The hypothesis of insubordination and three types of whexclamatives
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
Abstract
This paper provides evidence for Evans’ (2007) insubordination hypothesis w.r.t. whexclamatives. It investigates word order in matrix wh-exclamatives and their subordinate
correlates and tests matrix-subordinate asymmetry (the felicitousness of wh-words in matrix and
subordinate contexts). It establishes distinctions among three groups of wh-exclamatives. Group 1
comprises qualitative and quantitative wh-exclamatives, which together seem to be a basic crosslinguistic wh-exclamative pattern. The qualitative variety demonstrates several strategies of using
wh-words, some of which are exclamative-only and/or are sensitive to ellipsis of a gradable
adjective/adverb. Group 2 implies the semantic hierarchy w.r.t. the felicitousness of wh-words in
matrix exclamatives: ‘what’/‘who’/‘where’>‘when’>‘why’. Group 3 includes ‘which’, ‘what
kind’, ‘how’ (manner) exclamatives. Unlike Groups 1 and 3, Group 2 is subject to cross-linguistic
variation w.r.t. matrix-subordinate asymmetry. The paper suggests partial overlap between the
established classification of wh-exclamatives and the classification developed by Nouwen and
Chernilovskaya (2015) and have implications for an exclamative sentence type.
Key words: hypothesis of insubordination, syntax of exclamatives, semantics of exclamatives,
typology of exclamatives, embedded wh-exclamatives
1 Introduction
22
23
24
25
26
27
28
29
30
31
32
33
1.1 Theoretical background
There has been tacit agreement in the literature that sentences like (1a-b) are genuine exclamatives:
(1a) What a beautiful rainbow I saw yesterday!
(1b) How very beautiful this rainbow is!
Semantically, these examples convey the meaning that the rainbow under discussion is remarkably
beautiful,1 and the speaker expresses emotion (typically surprise)2 regarding this fact. Formally,
both sentences resemble subordinate interrogatives with respect to the word order and the whphrases they contain; however, unlike subordinate interrogatives, they have exclamative-only
elements: what a + NP in (1a) and how very + adjective in (1b).3
By saying “remarkably beautiful” I try to be as neutral as possible within the frame of existing exclamativity
approaches.
2
Surprise has been viewed as a necessary semantic ingredient of exclamatives in most existing exclamativity
approaches. Moreover, Rett and Murray (2013) ascribes exclamatives semantics of mirativity encoded grammatically.
However, it has been argued in the literature (e.g., by Badan and Cheng, 2015) that ego-evidentiality, which was
introduced in Marandin (2008), is an intrinsic characteristic of exclamatives rather than surprise.
3
Some papers argue that wh-exclamatives are free relatives (see Rett 2008a, 2008b, Koenig and Siemund 2007 among
others). However, either they confess that this issue is controversial and needs further research, or they do not comment
on the claim. Most of the papers (cf. Grimshaw 1979, Huddleston 1993, Michaelis 2001, Zanuttini and Portner 2003,
1
2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
Although every year the body of studies on exclamatives in various languages grows in
geometric progression, exclamatives still do not seem to be a well-established pragma-semantic
and/or syntactic category, neither in functional nor in formal approaches. One of the most
problematic issues crucial for the general theory of exclamatives is whether wh-exclamatives are
(non-)embeddable. In this paper, relying upon the data from a synchronic cross-linguistic study,
my goal is to provide further evidence for the hypothesis of insubordination w.r.t. wh-exclamatives
(cf. Evans (2007), Koenig and Siemund (2007, 2013)), which states that matrix wh-exclamatives
have diachronic origins as subordinate clauses. Moreover, I show that, as a consequence, the (non)embeddability problem is not applicable to wh-exclamatives. This requires the following change
in terminology which I consistently follow in this paper: the notion of embedded exclamatives
should be replaced with the notion of subordinate wh-clauses with exclamative readings.
To be more precise, in this paper, I investigate the data from eleven languages and justify
the hypothesis of insubordination w.r.t. wh-exclamatives by testing the following two observations
circulated in the literature.
The first observation was made by Elliott (1974: 233), who said that word order in matrix
wh-exclamatives is identical to word order in subordinate wh-exclamatives (or subordinate whclauses with exclamative readings, in my terms) as well as to word order in subordinate
interrogatives. I advocate this idea w.r.t. the following parameters: (i) obligatoriness of whmovement, (ii) the position of a (moved) wh-phrase, and (iii) obligatoriness of subject-verb
inversion. If the structures are identical, this is compatible with the hypothesis of insubordination,
i.e., the following scenario is plausible: matrix wh-exclamatives emerge from subordinate whclauses with exclamative readings which, in its turn, belong to the same syntactic class as
subordinate interrogatives.
The second observation originates with Elliott (1974: 232), who argued that a set of whwords acceptable in subordinate exclamatives (or subordinate wh-clauses with exclamative
readings, in my terms) forms a superset of wh-words acceptable in matrix exclamatives (so-called
matrix-subordinate asymmetry). In other words, if a language allows for a matrix wh-exclamative,
it allows for a subordinate wh-clause with an exclamative reading, but not vice versa. I show that,
cross-linguistically, this is true under certain circumstances. In doing so, I test, on the one hand,
exclamative constructions of wh-words and gradable adjectives or adverbs (hereinafter qualitative
wh-exclamatives) and, on the other hand, wh-exclamatives which involve the following wh-words:
inanimate object (‘what’), personal (‘who’), quantitative (‘how many/ much’), locative (‘where’),
temporal (‘when’), causal (‘why’), kind (‘what kind’), individual (‘which’),4,5 and manner (‘how’).
Firstly, I demonstrate that qualitative and quantitative wh-exclamatives are felicitous in
matrix and subordinate contexts in all the languages of the sample. They therefore seem to be
Evans 2007, Nouwen and Chernilovskaya (2015) among others) do not support this view and discuss wh-exclamatives
in comparison to wh-interrogatives, arguing either that such structures are interrogatives, or that they form a distinct
class of wh-clauses. The evidence for why wh-exclamatives are not likely to be free relatives comes from the fact that
subordinate wh-clauses with exclamative readings allow only for factive (uses of) matrix predicates but not for
predicates like eat, cf. (i)-(ii). Free relatives can follow verbs such as eat.
(i)
I am surprised at what you cooked. (exclamative reading, not a free relative)
(ii)
I will eat what you cooked. (non-exclamative reading, free relative)
4
I refer to wh-words with the meaning ‘which’ as individual since they typically have individual readings in terms of
Rett (2008a, 2008b, 2011). I do not suggest developing a new terminology for this class of words.
5
Actually, individual readings are possible for clauses with most, if not all, wh-words, cf.: I wonder what car Mary
bought – Mercedes or BMW; I wonder where you stopped – on Main Street or near the park.
3
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
prototypical instances of wh-exclamatives in the world’s languages. That is, if a language allows
for wh-exclamatives at all, it allows for qualitative and quantitative wh-exclamatives.
Secondly, I show that the felicitousness of kind, individual, and manner wh-exclamatives is
subject to cross-linguistic variation and, importantly, is identical in matrix and subordinate
contexts in all the languages of the sample.
Thirdly, I reveal that matrix-subordinate asymmetry observed in all other wh-exclamatives
is subject to cross-linguistic variation. Moreover, wh-exclamatives demonstrate an implicit
hierarchy ranging from the most to the least appropriate matrix clause cross-linguistically:
inanimate object/ personal/ locative > temporal > causal. To illustrate, if a language allows for
causal matrix wh-exclamatives, it should also allow for inanimate object, personal, locative, and
temporal matrix wh-exclamatives.
Generally, I suggest that each type of wh-words discussed so far is felicitous either in a
matrix wh-exclamative and a subordinate wh-clause with an exclamative reading, or in a
subordinate wh-clause with an exclamative reading only. Both states of affairs are compatible with
the hypothesis of insubordination. The former state of affairs suggests that a wh-exclamative might
have diachronically functioned as subordinate and at some point, via ellipsis of the main clause,
become conventionalized as a matrix clause. The latter state of affairs suggests matrix-subordinate
asymmetry, namely that a wh-clause with an exclamative reading has not yet been
conventionalized as a matrix clause.
Furthermore, I assume that, since qualitative and quantitative exclamatives are prototypical
instances of wh-exclamatives, they undergo insubordination faster than other types of whexclamatives. As this paper focuses on qualitative exclamatives, the question is whether they
obtain their own syntactic and semantic properties distinct from properties shared with
interrogatives. As I show, they demonstrate syntactically quite diverse strategies, with some of the
strategies being exclamative-only.
The paper is structured as follows. The remainder of this section is aimed at elaborating on
the aforementioned proposals and establishing hypotheses. Section 1.2 presents the methodology
of the study. Section 2 discusses word order of matrix wh-exclamatives and their subordinate
counterparts in comparison to word order of matrix and subordinate wh-interrogatives. Section 3
investigates semantic and syntactic properties of qualitative wh-exclamatives. Section 4 discusses
semantic and syntactic properties of other wh-exclamatives. Section 5 points out that the data
invites for several implications for the theory and typology of (wh-)exclamatives. Section 6
concludes the paper.
35
36
37
38
39
40
41
42
43
44
45
1.1.1 Embeddablility of wh-exclamatives?
The question of whether wh-exclamatives are embeddable is still being debated. Among those who
acknowledge embeddability of wh-exclamatives are Elliott (1974), Grimshaw (1979), Zanuttini
and Portner (2003), Koenig and Siemund (2007, 2013) among others. Among those who deny
embeddability of wh-exclamatives (and claim that wh-exclamatives are a matrix phenomenon) are
Huddleston (1993), Rosengren (1997), Lahiri (2000), d’Avis (2002), Abels (2005), Sæbø (2010),
Rett (2011) and others.
Evidence has circulated in the literature suggesting that the set of matrix predicates in which
exclamatives embed is tentatively restricted to factives, which semantically select true propositions
as their sentential complements (see Kiparsky and Kiparsky 1970), more precisely to four classes
of factives: perceptives, emotives, verbs of retaining or acquiring knowledge, and verbs of
4
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
communication (see examples and classifications in Huddleson (1993) for English; in Villalba
(2003), Castroviejo (2006) and Andueza (2011) for Catalan; in Ono (2006) and Yamato (2010) for
Japanese; in Beyssade (2009) for French; in d’Avis (2002) and Sæbø (2010) for German among
others). However, not every grammatical form of such verbs allows for an exclamative reading
even in one language, let alone cross-linguistically (cf. I know vs. I don’t know). Moreover, some
forms of non-factive predicates allow for an exclamative reading (e.g., you won’t believe). That is
to say, exclamatives do not seem to be easily embeddable. Their subordinate counterparts exhibit
more restrictions than other subordinate wh-clauses, in particular, subordinate interrogatives.
Furthermore, it is well-established that embedding in some factive verbs (e.g., know), whclauses have two readings: interrogative and exclamative, cf. (2a). Wh-clauses embedding in some
other factive verbs (e.g., surprise) have only exclamative readings, cf. (2b). Wh-clauses
embedding in non-factive matrix predicates (e.g., ask) have only interrogative readings, see (2c).6
(2a) John knows how tall Bill is.
Interrogative reading: ‘John knows the answer to the question “How tall Bill is”.’
Exclamative reading: ‘Bill is remarkably tall, the speaker expresses her surprise about this fact and
John knows that.’
(2b) John is surprised how tall Bill is.
Exclamative reading: ‘Bill is remarkably tall and the speaker expresses her surprise about this
fact.’
(2c) John asks how tall Bill is.
Interrogative reading: ‘John wants to know the answer to the question “How tall Bill is”.’
To account for such asymmetry, Grimshaw (1979) suggests that matrix predicates of the knowgroup select both wh-interrogatives and wh-exclamatives as their complements, whereas
predicates of the surprise-group take only wh-exclamatives as their complements and askpredicates allow only for wh-interrogatives. Starting from Huddleston (1993), this complementselection analysis has been criticized in the literature on the grounds that know-sentences with,
e.g., a who-clause (e.g., John knows who married) have only an interrogative reading. Therefore,
talking about subordinate wh-exclamatives presumably makes sense only w.r.t. exclamative-only
clauses like in (3) which, accordingly, allow for exclamative readings only.
(3) John {is surprised | knows} how very tall Bill is.
Last but not least, unlike interrogatives, exclamatives cannot occur in reported speech contexts, cf.
(4a-d). Exclamatives like (4a) cannot embed in the verb exclaim, cf. (4b); instead, their declarative
counterparts are used, cf. (4c). Interrogatives can embed in the verb ask; so they can occur in a
true reported context, cf. (4d).
(4a) How very tall Bill is!
6
Castroviejo and Schwager (2008) also observe that concealed questions and concealed exclamations (e.g., John
asked the height of the building and John could not believe the height of the building) are identical in form and the
difference between them comes from the semantics of matrix predicates.
5
1
2
3
4
5
6
7
8
9
10
11
(4b) *He exclaimed how very tall Bill is.
(4c) He exclaimed that Bill is very tall.
(4d) He asked how tall Bill is.
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
1.1.2 Hypothesis of insubordination vs. matrix verb operator
To the best of my knowledge, the hypothesis of insubordination was introduced in Evans (2007).
According to it, languages might display matrix clauses which originate as subordinate clauses.
To illustrate, the optative utterance If I only were there! might diachronically go back to the
construction If I only were there, I could solve the problem, where a main clause is elided and a
new construction becomes constructionalized. As Evans notes, a similar situation is observed in
some varieties of exclamatives, namely in clauses introduced with a complementiser (e.g., That he
should have left without asking me!, Evans 2007: 400), in main clause infinitives (e.g., To think
that I was once a millionaire!, Quirk et al. 1985 via Evans 2007: 401), and in wh-phrases (e.g.,
How they can bet on a bloody dog like that!, Evans 2007: 400).
Interestingly, Evans uses the term ‘ellipsis’ in the following (wide and non-standard) sense:
ellipsis allows for main clause elements which range from uniquely recoverable to non-uniquely
recoverable. In the case of diachronic insubordination of wh-exclamatives, ellipsis is understood
in a non-uniquely recoverable sense. For instance, a main clause in the utterance How they can bet
on a bloody dog like that! can be recovered in several ways: I don’t understand, I am amazed, etc.
A wide understanding of ellipsis in Evans’ paper is potentially subject to criticism. As
examples I consider two views developed within two quite different theoretical camps: Generative
Grammar (GG) and Construction Grammar (CG). Within GG, Grosz (2011) considers Evans’
analysis as a matrix clause deletion approach, and argues against this analysis. Instead, he proposes
a covert EX operator for optatives, with further possible extension to wh-exclamatives. Within
CG, Michaelis and Lambrecht (1996) point out that various exclamative constructions constitute
an abstract exclamative construction which has several semantic features. One of these features is
affective stance, which can be encoded by a matrix verb, by an interjection, or left unexpressed
(although can be inferred).7 It seems that Grosz’s EX operator and the unexpressed but inferred
affective stance as introduced by Michaelis and Lambrecht are similar phenomena. I am not sure
that both approaches contradict Evans’ analysis: Evans’ claim is much more cautious than a mere
automatic matrix clause deletion or unexpressed but inferred semantic component. Evans argues
for possible diachronic origins for matrix clause ellipsis, whereas Grosz treats Evans’ analysis as
though it would be synchronic and unpronounceable at PF level 8 (unexpressed affective stance is
also a synchronic phenomenon).
