Date: Mon, 28 Oct 1996 17:14:23 +0100 (MET)
From: Tanya Reinhart <Tanya.Reinhart@let.uva.nl>
To: Linguistics Conference <LINCONF@tamvm1.tamu.edu>
Subject: Dravidian Anaphora & comments on Hamilton.
Message-Id: <Pine.SOL.3.91.961028170930.18275B-100000@cclsun01>
Dravidian languages (Malayalam) were mentioned both by Lidz and by
Hamilton as posing a problem for the R&R reflexivity approach. In
fact, anaphora in these languages provides direct evidence for that
approach. It is also independently interesting, since it differs
in some respects than the better studied anaphora systems, so I would
like to survey these languages here in some detail, based on work
in progress with Martin Everaert and Eric Reuland.
HAMILTON WRITES:
"As promising as R&R's account of binding is, there
are at least two serious problems. First, R&R's
binding conditions predict that monomorphemic (alias
SE) anaphors should never be bound by a coargument
(i.e., assuming the predicate is not itself
intrinsically/lexically reflexive). This follows from
Condition B, since the reflexive predicate in this case
would not be reflexive-marked. Though this prediction
appears to hold for the Scandinavian languages on which
R&R focus, it is not clear that this pattern holds more
broadly. The Korean, Russian, and Malayalam anaphors
in (10-12), respectively,
A second problem for R&R's analysis is posed by
the Icelandic polymorphemic anaphor _sig sjalfur_.
Though both _sig sjalfur_ and its more commonly
discussed inverse counterpart _sjalfur sig_ (Note 3)
have emphatic usages not clearly subject to the binding
conditions, each may be used nonemphatically in certain
cases (and both exhibit local binding only). However,
the distribution of their nonemphatic uses differs
importantly. "
I will address both problems as I go along. We will see that
Scandinavian languages are not at all the marked case regarding the
first point. The second point is related, more broadly, to the
question of emphatic anaphors, which came up in this conference, and
especially, to Lidz' proposal that there may be two anaphora
interpretations of SELF anaphors. Malayalam may shed an interesting
light on this question.
==============
Tanya Reinhart
October, 1996
Based on work in progress with Martin Everaert and Eric Reuland.
DRAVIDIAN ANAPHORA AND EMPHATIC ANAPHORS.
Dravidian languages:
Kannada, (K) Malayalam, (M) Telugu, Tamil. SOV order.
1. INVENTORY:
Here is first a quick survey of the basic anaphora inventory these
languages all have in common. (I turn directly to where they differ.)
I will use English words with Dravidian word order. Most of the
original examples can be found in Jayaseelan's papers. Some are p.c.
-All have a pronoun (M: awan) and a SE anaphor (M: taan, similar form
in the other languages). No SELF anaphor as such.
-The distribution of the SE anaphor in the local context, is the same
as observed in R&R for Dutch: SE obeys condition B, as in (1a) (pointed
out in great detail in Jayaseelan (e.g. 1995), whose native language
happens to be Malayalam). The contrast observed in Dutch in between
object (1a) and ECM subject (1b) is found here as wall.
1) a) *Ram taan sees (Ram sees SE)
b) Ram [taan fool] considers (Ram considers SE fool)
c) *Ram [awan fool] considers. (Ram considers him fool)
Pronouns obey the chain condition, hence the contrast between (1b)
and (1c). The facts of (1) are confirmed in all Dravidian languages,
though they differ in the reflexivization system.
-The distribution of SE in the long distance context is similar to
Chinese (which differs from Scandinavian, on points b, c):
-a. Subject orientation and C-command obligatory.
-b. Occurs freely also in subordinate clauses.
-c. Blocking effects (matching features), like in Chinese.
-Full names: No condition C-effects with two full names. Yes:
condition B-effects. ( This is similar to what Lasnik found in
Vietnamese and Thai. I believe that the same is true for English
as well. This is an issue of rule I, of Grodzinsky and Reinhart, and
irrelevant for the present discussion).
FURTHER NOTES ON SE IN THE LOCAL CONTEXT:
The findings in (1) are in conflict with what Hamilton says about
Malayalam (quoting Mohanan 1982)). But there is ample evidence that
the relative judgment in (1), repeated below, is found not only in
all Dravidian languages, but also in Hindi (not Dravidian). According
to my p.c. checking of Korean, the same is found also there.
