It Cleft PDF

The English It-Cleft
Topics in English Linguistics

79
Editors
Elizabeth Closs Traugott

Bernd Kortmann
De Gruyter Mouton
The English It-Cleft

A Constructional Account
and a Diachronic Investigation
by
Amanda L. Patten
De Gruyter Mouton
ISBN 978-3-11-027780-7
e-ISBN 978-3-11-027952-8
ISSN 1434-3452
Library of Congress Cataloging-in-Publication Data
A CIP catalog record for this book has been applied for at the Library of Congress.
Bibliographic information published by the Deutsche Nationalbibliothek
The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie;
detailed bibliographic data are available in the Internet at http://dnb.dnb.de.
2012 Walter de Gruyter GmbH, 10785 Berlin/Boston
Cover image: Brian Stablyk/Photographers Choice RF/Getty Images
Printing: Hubert & Co. GmbH & Co. KG, Gttingen
Printed on acid-free paper
Printed in Germany
www.degruyter.com
Acknowledgements
This book is a revised version of my 2010 University of Edinburgh dissertation Cleft Sentences, Construction Grammar and Grammaticalization.
I would first like to thank my PhD supervisor, Nikolas Gisborne, and the
examiners of my thesis, Geoff Pullum and Kersti Brjars. I am immensely
grateful to the AHRC for funding my postgraduate studies. Without this
financial support, I could not have continued in education.
Among those who have discussed my work with me, I would especially
like to thank Graeme Trousdale and Regina Weinert. I am also grateful to
Roger Higgins for taking the time to correspond with me early on in my
studies and for encouraging me in my work on clefts. I must also thank
Catherine Ball, Nancy Hedberg, Bettelou Los and Javier Prez-Guerra for
helping me to access unpublished or not widely distributed works, as well
as Susan Pintzuk, Ann Taylor and Anthony Warner for introducing me to
the English historical parsed corpora series and for their guidance in its use.
Thanks also to Linda van Bergen and Meg Liang for help with translating
some of the historical data.
Above all, I thank Elizabeth Traugott for showing interest in my work
and for her guidance throughout the writing of this book. I am very grateful
for her encouragement and for her valuable comments on my written work
at various stages in the process. Many thanks also to Kirsten Brgen, Julie
Miess and Angelika Hermann at De Gruyter Mouton for help in producing
the finished piece.
ALP
Newcastle, 2012
Contents
Acknowledgements. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
vii
Chapter 1 Introduction and background. . . . . . . . . . . . . . . . . . . . . .

1. An outline of the project. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2. An overview of the literature on cleft sentences . . . . . . . . . . . . . .
2.1. The expletive approach. . . . . . . . . . . . . . . . . . . . . . . . . . .
2.2. The extraposition approach. . . . . . . . . . . . . . . . . . . . . . . .
3. A constructional approach to it-clefts. . . . . . . . . . . . . . . . . . . . . . .
4. A diachronic approach to it-clefts . . . . . . . . . . . . . . . . . . . . . . . . .
5. Methodology. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1
1
5
5
7
9
12
14
Chapter 2 A model of language structure and language change. . . .

1. Some basic assumptions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2. A constructional model of language structure. . . . . . . . . . . . . . . .
3. A constructional model of language change . . . . . . . . . . . . . . . . .
4. The application to it-clefts and copular constructions . . . . . . . . . .
16
16
17
21
25
Chapter 3 Specificational copular constructions. . . . . . . . . . . . . . . .

1. Different and competing analyses . . . . . . . . . . . . . . . . . . . . . . . . .
1.1. The equative approach . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.2. The inverse approach . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.3. A less formal approach . . . . . . . . . . . . . . . . . . . . . . . . . . .
2. Specification as (the inverse of) nominal predication . . . . . . . . . .
2.1. Specification and definite NP predicates . . . . . . . . . . . . .
2.2. Specification and inversion. . . . . . . . . . . . . . . . . . . . . . .
2.3. Capturing this account in cognitive and constructional
frameworks. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3. Accounting for the behaviour of indefinite NPs. . . . . . . . . . . . . . .
3.1. Specification and indefinite NP predicates. . . . . . . . . . . .
3.2. An account based on discourse requirements. . . . . . . . . .
3.3. An account based on definiteness. . . . . . . . . . . . . . . . . . .
4. Summarizing and extending the account. . . . . . . . . . . . . . . . . . . .
4.1. An overview of specificational NP be NP sentences . . . .
4.2. Positioning this account in relation to the literature . . . . .
27
27
28
30
32
34
34
38
42
47
48
49
51
56
56
58
viii
Contents
4.3. Other specificational copular constructions . . . . . . . . . . .

