Reconstructing reconstruction
Can we arrive at a unified picture?
Caroline Heycock
University of Edinburgh
Workshop on resumptive pronouns, Hebrew University, 2–5 July 2012
1
Introduction
A considerable amount of the most interesting recent discussion of the syntax/semantics of relatives
concerns various different types of RECONSTRUCTION effects. Evidence concerning the distribution
of these effects has been used to argue for a head-raising anaysis of relatives, or for the necessity to
postulate both a raising and a matching analysis.
By now a number of different reconstruction effects have been appealed to. In this talk I want to look at
a few of these as a contribution to establishing their value as diagnostics for syntactic reconstruction.
2
An argument that reconstruction is/can be syntactic
Scope reconstruction feeds Condition C (Heycock 1995, Fox 1999).
Fox (1999): scope reconstruction should be impossible in the structural configuration in (1).
(1)
[QP . . . r-expressioni . . . ]2 . . . pronouni . . . t2
Evidence that this is so:
(2)
a.
b.
(3)
a.
b.
How many stories does she want you to invent t?
(want > many; *many > want)
How many stories does she want you to reinvent t?
(want > many; many > want)
How many houses does John think you should build t?
(think > many; *many > think)
How many houses does John think you should rebuild t?
(think > many; many > think)
(4)
a. *[How many stories about Diana1 ]2 does she1 want you to invent t2 ?
b. [How many stories about Diana1 ]2 does she1 want you to reinvent t2 ?
(5)
a. *[How many houses in John1 ’s city]2 does he think you should build t2 ?
b. [How many houses in John1 ’s city]2 does he think you should rebuild t2 ?
1
3
Some reconstruction effects in relative clauses
• Variable binding: a pronoun/anaphor contained within the modified NP may be bound by a
quantifier within the relative:
(6)
a.
b.
I noticed the portrait of herselfi that every studenti had pinned up on the wall.
The book about hisi exploits that every candidatei brought with him annoyed the
interviewers.
• Condition A: an anaphor contained within the modified NP may be licensed by an antecedent
within the relative:
(7)
I admired the portrait of herselfi that Kahloi painted in 1950.
• “Low” readings for only, superlatives, including first and last, and ordinals (Bhatt 2002, Bhatt
and Sharvit 2005):
(8)
a.
b.
c.
That is the only book that John thinks Tolstoy ever wrote.
read as reporting John’s belief that Tolstoy wrote no books other than that one.
That is the first/last book that John thinks Tolstoy wrote.
read as reporting John’s belief that Tolstoy wrote that book before/after all the others.
That is the second book that John thinks Tolstoy wrote.
read as reporting John’s belief that Tolstoy wrote that book after completing one other
book.
• “Low,” de dicto readings for all modifiers (or the head) with respect to a predicate within the relative (argued in Hulsey and Sauerland (2006) to be the general case of the last type of effect, which
is itself argued to be just another case of binding reconstruction—binding of a world variable):
(9)
I went through the 30,000-word essay that he claimed to have written, and it turned out to
be only 20,000 words long.
• The head of the relative may form part of an idiom that is contained within the relative (Schachter
1973, Hulsey and Sauerland 2006):
(10)
John was satisfied by the amount of headway that Mary made.
Goal of this talk: to explore whether the last four kinds of reconstruction effects are all of a piece, and
whether the same kind of evidence for syntactic reconstruction in particular is available in relatives as in
e.g. questions.
2
4
Interactions between reconstruction effects
4.1
NPIs, Neg-Raising, and low readings for superlatives and only
Heycock (2005): the “low” readings for only and superlative adjectives discussed in Bhatt (2002) are
blocked in a range of environments:
• Low reading blocked by intervening negation (Bhatt 2002):
(11)
a.
b.
c.
This is the first book that John didn’t think that Antonia wrote. 6=
This is the book that John didn’t think that Antonia wrote first.
This is the first book that John denied that Antonia wrote. 6=
This is the book that John denied that Antonia wrote first.
That is the first book that few people said she read. 6=
That is the book that few people said she read first.
• Low reading blocked by intervening adverb:
(12)
a.
b.
This is the first book that we mistakenly thought that Antonia had written. 6=
This is the book that we mistakenly thought that Antonia had written before writing
all the other books.
This is the first book that people have occasionally thought that Antonia wrote 6=
This is the book that people have occasionally thought that Antonia wrote before
writing all the other books.
• Low reading unavailable with various predicates, including factives (13a), implicatives (13b),
weak and strong (as opposed to midscalar) deontic operators (13c):
(13)
a.
b.
c.
That is the only book that I know she likes 6=
That is the book that I know is the only one that she likes.
Those are the only people that he managed to insult. 6=
Those are the people s.t. he managed to insult only them (he successfully avoided
insulting others).
That is the only offence that he could / needed to claim to have committed. 6=
That is the offence s.t. he could/needed to claim not to have committed an offence
other than that.
Bhatt (2002): the NPI ever has to appear in the same clause as its licensor: this then disambiguates
where the superlative/only is reconstructed to:
(14)
a.
b.
the first/only/longest book that John said that Tolstoy had ever written.
Low reading only
the first/only/longest book that John ever said that Tolstoy had written.
High reading only
This account revised in Bhatt and Sharvit (2005):
(15)
Anna Karenina is the longest book that John said Tolstoy had ever written.
High reading: *; Low reading: OK
(16)
Anna Karenina is the longest book that John ever said Tolstoy had written.
High reading: OK; Low reading: *
3
Scenario A (“High” reading true, “Low” reading false):
John: “Tolstoy wrote Huckleberry Finn, Anna Karenina and Tom Sawyer. Tom Sawyer is the
longest of these.”
Anna Karenina is actually the longest among those books.
Scenario B (“High” reading false, “Low” reading true):
John: “Anna Karenina is the longest book Tolstoy wrote. He also wrote War and Peace and some
other shorter books.”
War and Peace is actually longer than Anna Karenina.
• The absence of the “high” reading when the NPI is in the embedded clause, as in (14a), (15) is
given an explanation in terms of the pragmatic theory of NPI licensing of Kadmon and Landman
(1993). No longer has to be stipulated that ever has to be a clausemate of its licensor (good,
because this constraint does not operate elsewhere). The licensing of the NPI and the unambiguous
low reading for the superlative are compatible with an account that involves reconstruction, but
they do not force it.
• The absence of the “low” reading when the NPI is in a higher clause in the relative, as in (14b),
(16), does depend on reconstruction: the superlative (Bhatt and Sharvit (2005) only discuss
superlatives in their paper) that is the licensor for the NPI no longer scopes over it from the
reconstructed position.
Alternative: (Heycock 2005) the only predicates—and environments—that allow a “low” reading are
those that support “Neg Raising” (NR)1 Roughly, the “low” reading arises because of the “Excluded
Middle” presupposition carried by NR predicates. Borrowing from the clear exposition of how this idea
could be executed in Bhatt and Sharvit (2005): if Anna Karenina is the longest book that John believes
Tolstoy wrote, there is a degree d such that John believes that Anna Karenina is d-long, but he doesn’t
believe of any other member of the comparison set that it is d-long. Because of the excluded middle
presupposition, it follows that John believes that every member of the comparison set is shorter than
Anna Karenina. This yields the low reading.
Relevant here: this account does not rely on syntactic reconstruction of the head+modifier(s).
Bhatt and Sharvit (2005) argue against this account on the basis that there is now no way to account for
the absence of the “low” reading in examples like (14b), (16).
But:
• It is not only the NPI ever that prevents the “low” reading: the same effect arises if we substitute
e.g. once, occasionally, or even mistakenly, foolishly or confidently:
(17)
Anna Karenina is the longest book that John/people {once / occasionally / mistakenly /
foolishly / confidently} said/thought Tolstoy had written.
High reading: OK; Low reading: *
This is as expected if the low reading arises from the same mechanism that underlies NR, since
NR is also blocked by these interveners:
(18)
a.
No one thought that Tolstoy would publish until he got an advance .
1
The notable exception to this is that say also allows this reading, while it is not a Neg-raiser. In Heycock (2005) it is argued
that this is possible just where say can be read as a kind of evidential, but Bhatt and Sharvit (2005), Hulsey and Sauerland (2006)
take this exception to invalidate the generalization. Bhatt and Sharvit (2005) argue that there are other exceptions, such as hope
and agree, but here their judgments and mine differ; this remains a matter for more thorough empirical investigation.
4
b. *No one {once / occasionally / mistakenly / foolishly / confidently} thought that Tolstoy would publish until he got an advance.
An account of this effect is given in Hegarty (2011). But it is not clear why reconstruction should
be blocked by adverbials.
• It has been observed that ever blocks NR (Lakoff 1969, Prince 1976, Gajewski 2005). (19) is from
Lakoff (1969), cited in Gajewski (2005), p. 19:
(19)
*I didn’t ever think that Bill would leave until tomorrow.
But if this is the case, the NR account predicts that the “low” reading will be unavailable in (14a),
(15). Thus this effect, while compatible with a reconstruction account (as we have seen) no longer
constitutes evidence for it.
Bhatt & Sharvit’s (2005) argument based on the unavailability of the “low” reading when there is an NPI
in the higher clause is an instance of the kind of argument for reconstruction based on the interaction
between two different reconstruction effects: in this case, the licensing of NPIs and the interpretation of
modifiers with respect to intensional predicates.
This is the kind of strategy that I will be pursuing in this talk: to attempt to determine whether reconstruction effects pattern together as expected.
As shown in (11)–(13), the “low” reading for superlatives/only is blocked in a number of different environments. Now our question is: do the same restrictions show up with respect to the other reconstruction
effects?
4.2
Low readings for superlatives, only
• Question: does low ever, taken in Bhatt (2002) on the basis of the readings of only and superlatives
to be diagnostic for reconstruction to a low position, force Condition C violations?
– First pass to establish an answer
(20)
Scenario: I want to claim that the photograph of Kahlo/Moss that we are looking at
is the only likeness, according to popular wisdom/me, that her mother could bear to
look at/that her agent has ever considered sufficiently flattering. So I say:
a. That is the only picture of Kahlo that they say her mother was ever willing to
look at.
b. That is the only picture of Moss that I believe her agent has ever considered
flattering enough.
(21)
Scenario: I want to claim that the photograph of Kahlo/Moss that we are looking at
is the only likeness, according to popular wisdom/me, that she herself could bear to
look at/considered sufficiently flattering. So I say:
a. That is the only picture of Kahlo1 that they say she1 was ever willing to look
at.
b. That is the only picture of Moss1 that I believe she1 has ever considered flattering enough.
The four sentences in (20), (21) were judged by 21 people (9/12 for each sentence): the mean
score on a scale of 1–5 was 4.85 for the examples in (20) and 4.81 in (21).
5
– But the acceptability of the examples in (21) could be because there is an intermediate position for a copy at the left periphery of the lowest clause, where the r-expression would still
not be c-commanded by the coreferential pronoun. So the probative examples have to have
a form more like (22):
(22)
a.
b.
That is the only picture of Moss1 that she1 thinks you should ever have published.
That is the only picture of Kahlo1 that she1 thought her mother would ever be
willing to have in the house.
To my ear, these are no worse than the examples in (21).
– Low reading for only 9 Condition C violation.
• Question: does low ever block anaphor binding from a higher position?
(23)
a.
b.
I finally saw the only picture of me that John has ever painted.
I finally saw the only picture of myself that John has ever painted.
(24)
a.
b.
I finally saw the only picture of me that they say John has ever painted.
I finally saw the only picture of myself that they say John has ever painted.
Again, (24b) does not stand out for me as less acceptable than (23b).
• Low reading for only 9 locality effect on binding of anaphor by higher antecedent
4.3
High readings of superlatives, only
• Conversely: does a necessarily high reading for a superlative/only induce a “trapping” effect?
(25)
a.
b.
At least one soldieri seems (to Napoleon) ti to be likely t i to die in every battle.
∀ > ∃, ∃ > ∀,
At least one solderi seems to himselfi ti to be likely ti to die in every battle.
∃>∀
Here we can recall that there are various environments where the low reading for superlatives/only
is blocked (see (11)—(13) above)
• The binding of anaphors is not blocked in the same environments:
(26)
That is the only picture that I doubt John will approve of 6=
That is the picture that I doubt is the only one that John will approve of.
(27)
That is a picture of himself1 that I doubt John1 will approve of.
• And in fact if ever does disambiguate the scope of only, we can find cases where only has a high
interpretation while the anaphor is bound by an element in a lower clause:
(28)
That is the only picture of himself1 that I ever thought Freud1 might sell.
• No trapping effect on anaphor binding from the interpretation of superlatives/only.
This could just be the result of an external attachment of only combined with a raising+reconstruction
analysis for picture of himself that I ever thought Freud might sell. What remains surprising however is the possibility of the anaphor binding given the blocking of the lower scope reading for
only.
6
4.4
Low binding of world variables
Reconstruction for low, “de dicto” readings is not restricted in the same way that low readings for
superlatives are:
(29)
That is the last/biggest picture that I fervently hope he will paint.
No low interpretation for last/biggest
(30)
The beautiful picture that I fervently hope he will paint may never come to be.
• Question: does a de dicto reading of an adjective modifier of the NP head force a Condition C
violation?
(31)
The “flattering” portrait of Moss1 that she1 heard Hirst had painted shocked her when she
actually saw it.
• Low reading for adjective 9 Condition C violation
• Question: Does a de dicto reading of an adjective modifier of the NP head block anaphor binding
from a higher position?
(32)
a.
b.
I saw the portrait of me that my sister said she had taken a year to paint, and I loved
it.
NO BINDING ; NO RECONSTRUCTION FOR ADJECTIVE
I saw the portrait of myself that my sister said she taken a year to paint, and I loved
it.
EXTERNAL BINDING ; NO RECONSTRUCTION FOR ADJECTIVE
(33)
Scenario: my sister claimed that she painted a flattering portrait of me. But I think she’s
made me look dreadful.
a. I saw the ‘flattering’ portrait of me that my sister said she had painted, and thought
it made me look like a witch!
NO BINDING ; RECONSTRUCTION FOR ADJECTIVE
b. I saw the ‘flattering’ portrait of myself that my sister said she had painted, and
thought it made me look like a witch!
EXTERNAL BINDING ; RECONSTRUCTION FOR ADJECTIVE
(34)
Scenario: Freud claimed to have spent a year painting a portrait of Will; Will doesn’t
think that it should have taken that long.
a. When Will saw the portrait of him that Freud claimed to have spent a year painting,
he was not that impressed.
NO BINDING ; NO RECONSTRUCTION FOR ADJECTIVE
b. When Will saw the portrait of himself that Freud claimed to have spent a year painting, he was not that impressed.
EXTERNAL BINDING ; NO RECONSTRUCTION FOR ADJECTIVE
(35)
Scenario: Freud claimed to have painted a sympathetic portrait of Will, but Will thinks
the portrait is hugely unflattering.
a. When Will saw the ‘sympathetic’ portrait of him that Freud claimed to have painted,
he almost blew a fuse.
NO BINDING ; RECONSTRUCTION FOR ADJECTIVE
7
Re#lexive binding and low readings for adjectives 5.0 4.5 4.0 Mean ra<ng 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0 No-‐binding; No Adj-‐recon Ext-‐binding; No Adj-‐recon No-‐binding; Adj-‐recon Ext-‐binding; Adj-‐recon Figure 1: Interaction of reflexive binding and low reading for adjectives
b.
When Will saw the ‘sympathetic’ portrait of himself that Freud claimed to have
painted, he almost blew a fuse.
EXTERNAL BINDING ; RECONSTRUCTION FOR ADJECTIVE
• Low reading for adjective 9 locality effect on binding of anaphor by higher antecedent
• Issue: “scare quote” intonation can license de dicto reading very freely:
(36)
a.
b.
When I saw the ‘flattering’ portrait of me by my sister, I was furious: she made me
look like a witch!
Freud’s ‘sympathetic’ portrait of Will was really a hatchet job, in my opinion.
So it could be that in fact (33b) and (35b) are acceptable because this more general process is
licensing the de dicto reading.
What about cases where there is no contra-indexing within the NP (Hulsey and Sauerland 2006)?
(37)
a.
b.
4.5
When I learned about the portrait of myself that my sister said she was about to start
painting, I was very surprised. And indeed, it turned out, as I suspected, that she was
actually planning something completely nonfigurative.
When I heard about the portrait of Moss1 that she1 thought her agent was commissioning from Hirst, I was very surprised. And indeed, it turned out that what was
commissioned was something completely nonfigurative.
Idioms
• Reconstruction for idiom intepretation also doesn’t seem to be blocked in the way that low readings for superlatives are:
(38)
That is the first book that few people said she read. 6=
That is the book that few people said she read before she read the others.
(39)
The is the kind of headway that few people (think we) can make.
8
(40)
This is the first book that we mistakenly thought that Antonia had written. 6=
This is the book that we mistakenly thought that Antonia had written before writing all
the other books.
(41)
We were foolishly delighted by the headway that we mistakenly thought we had made.
• Question: Does reconstruction for the purpose of idiom interpretation feed Condition C?
(42)
This represents the only headway on Lucy1 ’s problem that she1 thinks they have made so
far.
• Idiom reconstruction 9 Condition C violation
5
Dissociation of reconstruction effects
• So far we have seen:
– Low readings for superlatives do not disrupt “high” binding of anaphors, or induce Condition
C effects.
– De dicto readings for heads and modifiers do not disrupt “high” binding of anaphors, or
induce Condition C effects.
– Idiom interpretation does not induce Condition C effects
– Reconstruction for anaphor binding, de dicto readings, and idiom interpretation can be to
positions where low readings for superlatives are excluded.
• In contrast, we started with the observation that an argument for syntactic reconstruction is the
observation that scope reconstruction in questions induces Condition C effects: e.g. (4a) and (5a)
above; this is a specific case of a general assumption that only a single copy is interpreted.
• However, Lechner (1998) shows that also in questions there can be a dissociation between different
reconstruction effects.
Low scope reading for scrambled direct objects in German does not feed binding. Crucially, (43b)
allows the scrambled indefinite to be read with narrow scope with respect to the universal, but the
pronoun it contains cannot be bound by jedem Kandidaten.
(43)
a.
b.
weil sie jedem Kandidateni ein Bild von seinemi Auftritt
zeigte
since she every candidate a picture of his
appearance showed
since she showed every candidate a picture of his appearance [in the show]
weil sie [ein Bild von seinem∗i Auftritt]j jedem Kandidaten tj zeigte
since she a picture of his
appearance every candidate
showed
since she showed every candidate a picture of his appearance [in the show]
Similarly, the scrambled reciprocal in (44b,c) cannot be bound under reconstruction, even though
reconstruction for scope is possible, as evidenced by the fact that (44c) is scopally ambiguous:
(44)
a.
weil siei vielen Gästenj
einige Freunde
von einanderi/j vorgestellt haben
since they many guests.DAT some friends.ACC of each other introduced have
since theyi have introduced some friends of each otheri/j to many guestsj
b. *weil ich [einige Freunde
von einanderj ]k vielen Gästenj
tk vorgestellt habe
since I some friends.ACC of each other many guests.DAT introduced have
Intended: since I have introduced some friends of each otherj to many guestsj
9
c.
weil siei [einige Freunde
von einanderi/∗j ]k vielen Gästenj
tk vorgestellt
since they some friends.ACC of each other
many guests.DAT introduced
haben
have
since theyi have introduced some friends of each otheri/∗j to many guestsj
Reconstruction for scope 9 reconstruction for binding.
