Message-Id: <9610281823.AA27053@dragon.emich.edu>
Date: Mon, 28 Oct 96 10:29:57 EST
From: Robert Hamilton <HAMILTN@VM.SC.EDU>
Subject: Reply to Zlatic from Hamilton
To: Linguistics Conference <LINCONF@tamvm1.tamu.edu>
In-Reply-To: <Pine.SOL.3.95.961027011855.20387B-100000@piglet.cc.utexas.edu>
This is in response to Larisa Zlatic's question about my paper
"Reflexivity, Anaphoricity, or Polymorphemicity?" On Sun 27
Oct, Larisa Zlatic wrote the following:
>In your paper you state that "monomorphemic ana[phors] are subject to
>neither Conditin A (because they are not phrasal) nor Condition B (because
>they are not ppro[personal pronouns]" (p. 8). In order to explain the
>binding possibilities of monomorphemic reflexives in Scandinavian
>languages (which also have polymorphemic reflexives), you offer language
>specific 'predicate marking systems'. Does this mean that languages that
>have only monomorphemic reflexives (e.g. Slavic languages) are exempt from
>your universal binding theory, i.e. each such language has its own set of
>binding rules? If so, I don't understand what is 'universal' in your
>binding theory?
>
>--Larisa Zlatic
Let me first thank Larisa Zlatic for this question.
In response, it is important to note what "exemption" from
a binding condition (or multiple binding conditions) does NOT
mean:
(1) Exemption from a binding condition (or conditions) does NOT
mean that the binding condition(s) yield no empirical
prediction(s) regarding the binding patterns of the exempt
anaphor(s). For example, my formulation of the binding
conditions predicts that monomorphemic anaphors, which are
exempt from both condition A and B, are free to be both
locally and nonlocally bound (barring independent factors--
see below). This is a direct consequence of my formulation
of the binding conditions and has universal relevance
inasmuch as the binding conditions are part of universal
grammar. In this way, my binding conditions DO have
something to say about languages with only monomorphemic
anaphors; namely, that the basic pattern in such languages
(again, barring independent factors--see below) will be
the allowance of both local and nonlocal binding of anaphors.
(2) Exemption from a binding condition (or conditions) does NOT
entail that a language-specific binding system must be
invoked. In this regard, note that I posited a language-
specific system for Scandinavian even though the languages
in question have anaphors that are subject to Condition A.
The language-specific system is viewed as overlaying the
universal system and further restricting the binding
from what is allowed on universal grounds. I posit no
necessary connection between exemption from the binding
conditions and the presence of language-specific restrictions
on binding. (Let me note in passing that I find Lidz's
universal account of antilocal anaphors to be an appealing
alternative to my own language-specific account of these
phenomena in Scandinavian; however, I am not sure that his
Condition R can fully account for the Icelandic data I
presented in my paper. In general, Lidz's account seems
entirely complementary to my proposed universal
binding conditions.)
(3) One reason that it might (mistakenly) appear that my analysis
presents no universals with import for monomorphemic-anaphor-
only languages is that such monomorphemic anaphors may seem
to fall into a theoretical no-man's land in my analysis.
This is not really the case, however. The binding
conditions as I conceive of them are simply restrictions
that apply to certain structural configurations; these
conditions do NOT themselves define/constitute structural
binding as such or otherwise exhaust all that needs to be
said about structural binding (Zlatic did not necessarily
imply that I meant this as I've stated it here, but this
is something that may not have been clear in my paper
because I focused there strictly on a proper
characterization of the binding conditions). The binding
conditions are only one part of a larger binding theory,
which must characterize exactly what "binding" is and which
should account for other effects on structural binding that
do not clearly derive directly from the binding
conditions themselves (e.g., subject-orientation, the
blocking effect, tensed/overt agr clause opacity).
In my analysis, "binding" is characterized as structure-
sharing of INDEX values between referential nominal-
objects. Binding Condition A limits the possible
coindexations when certain structural configurations
obtain, whereas Conditions B and C prohibit certain
coindexations. Apart from these restrictions, however,
any coindexation is possible as far as the binding
conditions are concerned. There are other restrictions
on structural binding, though, that arise from the
mechanisms of the binding process itself. I had little
to say about these mechanisms in my conference paper
because they had little bearing on my formulation of the
binding conditions, but I am currently developing a detailed
proposal of these mechanisms within an HPSG framework as
part of my dissertation research. Though I'm not quite
ready to subject my broader binding theory to public
scrutiny, suffice it to say for present purposes that
the broader theory has much to say about the nonlocal
and local structural binding of monomorphemic anaphors,
including an account of subject orientation and opacity
effects such as the blocking effect that arise in some
languages. In a nutshell, I propose that a lexically
unspecified INDEX feature in the LOCAL | CONTENT of an
anaphoric morpheme always triggers a corresponding
NONLOCAL | REFL feature that will be passed up the tree
until discharged by a corresponding TO-BIND feature (I'm
using Pollard & Sag's terminology here, though P&S fail
to recognize the relevance of NONLOCAL feature mechanisms
to the nonlocal binding of monomorphemic anaphors). My
account of subject orientation and the blocking effect
is then built from this basic architecture in ways that
I will not go into here.
To summarize, the main concern about my proposal
seems to have been that my account is not really 'universal'
because the binding conditions do not apply to languages
that have only exempt anaphors. This concern is unfounded,
however, because (a) my binding conditions DO make predictions
about the binding possibilities of exempt anaphors, (b)
exemption from a binding condition (or conditions) does NOT
entail a language-specific binding system, and (c) my
broader binding theory (not treated in this conference
paper) has much to say about the structural binding of
monomorphemic anaphors that is not addressed directly
by the binding conditions.
Robert Hamilton
University of South Carolina
hamilton@sc.edu