To summarize, admitting the view that wh-exclamatives are embeddable suggests that they embed
in a rather restricted set of forms of exclamative-selecting predicates with a specific semantic
component of surprise. Such a complicated view can be replaced with the opposite position: whexclamatives originate as subordinate wh-clauses which serve as complements to the predicates of
surprise, and then become insubordinated whilst obtaining the semantics of suprise. In other words,
this leads to admitting the hypothesis of insubordination, which I am going to discuss now.
7
8
I thank an anonymous reviewer for drawing my attention to this feature.
I owe this observation to Nina Dobrushina (p.c.).
6
1
2
3
4
5
6
7
8
As a continuation of Evans’ (2007) proposal, Koenig and Siemund (2013) assume that some
minor sentence types introduced in Koenig and Siemund (2007), such as exclamatives or optatives,
are formed from subordinate structures with a diachronically elided matrix clause. The findings of
a more recent study on an exclamative sentence type reported in Siemund (2015) also accord with
the hypothesis of insubordination.
To justify the hypothesis of insubordination, I arrive at two groups of hypotheses. Both are
taken up in the next two sections.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
1.1.3 Word orders in matrix wh-exclamatives and their subordinate counterparts
As has been stated earlier in this paper, I verify Elliott’s (1974) observation on syntactic structures
of wh-exclamatives, wh-interrogatives, and their subordinate correlates. In doing so, I formulate
the following hypotheses:
31
32
33
34
35
36
37
38
39
40
41
42
1.1.4 Matrix-subordinate asymmetry
Starting from Elliott (1974: 232), it has been acknowledged that there is asymmetry between
matrix wh-exclamatives and their subordinate counterparts w.r.t. the use of wh-words. Consider
the following examples:
Hypothesis 1: Syntactic structures of matrix wh-exclamatives are analogous to syntactic structures
of subordinate wh-clauses with exclamative readings w.r.t. wh-movement and subject-verb
inversion;
Hypothesis 2: Syntactic structures of matrix wh-exclamatives and subordinate wh-clauses with
exclamative readings are analogous to syntactic structures of subordinate wh-interrogatives w.r.t.
the same parameters.
Starting from the remarks on English made in Elliott (1974: 233), the evidence for syntactic
structures of matrix wh-exclamatives being analogous to syntactic structures of their subordinate
counterparts and subordinate wh-interrogatives in various languages has been constantly growing
in research. However, to the best of my knowledge, there is no study which would systematically
verify this similarity w.r.t. the following syntactic parameters: (i) whether there is wh-movement
and, if yes, whether it is obligatory and to which position a wh-phrase moves; (ii) whether there is
subject-verb inversion and, if yes, whether it is obligatory.9 This paper aims to verify the two
aforementioned hypotheses.
(5a) How long he stayed!
(5b) Mary is surprised at how long he stayed.
(5c) # Whom he has invited!
(5d) John is surprised at who he has invited.
Languages such as English allow for matrix and subordinate how-exclamatives and what-aexclamatives (i.e. qualitative and quantitative wh-exclamatives). Furthermore, languages such as
9
Actually, these parameters have been proposed in Siemund (2001) and Dryer (2005) for the cross-linguistic study of
wh-interrogatives and their subordinate correlates.
7
1
2
3
4
5
6
English do not license personal (who), inanimate object (what), locative (where), temporal (when),
and causal (why) matrix wh-exclamatives. In contrast, their subordinate counterparts are felicitous.
Interestingly, in languages such as German, matrix-subordinate asymmetry disappears: the
aforementioned wh-exclamatives are possible as matrix and subordinate.10
German
(6a) Wie lang der geblieben
how
long
he
stay.PST.PTCP
be.PRS.3SG
‘How long he stayed!’ (Repp 2013: 65)
7
8
(6b) Maria ist erstaunt
John
is
surprise.PST.PTCP
wie lang der geblieben
ist!
how
be.PRS.3SG
long
he
stay.PST.PTCP
‘Mary is surprised at how long he stayed!’
9
10
(6c) Wen
who.ACC
der alles eingeladen
hat!11
he
have.PRS.3SG
all
invite.PST.PTCP
‘Look who he has invited!’ (d’Avis 2002: 5)
11
12
(6d) Erik ist erstaunt
Erik
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
ist!
is
surprise.PST.PTCP
wen
Maria eingeladen
hat!
who.ACC
Maria
have.PRS.3SG
invite.PST.PTCP
‘Eric is surprised at who Maria has invited!’ (ibid.)
Moreover, Italian allows for the following wh-words in matrix exclamatives: qualitative che
‘what’, quantitative quanti/quanto ‘how many/much’, chi ‘who’, cosa ‘what’ (inanimate object),
dove ‘where’, and quando ‘when’, cf. Zanuttini and Portner (2003: 67, ft. 31). Perché ‘why’ fails
to occur in a matrix exclamative,12,13 cf. Perché l’ha fatto! ‘Why he has done this!’ (lit.), although
this example is acceptable as a rhetorical question. More recently, Delfitto and Fiorin (2014: 12)
argued that quando ‘when’ is not readily acceptable in matrix exclamatives. This leads to
tentatively conclude that ‘why’ and ‘when’ matrix exclamatives might behave differently from
‘what’, ‘who’, and ‘where’ matrix exclamatives. Moreover, despite that the authors are not explicit
about whether matrix wh-exclamatives have subordinate counterparts, due to matrix-subordinate
asymmetry, this seems to hold true.
Last but not least, compare the following English and Dutch examples.
(7a) How Buck rides his horse! (Rett 2008a: 164, ex. (13b))
OKEvaluative reading: ‘Jan rides his horse beautifully/clumsily/etc.’
#Manner reading: ‘Jan rides his horse saddled/bare-backed/etc.’
According to Repp (2013), German matrix warum-clauses (‘why’ clauses) require some additional support, namely
a prosodic stress on a wh-word.
11
Remarkably, in Repp (2013), the same example is grammatical without alles ‘all’.
12
As Villalba (2008: 31) states, “we still lack the answer to the fact that why exclamatives are lacking universally”,
cf. also Sung (2015: 298, ft. 11). Taking into account the data from languages such as German, the claim that “why
exclamatives are lacking universally” does not hold true.
13
On the basis of Formosan languages, Sung (2015) claimed that ‘why’ matrix exclamatives do exist once the semantic
tests proposed by Zanuttini and Portner (2003) are applied to them. Be that as it may, Formosan ‘why’ exclamatives
differ structurally from ‘why’ exclamatives which involve the use of wh-words denoting cause.
10
8
1
2
3
4
5
6
7
8
9
(7b) # Which man I encountered in the street!
Dutch
(8a)
Jan
zijn
paard
berijdt!
how
Jan
his
horse
ride.PRS.3SG
‘How Jan rides his horse!’
OKEvaluative reading: ‘Jan rides his horse beautifully/clumsily/etc.’
OKManner reading: ‘Jan rides his horse saddled/bare-backed/etc.’
(Nouwen and Chernilovskaya 2015: 212)
(8b)
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Hoe
Welke
man
ik
net
op
straat
tegenkwam!
which
man
I
just
on
street
encounter.PST.SG
‘Which man I encountered just in the street!’ (lit.)
(ibid.: 205)
Interestingly, manner readings of ‘how’ as well as individual wh-words with the meaning ‘which’
are possible in Dutch but not in English matrix wh-exclamatives.14 Although the question of
whether their subordinate correlates have manner or individual readings has not been posed, it
seems to be the case due to the general principle of matrix-subordinate asymmetry.
Looking at the data presented above, I arrive at the following hypotheses:
Hypothesis 3: Qualitative and quantitative wh-exclamatives do not demonstrate matrixsubordinate asymmetry (i.e. a set of constructions in matrix wh-exclamatives is identical to a set
of constructions in subordinate wh-clauses with exclamative readings);
Hypothesis 4: Qualitative and quantitative wh-exclamatives are prototypical wh-exclamatives; in
other words, if a language allows for matrix wh-exclamatives, it allows for qualitative and
quantitative matrix wh-exclamatives;
Hypothesis 5: The following matrix wh-exclamatives have felicitous subordinate counterparts and
form an acceptability hierarchy: inanimate object/ personal/ locative > temporal > causal; e.g., if
a language allows for causal matrix exclamatives, it also allows for inanimate object, personal,
locative, and temporal matrix exclamatives, etc;
Hypothesis 6: Manner and individual wh-exclamatives have felicitous subordinate counterparts.
In my study, I also included kind wh-exclamatives, which involve the use of wh-words with the
meaning ‘what kind’, since I came across them in several languages of my sample. However,
initially I did not have any clear hypotheses for them: the only firm impression that I had was that
they should behave similarly to individual and manner wh-exclamatives.
14
Individual wh-words with the meaning ‘which’ is also present in Hungarian, cf. melyik (see Lipták 2006).
9
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
1.1.5 Qualitative wh-exclamatives
Additionally, my goal was to study qualitative wh-exclamatives which hypothetically represent
the core of wh-exclamatives (on a par with quantitative wh-exclamatives). Qualitative whexclamatives are more intriguing than quantitative from a cross-linguistic point of view since they
seem to show more variation in form and semantics. Basically relying on Ono (2002, 2006), who
studied Japanese wh-words nante, nanto, and nantoyuu in exclamatives, but delving a bit further,
I compile the following constructions to investigate:
Context (i): wh-word + NP with a gradable adjective in an attributive position (e.g., What a
beautiful dress my sister bought!);
Context (ii): wh-word + NP with an elided gradable adjective in an attributive position (e.g.,
What a dress my sister bought!);
Context (iii): wh-word + gradable adjective in a predicative position (e.g., How beautiful
this dress is!);
Context (iv): wh-word + gradable adverb (e.g., How fast my brother runs!);
Context (v): wh-word + an elided gradable adverb or a verb (e.g., How my brother runs!).
A few comments should be made here. To begin with, as the examples in brackets show, these
contexts seem to form two groups w.r.t. the distribution of wh-words: Contexts (i) – (ii) vs.
Contexts (iii) – (v).
Secondly, Siemund (2015) argues for the following two tentative co-occurrences in usage
data: on the one hand, a non-gradable NP with a gradable AP and, on the other hand, a gradable
NP without a gradable AP. In other words, non-gradable nouns do not allow for ellipsis of gradable
adjectives which modify them. Since the sentences I test have one and the same proposition ‘My
sister bought a beautiful dress’ with a non-gradable noun ‘dress’, I assume that non-elliptical
Contexts (i) and (iii) might be more preferable than their elliptical counterpart Context (ii).
Analogously, non-elliptical Context (iv) might turn out to be more felicitous than its elliptical
counterpart Context (v).
Taking these two ideas as a starting point, I arrive at the following two hypotheses:
Hypothesis 7: Qualitative wh-exclamatives differentiate between Contexts (i) – (ii) and Contexts
(iii) – (v) w.r.t. the use of wh-words;
Hypothesis 8: There are languages, where, for non-gradable nouns, non-elliptical Contexts (i),
(iii), and (iv) are considered as more appropriate than elliptical Contexts (ii) and (v).
Last but not least, in order to detect exclamative-only constructions like what a construction, I also
tested the aforestated five contexts in an interrogative environment.15,16
15
I leave aside exclamative-only how very + AP and alike.
The use of adjective in the interrogative version of Context (i) is non-neutral; the fact that the dress is beautiful is
presupposed.
16
10
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
1.2 Data collection and methodology
To test the eight hypotheses established so far, I conducted a pilot cross-linguistic study which
included eleven languages. The choice of languages was influenced by availability of the data
which could be obtained by contacting native speakers. In total, the language sample comprises
six language families (Altaic, Basque, Indo-European, Kartvelian, Korean, Semitic, and Uralic),
including two isolates – Basque and Korean (language classification is based on Lewis et al.
(2015)). The complete alphabetically-ordered list of languages is as follows: Basque, Bulgarian,
Estonian, Georgian, Hebrew, Hindi, Korean, Lithuanian, Ossetic, Russian, and Turkish.
To begin with, I generated a list of all the wh-words mentioned in the grammars of each
studied language. The list includes wh-words with the following meanings: quality (‘what’ + NP),
kind (‘what kind’), individual (‘which’), quantity (‘how many/much’), person (‘who’), inanimate
object (‘what’), evaluative/manner (‘how’),17 location (‘where’), time (‘when’) and cause (‘why’).
Afterwards, I generated the following preliminary list of sentences for each of the wh-words
(one sentence per condition) and asked some native speakers18 of the target languages to translate
them:
(i)
assertive (e.g., My sister bought a dress);
(ii)
matrix wh-interrogative (e.g., Did my sister buy a dress?);
(iii)
subordinate wh-interrogative (e.g., I want to know what dress my sister bought);
(iv)
matrix wh-exclamative (e.g., Wow, what a beautiful dress my sister bought!);
(v)
subordinate wh-exclamative (e.g., You won’t believe what a beautiful dress my
sister bought!).
Moreover, all the wh-words from the list were investigated in the following syntactic
positions: quality, kind, and individual wh-words in an attributive position with NPs; these NPs as
well as wh-phrases of quantity, person, inanimate object in argument positions (i.e. as subject and
direct object); wh-phrases of manner, location, time, and cause as adjuncts.
Importantly, matrix and subordinate wh-exclamatives were provided with relevant usage
contexts to elicit the speaker’s surprise. To illustrate, the matrix exclamative Wow, what a beautiful
dress my sister bought! was provided with the following context: My brother sees the dress that
our sister bought. He has not expected the dress to be so beautiful, and exclaims, addressing his
friend (after that the sentence to translate was given). Such contexts were quite necessary since
their use was aimed at avoiding question interpretations. Crucially, matrix wh-exclamatives were
introduced with the surprise interjection to yield an exclamative interpretation (cf. Rett (2011) who
suggests using items like the English wow as a good test for exclamatives). Accordingly, in the
examples discussed in the next sections, where appropriate, the attested sentences have surprise
interjections. I leave the question of eliciting other emotions for future research. This, however,
was not enough, since some matrix sentences (e.g., English Wow, why are you wearing orange
shoes!) even in exclamation-triggering contexts could still be interpreted as questions. To be
assured that I did not deal with questions at all, in my correspondence with language consultants I
asked them to judge whether target sentences can have the ignorance answer don’t know.
According to Zanuttini and Portner (2003), exclamatives cannot be followed with any answer. In
the question framework developed in d’Avis (2002), wh-clauses in exclamative environment
Two different interpretations – evaluative and manner.
All in all, the number of consultants I collected the data from was two or three people for almost all the languages,
except for Hindi, for which I found only one informant. Most of the consultants were naïve speakers; However, I also
addressed linguists, who were native speakers of particular languages.
17
18
11
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
cannot be followed with an ignorance answer. So, if a wh-clause was actually followed with such
an answer, this indicated a question; if not, it meant an exclamation.
After I had collected the initial list of all translated sentences, I generated a list of possible
word orders for each of the five types of sentences which contained gradable adjectives or adverbs
(qualitative constructions)19 and presented them to the language consultants. They had to either
approve of the generated sentences or reject them. If the sentences were accepted, I asked the
consultants whether the sentences still have the intended interpretations. The results of this work
are shown in Section 2.