4.3.1. Th-clefts as specificational copular sentences. . .
4.3.2. Wh-clefts as specificational copular sentences . .
4.3.3. All-clefts as specificational copular sentences. . .
4.3.4. A family of specificational copular sentences. . .
62
63
64
68
69
Chapter 4 It-clefts as specificational copular sentences. . . . . . . . . .

1. The English it-cleft. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.1. A discontinuous constituent account of it-clefts . . . . . .
1.2. Explaining the it-clefts pragmatic properties. . . . . . . . . .
1.2.1. Focus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.2.2. Presupposition . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.2.3. Exhaustiveness. . . . . . . . . . . . . . . . . . . . . . . . . . .
1.2.4. Contrast . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.3. Explaining the it-clefts structural properties . . . . . . . . . .
1.3.1. The behaviour of the cleft clause. . . . . . . . . . . . .
1.3.2. The evidence for VP constituency. . . . . . . . . . . .
1.3.3. The evidence from agreement . . . . . . . . . . . . . . .
1.4. Interim summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2. A comparison with expletive accounts of it-clefts. . . . . . . . . . . . .
3. A comparison with other extraposition accounts of it-clefts . . . . .
3.1. The early extraposition accounts of the 1970s . . . . . . . . .
3.2. The more recent discontinuous constituent accounts . . . .
3.3. A different extraposition account? . . . . . . . . . . . . . . . . . .
4. A comparison with other constructional accounts of it-clefts . . . .
71
71
72
78
79
80
82
85
87
88
93
96
101
102
107
107
110
114
115
Chapter 5 Other varieties of it-cleft . . . . . . . . . . . . . . . . . . . . . . . . .

1. Beyond the archetypal it-cleft . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2. Predicational (and proverbial) it-clefts. . . . . . . . . . . . . . . . . . . . . .
2.1. An expletive approach to predicational it-clefts. . . . . . . .
2.2. Predicational it-clefts and the inverse approach . . . . . . . .
2.3. Predicational it-clefts and the equative approach . . . . . . .
3. It-clefts with non-nominal foci. . . . . . . . . . . . . . . . . . . . . . . . . . . .
4. Informative-presupposition (IP) it-clefts . . . . . . . . . . . . . . . . . . . .
5. Summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
120
120
122
123
127
131
134
140
146
Chapter 6 The it-cleft and earlier periods of English . . . . . . . . . . . .

1. Beyond the present-day language system. . . . . . . . . . . . . . . . . . . .
2. The early history of the English it-cleft . . . . . . . . . . . . . . . . . . . . .
3. A restrictively modified pronoun? . . . . . . . . . . . . . . . . . . . . . . . . .
148
148
149
152
Contents
ix
4. An obligatorily extraposed relative clause?. . . . . . . . . . . . . . . . . .

5. An unusual pattern of agreement? . . . . . . . . . . . . . . . . . . . . . . . . .
6. The evidence from Old English gender agreement . . . . . . . . . . . .
7. The it-cleft as a relic from an earlier time . . . . . . . . . . . . . . . . . . .
8. Summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
157
162
170
177
183
Chapter 7 The it-clefts development over time. . . . . . . . . . . . . . . .

1. A diachronic investigation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2. The corpora, the search and the selection process . . . . . . . . . . . . .
2.1. OE presentational/impersonal sentences. . . . . . . . . . . . . .
2.2. Existential sentences with it . . . . . . . . . . . . . . . . . . . . . . .
2.3. The pattern I it am. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.4. Other constructions mistaken for clefts. . . . . . . . . . . . . . .
2.5. Interim summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3. Frequency information. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4. Changes to the clefted constituent . . . . . . . . . . . . . . . . . . . . . . . . .
5. Changes to the cleft clause. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6. Summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
184
184
184
186
190
191
192
193
194
196
204
211
Chapter 8 The it-cleft and constructional change. . . . . . . . . . . . . . .