1) a) *Ram taan sees (Ram sees SE)
a') *Ram awan sees (Ram sees him)
b) Ram [taan fool] considers (Ram considers SE fool)
c) *Ram [awan fool] considers. (Ram considers him fool)
In evaluating conflicting reports of the judgments, the crucial point
to consider is that also in Scandinavian, condition B violations are
always weak, relative to the chain condition violation with pronouns.
So if we compare a sentence with SE, like (1a) to the same sentence
with a pronoun, as in (1a'), (1a) is always much better. In R&R's
system this difference follows, first, since the pronoun in these
contexts violates both condition B and Chain, and, more importantly,
because the chain condition is viewed as a stronger -core syntax-
condition, compared to the weaker - semantic - condition B. The crucial
way to evaluate the Reflexivity predictions is to compare (1a) with
the judgment on the ECM structure (1b), where the weak condition B
is observed. In all languages I mentioned above, there is a clear
contrast here - (1b) is perfect, while (1a)- marginal. Probably, the
same is true in many other languages.
One of the reasons this has not been widely noted is that the
traditional BT cannot distinguish between objects and ECM subjects.
Hence the SE case (1a) was always only compared to the pronoun case,
(1a'), and the more relevant contrast (1a)-(1b) was not checked in
the literature reporting the availability of local binding of SE.
Another way to put it: condition B effects with SE are observable
only given the reflexivity condition B. The traditional BT-condition
B fails to distinguish objects and ECM subjects, since it was
formulated to capture the distribution of pronouns. Pronouns, indeed,
behave alike in these two contexts. For R&R this follows, since their
distribution is governed also by the chain condition, by which objects
and ECM subjects equally form a chain with the matrix subject. This,
in fact, was one of the major arguments in R&R for separating the
reflexivity and the chain conditions.
Hamilton suggests that condition B locality effects with SE are the
marked, or 'language-specific' case. However, the fact that it is
found in completely unrelated families of languages is significant.
I would say that it is the other way around: The reported lack of
this locality effects in Chinese and Japanese is the mystery to be
explained.
2. THE DRAVIDIAN ANAPHORA SYSTEM.
There is no SELF marking on arguments (No SELF anaphors) in any of
the Dravidian languages. Still, conditions A, B ('Reflexivity') are
clearly and strictly obeyed. There are two systems of dealing with
reflexivization:
a. Marking on V or agr: Morpheme 'koL' is the reflexive marker.
So everything is as in Dutch, except that the marking is on the verb.
b. Duplication, or 'protection from binding': The coindexed element
is embedded inside a more complex NP, thus avoiding a reflexivization
environment, and all issues of condition B and chain.
-Kannada, is purely of system a. (This is shown in Lidz' LI squib).
-Malayalam is purely of system b (no koL or any reflexive marker).
The whole issue of reflexivization is avoided, by never allowing
coindexation of co-arguments. (This is how Jayaseelan explains it.
He believes the same is true universally. SELF anaphors have the
function of protecting the NP from reflexivization (avant la lettre).
-Telugu combines both. The duplication system is different than
Malayalam, and there is also an inflectional reflexive marker koL.
While Kannada and Malayalam are easy to explain, Telugu is difficult,
and appears more 'unsystematic'. I will therefore not discuss it here.
(But I can send some summary of the facts and the conceptual directions
open to account for them, upon request.)
-Tamil falls under the duplication method. But I don't have the data
re whether it is like Malayalam or like Telugu.
3. INFLECTIONAL MARKING: Kannada
(For some actual sentences, and more detailed analysis, see Lidz
(1995) squib)
3.1. Reflexive marking and reflexivity condition A:
In R&R it was argued that reflexivization can, in principle, be marked
morphologically either on the verb or on the argument (SELF) (R&R,
p. 662). However, the only instance we had for marking on the verb
was intrinsic reflexivization. It turns out that Kannada is the full
fledged example of the later option: The only way to reflexive mark
the predicate is by marking the verb. This applies freely to any verb,
as in (2a).