Then, I asked language consultants to translate only qualitative matrix wh-exclamatives and
their matrix wh-interrogative counterparts in as many ways as possible. The goal of this was to
uncover the strategies of using qualitative wh-exclamatives in the languages under consideration
and, in particular, to reveal exclamative-only strategies.20 Afterwards, I tested both elliptical and
non-elliptical qualitative wh-exclamatives. To give an idea of this, e.g., I elicited Wow, what a
beautiful dress my sister bought! and, afterwards, asked whether eliminating the adjective beautiful
would still make the sentence felicitous. The same goes for adverbs. See Section 3 for the
discussion on both types of findings, especially Section 3.5 for the distinction between evaluative
and manner readings of ‘how’ exclamatives.
Last but not least, the question of why I chose the matrix predicate complex want to know
for subordinate wh-interrogatives and the matrix predicate form won’t believe for subordinate whexclamatives deserves special attention.
According to Elliott (1974) and Grimshaw (1979), matrix predicates which take wh-clauses
with exclamative readings as their sentential complements need to be factive (compare (2a) and
(2b)): both know and be surprised have this feature. This is not true of matrix predicates which
select wh-interrogatives as their complements (compare (2a) and (2c)): know is factive, whereas
ask is not. Remarkably, the set of factive matrix predicates in which wh-clauses with exclamative
readings embed varies greatly across languages. To illustrate, the Japanese translation equivalent
of English know is not subcategorised for wh-exclamatives (see Ono 2006).
Crucially, it is a verbal grammatical form, rather than a verbal lexical item (with a whole
paradigm), which selects for wh-exclamatives in Elliott’s and Grimshaw’s terms. For instance, the
predicate believe is not factive. However, its second person future form under negation You won’t
believe is factive and does select for wh-exclamatives. To illustrate, in (9a), the proposition ‘Mary
bought a car’ is true since it cannot be cancelled, whereas in (9b), the same proposition is false –
otherwise, the continuation would be infelicitous.
(9a) You won’t believe what a car Mary bought! # In fact, she did not buy anything.
(9b) You believe that Mary bought a car but, in fact, she did not buy anything.
19
For this purpose, I did not test all the strategies of qualitative wh-exclamatives discussed in Section 3. Instead, for
each language, I used the strategy which first came to mind by the language consultants while they were translating
the initial list of five sentences.
20
For this, I also tested matrix wh-interrogative counterparts of qualitative wh-exclamatives.
12
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
The form you won’t believe was intentionally chosen for subordinate surprise contexts.21
Importantly, according to Michaelis (2001), subordinate surprise contexts seem to predominantly
occur with this form throughout the world’s languages.22 Moreover, Huddleston (1993) studied
subordinate surprise contexts with this form as a main clause predicate.
As for subordinate wh-interrogatives, the matrix predicate form with the meaning ‘I want to
know’ seemed to be preferable for the current purposes. Since semantically it seeks an answer, it
licenses only interrogatives and is inappropriate with exclamative readings.
Furthermore, when relevant, our discussion is also based upon the data from Catalan, Dutch,
English, French, German, Classical Greek, Hungarian, Japanese, Spanish, and Swedish. While
discussing the data from these languages, I mostly rely upon the literature sources; in some cases,
I also provide judgments from language consultants.
Finally, since presenting all the collected data would suggest a much longer paper, in the
next sections, I limited the number of the examples due to the following reasons. First of all, I tried
to illustrate such distributions of items and constructions under consideration which were nontrivial. Secondly, I attempted to touch upon the data from all the studied languages.
2 Subordinate syntax of matrix wh-exclamatives
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
2.1 Preliminary remarks
In this section, on the basis of the sample language data, I verify Hypotheses 3 and 4 (both w.r.t.
qualitative wh-exclamatives) and subsequently Hypotheses 1 and 2.23 In doing so, I investigate the
following two construction schemas which involve gradable adjectives or adverbs (qualitative
constructions):
(i)
possessive adjective in an attributive position + NP + transitive verb + gradable adjective in
an attributive position + NP (e.g., My sister bought a beautiful dress);
(ii)
possessive adjective in an attributive position + NP + intransitive verb + gradable adverb
(e.g., My brother runs fast).24
These constructions were studied in all the five syntactic contexts: assertive, matrix whinterrogative, subordinate wh-interrogative, matrix wh-exclamative, and subordinate wh-clause
with an exclamative interpretation.
Matrix wh-interrogatives and matrix wh-exclamatives tested in my study can be exemplified
with the following sentences: What dress did my sister buy?; How fast does my brother run?; What
a (beautiful) dress my sister bought!; How fast my brother runs!. Subordinate wh-interrogatives
and subordinate wh-clauses with exclamative interpretations can be illustrated with such sentences
21
Although the predicate believe can take DPs as its complements (at least in English), its wh-complements are
clauses, cf. Michaelis (2001: 1046).
22
Indeed, all the languages of the sample demonstrated felicitous use of this form or its close modifications (you could
not believe, you would not believe) in subordinate wh-clauses with exclamative readings.
23
If Hypotheses 3 and 4 are not confirmed, it is not possible to test Hypotheses 1 and 2. The reason for this is that
word orders in the five types of sentences established in Section 1.2 are possible to investigate if in principle a language
allows for matrix wh-exclamatives.
24
The order of the elements in (i) and (ii) is dependent upon a particular language.
13
1
2
3
4
5
6
7
8
9
10
11
12
as I want to know what dress my sister bought; I want to know how fast my brother runs; You won’t
believe what a (beautiful) dress my sister bought!; You won’t believe how fast my brother runs!.
For the sake of brevity, if the word order of matrix wh-exclamatives and the word order of
their subordinate correlates are identical, examples are presented only with subordinate whexclamatives. The same goes for matrix wh-interrogatives and their subordinate counterparts.
The remainder of this section is structured in accordance with the three language groups
established w.r.t. wh-movement: Section 2.2 discusses languages with obligatory wh-movement
to the left periphery; Section 2.3 examines languages with obligatory wh-movement to the
preverbal position; Section 2.4 studies languages with optional wh-movement to the left periphery.
When relevant, subject-verb inversion is taken into account. Section 2.5 summarises the main
findings.
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
2.2 Obligatory wh-movement to the left periphery
Languages with obligatory wh-movement to the left periphery of a clause include Bulgarian,
Estonian, Hebrew, Lithuanian, and Russian.
In Bulgarian, Lithuanian and Russian, the word order in assertives is relatively free and
hinges upon the information structure of a sentence. However, the neutral combinations are as
follows: SVO, AdjN, and VAdv, see Rudin (1986) for Bulgarian among others; Mathiassen (1996)
for Lithuanian; Kovtunova (1976); Krylova and Khavronina (1976); Shvedova (1980) for Russian.
In matrix wh-interrogatives and their subordinate correlates, a wh-phrase moves to the left
periphery of a clause, see Rudin (1986) for Bulgarian; Shvedova (1980) for Russian.25 Thereby,
the neutral word order is subject-verb inversion, see Rudin (1986) for Bulgarian; Krylova and
Khavronina (1976) for Russian.26 In matrix wh-exclamatives and their subordinate correlates, a
wh-phrase undergoes movement to the left periphery as well. Subject-verb inversion is obligatory
in these three languages, see Kovtunova (1976); Krylova and Khavronina (1976) for Russian. The
following Lithuanian examples illustrate these facts:
Lithuanian
(10a)
Mano
my
ses-ė
nusipirk-o
graži-ą
suknel-ę.
sister-NOM.SG
buy-PRS(3)
beautiful-ACC.SG.F
dress-ACC.SG
‘My sister bought a beautiful dress.’
29
30
(10b)
Mano
brol-is
greitai
bėg-a.
my
brother-NOM.SG
fast
run-PRS(3)
‘My brother runs fast.’
31
32
(10c)
Nor-iu
žino-ti,
koki-ą
suknel-ę
want-PRS.1SG
know-INF
what-ACC.SG.F
dress-ACC.SG
nusipirk-o
mano
ses-ė.
buy-PRS(3)
my
sister-NOM.SG
‘I want to know what dress my sister bought.’
33
25
Here and elsewhere in Section 2, if some language fact is not provided with a reference, this means that I did not
find any literature sources to refer to and implied it from the witnessed data.
26
Throughout the whole Section 2, if a wh-phrase takes a position of a subject, subject-verb inversion does not occur.
14
1
(10d)
žino-ti,
kaip
greitai
want-PRS.1SG
know-INF
how
fast
bėg-a
mano
brol-is.
run-PRS(3)
my
brother-NOM.SG
‘I want know how fast my brother runs.’
2
3
(10e)
Tu
ne-patikė-si,
koki-ą
graži-ą
suknel-ę
you-NOM
NEG-believe-FUT.2SG
what-ACC.SG.F
beautiful-ACC.SG.F
dress-ACC.SG
nusipirk-o
mano
ses-ė!
buy-PRS(3)
my
sister-NOM.SG
‘You won’t believe what a beautiful dress my sister bought!’
4
5
(10f)
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Nor-iu
Tu
ne-patikė-si,
kaip
greitai bėg-a
mano
brol-is!
you-NOM
NEG-believe-FUT.2SG
how
fast
my
brother-NOM.SG
run-PRS(3)
‘You won’t believe how fast my brother runs!’
In modern Hebrew, the word order in assertives is different from the typical Semitic word order
(see Kautzsch 1910). It is relatively free and the basic combinations are SVO, NAdj, VAdv (see
Glinert 2005). In matrix wh-interrogatives, wh-exclamatives and their subordinate counterparts, a
wh-phrase moves to the left periphery of a clause. Subject-verb inversion is not obligatory;
however, the neutral variant is its absence (see Glinert 2005).
Finally, in Estonian, the word order of assertives is also relatively free: SVO, AdjN, VAdv
(see Erelt 2003). The basic rule is that a verb takes the second position. Wh-interrogatives are
derived via movement of a wh-phrase to the left periphery of a clause. In matrix and subordinate
wh-interrogatives, the verb takes the final position (see Erelt 2003, Erelt et al. 2007). According
to Erelt et al. (2007), the structures of wh-exclamatives are similar to the structures of whinterrogatives and subordinate wh-interrogatives as well.
Estonian
(11a)
Mu
õde
ost-is
ilusa
kleidi.
I.GEN
sister.NOM
buy-IPF
beautiful.GEN
dress.GEN
‘My sister bought a beautiful dress.’
21
22
(11b)
Mu
vend
jookse-b
kiiresti.
I.GEN
brother.NOM
run.PRS-3SG
fast
‘My brother runs fast.’
23
24
(11c)
Ma
taha-n
tea-da
mis
kleidi
mu
I.GEN
want-PRS.1SG
know-INF
what
dress.GEN
I.GEN
õde
ost-is.
sister.NOM
buy-IPF
‘I want to know what dress my sister bought.’
25
26
(11d)
Ma
taha-n
tea-da
kui
kiiresti
mu
I.GEN
want-PRS.1SG
know-INF
how
fast
I.GEN
vend
jookse-b.
15
brother.NOM
run.PRS-3SG
‘I want to know how fast my brother runs.’
1
2
(11e)
Sa
ei
usu
mis
ilusa
kleidi
mu
you-NOM
NEG
believe-PRS
what
beautiful.GEN
dress.GEN
my.GEN
õde
ost-is!
sister.NOM
buy-IPF
‘You won’t believe what a beautiful dress my sister bought!’
3
4
(11f)
Sa
ei
usu
kui
kiiresti
mu
vend
you-NOM
NEG
believe-PRS
how
fast
I.GEN
brother.NOM
jookse-b!
run.PRS-3SG
‘You won’t believe how fast my brother runs!’
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
2.3 Obligatory wh-movement to the preverbal position
In this section, I discuss the languages in which wh-phrases move to the preverbal position, namely
Basque, Georgian, and Ossetic. Moreover, a whole (wh-phrase and verb) complex can move to the
left periphery of a clause, although it is not obligatory.
According to Primus (2001: 868-869), who studies the syntax of wh-interrogatives in a
cross-linguistic perspective, the preverbal position in such languages is focal. To put it differently,
wh-phrases in interrogatives move to the fixed preverbal focal position (see also Lipták 2001).
However, a study of the connection between a focal position and wh-phrases is lacking.
According to Harris (1984) and Aronson (1990), the neutral basic word orders in Georgian
assertives are SOV, AdjN, and AdvV. Word orders in Ossetic are also relatively free (see Gagkaev
1956; Akhvlediani 1963; Bagaev 1982). However, the most typical are SVO and AdvV; AdjN is
fixed. Basque differs from Georgian w.r.t. the word order of noun and adjective in an attributive
position, which is NAdj in Basque (see Hualde and de Urbana 2003).
In matrix wh-interrogatives and their subordinate correlates of the three languages, a whphrase moves to the preverbal position and, together with the verb, can move to the left periphery
of a clause (see Harris 1984; Aronson 1990; Hewitt 2005 for Georgian; Hualde and de Urbana
2003; de Rijk 2008 for Basque; Ljutikova and Tatevosov 2009 for Ossetic). Similarly, in all the
three languages, the structures of matrix wh-exclamatives and their subordinate counterparts are
similar to the structures of matrix and subordinate wh-interrogatives (see Hualde and de Urbana
2003 for Basque). This is illustrated by the following examples from Georgian:
Georgian
(12a)
da-m
lamaz-i
ḳaba
iqida.
̣
my-ERG
sister-ERG
beautiful-NOM
dress.NOM
buy.PST.3SG>3
‘My sister bought a beautiful dress.’
29
30
(12b)
31
32
Čem-ma
Čem-i
dzma
sc ̣rapad
darbis.
my-ERG
brother-ERG
fast
run.PRS.3SG
‘My brother runs fast.’
16
(12c)
1
2
3
ḳaba
iqida.
̣
I.want
I.know.it.SUBJ
my-ERG
what
dress.NOM
buy.PST.3SG>3
sister-ERG
Minda
vicode
čem-i
dzma
ra
sc ̣rapad
darbis.
I.want
I.know.it.SUBJ
my-ERG
brother-ERG
what
fast
run.PRS.3SG
Ar
daijereb,
čem-ma
da-m
ra
lamaz-i
NEG
believe-FUT.2SG
my-ERG
sister-ERG
what
beautiful-NOM
ḳaba
iqida!
̣
dress.NOM
buy.PST.3SG>3
Ar
daijereb,
čem-i
dzma
ra
sc ̣rapad
darbis.
NEG
believe-FUT.2SG
my-ERG
brother-ERG
what
fast
run.PRS.3SG
‘You won’t believe how fast my brother runs!’
(12f’) Ar daijereb, ra sc ̣rapad darbis čemi dzma!
13
14
15
16
17
18
19
20
21
22
23
24
25
26
ra
‘You won’t believe what a beautiful dress my sister bought!’
(12e’) Ar daijereb, ra lamazi ḳaba iqida
̣ čemma dam!
(12f)
10
11
12
čem-ma da-m
‘I want to know how fast my brother runs.’
(12d’) Minda vicode, ra sc ̣rapad darbis čemi dzma.
(12e)
7
8
9
vicode
‘I want to know what dress my sister bought.’
(12c’) Minda vicode, ra ḳaba iqida
̣ čemma dam.
(12d)
4
5
6
Minda
2.4 Optional wh-movement to the left periphery
This section presents the data from languages which exhibit optional wh-movement to the left
periphery of a clause. Hindi, Korean, and Turkish belong to such languages.
