The Semantics Of Nominal Exclamatives 1 Introduction 2 The ...

The Semantics of Nominal Exclamatives
Paul Portner and Raﬀaella Zanuttini
1
Introduction
In this contribution we consider a type of exclamative construction in English which shows an
unusual pairing between syntactic form and semantic/pragmatic function. This is the nominal
exclamative, illustrated in (1):
(1)
The strange things that he says!
(2)
What strange things he says!
(1) is a noun phrase, and thus contrasts syntactically with (2), which is a clause. Yet the two
seem to be synonymous. We will argue that the noun phrase is not embedded in an elliptical
structure; rather, the phrase we see is all there is. And furthermore we will argue that (1) is not
just pragmatically equivalent to (2); the two are in fact semantically equivalent as well. This raises
the question, of central concern to this volume, of how a noun phrase achieves such a clause-like
function.
More speciﬁcally, we will argue that certain noun phrases containing a relative clause can have
either “clause-like” or “noun phrase like” meanings. When the meaning is clause-like, it is parallel
to that of the clausal exclamative in (2); that is, the noun phrase has the type of meaning of an
interrogative clause (a set of propositions, as in Karttunen 1977). We refer to such cases with the
term nominal exclamative. When the meaning is noun phrase like, it is that of an ordinary deﬁnite
NP. The syntactic distribution of nominal exclamatives with a clausal meaning is intricate, and
shows that an attempt to account for the two meanings in terms of ellipsis or pragmatic inference are
inadequate. This leads us to address the question of how the clause-like meaning is compositionally
derived. We note that only noun phrases that contain a relative clause may have the clause-like
meaning, and based on this we propose an account of (1) which crucially builds on the fact that
it contains a relative clause, in this case that he says. More precisely, we propose that the noun
phrases with a clause-like meaning share an abstract morpheme, present in the relative clause, with
the true clausal exclamatives like (2). In contrast, the noun phrase like meaning is not associated
with such a morpheme.
2
The category of exclamatives
Natural language categorizes sentences into a small number of universal clause types (Sadock &
Zwicky 1985).
Clause types constitute a grammatically determined pairing between form and
meaning. One of these is the “minor type” of exclamatives, and in labelling examples like (1) as
nominal exclamatives we intend to include them within this class. Because the fact that they are
1

exclamatives will become relevant at various points in our discussion, we begin by making precise
our reasons for considering them to be members of this type.
It is not always a straightforward matter to determine which sentences are to be classiﬁed as
members of a given clause type. This is so because sentences may be used with a variety of illocu-
tionary forces beyond that conventionally associated with their form. For example, a declarative
like It’s cold in here can be used, in the right context, with the illocutionary force conventionally
associated with imperatives, i.e. command. Nevertheless, the notion of clause type takes it that
a declarative is basically associated with asserting, and other illocutionary forces that it would
receive are pragmatically determined, for example by implicature. When it comes to exclamatives,
their conventional force is that of exclaiming. Certainly (1), like (2), can be used to exclaim. Our
question is whether (1) is conventionally associated with exclaiming, or whether it is usable in this
way through some indirect pragmatic mechanism.
In previous work (Zanuttini & Portner 2000, 2003; Portner & Zanuttini 2000), we have shown
that exclamatives are associated with certain semantic and pragmatic properties that can be used
to test for membership in the clause type of exclamatives. Here we will pick out a few of these
which make the case that (1) is truly an exclamative in this sense; please see the works cited for
more details.
All exclamatives are factive, and this leads to contrasts like the following:1
(3)
a. I know what strange things he says.
b. *I don’t know what strange things he says.
While (3)a is ﬁne, with an exclamative clause embedded under the factive predicate know, (3)b is
ungrammatical because negating I know conﬂicts with the exclamative’s presupposition of factivity.
We see the same pattern with nominal exclamatives:2
(4)
a. I certainly know the strange things he says.
b. *I don’t know the strange things he says.
Thus nominal exclamatives pass the ﬁrst test for exclamative status.
The second major property of exclamatives is that they generate a characteristic conventional
scalar implicature to the eﬀect that certain elements are at the extreme end of some contextually
determined scale.3 This leads to contrasts like the following:
1(3)b is acceptable on an irrelevant reading where the complement of know is a free relative. We can rule out the
free relative interpretation by adding a modiﬁer to the wh phrase: I (*don’t) know what incredibly strange things he
says.
