To: linconf@tamvm1.tamu.edu
Subject: Reply to Seely from Lidz
Date: Mon, 28 Oct 1996 10:51:39 -0500
From: Jeffrey Lidz <lidz@louie.udel.edu>
Message-Id: <9610281551.aa20527@stimpy.eecis.udel.edu>
In-Reply-To: <01IB2E87HAUQ95PJZH@EMUVAX.EMICH.EDU>
Dan Seely brings up two issues with respect to my paper. I will
address them in turn.
First, he asks how Condition R gives the strict-sloppy idendity
distinction. I think this is to be handled in a straightforward way.
A lexically reflexive predicate, when copied (or whatever the
appropriate description of the ellided vp is), will maintain its
reflexivity. That is, lexical information can not be changed via
whatever operation produces the ellided vp. Because the ellided vp is
forced to be reflexive, only the sloppy reading will be allowed. The
exact analysis of the near-reflexive cases depends on precisely how vp
ellipsis is to work. However, the basic observation is simpy that the
anaphors which allow a near-reflexive interpretation are independent
from their antecedents in a way that anaphors occurring in lexically
reflexive predicates are not. However this independence is to be
encoded, it follows that it is maintained in the vp ellipsis.
Relatedly, Seely asks whether vp ellipsis sentences will allow the
overt vp and the the elided vp to differ with respect to statue
interpretations. Thus, can (i)
(i) John defended himself better than Peter
have the following interpretations:
(a) John defended a statue of himself better than Peter defended
himself.
(b) John defended himself better than Peter defended a statue of
himself.
I find these readings hard to get (perhaps a dutch speaker could
comment on their status in dutch). It's not clear to me whether the
absence of such readings is due to the grammar or to discourse
conditions on vp-ellipsis. This may be like trying to view two necker
cubes at once and seeing them differently. While each form is
ambiguous, when viewed in parallel, they must be given the same
interpretation.
The second issue Seely brings up is the nature of variation in the
extent of lexical reflexivity allowed by a language. He comments tht
Dutch and Kannada appear to be different in that only a small subset
of verbs in Dutch can be lexically reflexivized while any verb in
Kannada can be lexically reflexivized through the affixation of the
verbal reflexive. Since one involves a morphological operation and
the other involves a lexical operation, aren't these languages
qualitatively different?
On the surface, this might seem to be the case. The interesting
question from this perspective is whether it is true that the function
of verbal reflexive is to morphologically reflexivize the predicate
(as argued in Lidz (1995)). The answer seems to be no. We arrive at
this answer by examining the full range of uses of the verbal
reflexive. Crosslinguistically, verbal reflexives are used in a range
of constructions which are not semantically reflexive. Such
constructions include inchoative constructions and certain possessive
cases, typically involving body-parts or inalienable possesion. The
following examples are from Kannada, but similar facts are found in
languages as diverse as Diyari (Pama Nyungan), Finnish (Finno-Ugric),
Imbabura (Quechua), Lithuanian (baltic), Spanish (romance), Yakut
(Turkic) and Yavapai (Yuman).
(ii) baagil-u much-i-koND-itu
door-NOM close-PP-REFL.PST-3SN
'The door closed'
(iii) hari kannu-gaL-annu tere-du-koND-a
hari eye-PL-ACC open-PP-REFL.PST-3sm
'Hari opened his eyes'
That these predicates are not semantically reflexive is clear. The
fact that these uses of reflexive morphemes are found consistently in
the languages that have them, makes it hard to offer any kind of
homophony story for them. Discussion of the precise analysis of why
these so-called reflexive elements are used in this range of
constructions cross-linguistically might take us too far off the track
(though i'm happy to give the analysis if the conferees are
interested), but the important thing to observe is that since these
morphemes are used in non-reflexive constructions it seems odd to say
that they reflexivize the predicate that they attach to. Rather, they
are the consequence of a particular structure which happens to arise
when a predicate is lexically reflexive (among other cases). Thus,
Dutch and Kannada are alike in that there is a lexical operation of
reflexivization (which happens to give rise to _koND_ in Kannada), but
differ only in the extent to which this operation can apply in the
lexicon.
-Jeff Lidz