The neutral word orders in Hindi and Korean assertives are SOV, AdjN, AdvV (see Rupert
and Weightman 2003 for Hindi; Lee 2004; Ramstedt 1939 [1968] for Korean). According to
Göksel and Kerslake (2005), in Turkish, SOV is also a basic word order, whereas AdjN and AdvV
are fixed. In Hindi matrix wh-interrogatives and wh-exclamatives, a wh-phrase stays usually in
situ but it can optionally move to the left periphery. The same happens in Korean and Turkish
matrix and subordinate wh-interrogatives (see Chang 1996 for Korean; Göksel and Kerslake 2005
for Turkish). Moreover, in all the languages, matrix wh-exclamatives and their subordinate
counterparts demonstrate the same behaviour as matrix and subordinate wh-interrogatives.
Consider the following examples from Korean:27
Korean
(13a)
Nay
enni-nun
yeyppu-n
tuleysu-lul
sa-ss-ta.
my
elder_sister-TOP
beautiful-PTCP
dress-ACC
buy-PST-IND
‘My sister bought a beautiful dress.’
27
28
(13b)
27
Nay
oppa-nun
ppalli
ttwi-n-ta.
I did not study which Korean verb suffixes can be used in wh-exclamatives and wh-interrogatives. I can only state
that in wh-interrogatives, various suffixes (-ni, -ci, -nunka among others) are appropriate depending on a particular
meaning they express. In matrix wh-exclamatives, the suffix -nunka is felicitous. The suffix -ci is used in subordinate
wh-interrogatives and subordinate counterparts of wh-exclamatives.
17
my
(13c)
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
nay
enni-ka
etten
tuleysu-lul
sa-ss-nun-ci
I-TOP
my
elder_sister-TOP
what
dress-ACC
buy-PST-ADNZ-APPER
al-ko
siph-ta.
know-CVB
want-APPER
Na-nun
nay
oppa-ka
elmana
ppalli
I-TOP
my
elder-brother-TOP
how_much
fast
ttwi-ess-nun-ci
al-ko
siph-ta.
run-PST-ADNZ-APPER
know-CV
want-APPER
Ne-nun
nay
enni-ka
elmana
yeyppu-n
you-TOP
my
elder_sister-NOM
how_much
beautiful-PTCP
tuleysu-lul
sa-ss-nun-ci
mit-ci
mos-ha-keyss-ci!
dress-ACC
buy-PST-ADNZ-APPER
believe-APPER
not_may-do-FUT-APPER
‘You won’t believe what a beautiful dress my sister bought!’
(13e’) Nenun elmana yeyppun tuleysulul nay ennika sassnunci mitci moshakeyssci!
(13f)
12
13
14
Na-nun
‘I want to know how fast my brother ran.’
(13d’) Nanun elmana ppalli nay oppaka ttwiessnunci alko siphta.
(13e)
9
10
11
run-PRS-IND
‘I want to know what dress my sister bought.’
(13c’) Nanun etten tuleysulul nay ennika sassnunci alko siphta.
(13d)
6
7
8
fast
‘My brother runs fast.’
1
2
3
4
5
elder_brother-TOP
Ne-nun
nay
oppa-ka
elmana
ppalli
you-TOP
my
elder_brother-NOM
how_much
fast
ttwi-ess-nun-ci
mit-ci
mos-ha-keyss-ci!
run-PST-ADNZ-APPER
believe-APPER
not_may-do-FUT-APPER
‘You won’t believe how fast my brother ran!’
(13f’) Nenun elmana ppalli ney oppaka ttwiessnunci mitci moshakeyssci!
2.5 Interim conclusion
Let us summarise the findings of this section.
First and foremost, the sample language data confirms Hypotheses 3 and 4 (both w.r.t.
qualitative wh-exclamatives). This means that, on the one hand, qualitative exclamatives do not
demonstrate matrix-subordinate asymmetry, and, on the other hand, they are prototypical whexclamatives; in other words, if a language allows for matrix wh-exclamatives, it allows for
qualitative wh-exclamatives, that is, for wh-exclamative constructions with gradable adjectives or
adverbs.
Secondly, the sample language data justifies Hypotheses 1 and 2. This means that w.r.t. the
two parameters — wh-movement and subject-verb inversion — syntactic structures of matrix whexclamatives are similar to syntactic structures of subordinate wh-clauses with exclamative
readings which, in its turn, resemble syntactic structures of subordinate wh-interrogatives.
So far, I have examined languages with structures of matrix wh-exclamatives, their
subordinate correlates and subordinate wh-interrogatives, such that they all are identical to
structures of matrix wh-interrogatives. According to Siemund (2001: 1012), generally, subordinate
18
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
wh-interrogatives are syntactically similar to matrix wh-interrogatives. However, there are wellknown exceptions found among the Germanic and Romance language groups (e.g., English,
French, and German).
According to Grimshaw (1979), Obenauer (1994), Haegeman (1997), and Radford (2006)
among others, English and French demonstrate matrix wh-exclamatives and their subordinate
correlates which are structurally similar to subordinate wh-interrogatives: a wh-phrase undergoes
movement to the left periphery of a clause and subject-verb inversion does not take place. The lack
of subject-verb inversion distinguishes them from matrix wh-interrogatives, see (14a-d) for
English, cited from Grimshaw (1979: 282).
22
3 Qualitative wh-exclamatives
(14a) How tall is John?
(14b) Fred knows how tall John is.
(14c) How tall John is!
(14d) It’s amazing how tall John is!
According to Radford (2006) and Holmberg (2011), in German, like in Estonian, a verb in a
subordinate clause (including subordinate wh-interrogatives) takes the final position in a clause.
In assertives and matrix wh-interrogatives, the verb moves to the second position. Structurally,
matrix wh-exclamatives and their subordinate correlates are similar to subordinate whinterrogatives.
23
24
25
26
27
28
29
30
31
32
3.1 Preliminary remarks
This section tests Hypotheses 6 (w.r.t. manner wh-exclamatives), 7 and 8. More precisely, here I
study Contexts (i) – (v) listed in Section 1.1.5 w.r.t. the use of wh-words with the following
meanings: quality, inanimate object, quantity, evaluation, and manner. Two of them – quantitative
and inanimate object wh-words – are studied in their literal meanings in Sections 4.2 and 4.3
respectively. Although kind and individual wh-words are also used in Сontext (i), they are
examined in Section 4 since they do not constitute qualitative wh-exclamatives.28
In what follows, I examine syntactic and semantic properties of the aforementioned whwords in matrix qualitative exclamatives and their interrogative counterparts.
33
34
35
36
37
38
39
40
3.2 Qualitative wh-words
Qualitative wh-words in both exclamatives and interrogatives are used in Contexts (i) and (ii). This
strategy is found in the following languages of the sample: Basque, Hebrew, Korean, Lithuanian,
Ossetic, Russian, and Swedish. In what follows, I discuss the data from Basque and Russian which
demonstrate quite interesting complications of this strategy.
In Russian, kakoj is used with an NP both with an elided or non-elided gradable adjective in
exclamatives and interrogatives.
28
Although a manner interpretation of case (v) is not qualitative, I consider it here due to the fact that, in some
languages of the sample, it is expressed with the same wh-word which also has an evaluative interpretation.
19
1
Russian
(15) Kak-oe
what-NOM.SG.N
2
3
4
5
6
7
8
9
kupi-l-a
moj-a
sestr-a!/?
dress-NOM.SG
buy-PST-SG.F
my-NOM.SG.F
sister-NOM.SG
Moreover, Russian has another qualitative wh-word kakov which exhibits a special syntactic
feature: it is only used in a predicative position followed by an NP with an elided or non-elided
adjective regardless of the (non-)gradability of the adjective both in exclamatives and
interrogatives. Moreover, it is stylistically non-neutral and in some cases even archaic.
Russian
(16a)
Kakov
by-l
prazdnik!/?
what.SG.M
be-PST.SG.M
party.NOM.SG
‘What a party it was!’ / ‘How was the party?’
(16b)
Kakov
etot
bolš-oj
načal’nik!/?
what.SG.M
this.NOM.SG.M
great-NOM.SG.M
chief.NOM.SG
‘What a powerful man he is!’ / ‘What can you say about this powerful man?’
Similarly, Hebrew eyze,29 Korean etten, Lithuanian kòks, Ossetic tsavær and Swedish vilken
behave like Russian kakoj.
As for Basque zein, in exclamatives it is only used in Context (i) and requires suffix -a
attached to the final word of an NP. In interrogatives, it is used both in Contexts (i) and (ii) and
requires the absence of the suffix -a.
Basque
(17a)
Zein
soineko
??
what
dress
beautiful-DF
nire
arreb-a-k!/*?
my
sister-DF-ERG
(ederr-a)
eros-i
zu-en
buy-PFV
AUX.TR.PST-PST.COMPL
‘What a beautiful dress my sister bought!’
21
22
(17b)
Zein
soineko
(eder)
eros-i
zu-en
what
dress
beautiful
buy-PFV
AUX.TR.PST-PST.COMPL
nire
arreb-a-k?/*!
my
sister-DF-ERG
‘What (beautiful) dress did my sister buy?’/ ‘What a (beautiful) dress my sister bought!’
23
24
25
26
27
28
29
plat’j-e
beautiful-NOM.SG.N
‘What a (beautiful) dress my sister bought!’ / ‘What (beautiful) dress did my sister buy?’
10
11
12
13
14
15
16
17
18
19
20
(krasiv-oe)
3.3 Inanimate object wh-words
In some languages, e.g. in English, inanimate object wh-words function either as NPs (e.g., You
won’t believe what she bought!, see Section 4.3) or as NP modifiers (e.g., What a clever boy!,
What a boy!). Interestingly, in my sample, inanimate object wh-words can be divided into three
groups: (a) those which have the same distribution as qualitative wh-words in exclamatives and
29
Hebrew eyze and Russian kakoj can also have individual and kind interpretations (see Sections 4.6 and 4.7
respectively).
20
1
2
3
4
5
6
7
8
9
10
interrogatives; (b) those which occur in some of the Contexts (i) – (iv) and are exclamative-only
and (c) those which are combinations of both strategies (a) and (b). Bulgarian kakvo, Estonian mis,
and Russian čto za ‘what for’ (lit.)30 are examples of group (a),31 whereas Hindi kyā, Ossetic tsæy,
Swedish vad, Russian do čego and Turkish ne illustrate group (b). Basque zer, Georgian ra, Ossetic
tsǝ instantiate group (c). I discuss the latter two groups below.
The Swedish wh-word vad ‘what’ can be used as a good illustration of group (b): it occurs
in Contexts (i), (iii), and (iv). In interrogatives, in all the listed contexts it is inappropriate. It is not
compatible in Context (ii) in either interrogatives or exclamatives.
Swedish (Delsing 2010: 17-21)
(18a)
stora
fötter
Vad
what
11
12
13
14
have.PRS
Vad
fötter
du
har!
what
leg.PL
you
have.PRS
Vad
dum
han
är!
what
stupid
he
be.PRS
‘How stupid he is!’
(18c’) *Vad dum är han?
‘How stupid is he?’
(18d)
23
24
25
26
27
28
29
30
31
32
33
you
‘What legs you have!’
(18b’) *Vad fötter har du?
‘What kinds of legs do you have?’
(18c)
19
20
21
22
leg.PL
har!
‘What big legs you have!’
(18a’) *Vad stora fötter har du?
‘How big are your legs?’
(18b)
15
16
17
18
big.PL
du
Vad
du
röker
ofta!
what
you
smoke.PRS
often
‘How often you smoke!’
(18d’) *Vad ofta röker du?
‘How often do you smoke?’
Turkish ne and Russian do čego, which is a prepositional genitive form of the inanimate object
wh-word čto, behave similarly. The Ossetic tsæy, which is a genitive form of tsǝ, is also used in
Contexts (i), (iii), and (iv) in an exclamative-only environment. However, tsæy may also be
appropriate in Context (ii) in exclamatives only, see (19a-d). Last but not least, Hindi kyā is used
in Contexts (i) and (ii) in exclamative-only constructions. As for its use in Contexts (iii) and (iv),
they are ruled out both in exclamatives and interrogatives. In this respect, Hindi kyā is quite distinct
from the rest of group (b).
Analogous constructions are witnessed in other European languages, e.g. in German, was für + NP. According to
Leu (2007), this combination is possible in all case forms in Germanic languages. Russian čto za combines with NPs
in nominative or accusative case but its use is expanding now other cases as well (see Kwon 2011).
31
They have the same distribution as Russian kakoj discussed in Section 3.2.
30
21
1
2
3
4
As for group (c), I illustrate its functioning with help of Ossetic tsǝ, whose distribution is
almost identical both in exclamatives and interrogatives, cf. Abaev (1974: 705) and (19e-h).
Ossetic
(19a)
(19b)
(19c)
ba-lχæd-t-a
mæ
χo!/*?
dress
PV-buy-TR-PST.3SG
my
sister
Tsæj
k’aba
ba-lχæd-t-a
mæ
χo!/*?
what.GEN
dress
PV-buy-TR-PST.3SG
my
sister
Me
‘fsǝmær tsæj
taʁd
zg’or-ǝ!/*?
my
brother
fast
run-PRS.3SG
what.GEN
‘How fast my brother runs!’ / ‘How fast does my brother run?’
9
10
(19d)
Tsæj
ræsuʁd u
atsǝ
k’aba!/*?
what.GEN
beautiful
this
dress
be.PRS.3SG
‘How beautiful this dress is!’ / ‘How beautiful is this dress?’
11
12
(19e)
Tsǝ
k’aba
ba-lχæd-t-a
mæ
χo!/?
what
dress
PV-buy-TR-PST.3SG
my
sister
‘What a dress my sister bought!’ / ‘What dress did my sister buy?’
13
14
(19f)
Tsǝ
ræsuʁd
k’aba
ba-lχæd-t-a
mæ
χo!/*?
what
beautiful
dress
PV-buy-TR-PST.3SG
my
sister
‘What a beautiful dress my sister bought!’ / ‘What beautiful dress did my sister buy?’
15
16
(19g)
Me
‘fsǝmær tsǝ
taʁd
zg’or-ǝ!/*?
my
brother
fast
run-PRS.3SG
what
‘How fast my brother runs!’ / ‘How fast does my brother run?’
17
18
(19h)
27
28
29
30
k’aba
beautiful
‘What a dress my sister bought!’ / ‘What dress did my sister buy?’
7
8
26
ræsuʁd
what.GEN
‘What a beautiful dress my sister bought!’ / ‘What beautiful dress did my sister buy?’
5
6
19
20
21
22
23
24
25
Tsæj
Tsǝ
ræsuʁd u
atsǝ
k’aba!/*?
what
beautiful
this
dress
be.PRS.3SG
‘How beautiful this dress is!’ / ‘How beautiful is this dress?’
The Georgian word ra has the same distribution as Ossetic tsǝ. The behaviour of the Basque whword zer is identical to the behaviour of Georgian ra and Ossetic tsǝ, except that it allows for
Context (i) not only in exclamatives but also in interrogatives. Note that in Basque interrogatives,
unlike exclamatives, NPs are not marked with a definite suffix.
3.4 Quantitative wh-words
The exclamative use of quantitative wh-words in their literal meaning (e.g., How many apples I
bought!) is examined in Section 4.2. Here, I discuss their functioning in exclamatives with gradable
adjectives or adverbs. Among the languages of the sample, they are felicitous in Basque,
Bulgarian, Hebrew, Hindi, Korean, Russian, and Turkish. In Georgian and Lithuanian, they are
22
1
2
3
4
5
6
7
8
9
used in a limited number of contexts, whereas in Estonian and Ossetic, according to native
speakers, they are employed neither in exclamatives, nor in interrogatives.