2As we noted in the introduction, the noun phrases which can function as a nominal exclamative can typically
also be used as an ordinary referential deﬁnite. In our examples, the referential reading is diﬃcult to get, and so they
seem ungrammatical, but similar examples will be acceptable on the (irrelevant) referential reading: I don’t know the
strange things he lectures about.
3The implicature is conventional, rather than conversational, because it does not arise due to reasoning based on
any sort of Gricean Cooperative Assumption. Rather, it is associated with this particular linguistic form. Note that
one could not explain the pattern we see with embedded exclamatives by appealing to the conversational use of the
corresponding non-embedded forms. When embedded, they do not have a conversational use of their own. Similarly,
as pointed out by Rob Stainton, Big guy! can be used as an exclamative, but when it’s part of a larger sentence, it
has no such conversatinal use of its own: I have met that big guy before.
2

(5)
a. It’s amazing what strange things he says.
b. *It isn’t amazing what strange things he says.
c. *Is it amazing what strange things he says?
Although an exclamative can be embedded under the predicate amazing, negating or questioning
this predicate gives rise to ungrammaticality. This is because negation denies the “amazingness”,
or extreme scalar quality, of the proposition; similarly, interrogative form casts into doubt its truth.
Since conventional implicatures are non-defeasible, the result is unacceptable. We ﬁnd the same
pattern with nominal exclamatives:4
(6)
a. It’s amazing the strange things he says.
b. *It isn’t amazing the strange things he says.
c. *Is it amazing the strange things he says?
Based on these tests, we conclude that the examples we are focusing on share the deﬁning semantic
and pragmatic properties of the exclamative clause type.
3
Properties of nominal exclamatives
The goal of this section is to go back to example (1) to show that it is indeed (a) a noun phrase and
(b) an exclamative with the same type of meaning as a clausal exclamative like (2). Demonstrating
the ﬁrst point is straightforward: we will show that both the internal structure and the distribution
of these phrases is that of an NP. As for the second point, subject agreement allows us to distinguish
subjects which have a clause-like meaning from those which have a noun phrase like meaning. Thus
in this section we argue that nominal exclamatives pattern with noun phrases in terms of their
structure and with clausal exclamatives in terms of their semantic type. In the next section we will
investigate how this unorthodox pairing of form and meaning can be compositionally derived.
1. A superﬁcial inspection of (1) certainly suggests that it is a noun phrase: it contains a deﬁnite
determiner, an adjective, a head noun and a relative clause. This string can certainly function
as a noun phrase in examples like (7):
(7)
We discussed the strange things that he says.
However, one can imagine other analyses of (1) in its exclamatory function; perhaps the
strange things has been topicalized from the object position of says. That is, it derives from
a structure like that he says the strange things by fronting of the deﬁnite NP. In this case the
presence of the complementizer would be unusual for English, but maybe this construction
is diﬀerent from other root clauses in the licensing of overt complementizers. We see a case
roughly of this sort in embedded contexts in Italian. In the following, a hanging topic precedes
an overt complementizer (Beninc`
a & Poletto 2004):
(8)
Mi
dispiace
[questo libro [che non ne
abbia parlato nessuno]].
to-me displeases this
book that neg of-it has
spoken nobody
4Example (6)c is similar to the grammatical example Are they amazing, the strange things he says? However, in
this case we have subject verb agreement, and this corresponds to the referential interpretation of the noun phrase;
see section 3 for discussion.
3

‘It bothers me that, as for this book, nobody has said anything about it.’
The embedded clause questo libro che . . . bears a passing similarity to example (1), and yet
it is undoubtedly a clause, and so we should consider whether (1) might be clausal also.
One piece of evidence that (1) is nominal rather than clausal comes from the parallelism it
shows with undisputed relative clauses:5
(9)
a. It’s amazing the strange things that/which/*who/∅ he says!
b. Two nice houses that/which/*who/∅ he would like to buy are on sale.
(10) a. It’s amazing the strange people that/which/who/∅ he invited!
b. Two cute girls that/which/who/∅ he met stole the cookies.
(11) a. It’s amazing the unbelievable trees that/which/*who/*∅ grow there!
b. Two of the small palms that/which/*who/*∅ grow there came from Australia.
