Date: Sun, 27 Oct 1996 00:23:57 +0200 (MET DST)
From: Tanya Reinhart <Tanya.Reinhart@let.uva.nl>
To: Linguistics Conference <LINCONF@tamvm1.tamu.edu>
Subject: Comments on Lidz
Message-Id: <Pine.SOL.3.91.961027002156.16738B-100000@cclsun01>
COMMENTS ON LIDZ FROM TANYA REINHART
Tanya Reinhart
Comments on Lidz' "The independence of syntactic and thematic binding"
Lidz argues that R&R reflexivity conditions should be modified in
a formulation he labels 'rule R', which deals only with lexical
(intrinsic) reflexives, and, hence, governs only the local occurrences
of SE anaphors ('zich). As for all other anaphora problems, including
the distribution of SELF anaphors, he suggests that the traditional
BT conditions A and B are retained as well.
At the conceptual levels Lidz argues that R&R are forced to resort
to stipulating the chain condition, which would be avoided by resorting
back to the full fledged BT conditions A & B. Note that the chain
condition R&R are assuming is the standard A-chain condition (which
excludes e.g. a chain with two case positions). We argue that this,
independently needed condition, is sufficient to capture also the
core issues of binding, and the residue which it can't capture is
captured by the reflexivity condition. Lidz suggest instead to keep
all three: a reflexivity condition for SE-anaphors, the full binding
conditions A and B and the chain condition (which is needed
independently of anaphora). Unless there is some body of facts that
can be captured by the three, and which is not also captured by just
two of these alone, it is hard to see the conceptual advantage of
assuming all three. So let us examine the empirical basis of this
move.
Lidz points out two problems which motivate his conclusions. Let us
consider them one by one.
LIDZ POINT 1: COMPLEMENTARITY.
Lidz says:
"...in sentences like (4). For R&R's theory, since there are no
reflexive predicates here, the binding theory should not apply. Thus,
R&R are forced to resort to an additional stipulation on properties
of chains in order to rule out (4c). While this condition does capture
the desired result, it fails to note the complementarity between
anaphors and pronouns which is the hallmark generalization of the
binding theory. It is quite surprising that the complementarity
between anaphors and pronouns that follows from R&R's binding theory
is maintained even in contexts in which their version of the binding
theory does not apply.
4) a. Max legt het boek achter zich
Max put the book behind self
'Max put the book behind him'
b. Max legt het boek achter zichzelf
Max put the book behind selfself
'Max put the book behind him'
c. *Max legt het boek achter hem
Max put the book behind him
'Max put the book behind him' "
COMMENTS:
In fact, (4c) is just as good in Dutch as its translation in English.
Specifically, (4c) happens to illustrate a well known problem with
the traditional BT-conditions A and B, namely that the complementarity
they entail between pronouns and anaphors often fails. (If you wish
to remain neutral about the judgment of the Dutch 4c, the English
equivalent is sufficient to make this point.) (4) is a context where
all three anaphoric forms are permitted. For the traditional BT
conditions, this is a mystery: They allow 4b, exclude 4c, and what
they could say about (4a), we will never know: If we defined 'zich'
as an anaphor, they will allow (4a), correctly, but by the same token,
they will entail that 'zich" has the same distribution as 'zichzelf'.
E.g. it would be allowed in (ia), and disallowed in (iia). Both results
are wrong. If we define it as pronoun, then why in (iiia) 'zich'
is allowed while a pronoun is ruled out by BT -B, will become the
mystery.
i a) *Max heard zich
b) Max heard zichzelf
ii a) Max heard [Lucie argue with zich]
b) *Max heard [Lucie argue with zichzelf].