Quantitative wh-words show strong cross-linguistic variation w.r.t. their use in Contexts (i),
(iii), and (iv). Generally, in Contexts (ii) and (v), that is, in the case of an elided adjective or adverb
in exclamatives or interrogatives, they are either inappropriate (see Bulgarian example (20a)) or
are interpreted differently: a wh-word expresses its literal quantity meaning, see Bulgarian
example (20b).
Bulgarian
(20a)
roklja
e
kupi-l-a
sestra
mi!/?
how_much
dress
be.PRS.3SG
buy-PRF-SG.F
sister
I.DAT
Kolko
‘How much fabric my sister bought for a dress!’ / ‘How much fabric did my sister buy for
a dress?’
10
11
12
(20b)
13
14
15
16
17
18
19
20
21
??
Kolko
bjaga
brat
mi!/?
how_much
run.PRS.SG
brother
I.DAT
‘How long my brother runs!’ / ‘How long does my brother run?’
Example (20a) illustrates that if the adjective is omitted, the countable noun is interpreted as
uncountable and the sentence seems to be infelicitous. Sentence (20b) with an elided adverb
exemplifies a quantitative interpretation of the wh-word kolko.
Moreover, the quantitative Bulgarian wh-word kolko is used in Contexts (i), (iii), and (iv)
both in exclamatives and interrogatives.
Bulgarian
(21a)
Kolko
how_much
roklja
e
kupi-l-a
sestra
mi!/?
beautiful-SG.F
dress
be.PRS.3SG
buy-PRF-SG.F
sister
I.DAT
‘What a beautiful dress my sister bought!’ / ‘What beautiful dress did my sister buy?’
22
23
(21b)
Kolko
bărzo
bjaga
brat
mi!/?
how_much
fast
run.PRS.SG
brother
I.DAT
‘How fast my brother runs!’ / ‘How fast does my brother run?’
24
25
(21c)
26
27
28
29
30
31
32
33
34
krasiv-a
Kolko
krasiv-a
e
tazi
roklja!/?
how_much
beautiful-SG.F
be.PRS.3SG
this
dress
‘How beautiful this dress is!’ / ‘How beautiful is this dress?’
Korean elmana, Russian skol’ and naskol’ko, Turkish ne kadar demonstrate a similar use of
gradable constructions. The other languages show some deviations from this pattern.
Hindi kitnā is used in exclamatives in Contexts (i), (iii), and (iv). However, in interrogatives,
it only occurs in Context (iv). In Hebrew, kama is only used in Contexts (iii) and (iv) in
exclamatives and in interrogatives. Basque zeinen, which is a genitive form of the qualitative whword zein (de Rijk 2008: 245), and its shortened form zein32 are employed in all the three contexts
in exclamatives and are impossible in interrogatives. The use of Georgian quantitative wh-word
32
Remarkably, if zein is used in Context (i), its status is difficult to determine: it might be either a shortened form of
zeinen or the qualitative wh-word zein.
23
1
2
3
4
5
6
7
8
9
ramdeni in all the three possible contexts is interrogative-only. The Lithuanian quantitative whword kiek is used in Context (iv) in interrogatives only. Finally, the Estonian quantitative wh-word
kui palju and its Ossetic equivalent tsas/tsal/tsæjbærts are infelicitous both in interrogatives and
in exclamatives.
To summarise, only Bulgarian, Korean, Russian, and Turkish quantitative wh-words are
appropriate in all the three contexts in interrogatives and exclamatives (Hebrew and Hindi deviate
from this pattern). In other languages of the sample, such quantitative wh-words are exclamativeonly, interrogative-only, or are infelicitous.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
3.5 Evaluative/manner wh-words
Rett (2008a, 2008b) claims that ‘how’ clauses have only scalar evaluative interpretations in
exclamations, whereas in questions, they allow for both scalar evaluative and non-scalar, i.e.,
manner, interpretations. For example, how in the question How does Buck ride his horse? can have
both manner and scalar evaluative readings: Buck rides his horse bare-backed, saddled, etc
(manner reading); beautifully, dangerously, clumsily, etc (scalar evaluative reading), the examples
cited from Rett (2008b, 2011). However, being an exclamation, How Buck rides his horse! is
expressively correct if the speaker expected Buck, e.g., to ride clumsily and it turns out that he
rides beautifully.33 The same exclamative is not expressively correct if the speaker expects Buck
to ride, e.g., saddled. However, w.r.t. other parameters, exclamatives and interrogatives show
similar behaviour. Both of them are compatible with evaluative (and generally with gradable)
adjectives or adverbs: e.g., How beautifully does she sing?, How beautifully she sings!, How
beautiful is she?, How beautiful she is!.
Since scalar vs. non-scalar contrast is beyond the scope of this paper, in what follows, I focus
only on evaluative vs. manner interpretations of ‘how’ exclamatives. To differentiate them, I
developed the following methodology, which was inspired by the aforementioned observation
from Rett (2008a, 2008b): I asked consultants whether the question like How does your brother
dance? can be answered beautifully/ fast (evaluative reading) or barefoot (manner reading).
Additionally, I asked whether the speaker exclaiming, e.g., How your brother dances! can mean
beautifully/ fast (evaluative reading) or barefoot (manner reading).
Generally, the languages of the sample differ w.r.t. two parameters: (i) whether their
evaluative/manner wh-words allow for evaluative and/or manner interpretations in exclamatives
and (ii) whether, interpreted evaluatively, they are sensitive to ellipsis of a gradable adverb in
exclamatives and interrogatives.
As for the first parameter (i), Basque, Bulgarian, Hebrew, Estonian, Lithuanian, Korean,
Ossetic, Russian, Turkish (and also Dutch) allow for both evaluative and manner interpretations
of ‘how’ words in exclamatives. Importantly, native speakers agreed upon evaluative
interpretations of exclamatives more readily than upon manner readings. This fact is compatible
with the hypothesis that qualitative wh-exclamatives are prototypical instances of exclamatives.
33
As can be seen from the English example, an adverb is elided. Similarly, in French, such constructions are also
possible.
Comme
il
regrette
sa
décision!
How
he
regrets
‘How strongly he regrets his decision!’
(Marandin 2008: 438)
his
decision
24
1
2
3
4
5
6
7
8
9
10
11
12
13
Moreover, Swedish ‘how’ exclamatives, like their English counterparts, have evaluative
interpretations only. In contrast, Hindi allows for manner interpretations solely.34
W.r.t. the second parameter (ii), Bulgarian, Georgian, Lithuanian, Ossetic, Russian,
Swedish35 are not sensitive to it, that is, they use the same evaluative/manner wh-word for elliptical
and non-elliptical gradable adverbial constructions. Basque, Estonian, Hebrew, Hindi, and Korean,
by contrast, have a special wh-word for exclamative and interrogative adverbial elliptical contexts
which does not occur in adverbial non-eliptical gradable contexts. Consider the following
examples.
Ossetic wh-word kwǝd, w.r.t. the parameter (i), can have both evaluative and manner
interpretations and, w.r.t. the parameter (ii), interpreted evaluatively, is not sensitive to ellipsis of
a gradable adverb in exclamatives and interrogatives. Moreover, it can occur in Context (iii).
Ossetic
(22a)
(22b)
kwǝd
taχd
zg’or-ǝ!/?
my
brother
how
fast
run-PRS.3SG
Me
‘fsǝmær
kwǝd
zg’or-ǝ!/?
my
brother
how
run-PRS.3SG
Exclamative evaluative reading (elliptical structure): ‘The speaker is surprised how fast
(beautifully, etc) her brother runs’
Exclamative manner reading: ‘The speaker is surprised at the manner (jogging, hopping, etc) by
which her brother runs’
Interrogative evaluative reading (elliptical structure): ‘The speaker wants to know how fast
(beautifully, etc) her brother runs’
Interrogative manner reading: ‘The speaker wants to know what is the manner (jogging, hopping,
etc) by which her brother runs’
(22c)
25
26
27
28
29
30
31
32
33
34
35
36
‘fsǝmær
‘How fast my brother runs!’ / ‘How fast does my brother run?’
14
15
16
17
18
19
20
21
22
23
24
Me
Kwǝd
ræsuʁd
u
atsǝ
k’aba!/?
how
beautiful
be.PRS.3SG
this
dress
‘How beautiful this dress is!’ / ‘How beautiful is this dress?’
Russian kak behaves similarly; the only difference is that it combines with a specific class of
adjectives in a predicative position, adjectives of measurement (e.g., kak dolog ‘how long’, kak
velik ‘how great’, kak širok ‘how wide’). Bulgarian kak, Lithuanian kaip, and Georgian rogor show
some further restrictions: they are only possible in adverbial constructions in exclamatives and
interrogatives. Finally, the use of Turkish nasıl is only restricted to one context, that is, elliptical
adverbial interrogatives.
Swedish hur, by contrast, is used in the widest range of contexts in interrogative
environments: it is compatible in Contexts (i) and (iii) as well as in elliptical and non-elliptical
adverbial constructions. As for exclamatives, this is only possible in elliptical gradable adverbial
constructions.
34
35
In order to obtain qualitative interpretations, other strategies are used in Hindi (see Sections 3.3 and 3.4).
In Georgian only in exclamatives and in Swedish only in interrogatives.
25
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
Languages such as Basque, Estonian, Hebrew, Hindi, and Korean are sensitive to ellipsis of
wh-words in adverbial constructions. Basque nola, Estonian kuidas, Hebrew eyx, Hindi kaise, and
Korean ettehkey are employed in exclamative and interrogative adverbial elliptical gradable
constructions only. Estonian is a unique language in this regard. It differentiates between adverbial
elliptical gradable constructions (i.e. with evaluative interpretations) and adverbial non-elliptical
constructions with manner interpretations, on the one hand, and all other contexts under
consideration (i.e. contexts with non-elided adverbs, gradable adjectives in NPs or in a predicative
position), on the other; this differentiation holds for both exclamatives and interrogatives. In the
former, the wh-word kuidas is possible and the wh-word kui makes the sentence ungrammatical,
cf. (23a) vs. (23a’). In the latter, the reverse situation holds: kui is appropriate, whereas kuidas is
not, cf. the contrasts between (23b) vs. (23b’), (23c) vs. (23c’), (23d) vs. (23d’) respectively.
Estonian
(23a)
mu
vend
tantsi-b!/?
how
I.GEN
brother.NOM
dance.PRS-3SG
Exclamative evaluative reading (elliptical structure): ‘The speaker is surprised how beautifully
(fast, etc) her brother dances’
Evaluative manner reading: ‘The speaker is surprised at the manner (barefoot, etc) by which her
brother dances’
Interrogative evaluative reading (elliptical structure): ‘The speaker wants to know how beautifully,
(fast, etc) her brother dances’
Interrogative manner reading: ‘The speaker wants to know what is the manner (barefoot, etc) by
which her brother dances’
(23a’) *Kui mu vend tantsib!/?
(23b)
25
26
27
Kuidas
Kui
kiiresti
mu
vend
jookse-b!/?
how
fast
I.GEN
brother.NOM
run.PRS-3SG
‘How fast my brother runs!’ / ‘How fast does my brother run?’
(23b’) *Kuidas kiiresti su vend jookseb!/?
(23c) Kui
how
28
29
30
kleidi
mu
õde
osti-s!/?
beautiful.GEN
dress.GEN
I.GEN
sister.NOM
buy-IPF
‘What a beautiful dress my sister bought!’ / ‘What beautiful dress did my sister buy?’
(23c’) *Kuidas ilusa kleidi su õde ostis!/?
(23d) Kui
how
31
32
33
34
35
36
37
ilusa
ilus
see
kleit
on!/?
beautiful.NOM.SG
this.NOM
dress.NOM.SG
be.PRS.3SG
‘How beautiful this dress is!’ / ‘How beautiful is this dress?’
(23d’) *Kuidas ilus see kleit on!/?
Last but not least, if a language allows for manner matrix wh-exclamatives, it also allows for
manner wh-clauses in subordinate surprise contexts, and vice versa. In this respect, manner whclauses are similar to individual and kind wh-clauses (see Sections 4.6 and 4.7 respectively).
26
1
3.6 Interim conclusion
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
To begin with, the findings of this section partially support Hypothesis 7. That is, qualitative whexclamatives cross-linguistically distinguish between Contexts (i) – (ii) and Contexts (iii) – (v)
established in Section 1.1.5. The former contexts are marked with qualitative wh-words. The latter
contexts demonstrate further division: quantitative wh-words are used in Contexts (iii) and (iv),
whereas evaluative/manner wh-words occur in Contexts (iv) and (v). In addition, Contexts (iv) and
(v) might have further distribution: elliptical gradable constructions (i.e. with evaluative
interpretations) are marked with a wh-word distinct from a wh-word used in non-elliptical gradable
constructions. Moreover, the division between Contexts (i) – (ii) and Contexts (iii) – (v) is not
always strict: quantitative wh-words might also be used in Context (i), whereas inanimate object
wh-words might occur in all the contexts, except for Context (v). All said above suggests that, in
qualitative wh-exclamatives, qualitative wh-words have a narrower distribution than all other whwords whose distribution seems to depend not directly upon the established five contexts but upon
the presence/absence of gradable adjectives or adverbs.
Secondly, the findings of this section support Hypothesis 8. That is, in some languages, nonelliptical Contexts (i), (iii), and (iv) are considered as more appropriate than elliptical Contexts (ii)
and (v).
Thirdly, Hypothesis 6 w.r.t. manner wh-exclamatives is only partially confirmed:
subordinate wh-clauses with manner exclamative readings are felicitous if and only if manner
matrix wh-exclamatives are felicitous.
Fourthly, several strategies might coexist in the same language. For instance, in Basque, all
the four strategies of qualitative wh-exclamatives discussed in this section are used, whereas
Estonian has only two strategies: inanimate object and evaluative/manner.
Last but not least, presumably, some of the discussed strategies are cross-linguistically more
consistent than the others. To illustrate, among the languages I studied, the strategy of inanimate
object wh-words is the most widespread, whereas the strategy of qualitative wh-words is restricted
to a smaller number of languages. In order to draw any solid statistical and theoretical implications
more languages should be involved.
30
4 Other wh-exclamatives
31
4.1 Preliminary remarks
32
33
34
35
36
37
38
39
40
41
42
In this section, my goals are twofold: on the one hand, to test Hypotheses 3 and 4 (both w.r.t.
quantitative wh-exclamatives) and, on the other hand, to test Hypotheses 5 and 6 (the latter w.r.t.
individual wh-exclamatives).36 In doing so, I examine behavior of the following types of whexclamatives: quantitative (‘how many/ much’), personal (‘who’), inanimate object (‘what’),
locative (‘where’), temporal (‘when’), causal (‘why’), and individual (‘which’). In addition, I
studied kind wh-exclamatives (‘what kind’), which seem to behave similarly to individual whexclamatives. Importantly, in all such languages, qualitative wh-exclamatives are felicitous (see
Section 3).
A few technical comments should be made here. Firstly, if felicitous, only matrix whexclamatives are given (in all such cases, their subordinate counterparts are supposed to be
felicitous as well). Otherwise, both infelicitous matrix wh-exclamatives and their felicitous
36
Manner wh-exclamatives have been studied on a par with evaluative wh-exclamatives in Section 3.5.
27
1
2
3
4
5
subordinate counterparts are provided. Secondly, although the study was aimed at collecting data
for both subject and direct object positions for inanimate object and personal wh-words, in what
follows, I consider the subject position for personal wh-words and the direct object position for
inanimate object wh-words.
