Primary object languages & pronouns

Matthew Dryer dryer at BUFFALO.EDU
Wed Apr 23 17:00:27 UTC 2003

Contrary to what Dan Everett says, the approach if my 1986 paper (Primary
objects, secondary objects, and antidative) would not analyse Wari' as a
language with obligatory antidative.  In fact, Wari' is exactly the sort of
language I proposed the notions of primary object and secondary object for.

Although I still believe that the notions of primary object and secondary
object are a useful way of viewing languages like Wari', I no longer think
the notion of antidative is a useful notion for describing languages like
English where we find two constructions for semantically ditransitive
clauses.  Rather, the two constructions should simply be analysed as
alternative syntactic frames, not related by rule.  However, even in the
framework of my 1986 paper, I would not analyse any language as having an
obligatory antidative, since one of the primary purposes of the notions of
primary and secondary object is to provide a way to avoid describing
languages like Wari' as involving an obligatory dative rule, a popular
approach to such languages both within relational grammar and outside
relational grammar (e.g. in the work of Givon).

I also would argue that Dan's use of the term 'patient' is possibly
Eurocentric.  He applies this term both to the single object in
monotransitive clauses ('I hit HIM') and to the thing other than the
recipient in ditransitive clauses ('I gave Mary THE BOOK').  This makes it
look like there is a mismatch between semantic roles and object categories
in a language with a primary object - secondary object distinction.  While
I think that the direct object - indirect object distinction aligns more
closely with semantic roles than the primary object - secondary object
distinction, I think one should be careful about using semantic labels like
'patient' in this way.  It is a bit like saying that an ergative language
uses the ergative case for transitive agents and the absolutive case for
intransitive agents.

Finally, it isn't clear to me what Dan is asking.  When he says that the
recipient cannot be a pronoun, does he mean that it cannot be an
independent pronoun, but must be realized entirely by the verb morphology?
There are certainly languages where this is true, not only for recipients,
but also for subjects and/or objects.

Matthew Dryer

--On Wednesday, April 23, 2003 3:33 PM +0100 Dan Everett
<dan.everett at MAN.AC.UK> wrote:

>>> Folks,
>>> Wari', Amazonian, shows agreement and syntax typical of what Dryer
>>> (1986) has treated as obligatorily anti-dative or Van Valin & La
>>> Polla (1997, 270ff) treat as a 'primary-object pattern'. That is, in
>>> simple transitive clauses the AGENT and PATIENT both trigger
>>> agreement on the verb. In di-transitive clauses, however, it is the
>>> RECIPIENT/GOAL which triggers/governs agreement on the verb. The
>>> PATIENT argument in these clauses appears as the object of a
>>> preposition (Wari' is V-IO-O-S). Pronouns in Wari' may not bear the
>>> RECIPIENT role. My question is this: Are there other languages like
>>> this? Some hypothetical examples of what I mean are:
>>> (1) a. I hit him.
>>>      b. Bill hit me.
>>>      c. Mary saw you.
>>> In 1a-c, the verb would agree in Wari' with both subject/agent and
>>> object/patient, regardless of whether these are NPs or pronouns -
>>> they may also be zero, but the verb will still show agreement.
>>> (2) a. I gave Mary of the book. (I gave the book to Mary) - VERB
>>> agrees with 'Mary' and 'I'.
>>>      b. *I gave her of the book. (Even though the verb agreement will
>>> be for 1 person singular and 3 singular feminine)
>>> Again, does anyone know of other languages with this pattern?
>>> -- Dan
> ********************
> Daniel L. Everett
> Professor of Phonetics and Phonology
> Department of Linguistics
> University of Manchester
> Manchester, UK
> M13 9PL
> Phone: 44-161-275-3158
> Department Fax: 44-161-275-3187
> 'Speech is the best show man puts on' - Whorf

More information about the Funknet mailing list