2a) Rama SE love-koL (Rama self-loves SE; Rama loves himself)
b) *Rama Mary love-koL-s (=Rama self-loves Mary)
c) *Rama thinks-koL that he is a fool.
koL has the sole function of reflexive marking. The way reflexivity
condition A is active is illustrated in (2b): Since a reflexive marked
predicate is not reflexive, the sentence is out. (2c) shows that koL
is indeed a reflexive marker, and not some general anaphora marker.
3.2. Conditions B and chain:
The following chart summarizes how the facts are derived in the
Reflexivity framework:
B Chain
3a) *Rama loves SE * ok
b) *Rama loves him * *
c) Rama koL-love SE ok ok
b) *Rama koL-love him ok *
4a) Rama believes [SE fool] ok ok
b) Rama koL-believe [SE fool] ok ok
(4b) deserves a comment: It poses no problem to Reflexivity condition
B, but the question is why it is allowed by condition A, given that
the matrix predicate is reflexive-marked and still, no two of its
direct arguments are coindexed. Recall that for independent other
problems, condition A applies in R&R to syntactic rather than semantic
predicates. The definition of syntactic predicates incorporates ECM
subjects. Rama and SE are, therefore, coarguments of a syntactic
predicate, even though not of a semantic one. So this syntactic
predicate is appropriately reflexive.
3.3. Some implications.
The morphological reflexive-marking of the verb in Kannada completes
the similarity between reflexivization and agreement systems: Just
as agreement can be either marked on the argument or on the verb,
so can reflexive marking.
As far as I know, Kannada provides also a confirmation of another
prediction in R&R. A question raised by R&R (p. 662) is whether one
reflexive marking is sufficient to license the coindexation of more
than two arguments, e.g. in the case of a three-place predicate.
Based on reflexive-marking of the argument, they conclude that this
is the case, and define reflexive-marking accordingly. Again, they
did not have any example of morphological marking of the verb.
According to some notes I have from R. Amritavalli (p.c), the following
is, indeed, fine in Kannada:
5 Rama tannannu tanage koLTukonDa
SE-acc SE- dat gave-koL
Rama gave himself to himself
Without the koL morpheme the sentence is out. For control, it is
necessary to check that the dative tanage is not always allowed to
be coindexed with either the subject or the object independently of
koL. (The equivalent paradigm with Dutch zelf is found in R&R (21).
I understand that taking the koL out, and replacing the accusative
tannu-SE with, say, man...= 'me', the sentence will be out.
3.4. Unsolved problems:
We have no solution for the NP issue discussed in Lidz' squib. More
generally, not enough work has been done, in the Reflexivity framework,
on SE anaphora inside NPs.
4. THE PROTECTIVE WAY: MALAYALAM.
(Detailed survey and examples can be found in Jayaseelan's (1995)
and previous papers.)
4.1. The basic facts.
-All NPs (Pron, SE, Proper name, no data given on Definite
descriptions) can occur with the suffixed tanne:
6) -SE: taan -- taan-tanne
-pron: awan --awan-tanne
-P names: Rama -- Rama-tanne
-With locally coindexed NPs (which would otherwise be co-arguments
of the same predicate), occurrence of tanne is obligatory - None of
the left members of (6) can occur alone in this context.
-All three right members of (6) can occur locally. I.e. the suffixed
tanne exempts from both condition B and Chain.
-The tanne does not induce locality. All (right and left) members
of (6) can occur in the long-distance context. The distribution of
the complex (right) unit is determined by the original anaphoric
element:
-taan-tanne, like the single SE taan, is subject oriented,
requires a c-commanding antecedent, and shows blocking effects.
-awan-tanne, like awan, is not subject oriented and does not
require c-command.
4.2. Jayaseelan's account:
-tanne is some kind of a focused element.
-It exempts from condition B (and chain), because the relevant DP
is no longer coindexed locally with its antecedent (just as in John
likes his mother, John and his are not locally coindexed.) The
internal structure of the DP is (7), with tanne as D. The specifier
DP is not coindexed with the top DP (just as in 'his father', the
pronoun 'his' is not coindexed with the whole NP). In (8), the top
DP is a coargument with the subject, but not coindexed with it, hence
all is fine.
7 DP
DP(i) D'
taan/awan D
tanne
8) Rama(i) [[taan](i)-tanne] loves.
COMMENT: It is not fully clear that the idea of tanne as D can be
maintained. Haripasad (dissertation) noted that in Telugu (which
is independently different) the suffix can occur after PP, as in (9)
(which is [kind of] Telugu, hence the koL])
9) Jaanu [[PP tana-too] tannu] spoke-koL.
with
Jaanu spoke with SE(self)
Jayaseelan (pc) acknowledged that the same is true for Malayalam.
Alternatively, the structure could be some sort of an adjunction of
the tanne to DP, PP, etc. In any case, it should have, for Malayalam,
the property of blocking co-arguments coindexation and chains. (Telugu
is more complex, as I mentioned.)
4.3.Condition A.
So far, then, there is just no reflexivization in Malayalam. Note
that if the inflectional marker koL existed in the language, we would
have gotten a violation of condition A with the complex anaphors,
as in (10).
10) Raama(i) [[taan(i)]-tanne] loves-koL
(Since koL reflexive-marks the predicate, there should be coindexation.
But there is no coindexation of co-arguments in (10).) However, there
is no koL in Malayalam. So the language just chose to avoid the issue
of reflexivization altogether (a point discussed also in Lidz' article
to this conference). Reflexivity Condition A is appropriately met,
since there are no reflexive-marked predicates in this language.
5. IMPLICATION FOR EMPHATIC ANAPHORS.
What we saw is that there are two systems of dealing with the
reflexivity requirements: One is to reflexive mark the predicate,
either on the verb, or on the argument. The other, is to embed the
anaphoric argument, thus avoiding a reflexive predicate altogether
(Malayalam). Jayaseelan argues that in fact, there is universally,
only one system, namely, the second (Malayalam) type. The SELF in
English and other languages is just a focus-emphatic marker, exempting
pronouns from condition B. If this is true, then condition A could
be dispensed with, and we only need reflexivity condition B. I think
that although, possibly, this line could be further developed, it
may be far-fetched, at the present stage of our understanding of the
problems (NOTE 1). If true, it would be very difficult to explain
why it is so hard to get this focus marking in, say, English, when
the anaphor is an argument. Specifically, why is it so much easier
to find the anaphor in the standard local context (10) than in (11a)?.
(In Malayalam, no such contrast is found.) Further, why do we find
the familiar contrast in (11a,b)?. (If anything, it should be easier
for the anaphor to be focus in (11a) than in (11b). Still it is very
difficult to get (11a), even as focus.) In this respect, English is
different from Malayalam, which allows complex anaphors in equal ease
in all syntactic contexts.
10 Max praised himself.
11 a) *Max wants us to praise himself
b) Max wants us to praise Lucie and himself.
But the general finding of Jayaseelan, namely that there is another
form and function of complex anaphors is very important. Possibly,
there are, universally, two ways anaphors may satisfy the reflexivity
requirements, and accordingly, even in languages with real SELF
anaphors, like English, there may be two syntactic functions of SELF -
as reflexive-marker and as a 'protector' of an index. When both are
available, the second is used for 'emphatic' purposes. I mentioned
in my previous comments on Lidz that in R&R we argued that all
instances of emphatic SELF anaphors in an argument position, are foci,
in the standard sense, and we were forced to assume they undergo focus
movement at LF, which then exempts them from condition A. But
independently of this problem, I don't really believe that foci must
move at LF. So, it would be nice if an alternative account could
be found.
Baker (1994) proposes a distinction between anaphoric and "non
anaphoric intensive use".(NOTE 2). The latter is generated as he
himself, in the nominative, but is realized as himself in the
accusative (Hence the confusion, since it is invisible in the
accusative.) The idea that some accusative anaphors are just another
form of the he himself emphatic form was proposed also by Bickerton
(1987), who argued, if I remember correctly, that the pronoun or pro
is deleted in the accusative.
Under this analysis, then, emphatic English forms may be viewed as
having the same structural properties as those of Malayalam. Note,
however, that since SELF does have a reflexive-marking function, the
contribution of the syntactic embedding in the complex (emphatic)
anaphor is not so much to 'protect' the pronoun but to 'protect'
the SELF from reflexive-marking. In a structure like 'he-himself'
the SELF is too deeply embedded to reflexive mark the verb of which
it is an argument. For this reason, such anaphors can occur also in
a context like (11a), without reflexive-marking the verb. Nevertheless,
there should be clear contextual reason to prefer this more complex
way to express anaphora. As I said in my previous reply to Lidz, I
think this reason is that it is easier to stress a complex anaphor
than a pronoun. When the reflexivity conditions allow the use of
a normal anaphor, as in (11b), this normal anaphor can be used for
stress, perspective, or other reasons. But when it cannot occur,
only the emphatic form can be used to enable the anaphor to carry
stress, without, at the same time, reflexive-marking the verb
incorrectly.
But along with protecting the SELF from reflexive-marking, the complex
embedded form still has the same effect, noted by Jayaseelan, of
protecting also the embedded pronoun ('he' in 'he-himself') from
condition B (and Chain). Hence, in the local context, it can be used
to obtain coreference without creating a reflexive predicate. This
was the case in Lidz example of ellipsis ('Max defended himself better
than Bill did), which I discussed in the previous round.
To my recollection, Baker does not substantiate his syntactic analysis
very deeply, so it would be useful to find more evidence for a
syntactic ambiguity in English, along the lines he proposes.
(Otherwise, we run into the danger of vacuity: if we assume ambiguity,
without clear tests for it, we will be always right, but in the wrong
way.) Specifically, more work is needed on how the analysis still
entails the contrasts in (10)-(11). Nevertheless, the idea is worth
pursuing.
I don't have much more to say on how to argue for such a structural
distinction, but at first glance, it seems that Hamilton's finding
on Icelandic may be relevant here. In Icelandic, there are two forms
of the SELF anaphor: 'sjalfur sig' and 'sig sjalfur'. Hamilton notes
that they do not have the same distribution. Since there are two
forms here, there is room for checking whether they could each belong
to a different type of those outlined by Baker.
HAMILTON SAYS:
"Unlike _sjalfur sig_ (which in all
relevant respects behaves as predicted by R&R's
system), _sig sjalfur_ may not be bound by a coargument
(_sig sjalfur_ is possible in (13) only on an emphatic
reading). (Note 4) This is so even though _sig
sjalfur_ and _sjalfur sig_ both contain the putative
SELF morpheme _sjalfur_: (Note 5)
(13) Jon(i) talar oft vid sig sjalfan(*i)/sjalfan
sig(i)
John talks often to himself
However, like _sjalfur sig_, _sig sjalfur_ may be
nonemphatically locally bound when it is the object of
an adjunct _about_-PP (see below on the adjunct status
of _about_-PPs):
(14) Maria(i) taladi vid Jon um sig sjalfa(i)/sjalfa
sig(i)
Mary talked to John about herself
Neither _sig sjalfur_ nor _sjalfur sig_, however, may
be (nonemphatically) locally bound when the object of
some other adjunct PP:
(15) Jon(i) sa snak nalaegt ser sjalfum(*i)/sjalfum
ser(*i)
John saw snake near himself "
COMMENT:
Given this data-base, it seems that only the SELF-SE form (_sjalfur
sig_ ) is the reflexive-marking SELF anaphor in Icelandic. Hence,
it can occur in the reflexivization environment (13). Given the logic
of Jayaseelan and Baker, the syntax of the second form (SE-SELF)
is different, so the index of the whole NP is not the same as the
index of 'sig'. When this second form appears in the local context
(13), all SELF does is to protect the index of sig from being co-bound.
Since icelandic, unlike Malayalam, also allows actual coindexation
in this context, there should be a reason why it is avoided, in favor
of the SE_SELF form. The most easily accessible interpretation is
that the emphatic reading is desired, as observed by Hamilton. But
perhaps it could also be used to obtain coreference rather than
variable binding, in contexts like those discussed by Lidz. In any
case, no reflexive predicate is formed in this case. In non -reflexive
contexts, as (14), the reflexivity conditions say nothing re which
of the two should occur, since neither violates any condition. Hence,
it is up to discourse to decide which of these is relevant, and whether
the use is emphatic or not, cannot be determined in isolation.
Note, since this was brought up, that the problem in (15) is
independent. It indicates a certain shortcoming of the treatment of
PP anaphora in R&R. In fact, (15) is not so good (non-emphatically)
also in English. An excellent alternative analysis of PP's in the
reflexivity framework is offered by Beit Arie (1994), which correctly
rules (15) out (unless it is emphatic).
========================
NOTE 1. The desire to eliminate condition A from the reflexivity
framework has very good reason: Condition A looks more syntactic
than condition B. First, it has to be stated in terms of syntactic,
rather than semantic, predicates, and next, it is stronger than
condition B -yielding much worse derivations when violated.
Nevertheless, I don't think the line could be assuming the Malayalam-
type protection system universally. Rather, the alternative seems
to be reducing condition A to the chain condition, which applies at
very similar environment. This line was proposed and fully executed
by Fox (1993).
NOTE 2. Baker argues that the distribution of 'logophors' is freer
in British than American English, and their emphatic use in argument
position is not always reducible to focus in the strict sense, as
proposed in R&R. He notes also that emphatic anaphors occur in an
argument position of a verb in British English more frequently than
assumed by R&R. (He gives examples from Jane Austin, in his (14).
In fact, similar facts can be found in the Zribi-Hertz collection.
Though, statistically, they are rare in her collection.)
============
-Baker, C.L. (1994) "Locally free reflexives, contrast and discourse
prominence in British English" Presented at the Annual Meeting of
the Linguistic Society of America, Boston. (ms. University of Texas,
Austin.)
-Beit-Arie, Oren (1994), "PP structure and anaphora", ms. Tel Aviv
University. Obtainable by e-mail: beitarie@aleph.co.il
-Bickerton, D. "He himself - Anaphor, pronoun, or ...?" Linguistic
Inquiry 18, 345-348, 1987.
-Fox, Danny (1993) "Chain and binding: A modification of Reinhart
and Reuland's "Reflexivity". Ms. MIT. (available at request:
fox@mit.edu)
-Jayaseelan, K.A. (1995) "Anaphors as Pronouns," ms. CIEFL: Hyderabad.
(Also earlier manuscripts).
-Lidz, J. (1995) "Morphological Reflexive Marking: Evidence from
Kannada," Linguistic Inquiry 26:705-710.
--Tanya ReinhartFrom daemon Mon Oct 28 11:20:35 1996
Received: from TAMVM1.TAMU.EDU by dragon.emich.edu; (5.65v3.2/1.1.8.2/27Aug96-0302PM)
id AA23842; Mon, 28 Oct 1996 11:20:29 -0500
Received: from TAMVM1.TAMU.EDU by tamvm1.tamu.edu (IBM VM SMTP V2R2)
with BSMTP id 3921; Mon, 28 Oct 96 10:19:34 CST
Received: from TAMVM1.TAMU.EDU (NJE origin LISTSERV@TAMVM1) by TAMVM1.TAMU.EDU (LMail V1.2a/1.8a) with BSMTP id 4193; Mon, 28 Oct 1996 10:19:34 -0600
Return-Path: <lidz@LOUIE.UDEL.EDU>
Received: from TAMVM1 (NJE origin SMTPE@TAMVM1) by TAMVM1.TAMU.EDU (LMail
V1.2a/1.8a) with BSMTP id 4148; Mon, 28 Oct 1996 10:18:06 -0600
Received: from eecis.udel.edu by tamvm1.tamu.edu (IBM VM SMTP V2R2) with TCP;
Mon, 28 Oct 96 10:18:04 CST
Received: from localhost.udel.edu by stimpy.eecis.udel.edu id aa21897;
28 Oct 96 16:09 GMT
To: linconf@tamvm1.tamu.edu
Subject: reply to Reinhart-part one
Date: Mon, 28 Oct 1996 11:09:29 -0500
From: Jeffrey Lidz <lidz@louie.udel.edu>
Message-Id: <9610281609.aa21897@stimpy.eecis.udel.edu>
Status: RO
Tanya Reinhart brings up a wide range of interesting issues with
respect to my paper. Because there is so much to say, I will address
them each in a different posting. This first posting addresses her
comments about the chain condition.