c) Max heard [Lucie argue with him]
iii a) Max heard [zich argue with Lucie]
b) *Max heard [him argue with Lucie]
It takes a particularly creative imagination to observe any consistent
complementarity pattern in these combined examples, that could be
entailed by BT A, B. The traditional BT, with its pronoun/anaphor
typology was conceived for English, where there are only a pronoun
and a SELF anaphor. But even in this impoverished environment, it
predicts complementarity where it does not exist, as in (4b,c). In
the 'Reflexivity' framework, (4) follows trivially, since neither
condition applies (i.e. all are satisfied): The two coindexed NPs
are not co-arguments, hence there is no reflexive predicate, and
reflexivity-condition B allows (4a) and (4c). There is also no A-chain
here (as witnessed by the fact that NP movement, e.g. passive, is
not possible from the syntactic position of the pronoun). Hence the
chain condition says nothing about (4c), i.e. it is allowed (and no
special stipulation is needed for that result). ((ia) is ruled out
by reflexivity condition B; (iib) is ruled out by reflexivity condition
A, and (iiib) is ruled out by the chain condition.)
Many other contexts, where the complementarity entailed by traditional
BT fails, are discussed by R&R (as well as by many others). To
evaluate further Lidz' claim in point 1, it would be useful to get
some concrete examples where those instances of complementarity which
happen to be captured correctly by the BT-conditions A & B are not
captured also by R&R.
While "the hallmark generalization of the binding theory", namely
the complete complementarity of pronouns and anaphors, is empirically
wrong, our point in R&R is that there is a strong generalization
regarding where strict complementarity does, indeed, always hold -
a generalization missed by traditional BT. The domain in which
pronouns are complementary with both SE and SELF anaphors is precisely
the syntactic domain in which NP-movement is allowed. This is what
we call the A-chain domain. Coached in terms of the present
conference, this domain is statable (only) in terms of purely
'geometric' or configurational properties. Abstracting away from
issues of Theta roles and argument structure, if we look at the
(geometrical properties of the) structures below, we see that precisely
where an NP trace can occur, SE or SELF are allowed, but a pronoun
is excluded.
iv a) Max was washed t.
b) Max washed himself.
c) Max washed zich
d) *Max washed him.
v a) Max was expected [t to win]
b) Max expects [himself/SE to win]
c) *Max expects [him to win]
vi a) Max was heard [t arguing with Lucie]
b) Max heard [himself/SE argue with Lucie]
c) *Max heard [him argue with Lucie]
This correlation between NP movement and anaphora was, actually, among
the first anaphora facts to be discovered, and first it was assumed
that the same conditions must govern both (chomsky 1973). However,
as years went by, the BT got increasingly complex, mainly since it
attempted to capture also properties of logophoric anaphora. In the
process, it became, eventually, impossible to derive the conditions
on NP movement and local anaphora uniformly. Hence, the two got
separated. While NP movement is captured, universally, by the chain
condition (which, in earlier implementations, fed the ECP), the BT
is doing what it can to capture the intricate relations of pronouns
and SELF-anaphors in English.
Our conclusion in R&R was that the chain condition, in fact, reflects
a real universal generalization, which is sufficient to capture the
core aspects of local anaphora: An A-chain (=coindexation in the A-
chain syntactic domain) must have a fully specified NP (+R) at its
head, and cannot have such NP at its tail. The residue not captured
this way is the specific behavior of reflexive predicates. While
the later is an issue of argument structure, the core generalization
of local anaphora remains structural: What the structures in (iv -vi)
have in common is their configurational relations, rather than their
argument structure. (This is explained in detail in R&R's
'Reflexivity').
Note, finally, that the absence of the predicted complementarity is
just one of the empirical problems with the BT-conditions A and B,
which are surveyed in R&R. Since Lidz assumes the reflexivity
condition for SE anaphors, he captures, indeed, their distribution
correctly. But since he resorts back to the BT-condition A for SELF
anaphors, the same old problems (pointed out also in Pillared and
Sag) show up again. E.g. how is the contrast in (i) supposed to
follow, given that version of condition A?
vii) a) *Max boasted that the queen invited himself for tea.
b) Max boasted that the queen invited Lucie and himself for
tea.
Of course, if there is no upper bound on the mumbler of rules we can
introduce to face every new problem, some story could be invented.
But if we just forget about the traditional-BT A and B, nothing needs
to be invented here. (By the reflexivity conditions (ia) is out since
'himself' reflexive-marks the predicate and reflexivity condition
A requires that reflexive marked predicates be reflexive. But in
(ib) the anaphor is not a direct argument of the verb, hence it does
not reflexive-mark the predicate.)
LIDZ' POINT 2: TWO TYPES OF REFLEXIVE PREDICATES.
Lidz says:
"Finally, and most seriously, R&R's theory predicts that
reflexive-marked predicates are semantically uniform since the two
types of reflexive-marking are collapsed in the binding conditions.
However, predicates which are reflexive-marked by SELF anaphors have
different semantic properties than predicates which are lexically
reflexive-marked. That is, the one-to-one mapping between the semantic
property of being reflexive and the formal property of being
reflexive-marked breaks down when we examine the semantic properties
of these two types of predicate more closely. It is from the
perspective of this third problem that we will be able to see how
the theory of reflexive predicates is independent of the theory of
anaphora."
COMMENTS:
The problem Lidz' discusses here is very interesting. It has been
noted that SELF anaphors in reflexive predicates have properties
different from intrinsic reflexivization. (In English this has been
noted with the difference between 'The boys washed themselves" and
"the boys washed". Another example Lidz discusses is the Madame
Tussaude case, where the SELF anaphor can be used to refer to a statue
depicting the antecedent, while SE cannot). This ties in with a
broader set of interesting questions, but we should note, at the start,
that it has nothing to do with the reflexivity conditions.
Lidz conclusion that "R&R's theory predicts that reflexive-marked
predicates are semantically uniform since the two types of
reflexive-marking are collapsed in the binding conditions" is, in
fact, too quick. The binding conditions regulate only anaphora options,
and the fact that two structures are governed by one rule, does not
normally entail that they are synonymous. E.g. the fact that two NPs
get nominative case does not entail that they must have the same
thematic role.
Let us look at the semantic analysis Lidz offers for the difference
under consideration:
"What the Madame Tussaud examples illustrate is that SELF anaphors
allow an interpretation in which the anaphor is referentially
dependent on its antecedent but not necessarily identical with it.
That is, the anaphor can pick out an entity from the domain of
discourse which is related to and similar to its antecedent. We will
call such anaphors "Near-Reflexives."
(12) a. Lx [P(x,x)] (semantic reflexive)
b. Lx [P(x, f(x))] (Near Reflexive)
[L = lambda]
Since lexically reflexive predicates do not allow a non-identity
interpretation, it follows that they are semantically reflexive, i.e.,
that they have the representation (12a). Predicates which are
reflexive-marked via a morphologically complex anaphor do permit the
non-identity interpretation, and so implicate the semantic
representation (12b)."
In fact, Lidz is assuming here two types of coindexation. The same
idea has received different formulations in Fiengo and May's double
indexation, Higginbotham's linking conventions, and Heim's concept
of guises, captured with inner and outer indices. While these studies
showed this with pronouns, Lidz' important contribution is in showing
that the same can be found also with SELF anaphors in the local, (even
in the reflexivization) domain. Let us assume, for the present
discussion, that the double indexation idea is on the right track.
What does it tell about the reflexivity conditions? In fact, nothing
more than we want it to tell us. What (12a and b) have in common is
that the two arguments of the verb share an index (x). We may take
this shared property into the definition of reflexive predicates,
which will now read: ' A predicate is reflexive iff two of its
arguments share an index' (rather than 'are coindexed', in the original
formulation). Thus, with respect to the reflexivity conditions the
two types of reflexive uses behave alike. In addition, the SELF
anaphor has the property that it can be interpreted as in (12b).
Or, closer to the spirit of Lidz, and the studies cited, all anaphoric
elements have this property, with the exception of lexical (intrinsic)
reflexivization, where a theta role is absorbed in the lexicon, hence
it allows only the stricter interpretation (12a). We may conclude
that anaphora ambiguity of the type Lidz formulates in (12) (single
or complex index), is possible whenever an anaphoric element (including
a pronoun) occupies a free theta position. The studies mentioned above
show how all binding conditions, in all frameworks, can be stated
so that they do allow these two possible interpretations of anaphora.
The next question is, of course, whether there is any fact that Lidz'
formulation can capture, which R&R, modified as above, could not?
Note that the problems Lidz mentions all evolve the interpretation
of reflexives, and not the question whether anaphora is permitted
or not. Hence, it follows almost by definition, that the relevant
interpretation could follow from the modified R&R definition, to the
same extent that it follows from Lidz' reformulation. Let us consider
one example:
LIDZ WRITES:
"A second place where R&R's theory makes the wrong prediction is in
comparative deletion constructions. Here again, all reflexive-marked
predicates are predicted to behave alike but they do not. The
lexically reflexive-marked predicate allows only a sloppy
interpretation while the syntactically reflexive-marked predicate
allows either a strict or sloppy reading (Sells, Zaenen and Zec
1987):[4]
(11) a. zij verdedigde zich beter dan Peter
she defended self better than Peter
'She defended herself better than Peter defended himself'
'*She defended herself better than Peter defended her'
b. zij verdedigde zichzelf beter dan Peter
she defended herself better than Peter
i ) 'She defended herself better than Peter defended himself'
ii) 'She defended herself better than Peter defended her'"
COMMENT:
In his own analysis, later on, Lidz (probably) assumes that the
availability of the double indexing in (12b) for the SELF anaphors
should be sufficient to license the strict interpretation in (11b).
If this is so, then the same follows from what we just said. Since
in (11a) there is only "single" index, because of lexical
reflexivization, it has only the bound (sloppy) reading of ellipsis,
but the SELF in (11b) allows both a single and a complex index.
Personally, I am not convinced that double indexation is the correct
approach to the problem. If that's what is happening here, then why
is it so hard to get the strict reading all over the place, e.g. also
in the simple VP ellipsis in (vii).
vii) Max praised himself and Felix (did) too.
In R&R we take a different line on this same question (section 3.3.)
We argue that SELF anaphors are not different than pronouns in allowing
both the bound and the coreference interpretation. It is only when
they occur as arguments of a predicate that coindexation, and hence,
the bound variable interpretation is forced (by condition A). But
even in that position, they have also the option of being interpreted
as foci (one of their logophoric uses). Technically, under our specific
implementation there, they are raised to focus position, in this case,
hence they no longer reflexive mark the predicate, hence, condition
A says nothing, and the interpretation of anaphora as binding or
coreference is free. Since this is possible only when they are foci,
this means that getting the strict reading in cases like (vii) is
only possible when it is appropriate to view them as foci. Already
Sag (1976) has noted that the strict reading in (vii) is not fully
out, but it is harder to get. If it requires special context
justification, this follows.
Under this line, what makes SELF anaphors different from the SE ones
is their ability to carry stress, hence function as foci. In this
context, I also have a query regarding the whole line that Lidz offers
for why SE anaphors do not allow this type of (double indexing)
interpretation.
LIDZ SAYS:
" 14b. ze zag zich in een griezelige hoek staan
she saw self in a creepy corner stand
'She saw herself (=reflection, *statue)
in a creepy corner stand.'
The Near-Reflexive interpretation is ruled out in (14b) because the
predicate is lexically reflexive and so must be semantically reflexive,
by the right-to-left implication of Condition R."
QUESTION.
I do not understand how this works. Which predicate is lexically
reflexive here? Certainly not 'see'. Suppose even it was 'see', 'zich'
is not an argument of this predicate.? The major claim of Lidz is
that what determines the availability of the "near-reflexive" reading
is the absence of a lexically reflexive predicate. It seems,
therefore, that (14b) should allow the 'near-reflexive' reading.
Since it nevertheless does not, this points at the direction outlined
above, that the difference is in the stress options. Nevertheless,
Lidz observation about lexically reflexive predicates remains correct:
When there an intrinsic reflexivization process (as in 'Max behaved
himself') the stressing option cannot be realized also if the anaphor
is a SELF one. (When an NP does not even occupy a real theta position
it cannot, independently, be the focus). More could be said about
emphatic anaphors, and perhaps I will get a chance to come back to
this.
Lidz brings up also very interesting facts of Malayalam. I will try
to get back to those in a day or two.