so-called "pro-drop" languages

Daniel L. Everett dan.everett at MAN.AC.UK
Tue Feb 10 10:22:39 UTC 2004


Folks,

There are a couple of issues in this discussion that result from
underlying issues of some importance. The first, which have been
commented on before here, is the role of terminology in typological
discussion/analysis. The second is the function of referent-tracking
and the role of pronouns, clitics, and agreement in expressing this
function.

With regard to the first issue, the term  'pro-drop', as many have
pointed out, whatever its original role in Luigi Rizzi's development of
one of the first supposed 'parameters' of Principles & Parameters
theory, has become a mnemonic device for talking about languages that
omit subjects and other arguments. The original idea behind pro-drop
was that if a language's inflectional system was 'sufficiently rich'
then pro-drop would be licensed. So, French and English lack it,
because their agreement morphology is 'too weak' but Spanish and
Italian have it because their agreement morphology is strong. This neat
view had to be abandoned as soon as people looked East and noticed that
pro-drop is common in Asian languages that lack agreement morphology
entirely. So a new hypothesis suggested that pro-drop was possible when
a language's agreement morphology was 'consistent', i.e. either
inflection was manifested with all forms (Italian) or none (Mandarin).
English has agreement with some forms ( -s), but is inconsistent.
However, the number of auxiliary hypotheses required to sustain this
enterprise has grown to such an extent that to many it has lost its
interest, if such an approach ever had any interest. But the term has
remained to refer to languages which allow argument omission. I don't
see any problem with that, even though most languages fall into this
category, so long as we realize that the best typology is not based on
trying to determine how closely this or that language fits Procrustes's
term for this or that, but on the best analysis of this or that
form-function pairing in a particular language.

With regard to the second issue, there are interesting relations
between pronominals in referent-tracking which are expressed in
implications e.g. the following two from my book, Why there are no
clitics:

(i) If object agreement --> subject agreement;

(ii) If subject clitics --> object clitics

This holds for most nominative-accusative languages. To revise it to
account for a much wider range of languages, at least in principle, it
would have to be stated in term of Privileged Syntactic Argument  (in
the RRG sense), e.g.:

(i') If non-PSA agreement --> PSA agreement

(ii') If PSA  clitics --> non-PSA clitics

This means that if a language has verb agreement with an argument, it
will first prefer the argument that is the syntactic/semantic pivot
(PSA will vary from language to language, but it will roughly
correspond to nominative argument in nom-acc languages say).

This follows in the model developed in the reference above because verb
structure will favor general specification for the PSA over the non-PSA
argument in structure. Under this view, the principal difference
between pronouns, clitics, and affixes is structural:

(iii) [Xpronominal features] = pronoun

(iv) [X-root[Xpronominal features]] = clitic

(v) [X-root pronominal features] = affix

That is, a free-standing/morphologically separate set of pronominal
features will be viewed by the phonology as a pronoun. A set of
pronominal features adjoined (and thus in no necessary  semantic
relation) to a host will be treated phonologically as a clitic. And a
set which is a structural sister to the root will be treated as an
affix. The iconically closer relation in (v) is most appropriate for
tracking the PSA and this derives (i). The more distant relation (iv)
is  most appropriate for non PSAs and thus derives (ii).


Ultimately, of course, whether readers find my own proposal here at all
interesting is not the issue. I state this only by way of illustration.
The issue, which most on this list recognize, is that form of referent
tracking is best understood in conjunction with function. Function is
not always necessary to understanding form and is rarely, if ever,
sufficient. But it is always useful and often necessary.

Dan


---------------------------------------------
Daniel L. Everett
Postgraduate Programme Director
Professor of Phonetics & Phonology
Department of Linguistics and English Language
University of Manchester
Manchester M13 9PL UK
Fax: 44-161-275-3187
Phone: 44-161-275-3158
http://ling.man.ac.uk/info/staff/de/

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: text/enriched
Size: 4540 bytes
Desc: not available
URL: <http://listserv.linguistlist.org/pipermail/lingtyp/attachments/20040210/8d92a493/attachment.bin>


More information about the Lingtyp mailing list