comments on "Questions": Pro-Drop

[D. N. S. Bhat]

This is with reference to Allen R. Kings comments on my earlier comments:

The point that I was trying to make, in my earlier comments, was that there
would always be some languages that remain outside the sphere of a given
typological classification.That is, typological classifications could be
hierarchic.

For example, it was thought at one time that the ergative-accusative
distinction can be applied to all languages such that any given language
can be regarded as using either an ergative system or an accusative system,
or a mix of the two.

However, R.M.W.Dixon, in his 1994 book on ergativity has come to the
conclusion that the typology can be applied only to one set of languages,
namely the ones which grammaticalize the representation of core relations.
There is an entirely different set of languages that do not grammaticalize
these relations (like Kannada, Manipuri, etc.) for which the typology is
unapplicable.

I think the Pro-Drop/non-Pro-Drop typology is similar. We can think of a
set of languages (set A) which organize their sentences as consisting of a
predicate and a set of arguments. Some of the type A languages (called
Pro-Drop) allow any of these arguments to be left unspecified in a
sentence, whereas some (called Non-Pro-Drop) need a minimum of core
arguments to be obligatorily specified in the sentence.

There could be a different set of languages (set B) which organize their
sentences in an entirely different fashion, as has been suggested by
several grammarians who have actually worked on such languages. The
sentences would consist of an obligatory predicate that includes
representations of core arguments in the form of pronominal markers. They
might contain additional optional constituents that are like the
complements of familiar languages in the sense that they might modify or
restrict the meaning of either the predicate itself or of one of the core
arguments that it contains.

The crucial difference between Set A and Set B languages is that, in the
latter case, there is no distinct category of nouns. Any word can occur as
a predicate, and can take all the inflectional markers that can occur with
a predicate. Further, the complements that occur with the predicate in a
sentence are only the "non-finite" forms of predicates. They do not involve
any special class of nouns, nor do they manifest any of the characteristics
that we can assign to noun phrases in Set A languages. Functionally, they
are not used for identifying the participants of the event that the
predicate denotes, as in set A languages. (Yimas, I think, has a distinct
category of nouns, and hence it does not belong to this set.)

If this characterization of set B languages is acceptable, (i.e. if we can
concede that there do occur such languages), then we cannot apply the
typological distinction between pro-drop/non-pro-drop languages to set B
languages in the way in which we define that distinction. That is, we will
have to concede that such languages remain outside the sphere of our typology.

It might be possible, further, that set B languages also show a typological
distinction in which some languages allow their predicates to occur without
specifying one or the other of the core arguments. (That is, the use of
pronominal markers in the predicate might be optional in some of them.) We
might perhaps find some that allow the subject marker also to be left
unspecified. Only such a typological distinction, if it does exist, would
be comparable with the pro-drop/non-pro-drop distinction of set A languages.

Our ultimate goal in establishing these typological distinctions is to
discover explanations to those differences. I do not think we can succeed
in this endeavor if we force a particular typology to all languages. We
should always be willing to keep aside languages which do not fit in with a
typology. In the above context, for example, the optionality of complements
(non-finite forms of predicates) in set B languages appears to be quite
different from the optionality of arguments (noun phrases) in set A
languages. Forcing the two into a single typology by identifying the
complements of the former with the noun phrases of the latter would
therefore have the effect of masking the possible underlying explanation.

D.N.S.Bhat
533, 15th Main, 5th Cross,
Saraswatipuram,
Mysore 570 009