(12) a. It’s amazing the wierd friends that/?which/who/*∅ visit him all the time!
b. Two British friends that/?which/who/*∅ teach at Oxford asked questions.
In (9), we see that the nominal exclamative containing an inanimate head noun and an object
gap allows that, which, and ∅; these possibilities are exactly the same as those in the clear
relative clause. Similar patterns can be seen with the animate noun/object gap, an inanimate
noun/object gap, and animate noun/object gap cases in (10)–(12).
The possibility of wh words in examples like (10)a poses a signiﬁcant problem for the view
that this is actually a clausal structure. If the strange people has been moved from the object
position, there is no source site for which or who. If, however, we take the wh word to have
been moved from object position, this requires that we treat the strange people as unmoved
left-dislocated element. Since left-dislocation in English always requires the presence of a
resumptive pronoun, this analysis is ruled out.6 Thus, the obvious conclusion from the data
in (9)–(12) is the right one: nominal exclamatives are simply noun phrases containing a
relative clause.
This conclusion is further supported by examining the external distribution of nominal excla-
matives. We must restrict our attention to those predicates which support the exclamative
interpretation, but if we do we see that nominal exclamatives only occur as the complements
of predicates which subcategorize for a noun phrase. Grimshaw shows that, among those
predicates that embed exclamatives, some subcategorize for both noun phrases and clauses,
whereas others only take clauses (data from Grimshaw 1979, section 4.1):
(13) a. John couldn’t believe what a height the building was. (clausal complement)
b. John couldn’t believe the incredible height of the building. (nominal complement)
(14) a. I don’t give a damn what an outrageous size his salary is. (clausal complement)
b. *I don’t give a damn the outrageous size of his salary. (nominal complement)
5We embed the (a) examples under it’s amazing to make sure they are read as exclamatives rather than ordinary
deﬁnite NPs. Other exclamative-embedding contexts, like I am amazed at, allow both exclamative and ordinary NP
complements. Thus, I am amazed at the strange people that he invited could have an embedded-exclamative meaning
(= ‘I am amazed at what strange people he invited’) and a referential complement meaning (= ‘I am amazed at
certain people, namely the strange people that he invited’; i.e., I’m amazed at some aspect of these people, though
not necessarily at the fact that he invited them).
6One might consider the possibility that the wh word acts as the resumptive pronoun, but there would be no other
instances of such a use in English. Moreover, the cases which lack a wh word would have no overt resumptive pronoun.
This means we’d have to say that the initial material is a topic when there is no wh word and a left-dislocated element
when there is, a complicated possibility for which there is no evidence.
4

Both couldn’t believe and give a damn support the exclamative interpretation; that is, they
‘s-select’ exclamatives in Grimshaw’s terms. However, whereas couldn’t believe subcategorizes
for both clauses and noun phrases (‘c-selects’ these categories, in Grimshaw’s terminology),
give a damn only takes clauses. As would be predicted if they are noun phrases, nominal
exclamatives are only possible with predicates of the ﬁrst type:
(15)
John couldn’t believe [the strange things she said].
(16)
*I don’t give a damn [the strange things she said].
A potential problem with our view is the fact that the complement of (15) is very similar to
what Grimshaw calls ‘concealed exclamatives’, following Elliott (1974) and ultimately Baker
(1970). A concealed exclamative is a noun phrase which fulﬁlls the s-selection requirements of
a predicate that takes an exclamative complement (likewise there exist concealed questions,
examples from Grimshaw 1979:299):
(17)
John asked the height of the building. (concealed question)
(18)
John couldn’t believe the height of the building. (concealed exclamative)
Notice that these examples are synonymous with John asked how high the building was and
John couldn’t believe how very high the building was, respectively. Thus one might wonder
whether we are justiﬁed in treating examples like (1) as a distinct class of ‘nominal exclama-
tives’ separate from concealed exclamatives.7 There is a crucial diﬀerence between the two
constructions. The examples which are the focus of this paper, i.e. (1) and (15), can stand
alone with exclamative sentential force, whereas the concealed exclamative in (18) cannot. So,
although #The height of the building! can be used to exclaim, it is elliptical, in the sense that
it feels like a sentence fragment. It refers to something to which we attribute a contextually
provided property. In contrast, examples likes The strange things he says! are not perceived
as elliptical in the same way. As we will see in the discussion below, the nominal exclamatives
have a full exclamative meaning on their own, and in this way contrast with ordinary noun
phrases used as concealed exclamatives or with the indirect illocutionary force of exclaiming.
