in "Les
Romains ils sont fous" both "les Romains" and "ils" are subjects is a
no-no for most of us, because... well because that's the way it has always
been. We take our task to be that of figuring out which of the two is
the real subject, so we can close the debate. But maybe we just aren't
asking the right questions.

I hope to have opened a can of worms with these remarks. I apologize
for their length.

Knud Lambrecht
UT Austin
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue