Attributive Wrong
1. Larson (2000) observes that attributive wrong operates non-locally on material external to the
containing noun phrase. (1), for example, implies my opening of a bottle that it was wrong for me to
open, suggesting that the semantic scope of wrong includes the main clause. Larson credits this non-
locality to antecedent-contained deletion within the adjective phrase, positing ellipsis of an infinitival
complement of wrong under identity with the main clause after scoping of the containing noun phrase.
This analysis assigns (1) the logical form (2). In this paper we present observations that are unexpected
under an ellipsis analysis and propose an alternative account under which non-locality is due to wrong
being the main functor in its clause.
2. Sentence (3) can be read as implying that you closed a bottle that it was wrong for you to close, but
not that you closed a bottle that it was wrong for me to open. Under the ellipsis analysis, this indicates
that the elided complement of wrong must take a clause-internal antecedent and that the logical form in
(4) is ill-formed. This ill-formedness cannot be credited to a general restriction on ellipsis, not even a
general restriction on antecedent-contained deletion. In related discussion, Kennedy (1998) observes that
(5) can be read as implying that she met every alien Jones saw, indicating that antecedent-contained verb
phrase ellipsis permits logical forms parallel to (4). It is not clear, therefore, how an ellipsis analysis can
prevent non-local wrong from permitting external antecedents.
3. With THE and WRONG abbreviating the denotations of the and clause-embedding wrong, (6) below
describes the truth-conditions of (2) in the actual world w0. We present several concrete scenarios to argue
that, however THE and WRONG are defined, these truth-conditions do not meet intuitions on (1). We begin
with a scenario for which (6) seems correct. In (7), I opened the unique bottle that I was not supposed to
open. Under plausible construals of THE and WRONG, (6) is true in this scenario. (1) is judged true as well.
But now consider scenario (8). In (8) there are two bottles that I was not supposed to open, Therefore, if
THE contributes uniqueness, (6) is presumably not true in (8). However, (1) is judged true in (8). This may
suggest that the in noun phrases with non-local wrong is semantically indefinite. This is not implausible.
Szabolcsi (1986) proposes that the in certain superlatives is semantically indefinite, and Larson (2000)
detects a semantic similarity between wrong and superlatives. However, we present evidence that the in
(1) contributes uniqueness after all.
4. If THE is a mere existential, then (6) does not differentiate between scenarios (7) and (8) above and
scenarios (9) and (10) below, as it is true in all of these scenarios.
In fact, however, we have difficulty in judging (1) either true or false in (9) and (10). This judgment
suggests that THE contributes a uniqueness presupposition and that (6) is to be replaced with the truth-
conditions in (11), according to which it was wrong for me to open the unique bottle that I actually
opened.
5. We do no see a way to derive these truth-conditions in an analysis that solely relies on ellipsis. But,
not surprisingly, they can be derived by instead stipulating a semantics under which wrong is the main
functor in the clause. Assuming the logical form for (1) in (12), the denotation for wrong in (13) derives
(11) in a straightforward way.
6. Apart from deriving (11), the lexical analysis based on (13) straightforwardly derives the judgment
on (3) discussed above. It cannot in principle derive an unattested reading for (3) according to which you
closed a bottle that it was wrong for me to open. The lexical analysis moreover excludes as type-
mismatches cases where non-local wrong follows, or is coordinated with, another attributive adjective
such as green. This is correct, as illustrated by the contrast between (14) and (15). It is not clear to us how
an ellipsis analysis can derive this contrast.
7. A lexical account for non-local wrong can accommodate yet another observation that does not seem
derivable through ellipsis. In scenario (16) below, it was wrong for me to open the unique bottle that I
actually opened. So (11) is true in (16). However, (1) is not in fact judged true in (16). Apparently, (1) not
only implies existence of a bottle that I opened even though I should not have, but also existence of a
bottle that I did not open even though I should have. This implication does not seem derivable in an
ellipsis analysis. In contrast, strengthening the denotation in (13) appropriately is unproblematic.
-1-
Examples
(1) I opened the wrong bottle.
(2) the 1[wrong [for me to open e1] bottle] 2[I opened e2]
(3) I opened the wrong bottle and you closed the wrong bottle.
(4) the 1[wrong [for me to open e1] bottle] 2[I opened e2] and
the 1[wrong [for me to open e1] bottle] 2[you closed e2]
(5) Smith saw every alien I did and she met every alien Jones did.
(6) THE(x. WRONG(w0)(w. I opened x in w) & x is a bottle in w0)(x. I opened x in w0)
(7) You have two bottles of wine, one red and one white, and you ask me to open the red and leave
the white closed. I open the white and leave the red closed.
(8) You have three bottles of wine, one red and two whites, and you ask me to open the red and leave
the whites closed. I open one of the whites and leave the red closed.
(9) You have three bottles of wine, one red and two whites, and you ask me to open the red and leave
the whites closed. I open the red and one of the whites, leaving the other white closed.
(10) You have three bottles of wine, one red and two whites, and you ask me to open the red and leave
the whites closed. I open both of the whites, leaving the red closed.
(11) THE(x. I opened x in w0 & x is a bottle in w0)(x. WRONG(w0)(w. I opened x in w))
(12) the [wrong bottle] 1[I opened e1]
(13) Pet.D(et)((et)t).Qs(et).D(x.Q(w0)(x) & P(x))(x.WRONG(w0)(w.Q(x)(w)))
(14) I opened the wrong green bottle.
(15) a. * I opened the green wrong bottle.
b. * I opened the wrong and green bottle.
(16) You have two bottles of wine, one red and one white, and you ask me not to open either. I open
the white and leave the red closed.
References cited
Kennedy, Christopher: 1998, `Local dependencies in comparative deletion', in K. Shahin, S. Blake, and
E. Kim (eds.), Proceedings of WCCFL 17, CSLI, Stanford.
Larson, Richard: 2000, `ACD in AP?', paper presented at WCCFL 19, Los Angeles, CA.
Szabolcsi, Anna: 1986, `Comparative superlatives', in N. Fukui et al. (eds.), Papers in Theoretical
Linguistics, MITWPL, vol. 8, MIT.
-2-