2. We have just shown that nominal exclamatives really are nominal: they have the syntax of
noun phrases. The other side of our claim is that they have the sort of meaning which is
typically associated with a clause, and are essentially synonymous with clausal exclamatives.
Intuitively, as already mentioned, this way of looking at nominal exclamatives is supported
by the fact that they do not feel elliptical when uttered out of the blue. In this respect, they
contrast with ordinary noun phrases used with the function of exclaiming:8
(19)
The silly questions these students ask!
(20)
Those silly questions!
One might wonder if the reason the noun phrase in (20) feels elliptical is that it lacks suﬃcient
informational content. However, there are cases in which an ordinary noun phrase (lacking a
relative clause) has the same content as a nominal exclamative, but it still feels incomplete:9
7In fact, Elliott (1974) calls examples like (15) concealed exclamatives. Against this, we are arguing that the two
kinds of examples need to be distinguished, even though they are both syntactically noun phrases.
8Bare wh phrases like What silly questions! are more able to stand on their own than deﬁnites lacking a relative
clause. Perhaps they are elliptical for clausal wh exclamatives (which, in that case, would be seen as tolerating ellipsis
better than nominal exclamatives and declarative clauses).
9It’s interesting to note that (22) is much worse with the than with those. We are not sure why this would be the
5

(21)
The strange people who come from Italy!
(22)
The/those strange Italians!
Data of the kind above suggest that only deﬁnite NPs which contain a relative clause can be
true nominal exclamatives; that is, only they can have exclamative sentential force. But while
it is certainly important that speakers’ semantic intuitions support a diﬀerence of this sort,
we would like to have some corroborating grammatical data. We ﬁnd such data in agreement
facts with predicates like be amazing. To begin, ordinary noun phrases trigger agreement
with be amazing when they are in subject position. When they are in post-predicate position,
they act as right-dislocated arguments, and agree with the predicate via a pronoun in subject
position:
(23)
Those Italians are amazing.
(24) a. They’re amazing, those Italians.
b. *It’s amazing, those Italians.
Nominal exclamatives behave like ordinary noun phrases when in subject position, but in
post-predicate position, they are pronominalized by it and do not trigger agreement:
(25)
The strange people who come from Italy are amazing.
(26)
It’s amazing the strange people who come from Italy.
It is possible to have an agreeing pronoun and copula with the strange people who come from
Italy in post-predicate position, but the meaning is diﬀerent from (26):
(27)
They’re amazing, the strange people who come from Italy.
(27) says that the people themselves are amazing. In contrast, (26) does not imply that
the people themselves are amazing. Rather, it is some aspect of the information that there
are such people who come from Italy which is amazing; for example, one might be amazed
that they are strange to such a high degree or that so many of them are strange. As these
paraphrases make clear, the contribution of the strange people who come from Italy is a group
of individuals in (27) and some sort of propositional meaning in (26). That is, (27) (like
(25)) contains an ordinary noun phrase followed by a relative clause whereas (26) contains a
nominal exclamative.
The lack of agreement in (26) is reminiscent of the behavior of clauses linked to the subject
argument:
(28) a. It’s amazing what strange things he says.
b. *They are amazing what strange things he says.
(29)
It’s surprising that they won.
(30)
It’s a good question whether they came to the party.
We attribute the fact that clauses do not trigger agreement to the fact that their meaning is
propositional rather than noun phrase-like (referential or quantiﬁcational). More speciﬁcally,
when in subject position they cannot trigger agreement because they lack agreement features
(person and number); in post-predicate position, they are pronominalized by it and therefore
are associated with a third person singular verbal form. Based on this, we can provide an
case, but it does show that there is something special about the nominal exclamative. Moreover, those strange people
who come from Italy seems incomplete, and no longer has the feel of a nominal exclamative
6

explanation of the data in (25)–(27). On our view, the string the strange people who come
from Italy has a single structural analysis, but can have either a propositional or a nominal-
type interpretation. In example (26) the phrase in post-predicate position has a clause-like
meaning, and hence does not trigger agreement. More speciﬁcally, we assume that because
it is propositional, it is pronominalized by it and hence is associated with the third singular
form of be. In (27) it has a nominal meaning, and so is pronominalized by they, yielding plural
agreement.
In contrast to the post-predicate cases, as seen in (25) the preverbal NP always triggers
agreement. This is simply because it has person and number features and is in the right
syntactic conﬁguration for agreement to occur. Note that, in the presence of agreement, the
clausal interpretation is ruled out; this leaves us with only the nominal one. The question is
why there should be such a connection between agreement and interpretation. Of course this
question goes beyond the present case. It involves asking why in general constituents which
have a propositional interpretation (canonically, clauses) always cooccur with third person
singular verb and pronoun forms, plausibly the default forms. We do not seek to address this
question here.
Note that the pattern just described cannot be accounted for under an analysis which treats
the strange people who come from Italy as syntactically ambiguous between an agreement-
triggering phrase (that is, a noun phrase) and a non-agreement-triggering one (e.g. a clause).
Under such a proposal, one could not explain why agreement is required in (25), since we
would have the option of having the non-agreement-triggering form in subject position. This
point also argues against an ellipsis account of nominal exclamatives. One would have to
propose that ellipsis is impossible in subject position, given that we always get agreement
and a non-propositional interpretation, but as far as we can see this restriction would have
to be simply stipulated. Of course the present approach is incomplete as well, since we
don’t fully understand the connection between agreement and semantic type, but at least
the generalizations we appeal to are independently motivated. In general, a noun phrase in
subject position triggers agreement, while phrases with propositional meaning never trigger
agreement.
These requirements cannot both be met in a sentence containing a nominal
exclamative in subject position, and so the grammar rules out this possibility.
4
Compositional interpretation of nominal exclamatives
Having shown that nominal exclamatives are structurally noun phrases but semantically clause-
like, we now turn to the question of how this unusual pairing of form and meaning can arise.
Given their syntactic parallelism with interrogatives, we assume that exclamatives have meanings
of the same type as questions (cf. Portner & Zanuttini 2000, Section 5). That is, they denote
sets of propositions (Hamblin 1973, Karttunen 1977, etc.) Of course there are diﬀerences between
interrogative and exclamative denotations, but this won’t be our focus here. These diﬀerences are
of a pragmatic nature; in particular, exclamatives and interrogatives have diﬀerent sentential forces.
The simplest option is perhaps to postulate an empty operator of some sort which converts an
ordinary deﬁnite NP meaning into the appropriate question-like meaning. A traditional syntactic
analysis of (31)a would be something like (31)b:
(31) a. The things that are in this bag!
7

b. [DP the [NP things [CP whichi C [IP ti are in this bag]]]]
Assuming that this noun phrase is referential, an operator with the meaning in (32)a would convert
this into the appropriate set of propositions in (32)b:10
(32) a. [λf . {p : w0∈p & ∃y[p = {w : y ≤ f(w)}]}]
b. {p : ∃y[p is true and p = ‘y is one of the things in this bag’]}
However, such an analysis does not explain why the relative clause is required in true nominal
exclamatives. A better strategy would be to build the semantic analysis somehow on the presence
of the relative clause.
We will pursue an alternative based on Kayne’s (1994) analysis of relative clauses. Kayne’s
approach is promising because it takes noun+relative clause sequences to actually be full clauses
(CPs) which are the complement of the determiner. A wh phrase containing the “head noun” begins
within the IP and raises to the speciﬁer of CP. The key diﬀerence from the traiditonal account is
the idea that the noun which the relative clause modiﬁes originates internal to it.11
(33)
[DP the [CP [things which]i C [IP ti are in this bag]]]]
We would like to exploit the basically clausal nature of this analysis to derive the propositional
meaning of nominal exclamatives. This would also allow us to explain why nominal exclamatives
must contain a relative clause, since noun phrases lacking relative clauses don’t have this kind of
structure.
