Deriving Concealed Questions from the Semantics of the Predicate
The underlined DPs in (1a) and (2a) are known in the literature (Heim 1979, Romero 2004) as `con-
cealed questions' (CQ) because they can be paraphrased by the corresponding embedded questions in
b. In this paper, I argue that these DPs are definite descriptions denoting properties, and that the so-
called CQ-reading derives from semantic restrictions imposed by the embedding predicate.
   Prima facie, the class of predicates that allows for CQ-readings seem to overlap with the class of
predicates embedding questions. This parallel, however, breaks down in direct perceptual reports. See,
for instance, is a question-embedding verb. However, when see is used in an epistemically neutral
perceptual report (in the sense of Barwise (1981)) as in (3), it doesn't admit a CQ-reading (3'). Notice
that when see is used in an epistemically positive sense (4), the CQ-reading is available.
    I propose that CQ-readings only arise with Factive-Intensional predicates (FIP) because they
select for an argument that characterizes both the external (res, Lewis (1979)) and the internal
(descriptive) content of the attitude (see also Kratzer 2000). The intuitive analysis is sketched in
(5). In direct perceptual reports like (3), see selects only for an external event or situation. Therefore,
the lack of descriptive content is responsible for the absence of the CQ-reading.
My account assumes that definite descriptions can denote properties instead of individuals. Mikkelsen
(2004) argued that this is the case for subjects of specificational clauses. In contrast to the predica-
tional clause in (6), the use of `it' in (7) indicates a property-type interpretation of the definite descrip-
tion. Notice that the same `test' can be applied to the CQ in (8).
   If property-denoting DPs can have CQ-readings, we predict that indefinite-descriptions can too. In-
tuitively, this seems correct, as shown by (9) and its paraphrase (10). Interestingly, once we consider
indefinites as well, embedded question paraphrases do not capture the meaning of the CQ-reading
anymore, as shown in (11). To the extent that (11) is good, it implies that John knows the exhaustive
list of doctors that can help you. On the other hand, in order for (10) to be true, John doesn't need to
have an exhaustive list of doctors in mind. Indefinite descriptions thus show support to the idea that
CQ-readings are not concealed questions in the literal sense.
   Not every property-denoting DP has an easy-identifiable CQ-reading, as shown in (12). Heim(1979)
and Romero(2005) have argued that CQ-readings are only possible with DPs that denote individual
concepts (). The notion of individual concept, however, is too unconstrained to rule out exam-
ples like (12), since in principle we can treat any definite description as a function from possible
worlds to individuals. Notice, moreover, that if we add a Relative Clause (RC) to (12), as in (13), the
sentence becomes completely acceptable and has a CQ-reading. I will argue that this happens because
the RC in (13) is providing an informative property that was missing in (12). Intuitively, (12) is unac-
ceptable because, by uttering that, one would express the odd uninformative statement `I know of this
entity that it is shoes'. By adding a relative clause in (13), on the other hand, we are contributing an
informative property (`being the shoes that you like') and turn the statement into an informative one.
   Finally, I address certain ambiguities discussed in Heim (1979) and Romero (2005). Imagine (14) is
uttered in a context where the price of a particular car is at issue. In this context, (14) might be inter-
preted as (14a), but also as (14b). The analysis I am proposing can easily account for reading (b), as
paraphrased in (15), but not as easily for reading (a). Examples like these, motivated Heim to consider
the relevant property to be contextually determined and Romero to argue for an intrinsic ambiguity of
the predicate. I will argue that another, independently motivated analysis is possible. Following Fox
(2002), I will show that reading (a) can be accounted for by the fact that RCs can be late merged. By
result of late-merge, the RC is not under the scope of the intensional predicate and it cannot contribute
to the determination of the relevant property. In support of this hypothesis, notice that when we force
extra-position by inserting an adverb between the head and the RC, reading (b) is no longer available
for (14), as shown in (16), and examples like (13) are no longer grammatical, as shown in (17).
   To sum up, I proposed an account that derives CQ-readings from semantic properties of FIP. This
solution captures the restricted distribution of CQs, and it correctly characterizes the non-exhaustive
meaning of an indefinite CQ. I showed how RCs can provide relevant information that rescues other-
wise uninformative statements. Finally, I showed how extraposition of a RC can affect the derivation
of CQ-readings, providing a new analysis of Heim's ambiguities that relies on independently moti-
vated assumptions about extraposition.

 Examples

1a. Kim knows the governor of California. b. Kim knows who the governor of California is.
2a. She revealed the winner of the contest. b. She revealed who the winner of the contest was.

3. The janitor actually saw the culprit leaving the scene of the crime, but at that time he didn't know
he was witnessing a crime.
3' The janitor actually saw the culprit...  The janitor saw who the culprit was.
4. I see your point. (`I see what your point is')

5. Kim knows the governor of California =
   Kim knows of Arnold Schwarzenegger that he is the governor of California

External content                                     Internal content
There is an actual x of which P holds in w0.         Kim believes de re of x that it has property P.


6. The winner of the contest is Iranian. Isn't she/ *it?                      PREDICATIONAL
7. The winner of the contest is Susan. Isn't it/ *she?                      SPECIFICATIONAL
8. I know the winner of the contest. It's Susan/ *She is Susan.                     CQ

9. John knows a doctor who can treat your illness.
10. John knows of a certain person that he is a doctor who can treat your illness.
11 John knows who is a doctor that can treat your illness.

12. ?? I know the shoes. (#I know what the shoes are.)
13. I know the shoes you like

14. John was told the price that you were told.
(a) John and you were told the same price for that car, say $10,000.
(b) John was informed of the price they told you. He might have been told, for example, that you were
     told the price of the car was $10,000.
15. John was told of $10,000 that was the price (of the car) that you were told.

16. John was told the price yesterday that you were told.
17. #? John found out the shoes yesterday that you like.


References: Barwise, J. 1981: "Scenes and other Situations." The Journal of Philosophy 78, 369-
397, Fox, D. 2002: "Antecedent Contained deletion and the copy theory of movement, Linguistic In-
quiry 34, Heim, I. 1979. "Concealed Questions", in R. Bäuerle, U. Egli and A. von Stechow (eds.),
Semantics from different point of views, Berling: Springer-Verlag. Pg 51-60, Kratzer, A. 2002.
"Facts: Particulars of information units?" Linguistics and Philosophy, 25:655­670, Lewis, D.K. 1979:
"Attitudes De Dicto and De Se". Philosophical Review 88, Mikkelsen, L.2004: "Specificational sub-
jects- a formal characterization and some consequences". Acta Linguistica Hafniensia 36, Romero,
M. 2005 "Concealed Questions and Specificational Subjects" submitted to Linguistics and Philoso-
p h y.