6
4.2 Quantitative wh-exclamatives
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
Since quantitative wh-exclamatives (‘how many/much/few’) in the most comprehensively studied
languages such as English or German involve ‘how’ word and, therefore, exhibit the same pattern
of exclamatives as ‘how’ + gradable adjectival or adverbial structures, to the best of my
knowledge, so far they have not been studied cross-linguistically. However, in many languages,
quantitative exclamatives use a distinct wh-word.
It seems to be a well-known fact that languages divide into groups w.r.t. how they encode
basic quantity meanings: with one word (e.g., Estonian mitu or Russian skol’ko) or with two words
(e.g., English how many, how much or German wie viel ‘how many/much’). In languages with one
word for a basic quantity meaning, using two words is marked and expresses large quantity (e.g.,
Estonian kui palju and Russian kak mnogo ‘how many/much’), whereas in languages with two
words for a basic quantity meaning, the use of two words is neutral and does not necessarily
express the meaning of large quantity, although it may have this meaning as well. Importantly, in
both groups of languages, small quantity meaning is marked, e.g., is expressed with two words (cf.
English how few and how little, Estonian kui vähe ‘how few/little’, and Russian kak malo ‘how
few/little’).
In this paper, I primarily studied sentences expressing basic quantity meanings which were
witnessed in all the languages of the sample in matrix wh-exclamatives and their subordinate
counterparts. As for large quantity, I found that exclamatives exhibit differences between the whwords for basic and large quantity. For instance, in Estonian exclamatives mitu with the basic
quantity meaning is not used, whereas kui palju with the large quantity meaning is felicitous.
However, so far I have examined only the use of those wh-words which denote basic quantity and
leave the findings of large and small quantity exclamative contexts for future research.
I also paid attention to the distinction between countable and uncountable quantity.
Interestingly, Turkish was the only language from the sample which has two quantitative whwords for countable and uncountable quantity: kaç ‘how many’ and ne kadar çok ‘how much’
respectively. The following examples illustrate their use:
Turkish
(24a)
kaç
tane
elma
al-mış-ın!
INTERJ
how_many
item
apple
buy-PST.INFER-2SG
‘Wow, how many apples you bought!’
35
36
(24b)
37
38
Vay,
Vay,
ne kadar
çok
balık
al-dı-m!
INTERJ
how_much
much
fish
buy-PST-1SG
‘How much fish I bought!’
28
1
2
3
4
5
Hebrew is a language which exhibits remarkable syntactic features in quantitative matrix whexclamatives: the proclitic complementiser še, according to the language consultants, is rather
optional.37
Hebrew
(25) Vau,
INTERJ
6
7
8
9
10
11
kama
tapux-im
(še-)ha-yelad-im
axl-u!
how_many
apple-PL
COMP-DF-child-PL
eat.PST-3PL
‘How many apples the children ate!’
Besides the languages I studied, quantitative exclamatives are present in the Romance languages
(see Marandin (2008) for French, Olbertz (2009) for Spanish, Castroviejo (2006) for Catalan,
Zanuttini and Portner (2003) for Italian), and in Classical Greek (see Faure (2012)).
12
4.3 Personal and inanimate object wh-exclamatives
13
14
15
16
17
18
19
Languages such as Basque, Bulgarian, Dutch, Estonian, Georgian, German, Hebrew, Hungarian,
Lithuanian, Ossetic, Russian, and Turkish allow for personal and inanimate object wh-words in
matrix exclamatives. To illustrate, Turkish exclamatives demonstrate felicitousness of the
following wh-words: kim ‘who’ and ne ‘what’. Both are also appropriate in subordinate surprise
contexts.
20
21
Turkish
(26a)
biz-e
yarın
kim
gel-ecek!
INTERJ
we-DAT
tomorrow
who
come-FUT.3SG
‘Wow, the person that will come to us tomorrow!’ (‘Wow, who will come to us tomorrow!’, lit.)
(26b)
22
23
24
25
26
27
28
29
30
31
Vay,
Vay,
anne-m
ne
piş-ir-miş!
INTERJ
mother-my
what
cook-PASS.PTCP-INFER.3SG
‘Wow, the stuff mom cooked!’ (‘Wow, what my mother cooked!’, lit.)
In Hebrew, the proclitic complementiser še attaches to the linearly initial wordform in a clause
(Shlonsky 1988) and is obligatory in exclamatives (Sharvit 1999).38 However, according to the
judgements of the native speakers, the complementiser in matrix exclamatives with personal and
inanimate object wh-words is rather optional. Remarkably, in subordinate contexts, the
complementiser is less felicitous. For the sake of brevity, only two examples – with the inanimate
object wh-word ma in matrix and subordinate use – are given below.
Hebrew
(27a)
Vau,
ma
(še-)ima
af-ta!
INTERJ
what
COMP-mother
bake.PST-3SG.F
‘Wow, the stuff mom baked!’ (‘Wow, what mother baked!’, lit.)
32
33
37
Interestingly, in personal and inanimate object matrix wh-exclamatives, the complementiser is optional as well,
whereas in locative and temporal matrix wh-exclamatives it is obligatory (see Sections 4.3 and 4.4).
38
On the basis of this and Paduan che-exclamatives discussed in Zanuttini and Portner (2003), Rett (2008b) pointed
out that wh-exclamatives are free relatives rather than interrogatives. See also footnote 3.
29
(27b)
1
2
3
4
5
6
7
Ata
lo
taamin
ma
(??še)-ima
af-ta!
you.M
NEG
believe.FUT.2SG.M
what
COMP-mother
bake.PST-3SG.F
‘You won’t believe what mom baked!’
The languages which do not allow for personal and inanimate object wh-words in matrix
exclamatives are English, Hindi, Classical Greek, Korean, and the Romance languages (Catalan,
French, Italian, and Spanish). The next examples from Hindi illustrate this.
Hindi
(28a)
Tum,
višwās nahīn
kar-ogī
ki
kaun
nāc-t-ā
you.HON
belief
do-FUT.2PLF
COMP
who
dance-PRS.PTCP-SG.M
NEG
h
rah-ā
rāt-b ar!
stay-PST
night-all
‘You won’t believe who was dancing all night!’
8
9
(28b) Tum,
you.HON
višwās nahīn
kar-ogī
ki
main
kyā
lāyā!
belief
do-FUT.2PLF
COMP
I
what
bring.PST.M.SG
NEG
‘You won’t believe what I brought!’
10
11
12
4.4 Locative and temporal wh-exclamatives
13
14
15
16
17
18
Languages such as Bulgarian, Georgian, German,39 Hungarian, and Ossetic allow for locative and
temporal wh-words in matrix exclamatives. To illustrate, Ossetic exclamatives demonstrate
felicitousness of kæm ‘where’ and kæd ‘when’. These wh-words are also appropriate in
subordinate surprise contexts.
Ossetic
(29a)
kæm
uyd-tæn
abon!
INTERJ
where
be-PST.INTR.1SG
today
‘Wow, the place I have been today!’ (‘Wow, where I have been today!’, lit.)
19
20
(29b)
21
22
23
24
25
26
27
O,
O,
kæd
kaf-dzystæm!
INTERJ
when
dance-FUT.1PL
‘Wow, at what time we will dance!’ (‘Wow, when we will dance!’, lit.)
Basque, Estonian, Lithuanian, Russian,40 and Turkish allow only for locative wh-words. In
subordinate surprise contexts, both locative and temporal wh-exclamatives are appropriate, cf. the
next examples from Basque.
Basque41
(30a) Ez
NEG
du-zu
sines-tu-ko
non
egon
AUX.TR-2SG.A
believe-PFV-PROSP
where
be.PFV
Some of the native speakers preferred the use of alles in matrix exclamatives, cf. Wo ich heute ??(alles) war! ‘Where
I have been today!’ (lit.). Compare also the contrast between (6c) and footnote 11.
40
In Russian, temporal wh-exclamatives are possible only if they are introduced with particles, e.g., nado že, cf. also
causal wh-exclamatives in Section 4.5.
41
In the case of matrix exclamatives, the native speakers suggested to use an exclamative particle, e.g., hara or a.
39
30
1
2
naiz-en
gaur!
AUX.1SG-COMPL
today
‘You won’t believe where I was today!’
(30b) *(Ez
NEG
du-zu
sines-tu-ko)
noiz
dantza-tu-ko
AUX.TR-2SG.A
believe-PFV-PROSP
when
dance-PFV-PROSP
du-gu-n!
AUX.TR-1PL.A-COMPL
3
4
5
6
7
8
9
10
11
12
‘You won’t believe when we will dance!’
In Hebrew, unlike personal and inanimate object wh-exclamatives, the proclitic complementiser
še is obligatory in locative and temporal wh-exclamatives (otherwise, they are interpreted as
questions). Remarkably, in subordinate contexts, the complementiser is worse (cf. Section 4.3).
For the sake of brevity, below I provide examples with ejfo ‘where’.
Hebrew
(31a)
??
ejfo
(še-)ha-xaver-im
šelxa
rakd-u!
INTERJ
where
COMP-DF-friend-PL
your
dance.PST-3PL
/*Vau,
‘Wow, the place your friends danced!’ (‘Wow, where your friends danced!’, lit.)
(31b) Ata
you.M
13
lo
taamin
NEG
believe.FUT.2SG.M
ejfo
(še-)ha-xaver-im
šelxa
rakd-u!
COMP-DF-friend-PL
your
dance.PST-3PL
where
‘You won’t believe where your friends danced!’
14
15
16
17
English, Classical Greek, Hindi, Korean, and the Romance languages (Catalan, French, Italian,
and Spanish) do not allow for locative and temporal wh-words.
18
4.5 Causal wh-exclamatives
19
20
21
22
23
24
25
26
27
28
In all the languages of the sample, causal wh-words (i.e. with the meaning ‘why’) are infelicitous
in matrix exclamatives. To make them felicitous, some languages require additional support: either
particles or prosodic stress are involved. To illustrate, Russian počemu-exclamatives are possible
only when preceded with a particle (e.g., nado že). Beyond the languages of the sample, in German,
warum is felicitous in matrix exclamatives only if the main prosodic stress falls on it, see Repp
(2013).
By contrast, in all the languages of the sample, causal wh-words occur in subordinate
surprise contexts. To illustrate, Georgian raṭoms ‘why’ is possible only in such an environment.
Georgian
(32) ??/*(Ar
NEG
29
30
daijereb,)
raṭoms
ӡinavs
čem-s
važ-s
coṭa!
believe.FUT.2SG
why
sleep.he_DAT
my-DAT
son-DAT
little.NOM
‘You won’t believe why my son sleeps so little!’
31
1
2
3
4
The infelicitousness of causal matrix wh-exclamatives is further supported with the data from
Hungarian, Italian, and the North-Caucasian languages, cf. Lipták (2006), Zanuttini and Portner
(2003), and Kalinina (2011) respectively.
5
4.6 Individual wh-exclamatives
6
7
8
9
Wh-words with the meaning ‘which’ are typically appropriate in interrogatives but not in
exclamatives. They occur in Contexts (i) and (ii). The following Hindi sentences illustrate this:
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
Hindi
(33)
Mer-ī
bahan
ne
kauns-ī
(sundar)
pošāk
xarīd-ī?/*!
my-SG.F
sister
ERG
which-SG.F
beautiful
dress
buy-PST.F.SG
‘Which (beautiful) dress my sister bought!’ (lit.) / ‘Which (beautiful) dress did my sister buy?’
English which, Estonian missugune, Georgian romeli, Lithuanian kurì, Russian kotoryj, and
Turkish hangi exhibit the same pattern. In languages such as Basque and Korean, qualitative whwords might have an individual meaning in interrogatives; however, none of the language
consultants considered qualitative wh-words with individual meanings appropriate in
exclamatives.42
The following languages from the sample allow for individual meanings of qualitative whwords: Bulgarian (koj), Estonian (milline),43 Hebrew (eyze), Korean (etten), and Russian (kakoj)
respectively. Consider the following situation for Russian: A is going to buy a new car. A and B
visit a carshow. The next day B finds out which car A has bought and is surprised that A has bought
a particular car. B exclaims (34). In this context, kakoj has an individual meaning: B is surprised
that A has chosen a particular car from the set of the cars known to both of them.
Russian
(34) Ogo
INTERJ
25
26
27
28
29
30
31
32
33
34
35
36
kaku-ju
mašin-u
ty
kupi-l!
what-ACC.F
car-ACC.SG
you
buy-PST[SG.M]
‘Wow, the car you bought!’ (‘Wow, what a car you bought!’, lit.)
Remarkably, if B exclaimed Kakuju krasivuyu mašinu ty kupil! ‘What a beautiful car you bought!’,
kakoj would have a quality meaning rather than an individual meaning.
Remarkably, as in the case of manner exclamatives (see Section 3.5), subordinate
counterparts of individual matrix exclamatives are felicitous if and only if individual matrix
exclamatives are felicitous.
In fact, any wh-word can have an individual meaning on condition that it is felicitous in a
context in which the answer is restricted to a given set of alternatives. Consider a question context:
I want to know where John lives now, in Canada or in Mexico. I leave the question of whether
other wh-words, besides qualitative wh-words, can have an individual meaning in exclamatives
for future research.
42
According to Rett (2008a, 2008b), English what-exclamatives are also infelicitous in individual scenarios: e.g.,
#What cards he picked!, interpreted as if he picked diamonds.
43
Interestingly, Estonian exhibits two ‘which’ words milline and missugune; according to native speakers, the former,
unlike the latter, is used in exclamatives.
32
1
2
3
4
5
6
7
8
9
10
11
12
13
4.7 Kind wh-exclamatives
Kind wh-words (i.e. with the meaning ‘what kind’: e.g., from what material an object is made; to
which kind it belongs; etc) do not constitute a wide-spread strategy. They modify an NP and,
importantly, require the absence of an adjective. The strategy is found in the following languages
of the sample: Basque, Georgian, Korean, and Turkish. Moreover, according to Lipták (2006), it
is witnessed in Hungarian (milyen). Hebrew and Russian allow for kind interpretations of
qualitative wh-words.
Consider, for example, Basque. It has the wh-word nolako and the fixed wh-word complex
zer nolako. Both are used with an NP in exclamatives and interrogatives and prohibit an adjective.
Note that not only in exclamatives but also in interrogatives, an NP has the suffix -a.
Basque
(35a)
soineko
(ederr-a)
eros-i
zu-en
what
dress
beautiful-DF
buy-PFV
AUX.TR.PST-PST.COMPL
nire
arreb-a-k?/!
my
sister-DF-ERG
‘What kind of a (beautiful) dress did my sister buy?’
‘What kind of a (beautiful) dress my sister bought!’ (lit.)
14
15
16
(35b)
Nolako
soineko-a
eros-i
zu-en
what
dress-DF
buy-PFV
AUX.TR.PST-PST.COMPL
nire
arreb-a-k?/!
my
sister-DF-ERG
‘What kind of a dress did my sister buy?’
‘What kind of a dress my sister bought!’ (lit.)
17
18
19
(35c)
*Zer
nolako
soineko
(ederr-a)
eros-i
zu-en
what
what
dress
beautiful-DF
buy-PFV
AUX.TR.PST-PST.COMPL
nire
arreb-a-k?/!
my
sister-DF-ERG
‘What kind of a (beautiful) dress did my sister buy?’