Before tackling the exclamative interpretation of (33), perhaps it would be helpful to consider
how an ordinary referential interpretation could be derived for such a structure. Within the CP, we
have two set-denoting expressions: things and ti are in the bag.12 These two sets are intersected,
either by virtue of the meaning of which, which would then simply be predicate conjunction, or
through a default rule of predicate conjunction (cf. for example, Heim & Kratzer 1998). In the
latter case, which would be treated as semantically vacuous. The whole CP would then denote a
set of individuals, here the things which are in the bag, and the determiner would take this set as
argument in the ordinary fashion.
The interpretation of nominal exclamatives can be attributed to an additional morpheme E
somewhere in the structure. Though various are conceivable (E in D, adjoined to CP, or adjoined
to DP, among others), we will pursue the idea that E is part of the wh phrase. The motivation for
this is the fact that wh phrases in clausal exclamatives often have a unique form, diﬀerent from
their interrogative counterparts. For instance, the a in what a dog and very in how very tall are
markers of exclamative status and could be identiﬁed with a similar E morpheme. A wh word with
10This operator gives the set of true propositions of the appropriate form. Alternative meanings for questions could
be handled with appropriate changes to our operator. What we have to say could work as well if deﬁnite NPs are
quantiﬁcational, though we will consider only the referential analysis for simplicity. Note that we assume that the
operator combines with the intension of the NP meaning, corresponding to the variable f.
11Assuming that we start out with which things as a constituent, Kayne derives the order things which by moving
things from the complement to the speciﬁer position of a QP headed by which: [QP thingsi [Q which ti]]. We omit
this detail for simplicity. An unresolved issue is how the precise form of the wh phrase is chosen, though this is a
problem that aﬄicts traditional accounts of relative clauses as well. For instance, why is a null operator available in
object relatives but not object interrogatives?
12Various techniques could be used to create a set out of the IP ti are in the bag. In terms of Heim & Kratzer’s
system, the index i from the wh phrase would be adjoined to C’, and this would trigger a rule of Predicate Abstraction.
8

the following interpretation achieves the correct result:
(34) a. [ whichEXCL ] = [λPλQ . {p : w0∈p & ∃x[p = {w : P(w)(x)=1 and Q(w)(x)=1}]}]
b. {p : w0∈p & ∃x[p = {w : x is a thing in w and x is in the bag in w}]}
If we take the wh words in ordinary relative clauses to be meaningless, (34)a would simply be the
interpretation of E. If, on the other hand, they contribute predicate conjunction, we can still give
a compositional interpretation, though E’s meaning would be slightly more abstract.13 In either
case, the wh word denotes a function which combines with two properties: that denoted by the
noun (represented by P) and that denoted by the IP (represented by Q). The value of this function
is then the set of true propositions of the form ‘x is a P and x is a Q’.
The analysis that we have given provides a clausal meaning to the nominal exclamative. More-
over, by adopting Kayne’s view of relative clauses, the mapping between syntactic categories and
semantic types is better motivated than under the alternative. The relative clause CP denotes a set
of propositions, and the wh word denotes a function from common noun meanings and predicates
to such a set of propositions. These are completely normal meanings for CPs and wh words to have.
In particular, we provide the relative wh word with the type of meaning normally associated with
interrogative wh words in this case, and this explains the synonymy between clausal and nominal
exclamatives. In contrast, a traditional structural analysis of relative clauses could make use of an
operator like that in (34)a. However, this would result in the relative clause CP denoting a function
from common noun meanings to sets of propositions—an unprecedented semantic type for CPs.
One issue that comes up at this point is the role of the determiner the in nominal exclamatives.
Given what we have said, the sister of the denotes a set of propositions, and this is not the ordinary
type to be the argument of a determiner. We would like to suggest that an answer to this puzzle
resides in a point that we made earlier, that exclamatives are always factive. We propose that the in
this case marks factivity. This idea is pretty natural from the perspective of a theory of deﬁniteness
which makes the sole function of the to be that of marking a familiarity presupposition (Heim 1982,
for example). Factivity has a clear intuitive relation to familiarity, and so it’s plausible that the
function of the could be extended to marking factivity.14 Moreover this idea lets us explain why no
other determiner is possible in nominal exclamatives.
There are other possible ideas for how factivity is marked in nominal exclamatives. In other
work, we have proposed that factivity is marked in clausal exclamatives by a null operator in the
CP domain. Such an operator could exist in nominal exclamatives as well, presumably in the
relative clause which is obligatorily present. We are not aware at this point of empirical evidence
that would let us decide between these alternatives.