• Lechner concludes that we need both syntactic reconstruction (deletion of higher copies) and
semantic reconstruction (via higher type traces), and that the data above show that the two are
independent.
• Problem as I understand it for the relative clause case at hand is that at least as analysed in Hulsey
and Sauerland 2006 the “low” readings of the de dicto adjectives and of superlatives is also a case
of binding (of world variables). Thus the dissociation between these effects and e.g. anaphor
binding or Condition C is still unexpected.
6
How do these effects interact with extraposition?
• Does extraposition distinguish between reconstruction for binding and other cases of “low” readings? Hulsey and Sauerland (2006): extraposition allows only the “matching” derivation for relative clauses, and hence is predicted not to allow relatives showing any of the reconstruction effects
they discuss. Their examples in (45), as reported in Hulsey and Sauerland (2006) bear out this
prediction and hence constitute an indirect argument that many relatives are derivationally ambiguous:
– Idioms:
(45)
a.
b.
Mary praised the headway that John made.
Mary praised the potroast that John made.
(46)
a. *Mary praised the headway yesterday that John made.
b. Mary praised the potroast yesterday that John made.
– Reflexive binding:
(47)
a.
b.
I saw the picture of himselfi that Johni liked.
I saw the picture of Clinton that John liked.
(48)
a. *I saw the picture of himselfi yesterday that Johni liked.
b. I saw the picture of Clinton yesterday that John liked.
– Superlatives:
(49)
a.
b.
I read the first novel that John said that Tolstoy had ever written.
I read the first novel that John ever said that Tolstoy had written.
(50)
a. *I read the first novel last week that John said that Tolstoy had ever written.
b. I read the first novel last week that John ever said that Tolstoy had written.
• A preliminary attempt to test/extend this paradigm:
– De re /De dicto adjectives
(51)
N O RECONSTRUCTION ; R ELATIVE IN SITU
10
a.
b.
Yesterday I read the savage review that you told me Smith had written, and to
be honest I thought it should never have been printed.
This morning I did a read-through of the 30,000-word thesis that your student claimed to have written without assistance, and I’m afraid it looked suspiciously familiar.
(52)
N O RECONSTRUCTION ; R ELATIVE EXTRAPOSED
a. I read the savage review yesterday that you told me Smith had written, and to
be honest I thought it should never have been printed.
b. I did a read-through of the 30,000-word thesis this morning that your student
claimed to have written without assistance, and I’m afraid it looked suspiciously familiar.
(53)
R ECONSTRUCTION ; R ELATIVE IN SITU
a. Yesterday I read the ‘savage review’ that you told me Smith had written, and
actually I thought it was quite fair.
b. This morning I did a word-count of the ‘30,000 word thesis’ that your student
claimed to have submitted, and it came in at 40,000 words.
(54)
R ECONSTRUCTION ; R ELATIVE EXTRAPOSED
a. I read the ‘savage review’ yesterday that you told me Smith had written, and
actually I thought it was quite fair.
b. I did a word count of the ‘30,000-word thesis’ this morning that your student
claimed to have submitted, and it came in at 40,000 words.
– Anaphor binding
(55)
N O RECONSTRUCTION ; R ELATIVE IN SITU
a. I had to laugh when I read this morning the anecdote about his wife that he put
in his column.
b. I love Tracey’s work! I was so pleased when at last I got to see the sketch of a
tiger that she had once painted on the wall of her house.
(56)
N O RECONSTRUCTION ; R ELATIVE EXTRAPOSED
a. I had to laugh when I read the anecdote about his wife this morning that he put
in his column.
b. I love Tracey’s work! I was so pleased when I got to see the sketch of a tiger
at last that she had once painted on the wall of her house.
(57)
R ECONSTRUCTION ; R ELATIVE IN SITU
a. I had to laugh when I read this morning the anecdote about himself that he put
in his column.
b. I love Tracey’s work! I was so pleased when at last I got to see the sketch of
herself that she had once painted on the wall of her house.