‘What kind of a (beautiful) dress my sister bought!’ (lit.)
20
21
22
(35d)
23
24
25
*Nolako
Zer
nolako
soineko-a eros-i
zu-en
what
What
dress-DF
AUX.TR.PST-PST.COMPL
nire
arreb-a-k?/!
my
sister-DF-ERG
buy-PFV
‘What kind of a dress did my sister buy?’
‘What kind of a dress my sister bought!’ (lit.)
33
1
2
3
4
5
6
7
Georgian rogori, Hebrew eyze, Hindi kaisā,44 Hungarian milyen, Korean mwusun, Russian kakoj,
Turkish nasıl,45 behave similarly, i.e., they can have the kind meaning and are only felicitous with
an NP and no adjective in exclamatives and interrogatives.
Remarkably, as in the case of manner and individual exclamatives (see Sections 3.5 and 4.6
respectively), subordinate counterparts of kind matrix exclamatives are felicitous if and only if
kind matrix exclamatives are felicitous.
8
4.8 Interim conclusion
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
In this section, I have examined the use of wh-words in matrix wh-exclamatives and in subordinate
wh-clauses with exclamative readings. The main findings are as follows.
Firstly, Hypotheses 3 and 4 (both w.r.t. quantitative wh-exclamatives) are confirmed. In
other words, like qualitative (see Section 3), quantitative wh-exclamatives do not demonstrate
matrix-subordinate asymmetry and represent prototypical instances of wh-exclamatives.
Secondly, the findings only partially support Hypothesis 6 w.r.t. individual whexclamatives. That is, as in the case of manner wh-exclamatives (see Section 3.5), subordinate whclauses with individual exclamative readings are felicitous if and only if individual matrix whexclamatives are felicitous. The same goes for kind wh-exclamatives. In addition, the languages
of the sample are divided into those which tolerate the use of individual interpretations of some
qualitative wh-words with a bare NP (i.e. without an adjective) in matrix exclamatives and their
subordinate counterparts and those which do not. The same goes for kind wh-words and kind
interpretations of some qualitative wh-words with a bare NP.
Thirdly, Hypothesis 5 is confirmed. That is, the following matrix wh-exclamatives have
felicitous subordinate counterparts and form an acceptability hierarchy: inanimate object/
personal/ locative > temporal > causal wh-exclamatives. In other words, if a language has causal
wh-exclamatives, it also has inanimate object, personal, locative, and temporal wh-exclamatives;
if a language has temporal wh-exclamatives, it also has inanimate object, personal, locative, and
temporal wh-exclamatives. This, however, does not suggest that inanimate object, personal, and
locative wh-exclamatives are prototypical wh-exclamatives, like qualitative and quantitative whexclamatives. The study revealed that the languages of the sample can be divided into (i) those
which allow for wh-words denoting a person, an inanimate object, and a location in matrix
exclamatives, and (ii) those which do not license the use of them in such an environment (although
their subordinate counterparts are felicitous). In addition, the languages of the group (i) are split
into those which allow the use of wh-words denoting time in matrix exclamatives and those which
do not, whereas the languages of the group (ii) do not license this use. Last but not least, the use
of wh-words denoting cause either requires additional support (particles or prosodic stress) or is
infelicitous in matrix exclamatives.
For the time being, I have a very tentative explanation for these observed facts. The hierarchy
seems to be related to an event structure (and presumably to semantic roles): generally, core
participants of an event are more privileged than non-core participants (non-core elements reveal
further hierarchy).46 Typically, person and inanimate object have cognitively more privileged
Hindi has a homonymous wh-word kaise ‘how’, see Section 3.5.
Turkish has a homonymous wh-word nasıl ‘how’, see Section 3.5.
46
Grammar reflects this semantic distinction: core elements are grammatically encoded as arguments, whereas noncore elements are grammatically encoded as adjuncts.
44
45
34
1
2
3
4
5
6
7
8
9
statuses than time and cause. As for locative wh-exclamatives, in most languages of my sample,
they reside to the left of the implicit hierarchy grouping together with personal and inanimate
object wh-exclamatives. Interestingly, languages such as Hebrew grammatically encode locative
and temporal wh-exclamatives in a different way than personal and inanimate object whexclamatives. This suggests that, in this or that way, wh-exclamatives are sensitive to the semantic
distinction between core vs. non-core participants.47 However, for the time being, I do not have a
clear idea of what this entails. Future research of exclamatives should definitely involve semantic
roles.
10
5 Discussion
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
The language data studied in this paper supports Evans’ (2007) hypothesis of insubordination of
wh-exclamatives, namely that subordinate contexts with exclamative readings are diachronically
primary and matrix wh-exclamatives originate via ellipsis of main clauses.
From a grammaticalisation point of view, the direction of change from complex structures
to simple structures seems to be unexpected. However, as Traugott and Heine (1991: 6–7),
“virtually nothing is exceptionless, and there are of course instances of change in languages that
are counterexamples of tendencies that can be characterized as “less>more grammatical”, “main
clause>subordinate clause”. Following them, Evans (2007: 422) argues that “insubordination is an
important phenomenon because of the unusual way the direction of diachronic change runs: from
subordinate clause to main clause, from morphosyntax to discourse, and (in its initial stage) from
grammar to pragmatics.”
The hypothesis of insubordination seems to accord with the idea that “the exclamatory
meaning […] is attributable to the matrix you won’t believe […] and hence is perfectly consistent
with the complement being interrogative” (Huddleston 1993: 179). A possible scenario would be
that an exclamatory meaning is somehow transfered from a matrix predicate to a subordinate
clause. Presumably, insubordination takes place via ellipsis of factive predicate forms such as you
won’t believe or imperative look, although I do not claim that these two forms are the only possible.
Importantly, such predicate forms convey not only their “primary” illocutionary forces but also
the speaker’s evaluation. To illustrate, in the utterances You won’t believe what a car Mary bought
or Look what a car Mary bought, the main clause predicate forms convey not only assertive and
imperative illocutionary forces respectively, but also the speaker’s evaluation of Mary’s car. In
contrast, there are predicates which allow for subordinate wh-clauses with exclamative readings
(e.g., know as in Fred knows what a car Mary bought) and which do not necessarily preserve the
speaker’s evaluation of John’s tallness; rather, it can be the main clause subject (i.e. Fred) who
evaluates a given fact. Therefore, presumably, insubordination does not take place via ellipsis of
factive predicate forms predicates such as know.
The findings of this paper have some implications for the classification of wh-exclamatives
in particular and for the typology of exclamatives in general.
47
Few of the languages studied here show felicitousness of causal matrix exclamatives which require special contexts;
a plausible explanation I suggest is that cause is the most peripheral among non-core participants. To put it differently,
cause is cognitively less important than person, object, and even location or time. The reason is that cause is not
directly associated with an event. Actually, it can constitute another event and, grammatically, it can be expressed
with a separate clause.
35
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
5.1 Implications for the classification of wh-exclamatives
The study reported in this paper established three types of wh-exclamatives. The first group
comprises qualitative and quantitative wh-exclamatives which seem to be prototypical, or basic,
matrix wh-exclamatives cross-linguistically. In other words, if a language allows for matrix whexclamatives, it has qualitative and quantitative matrix wh-exclamatives. Moreover, qualitative
wh-exclamatives exhibit the most ramified network of strategies and presumably is the main
source of exclamative-only constructions.
Personal, inanimate object, locative, temporal, and causal wh-exclamatives constitute the
second group. Unlike qualitative and quantitative wh-exclamatives, they are subject to matrixsubordinate asymmetry which is regulated with the implicit semantic hierarchy which has much
in common with the distinction between core vs. non-core event participants.
The third group consists of manner, individual, and kind wh-exclamatives. This group does
not demonstrate matrix-subordinate asymmetry, that is, a set of manner, individual, and kind whwords in matrix clauses is identical to a set of the same wh-words in subordinate clauses.
The proposed classification of wh-exclamatives partially resembles the classification
proposed in a recent study by Chernilovskaya and Nouwen (2012), Nouwen and Chernilovskaya
(2015). The authors suggested that wh-exclamatives are divided into two types: (i) those which
express noteworthiness of an individual level (i-level, a referent of a wh-word is noteworthy) and
(ii) those which convey noteworthiness of a proposition level, or, roughly speaking, event level
(e-level, a proposition refered to in an exclamative clause is noteworthy).48,49 Wh-exclamatives of
the former group are called scalars, whereas wh-exclamatives of the latter group are coined as nonscalars. Scalars can be instantiated with help of English what a construction, whereas non-scalars
can be illustrated by virtue of German (6c) example. Accordingly, languages differ w.r.t. whether
scalars alone or scalars and non-scalars together are possible as matrix (cf. English vs. German).
Moreover, the authors argue that both scalars and non-scalars are of a different morphosyntactic
nature. For instance, scalars, unlike non-scalars, are syntactically reducible, cf. English What a
cake! vs. *Who!; German (6c) vs. *Wen alles! ‘Who all!’, see Siemund (2015), who provided
independent corpus quantificational evidence for the syntactic reducilibity of English what a
constructions. All in all, the authors suggest that scalars are non-standard wh-constructions,
whereas non-scalars resemble interrogatives.
At first sight, it seems that my qualitative and quantitative wh-exclamatives correspond to
Nouwen and Chernilovskaya’s scalars, whereas my other types of wh-exclamatives correspond to
Nouwen and Chernilovskaya’s non-scalars.50 However, I did not touch upon the semantic (non)scalar status of the three types of wh-exclamatives established in this paper, and further research
should definitely be directed towards more careful and detailed examination of these two
classifications.
38
39
40
5.2 Implications for the typology of exclamatives
As for the syntactic typology of exclamatives in general, cross-linguistically, matrix exclamative
structures are divided into five quite diverse syntactic strategies (for an overview see Michaelis
48
For a full semantic justification of this approach, I refer the reader to Nouwen and Chernilovskaya (2015).
Nouwen and Chernilovskaya (2015) claim that the dichotomy is not applicable to adverbial ‘how’ exclamatives.
50
‘How’-exclamatives is beyond the scope of this comparison.
49
36
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
(2001), Zevakhina (2013) among others): subordinate clauses, nominal phrases (with or without
relative clauses and nominalisations), wh-clauses, inversion clauses, cataphoric51 (so/ such)
clauses. Some of them are split into sub-strategies. For instance, subordinate clauses are further
split into ‘that’ clauses, infinitives, and participles. The next English exclamatives (36a-f) illustrate
the following strategies respectively: wh-clauses, inversion clauses, ‘that’ clauses, infinitives,
nominal phrases with relative clauses, and cataphoric clauses.
(36a) How beautiful the rainbow is!
(36b) Boy! Is syntax easy! (McCawley 1973)
(36c) That he should have left without asking me! (Quirk et al. 1985: 841)
(36d) To think that I would once be a millionnaire! (ibid.)
(36e) The incredible things he says! (Castroviejo and Schwager 2008: 178)
(36f) It is so hot! (Michaelis 2001: 1040)
However, there is a long-running debate on whether all the aforementioned constructions represent
an exclamative sentence type. The most controversial structures are inversion and cataphoric
constructions: e.g., McCawley (1973) and Michaelis (2001) advocate the idea that these structures
constitute an exclamative sentence type, whereas Huddleson (1993) and Collins (2004) cast doubt
on this. Inversion structures are identical to yes/no-interrogatives, whereas cataphoric
constructions are declaratives in form. Consequently, pace these two authors, the fact that such
structures function as exclamations52 is a pragmatic rather than syntactic issue.53
Another question is whether all the aforementioned strategies of exclamatives have
diachronic origins as subordinate structures. Evans (2007) argues only for wh-exclamatives, ‘that’
clauses, and infinitives. It is not clear whether the others — cataphoric, nominal, and inversion
exclamatives — undergo the process of insubordination. We can only notice that nominal and
cataphoric structures seem to be elliptical: cf. It is so cold [that I have to wear my heavy coat]54
and The shoes she is wearing [surprise me]. Inversion constructions are even harder to make any
predictions at all concerning subordination or ellipsis. Future research should definitely involve
not only the role of insubordination but also the role of ellipsis in the study of various exclamative
structures.
51
Michaelis (2001) calls them anaphoric, but I think a better variant, which was independently suggested to me by an
anonymous reviewer, is cataphoric since so/ such phrases refer to a subsequent elided part of a clause.
52
It is well-known that exclamations exhibit a pragmatic phenomenon and are opposed to other speech acts (e.g.
assertions and questions), whereas exclamatives demonstrate a syntactic phenomenon and are contrasted to other
sentence types (e.g. declaratives, interrogatives, and imperatives).
53
The radical view on exclamatives would also throw out subordinate structures and DPs from the basket of
exclamatives since they also have various functions, i.e., they are not exclamative-only strings in a strict sense. This
would leave us with the only strategy – wh-clauses. In fact, some of them are identical to subordinate interrogatives
in form. Should we throw them out as well? Doing so would lead to postulating exclamative-only structures (e.g., (1ab)) as a proper exclamative sentence type. Actually, it is too small in comparison to declaratives, interrogatives and
imperatives and, therefore, one can merely ignore it. Does this mean that there is no exclamative sentence type? There
is a semantic basis for claiming that exclamatives do constitute a separate sentence type. It has been argued that all
the five strategies of exclamatives considered above share common semantic properties (cf. Michaelis 2001,
Zevakhina 2013). What is needed is to provide syntactic evidence for attributing all the five strategies to one and the
same exclamative sentence type. Such evidence, as I suppose, might be insubordination, ellipsis or both.
54
I owe the observation that cataphoric constructions are elliptical to an anonymous reviewer.
37
1
2
3
4
5
6
7
8
9
5.3 Implications for exclamative-only constructions
The majority of exclamative structures we have discussed so far are identical to interrogatives in
form. However, exclamative-only sentences like (1a-b) are not identical to interrogatives.
Huddleston (1993) claims that they are truly embeddable exclamatives. Interestingly, only some
exclamative-only sentences have subordinate counterparts, whereas others do not, cf. (3) vs. (37ab) respectively.
German
(37a)
(37b)
10
11
12
13
14
15
16
17
18
19
20
21
22
Was/wie
bist
du
groß
geworden.
what/how
be.PRS.2SG
you
Big
become.PASS.PTCP
‘How big you have become!’
*Irre
was/wie
bist
du
groß
geworden.55
be.mistaken.PRS.1SG
be.PRS.2SG
you
big
become.PASS.PTCP
what/how
‘It is unbelievable how big you have become!’
(Sæbø 2006: 4)
For the time being, the distinction between exclamative-only constructions with subordinate vs.
insubordinate correlates is hard to explain. It is also not clear whether the hypothesis of
insubordination is applicable to such constructions. My two sole observations are as follows.
Firstly, at least in English, exclamative-only constructions with subordinate correlates and
exclamatives which are identical in form to interrogatives occur in the context of the same range
of matrix predicates. Secondly, both groups of exclamatives have the same subordinate syntax
(w.r.t. wh-movement and subject-verb inversion). At least, this does not contradict the hypothesis
of insubordination. At any rate, it would be good to find diachronic evidence for the
insubordination process of such exclamatives.56
23
6 Conclusion
24
25
26
27
28
29
30
31
On the basis of the data from eleven genealogically related and unrelated languages, this paper
provided further evidence for the hypothesis of insubordination w.r.t. wh-exclamatives and
established three cross-linguistically consistent groups of wh-exclamatives.
