reflexives in subject position

[vanvalin at ACSU.BUFFALO.EDU]

Toba Batak provides a good example of a reflexive in subject position.
Toba Batak is VOS and has the Western Austronesian subject-only
constraint on extraction and other phenomena; only the final NP can be
questioned, omitted, etc.  Consider the following examples, from
Schachter 1984, Shugamoto 198.:

(1) a. Mang-ida    si   Ria  si      Torus.
                     ACTive-see PNM       PNM                   ŒTorus sees Ria.¹

           b.   Di-ida     si Torus si Ria.
                     PASS-see
                     ŒTorus sees Ria.¹
           c.   Mang-ida si Torus si Ria
                     ACTIVE-see
                    ŒRia sees Torus.¹
           d.   Di-ida  si Ria si Torus.
                     PASS-see
                     ŒRia sees Torus.¹

(2)     a.      Mang-ida   diri-na       si Torus.
                      ACTIVE-see self-3sgPOSS                            ŒTorus sees himself.¹

           b.   *Di-ida      diri-na      si Torus.
                      PASSIVE-see self-3sgPOSS             *ŒHimself sees Torus.¹
            c.  Di-ida      si Torus diri-na.
                      PASSIVE-see            self-3sgPOSS   ŒTorus sees himself.¹

            d.  *Mang-ida si Torus diri-na.
                       ACTIVE-see        self-3sgPOSS       *ŒHimself sees Torus.¹

The voice system, briefly illustrated in (1), signals whether the O
(internal argument) and S (external argument) are actor and undergoer,
respectively; if the prefix is mang-, the S is actor and the O undergoer,
and if the prefix is di-, the O is actor and the S is undergoer, as the
translations show.  Reflexives are illustrated in (2), the relevant
examples are (c, d): the reflexive is in subject (external argument)
position in these two examples, and when the voice is marked by di-, the
sentence is perfectly grammatical.  If one takes the ang-NP in Tagalog to
be the subject, as e.g. Kroeger argues, then it too has constructions
like (2c) in which the antecedent is marked by ng and the reflexive by
ang.

Robert Van Valin

All three accounts correctly predict the grammaticality of (9a) and the
ungrammaticality of

***************
Robert D. Van Valin, Jr.                  Tel 716 645-2177, ext. 713
Professor & Chair                         Fax 716 645-3825
Department of Linguistics
609 Baldy Hall
State University of New York at Buffalo
Buffalo, NY 14260-1030   USA              VANVALIN at ACSU.BUFFALO.EDU