(58)
R ECONSTRUCTION ; R ELATIVE EXTRAPOSED
a. I had to laugh when I read the anecdote about himself this morning that he put
in his column.
b. I love Tracey’s work! I was so pleased when I got to see the sketch of herself
at last that she had once painted on the wall of her house.
– Idioms
(59)
N O RECONSTRUCTION ; R ELATIVE IN SITU
11
Extraposition and reconstruction effects 5.0 4.5 4.0 Mean ra.ng 3.5 3.0 No reconstruc.on; In situ 2.5 No reconstruc.on; Extraposed Reconstruc.on; In situ 2.0 Reconstruc.on; Extraposed 1.5 1.0 0.5 0.0 Adjec.ves Anaphors Idioms Figure 2: Interaction of extraposition and reconstruction effects
a.
b.
Describe to me all the habits that you dislike in a partner.
Describe to me the progress that you observed.
(60)
N O RECONSTRUCTION ; R ELATIVE EXTRAPOSED
a. Describe all the habits to me that you dislike in a partner.
b. Describe the progress to me that you observed.
(61)
R ECONSTRUCTION ; R ELATIVE IN SITU
a. Describe to me all the habits that you want to kick.
b. Describe to me the headway that you think you have made this year.
(62)
R ECONSTRUCTION ; R ELATIVE EXTRAPOSED
a. Describe all the habits to me that you want to kick.
b. Describe the headway to me that you think you have made this year.
Results shown in the graph are from 21 people (9/12 for each sentence).
• From these preliminary data, extraposition does not appear to interact with “reconstruction” as
far as de dicto readings and idiom interpretations are concerned (the latter contra the findings in
Hulsey and Sauerland 2006); there is perhaps an indication though that extraposition does have an
effect on the possibility of binding an anaphor from within the relative clause.
7
Conclusion/agenda
• De dicto interpretations of adjectives, low readings of superlatives, and idiom interpretation do
not cluster together with reconstruction effects for binding.
• From this it seems that we should conclude that at least some of these effects should not be handled
via syntactic reconstruction.
• Still required: a much more thorough investigation and theoretical analysis of the differing “reconstruction effects.”
12
References
Bhatt, R. (2002). The raising analysis of relative clauses: Evidence from adjectival modification. Natural
Language Semantics 10, 43–90.
Bhatt, R. and Y. Sharvit (2005). A note on intensional superlatives. In E. Georgala and J. Howell (Eds.),
Proceedings of SALT XV, Ithaca, NY, pp. 62–78. Cornell University.
Cresti, D. (1990). Extraction and reconstruction. Natural Language Semantics 3, 79–122.
Fox, D. (1999). Reconstruction, binding theory, and the interpretation of chains. Linguistic Inquiry 30,
157–196.
Gajewski, J. R. (2005). Neg-Raising: Polarity and presupposition. Ph. D. thesis, MIT, cambridge, MA.
Hegarty, M. (2011). Neg-raising and duals. In Proceedings of Colloque a contrario.
Heycock, C. (1995). Asymmetries in reconstruction. Linguistic Inquiry 26.4, 547–570.
Heycock, C. (2005). On the interaction of adjectival modifiers and relative clauses. Natural Language
Semantics 13, 359–382.
Hulsey, S. and U. Sauerland (2006). Sorting out relative clauses. Natural Language Semantics 14,
111–137.
Kadmon, N. and F. Landman (1993). Any. Linguistics and Philosophy 16, 353–422.
Lakoff, R. (1969). A syntactic argument for negative transportation. In Proceedings of CLS 5, pp.
149–157.
Lechner, W. (1998). Two kinds of reconstruction. Studia Linguistica 52(3), 276–310.
Prince, E. (1976). The syntax and semantics of neg-raising, with evidence from French. Language 52,
404–426.
Schachter, P. (1973). Focus and relativization. Language (49.1), 19–46.
13
© Copyright 2026 Paperzz