To obtain a more comprehensive picture of how all the three types of wh-exclamatives
function in natural languages, more linguistic data should be further investigated. Moreover, to
further verify the hypothesis of insubordination, we need to study subordinate exclamatives
diachronically and establish a more precise range of matrix predicates as well as their semantic
and grammatical forms.
55
The ungrammaticality of this example can be explained by the fact that subordinate sentences generally require the
final position of the verb. Note that the following sentence, where the verb of a subordinate clause takes the final
position, is felicitous with wie but infelicitous with was. Remarkably, the adjective groß should follow wie, otherwise
the sentence is ungrammatical.
(i) Es
ist
irre
wie/*was groß du
geworden
bist!
it
be.PRS.3SG
be.mistaken.PRS.1SG
how/what
‘It is unbelievable how big you have become!’
56
I thank an anonymous reviewer for this remark.
big
you
become.PASS.PTCP
be.PRS.2SG
38
1
2
Conventions
3
4
5
6
7
8
9
??
10
– an example is not entirely rejected but also not readily accepted
– an example is almost inappropriate
* – an example is inappropriate
!/? – an example is appropriate both in exclamatives and interrogatives
!/*? – an example is appropriate in exclamatives but not in interrogatives
?/*! – an example is appropriate in interrogatives but not in exclamatives
??/*
Abbreviations
1
first person
F
feminine
PASS
Passive
2
second person
FUT
future
PFV
Perfective
3
third person
GEN
genitive
PL
plural
A
agent-like argument
of canonical
transitive verb
HON
honorative
PRF
perfect
ACC
accusative
IND
indicative
PROSP
prospective
ADNZ
adnominal modifier
INDF
indefinite
PRS
present
APPER
apperceptive (mood)
INFER
inferentive
PST
past
ATTR
atributiviser
IPF
imperfect
PTCL
particle
AUX
auxhiliary
LOC
locative
PTCP
participle
COMP
complementiser
M
masculine
PV
preverb
COMPL
subordinate clause marker
N
neuter
SG
singular
CVB
converb
NEG
negation
TOP
topic
DF
definite
NOM
nominative
TR
transitive
ERG
ergative
11
12
References
13
14
15
16
17
18
19
20
21
22
23
Abels, Klaus. 2004. Why surprise-predicates do not embed polar interrogatives. Linguistische
Arbeitsberichte 79. 203–221.
Abels, Klaus. 2005. Remarks on Grimshaw’s clausal typology. In Emar Maier, Corien Bary and
Janneke Huitink (eds.), Proceedings of Sinn und Bedeutung 9. 1–15.
Abels, Klaus. 2010. Factivity in exclamatives is a presupposition. Studia Linguistica 64(1). 141–
157.
Akhvlediani, Georgij S. (ed.). 1963. Grammatika osetinskogo jazyka. [The grammar of the Ossetic
language]. Ordzhonikidze: NII pri sovete ministrov Severo-Osetinckoj ASSR.
Andueza, Patricia. 2011. Rhetorical exclamatives in Spanish. Columbus, OH: The Ohio State
University dissertation.
Aronson, Howard. 1990. Georgian: a reading grammar. Columbus, OH: Slavica publishers.
39
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
Badan, Linda and Lisa Cheng. 2015. Exclamatives in Mandarin Chinese. Journal of East-Asian
Linguistics 24(4). 383–413.
Bagaev, Nikolaj K. 1982. Sovremennyj osetinskij jazyk [Contemporary Ossetic Language].
Ordzhonikidze: Ir.
Beyssade, Claire. 2009. Exclamation and presupposition. In Paul Égré and Giorgio Magri (eds.),
MIT Working Papers in Linguistics 60, 19–34.
Castroviejo, Elena. 2006. Wh-exclamatives in Catalan. Barcelona: Universitat de Barcelona
dissertation.
Castroviejo, Elena and Magdalena Schwager. 2008. Amazing DPs. Proceedings of SALT
[Semantics and Linguistic Theory] XVIII. Ithaca, New-York: CLC publications. 176–193.
Chang, Suk-Jin. 1996. Korean. Amsterdam: John Benjamins.
Chernilovskaya, Anna and Rick Nouwen. 2012. On wh-exclamatives and noteworthiness. In Maria
Aloni, Vadim Kimmelman, Floris Roelofsen, Galit Sassoon, Katrin Schulz, and Matthijs Westera
(eds.), Logic, Language and Meaning: 18th Amsterdam Colloquium. Lecture notes in Computer
Science vol. 7218, 271–280. Springer.
d’Avis Franz-Josef. 2002. On the interpretation of wh-clauses in exclamative environments.
Theoretical Linguistics 28(1). 5–32.
de Rijk, Rudolf. 2008. Standard Basque: a progressive grammar. Cambridge & London: MIT
Press.
Delfitto, Denis and Fiorin Gaetano. 2014. Exclamatives: issues of syntax, logical form and
interpretation. Lingua 152. 1–20.
Dryer, Matthew. 2005. Position of interrogative phrases in content questions. In Martin
Haspelmath, Matthew Dryer, David Gil & Bernard Comrie (eds.), The World Atlas of Language
Structures, 378–381. Oxford: Oxford University Press.
Elliott, Dale. 1974. Toward a grammar of exclamations. Foundations of language 11(2). 231–246.
Erelt, Mati. 2003. Estonian language. Tallinn: Estonian academy publishers.
Erelt, Mati, Tiiu Erelt, and Kristiina Ross. 2007. Eesti keele käsiraamat. Tallinn.
http://www.eki.ee/books/ekk09/index.php?p=5&p1=2&id=447 (accessed 27 February 2016)
Evans, Nicholas. 2007. Insubordination and its uses. In Irina Nikolaeva (ed.), Finiteness:
theoretical and empirical foundations, 366–431. Oxford: Oxford University Press.
Faure, Richard. 2012. The interaction between presupposition and focus: Classical Greek whexclamatives. Journal of Greek linguistics 12(2). 276 – 304.
Gagkaev, Konstantin E. 1956. Sintaksis osetinskogo jazyka [Syntax of the Ossetic language].
Ordzhonikidze: Severo-osetinskoe knizhnoe izdatel’stvo.
Glinert, Lewis. 2005. Modern Hebrew: an essential grammar. London: Routledge.
Göksel, Aslı & Ceila Kerslake. 2005. Turkish: a comprehensive grammar. London: Routledge.
Grimshaw, Jane. 1979. Complement selection and the lexicon. Linguistic inquiry 1 (2). 279–326.
Haegeman, Liliane. 1997. Introduction to elements of grammar. A handbook of generative syntax.
Dordrecht: Kluwer.
Harris, Alice. 1984. Georgian. In William Chisholm, Louis Milic & John Greppin (eds.),
Interrogativity: A colloquium on the grammar, typology and pragmatics of questions in seven
diverse languages, 63–112. Amsterdam: John Benjamins.
Hewitt, George. 2005. Georgian: a learner’s grammar. London: Routledge.
Holmberg, Andres. 2011. Verb second. In Tibor Kiss & Artemis Alexiadou (eds.). Syntax: an
international handbook of contemporary syntactic research. Berlin: Walter de Gruyter.
40
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
Hualde, José I. & Ortiz de Urbana. 2003. A grammar of Basque. Berlin & New York: Mouton de
Gruyter.
Huddleston, Rodney. 1993. Remarks on the construction You won’t believe who ED has married.
Lingua 91. 175–184.
Kalinina, Elena. 2011. Exclamative clauses in the languages of the North Caucasus and the
problem of finiteness. In Gilles Authier & Timur Maisak (eds.), Tense, aspect, modality and
finiteness in East-Caucasian languages, 161–201. Bochum: Brockmeyer.
Kautzsch, Emil. (ed.). 1910. Gesenius Hebrew grammar. Oxford: Clarendon Press.
Kiparsky, Paul & Carol Kiparsky. 1970. Fact. In Manfred Bierwisch & Karl Heidolph (eds.),
Progress in linguistics, 143–173. The Hague: Mouton.
Koenig, Ekkehard & Peter Siemund. 2007. Speech act distinctions in grammar. In Timothy
Shopen (ed.), Language typology and syntactic description, volume 1, 276–324. Cambridge:
Cambridge University Press.
Koenig, Ekkehard & Peter Siemund. 2013. Satztyp und Typologie. In Jörg Meibauer, Markus
Steinbach und Hans Altmann (eds.) Satztypen des Deutschen, 846–873. Berlin: Mouton de
Gruyter.
Kovtunova, Irina I. 1976. Russkij sintaksis: porjadok slov i aktual’noe chlenenie predlozhenija
[Russian syntax: word order and information structure of a sentence]. Moscow: Prosveshchenie.
Krylova, Olga A. & Serafima A. Khavronina. 1976. Porjadok slov v russkom jazyke [Word order
in Russian]. Moscow: Russkij jazyk.
Kwon, Kyong. 2011. What for diachronically. In Alexander Podobryaev et al. (eds.), Proceedings
of Formal Approaches to Slavic Languages Conference 20. Ann Arbor, MI: Michigan University
Press.
Lahiri, Utpal. 2000. Lexical selection and quantificatonal variability in embedded interrogatives.
Linguistics and Philosophy 23. 325–389.
Leu, Thomas. 2007. A note on what for split. In: Lisa Levinson and Oana Savescu-Ciucivara (eds.),
NYU Working Papers in Linguistics, volume 1. New York: New York University.
Lewis, M. Paul, Gary F. Simons, and Charles D. Fennig (eds.). 2015. Ethnologue: Languages of
the World, 18th edition. Dallas, TX: SIL International.
Lipták, Aniko. 2001. On the syntax of wh-items in Hungarian. Utrecht: LOT dissertation.
Lipták, Aniko. 2006. Word order in Hungarian exclamatives. Acta Linguistica Hungarica 53 (4).
343–391.
Ljutikova, Ekaterina & Sergey Tatevosov. 2009. The clause internal left edge: exploring the
preverbal position in Ossetian. Paper presented at the Third International Conference on Iranian
Linguistics, Paris III, Sorbonne Nouvelle, 11–13 September.
Marandin, Jean-Marie. 2008. The exclamative clause type in French. In Stefan Müller (ed.),
Proceedings of the HPSG08 Conference, 436–456. CSLI Publications.
Mathiassen, Terje. 1996. A short grammar of Lithuanian. Bloomington, IN: Slavica publishers,
Inc.
Michaelis, Laura. 2001. Exclamative constructions. In Martin Haspelmath, Ekkehard Koenig, Wulf
Oesterreicher, Wolfgang Raible (eds.). Language typology and language universals. An
international handbook, volume 2, 1038–1050. Berlin & New York: Walter de Gruyter.
Nouwen. Rick and Anna Chernilovskaya. 2015. Two types of exclamatives. Linguistic Variation
15(2). 201–224.
Obenauer, Hans-George. 1994. Aspects de la syntaxe A-barre: Effets d’intervention et mouvements
des quantifieurs. University of Paris 8 dissertation.
41
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
Olbertz, Hella. 2009. Mirativity and exclamatives in functional discourse grammar: evidence from
Spanish. In Evelien Keizer and Gerry Wanders (eds.), Web-papers in functional discourse grammar,
66–82.
Ono, Hajime. 2002. Exclamatory sentences in Japanese: a preliminary study. In Yukio Otsu (ed.),
Proceedings of the Third Tokyo Conference on Psycholinguistics, 305–326. Tokyo: Hituzi Syobo
Publ.
Ono, Hajime. 2006. An investigation of exclamatives in English and Japanese: syntax and sentence
processing. University of Maryland dissertation.
Primus, Beatrice. 2001. Word order typology. In Martin Haspelmath, Ekkehard Koenig, Wulf
Oesterreicher, and Wolfgang Raible (eds.), Language typology and language universals. An
international handbook, volume 2, 855–872. Berlin & New York: Walter de Gruyter.
Quirk, Randolph, Stanly Greenbaum, Geoffrey Leech & Jan Svartvik. 1985. A comprehensive
grammar of the English language. London: Longman.
Radford, Andrew. 2006. Minimalist syntax revisited.
http://faculty.ksu.edu.sa/dobaian/Courses/English%20427/A%20Minimalist%20Syntax%20Cour
sebook/A%20Minimalist%20Syntax%20coursebook%202006.pdf (accessed 22 February 2015)
Ramstedt, Gustaf. 1939 [1968]. A Korean grammar, 2nd edn. Oosterhout, The Netherlands:
Anthropological Publications.
Repp, Sophie. 2013. D-linking vs. degrees: Inflected and uninflected welch in exclamatives and
rhetorical questions. Interfaces of Morphology, 59–90. Akademie, Berlin.
Rett, Jessica. 2008a. Degree modification in natural language. University of Rutgers dissertation.
Rett, Jessica. 2008b. A degree account of exclamatives. In Tova Friedman and Satoshi Ito (eds.),
Proceedings of the 18th Semantic and Linguistic Theory Conference, 601–618. Fort Washington,
PA: CLC publications.
Rett, Jessica. 2011. Exclamatives, degrees and speech acts. Linguistics and Philosophy 34(5). 411–
442.
Rett, Jessica and Sarah Murray. 2013. A semantic account of mirative evidentials. Proceedings of
SALT 23, 453 – 472.
Rudin, Catherine. 1986. Aspects of Bulgarian syntax: complementisers and wh-constructions.
Columbus, OH: Slavica publishers Inc.
Sæbø, Kjell. 2006. Explaining clausal exclamatives. 3ieme Journée de Sémantique et
Modélisation. ENS, Paris.
Sæbø, Kjell. 2010. On the semantics of “embedded exclamatives”. Studia Linguistica 64(1). 116–
140.
Sharvit, Yael. 1999. Connectivity in specificational sentences. Natural language and linguistic
theory 7(3). 299–339.
Shlonsky, Ur. 1988. Complementiser-cliticization in Hebrew and the empty category principle.
Natural Language and Linguistic Theory 6 (2). 191–205.
Shvedova, Natalia Y. (ed.). 1980. Russkaja grammatika [Russian grammar]. Moscow: Nauka.
Siemund, Peter. 2001. Interrogative constructions. In Martin Haspelmath, Ekkehard Koenig, Wulf
Oesterreicher, Wolfgang Raible (eds.). Language typology and language universals. An
international handbook, volume 2, 1010–1028. Berlin, New York: Walter de Gruyter.
Siemund, Peter. 2015. Exclamative clauses in English and their relevance for theories of clause
types. Studies in Language 39(3). 698–728.
42
1
2
3
4
5
6
7
8
9
10
11
12
Sung, Li-May (2015). Why exclamatives in Budai Rukai. In Elisabeth Zaitoun, Stacy Teng and
Joy Wu (eds.), New Advances in Formosan Linguistics, 291–312.
Villalba, Xavier. 2003. An exceptional exclamative sentence type in Romance. Lingua 113. 713–
745.
Villalba, Xavier. 2008. Exclamatives: a thematic guide with many questions and few answers.
Catalan Journal of Linguistics: Exclamatives at the Interface 7. 9–40.
Yamato, Naouyki. 2010. The left periphery of Japanese exclamatives. Studia Linguistica 64(1).
55–80.
Zevakhina, Natalia. 2013. Syntactic strategies of exclamatives. The Journal of Estonian and
Finno-Ugric Linguistics 4 (2). 157–178.
Zanuttini, Raffaella and Paul Portner. 2003. Exclamative clauses: at the syntax-semantics
interface. Language 79 (1). 39–81.