13This proposal raises the question of why nominal exclamatives cannot have the clausal exclamative-type of wh
phrase: what a and how very. Presumably this reduces to the broader question of why the wh phrases in relative
clauses diﬀer in form from their interrogative counterparts (cf. footnote 11).
14On other approaches to the deﬁnite article, where the would have some semantic function in addition to marking
a deﬁniteness presupposition, we could still see its use in marking factivity as an extension of its usual function,
though the shift would be greater.
9

5
Conclusion
Our central contribution has been to argue that nominal exclamatives like The strange things that he
says! are noun phrases but have a semantic content of the kind more typical of clauses. In particular,
we hypothesize that they have the semantic type of wh clauses, i.e. denoting a set of propositions.
They are not derived by ellipsis from full clauses, nor is their similarity to clausal exclamatives
to be explained by a pragmatic process like implicature. This combination of properties may be
at ﬁrst glance surprising, since it shows that grammar is not constrained by simple correlations
between syntactic category and semantic type like noun phrase → entity/predicate/quantiﬁer and
sentence → proposition/set of propositions. This raises the question of whether this mapping is
irregular and arbitrary.
Towards answering this question, we have argued that Kayne’s (1994) syntactic analysis of relative
clauses provides a better foundation than more traditional views for explaining the pattern of
nominal exclamatives found in English. In particular, the fact that, according to Kayne’s view, a
noun phrase with a relative clause ultimately consists of a determiner taking a clausal complement
helps us see why nominal exclamatives always contain relative clauses. Hence, the fact that nominal
exclamatives constitute an exception to the typical relationship between syntactic category and
semantic type is not as surprising as it seems at ﬁrst. The interpretation of these forms derives
from the presence of clausal structure in combination with a hypothesized exclamative morpheme.
This leads to a view of the syntax/semantic interface which takes the category/type mapping to
be derived compositionally from the internal makeup of the syntactic forms. The mapping is not
arbitrary; it is motivated “from the bottom up”. Regularities exist in the mapping because the
internal structure of categories shows consistency, but exceptions can arise when some aspect of
the structure is unique. Nominal exclamatives provide such a case.
References
Baker, C. L. 1970. Notes on the description of English questions: The role of an abstract question
morpheme. Foundations of Language 6:197–219.
Beninc`
a, Paola and Cecilia Poletto. 2004. Topic, focus and V2: Deﬁning the CP sublayers. In The
Structure of CP and IP: The Cartography of Syntactic Structures, 52–75. New York and Oxford:
Oxford University Press.
Elliott, Dale E. 1974. Toward a grammar of exclamations. Foundations of Language 11:231–246.
Grimshaw, Jane. 1979. Complement selection and the lexicon. Linguistic Inquiry 10(2):279–326.
Hamblin, C.L. 1973. Questions in Montague English. Foundations of Language 10:41–53.
Heim, Irene. 1982. The Semantics of Deﬁnite and Indeﬁnite Noun Phrases. Amherst: GLSA.
Heim, Irene and Angelika Kratzer. 1998. Semantics in Generative Grammar . Blackwell.
Karttunen, Lauri. 1977. Syntax and semantics of questions. Linguistics and Philosophy 1:3–44.
Kayne, Richard S. 1994. The Antisymmetry of Syntax . Vol. 25 of Linguistic Inquiry Monographs.
Cambridge, Mass.: MIT Press.
Portner, Paul and Raﬀaella Zanuttini. 2000. The force of negation in Wh exclamatives and in-
terrogatives. In L. Horn and Y. Kato (Eds.), Studies in Negation and Polarity: Syntactic and
Semantic Perspectives, 201–39. New York and Oxford: Oxford University Press.
Sadock, Jerrold M. and Arnold Zwicky. 1985. Speech act distinctions in syntax. In Timothy
10

Shopen (Ed.), Language Typology and Syntactic Description, 155–196. Cambridge: Cambridge
University Press.
Zanuttini, Raﬀaella and Paul Portner. 2000. The characterization of exclamative clauses in Paduan.
Language 76(1):123–32.
Zanuttini, Raﬀaella and Paul Portner. 2003. Exclamative clauses: At the syntax-semantics inter-
face. Language 79(3):39–81.
11