Ergativity and objects in Spanish

[Diego Quesada]

On Sat, 14 Feb 1998, Jon Aske wrote:

> About the correlation between S and O and postverbal position and A and
> preverbal position in Spanish, I repeat I think this is a rather weak
> correlation.   In my experience, overt (full nominal) objects (O's) tend to
> be foci, as do about half of overt intransitive subjects (S's).   Not so
> much overt A's, however.  This explains the fact that overt S's are
> postverbal more often than overt A's are, since foci are predominantly
> postverbal in Spanish.
        Having made no counts myself I have nothing to say here. Still,
just on paper -that is, regardless of the information structure of the A,
S, and O referents- the fact that one-argument structures tend to be
postverbal while two-argument structures tend to be SVO matches
-accidentally if you like- the ergative pattern. This, I assume, was
Clancy's original proposal. Now, I'd be surprised if such a pattern were
recurrent, frequent, and likely to be categorically predicted.
For one, as Jon suggests:
        > word order is not a mechanism for coding grammatical
        > relations in Spanish at all.
Second, Spanish, as many other languages, tends to make a basic
distinction between the word order of sentence at discourse-onset
(thetic, in Lambrecht's 1994 terminology), with full (overt) NPs, and the
rest; the former do tend to be S/A-initial; the others... well, it takes
all kinds. This fact, rather than supporting Clancy's claim, speaks for a
discourse-determined word order pattern, as determined basically by topic
continuity. In my work on Teribe, Rama, and Boruca -all three Chibchan
languages of Central America- a similar pattern appears (only that the
word-order types in running discourse are not as flexible as in Spanish):
SOV discourse-initially, OSV/OVS -almost- elsewhere. Thus, despite the
arroz-con-mango style of word order in Spanish, it seems to be in line
with what occurs in many (?) other languages of different affiliations.


> b) Subjects (and all other settings) may also be postverbal, not because
> they are foci, but because they are topics in an assertion with a very
> salient foci (cf. exclamations, e.g. *Siempre viene Juan los viernes!* "Juan
> ALWAYS comes on Fridays"; *Cuando viene Juan?* "when does Juan come).  These
> "antitopics" do not receive an information accent, such as foci and
> left-dislocated topics do.
        Is the double-starring signalling ungramaticality/unacceptability
or what? In any case, I'd tend to say the first of those sentences with
Juan at the left (Juan siempre...). The other one sounds OK.


> >         (1) La amenaza de confiscarles los indios...
> [the threat of confiscating the Indians from them]
>
> (For those who are not familiar with the language, the interesting thing
> here is that there is no (so-called personal) "a" before "los indios".)
>
> >         (2) La llegada tan esporadica de los barcos a los puertos de
> >             Honduras no solo afecto a las exportaciones, sino tambien
> >             a la importacion de productos europeos.
> [the very sporadic arrival of ships to the Honduran harbors didn't only
> affect [a] (the) exports, but also the imports, of European products]
>
> (The interesting thing here is that the (inanimate) object of affect has the
> "a".)
>
> Surely these are interesting cases in that they don't follow the traditional
> rule, according to which human direct objects bear the A and non-human
> objects don't.
        There you go! I have lots of cases -from both oral and written
language- that show that presence/absence of 'a' has every time less to do
with the human/animate status of the direct object referent. Typical in
this respect are cases of left-dislocation where the D.O. is not
'a'-marked:

        (3) Juan lo vi ayer
            Juan 3sg.masc see-perf yesterday

***WITHOUT*** a pause between Juan and the rest of the sentence (this is
part of a long story I won't go into). Just as there are cases of
inanimate D.O.'s which are 'a'-marked:

        (from Suner 1989)
        (4) Lo van a empujar al omnibus.
           3sg.masc go-3pres.pl a-3def.sg bus

Besides, cases as those in (5a-d) are more than common:

        (5a) Busque abogado si no quiere que lo guarden
             Look for-imper. lawyer if [2sg] neg want-2sg that 3sg.
                                                put-3subj.pl in jail
            'You'd better get yourself a lawyer if you don't want to end
                up in jail'

        (5b) Busque un abogado...

        (5c) Busque a un abogado si no...

        (5d) Busque abogados si no...

        (5e) ? Busque a unos abogados si no...

        (5f) ? Busue unos abogados si no...

Clearly, being human is not a necessary condition for 'a'-marking. The key
to the puzzle lies in that 'a' is not a marker of grammatical relations
anymore. The system is moving to a different pattern.

> I think there is another reason for not using the A in the first
> example, which would also be a reason for using the A in the second example,
> namely the fact that confiscar (confiscate) typically takes an inanimate
> direct object (merchandise) and that afectar (affect) typically takes a
> human direct object, and thus the absence vs. presence of the A is somewhat
> automatic (grammaticalized).
>
> The possibility of dropping the A in the first example would be due to the
> fact that the object of confiscate is typically some type of merchandise and
> merchandise is not marked with A, be it human or not.  (But if it wasn't
> Indians but, say, relatives, the personal A would not have been ommitted.)
        Could be.

> The addition of the A in the second example, which by the way doesn't sound
> all that great to my non Costa Rican ears, I can only see as an extension of
> the prototypical case in which the affected party is human.
        Could be too, thereby partially confirming the old rule of [+
human] referents, but which is giving way to a new shift in the language.

Diego