(A) Reinhart says:
"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."
My Reply:
The issue here is not whether the chain condition is independently needed,
but whether it is appropriate to apply it to the distribution of
anaphors. The first question to be addressed is whether RR's chain
condition is the same as the standard chain condition. If the standard
chain condition excludes chains with two case positions, then it is not
clear to me how this is meant to apply to anaphor-antecedent chains in
which both members of the chain are overtly marked with a morphological
case, as we find in Kannada:
(a) Hari-(yu) tann-annu hode-du-koND-a
Hari-(NOM) self-ACC hit-PP-REFL.PST-3SM
'Hari hit himself'
A chain condition which excludes two case-marked elements in a chain
should exclude (a) if the anaphor-antecedent relation is a chain relation.
Thus, it would seem that the anaphor-antecedent relation is not a chain
relation.
The second issue is whether RR's chain condition is even empirically
accurate. RR's chain condition (as it is stated) is a condition on the
referentiality of the members of elements in the chain relation. In
particular, they state that only one member (the head) of a chain can be +R
(where +R means bearing an independent reference). It
follows from this that pronouns cannot be the tail of a chain since
pronouns are +R. This prediction is not borne out. For example in
Kannada, we can replace the anaphor in (a) with a pronoun, without altering
the interpretation:
(b) Hari avan-annu hode-du-koND-a
Hari him-ACC hit-PP-REFL-3SM
'Hari hit himself'
Similar cases are found in various northern Italian Dialects, like
Piedmontese (Luigi Burzio, personal communication) and Padovano
(Christina Tortora, personal communication), given here:
(c) Gianni se varda lu
Gianni REFL sees him
'Gianni sees himself'
These facts suggest that R&R's chain condition is not at work in
determining the conditions on anaphora and thus that Reinhart's claim that
a theory with the chain condition and no binding conditions is more
parsimonious than one with both does not hold. If there is a chain
condition on A-movement, the anaphor-antecedent relation does not appear
to be subject to it.
>From the perspective of the paper under discussion, these facts
suggest that the binding conditions do not apply if Condition R is
satisfied (i.e., since condition B would presumably be violated in
(b-c). The details of getting this to work are not entirely
clear. However, the generalization seems to be that if the
interpretative component can avoid applying the binding conditions (by
providing an interpretation to the predicate directly from the
lexicon), then it should.
(B) In discussing my examples:
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' "
Reinhart says:
"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.)
I Reply:
The proper analysis of (4c) will have to remain controversial.
Putting aside our differences about the Dutch judgement, Hestvik
(1990) also reports the judgement i give for Dutch for Norwegian. The
fact that English differs from Norwegian suggests that there is
something more complicated going on here than either of our theories
can immediately explain. (i.e., I predict Norwegian and my
informant's version of Dutch, but not English. Reinhart predicts
English and her version of Dutch, but not Norwegian.)
As for how the traditional BT deals with (4) and also Reinhart's (i-iii) seems
clear from the perspective of Condition R and the Head-movement theory of
reflexives. As Reinhart observes, the grammaticality of (4a) would lead us
to thinking that _zich_ is an anaphor under the traditional BT and thus that
it should pattern with _zichzelf_ in all cases. However there are
additional factors which mask this pattern. (ia) is ruled out (in
contrast with (ib)) in
violation of Condition R (or even R&R's condition B), despite the fact
tat the standard BT will allow it. (iia) is
grammatical (in contrast with (iib))
because certain monomorphemic anaphors may undergo head movement at LF,
raising into the domain of the antecedent (cf. Pica, Cole and Sung, etc.)
(iiia) is expected under the usual complementarity predicted by the
traditional BT.
(c) Admittedly, the close similarity between the domain of A-movement
and the domain of anaphora needs to be accounted for, as does the
nature of the logophoric examples pointed out by Pollard and Sag and
R&R. The question here is whether the chain-condition can explain the
facts given above (i think it can't, without serious modification,
perhaps the right move, but perhaps not). As for the logophoric
cases, I really have nothing to add.
-Jeff Lidz