1. The two kinds of constructional change. . . . . . . . . . . . . . . . . . . . .
2. A grammatical constructionalization account . . . . . . . . . . . . . . . .
3. Some alternative explanations . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.1. An impersonal account . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.2. A Celtic account . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.3. A word order account . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.4. Interim summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4. Why do it-clefts undergo a construction-specific development?. .
4.1. Why do it-clefts develop a construction-specific range of
foci?. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.2. Why do it-clefts develop construction-specific discourse
functions? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.3. Summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
212
212
213
218
219
221
222
224
225
Chapter 9 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
243
Corpora and data sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
248
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
249
Index. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
266
225
232
241
Chapter 1
Introduction and background
1.
An outline of the project
It-clefts are interesting for a number of reasons. For one thing, they have a
non-standard structure which appears not to conform to the general rules of
the language. If we take a look at the example in (1), we can see that itclefts have four main components: the introductory pronominal it, a form of
the copular verb be, a postcopular phrasal element and a sentence-final
clause.
(1)
[It] [was] [Frank] [that complained]
From this example, we can see that (for the linguist) the it-clefts syntactic
configuration is difficult to make sense of. The that-clause is structured
internally like a restrictive relative: it contains a gap (in subject position)
which corresponds to the constituent that precedes the clause. However,
proper names, such as Frank, are full noun phrases. As such, they cannot
normally be modified by restrictive relative clauses. So how does this
clause relate to the postcopular element, if at all? Can we really call this a
restrictive relative clause? If so, what does it modify? If not, are we dealing
with a clausal structure that is unique to the it-cleft? Equally problematic is
the role of initial it. Is this an expletive dummy subject and if so, why is it
there? Does it operate as a syntactic placemarker and if so, for which element? Or is the constituent it related in a different way to other elements in
the sentence?
In addition, it-clefts have a number of unusual semantic, pragmatic and
discourse-functional properties. These are particularly interesting since it is
not immediately clear which elements in the cleft structure contribute to the
meaning of the construction. For example, the it-cleft is a focusing construction. The primary informational content is placed in the syntactically
marked postcopular focal position and is often given primary stress, see
(2).1 However, it is not at all obvious why this particular syntactic configuration should be chosen as a focusing device. Is this its primary function?
(2)
It was FRANK that complained
A further property of the it-cleft is that it exhibits exhaustiveness effects.

For example, in (2) we assume that Frank was the only person who complained on this occasion. They are also presuppositional; the information in
the sentence-final clause is not asserted and is preserved under negation.
For instance, in example (3), we are told that Frank didnt complain but we
are left with the presupposition that somebody (else) did complain. This
begs the question, where do these pragmatic meanings come from? Which
elements contribute to them?
(3)
It wasnt Frank that complained
Cleft sentences also have a specificational (or identifying) meaning. For

some authors, specificational meaning involves a value-variable relationship (see Higgins 1979; Declerck 1988). For example, (2) identifies Frank
as the value for the variable described by the sentence-final clause, x
complained. However, for others, specificational meaning is attributed to a
special use of the copular verb. So does be have a specificational meaning
in the it-cleft? If not, where does the specificational meaning of it-clefts
come from?
To a large extent, how these questions are answered (and perhaps
whether they are even asked at all) depends on how we think it-clefts relate
to other constructions, or configurations, in the language. Most approaches
to it-clefts fall into two broad categories: those that understand it-clefts in
relation to simple subject-predicate sentences, such as (4), and those that
relate it-clefts to other specificational copular sentences, such as (5).
(4)
Frank complained
(5)
The one that complained was Frank
1. I use the term focus to refer to a unit of information structure where the assertion differs from the presupposition (see Chapter 4, Section 1.2.1). Although focus is often marked by intonation, Lambrecht (1994: 208) observes that accent
placement and focus marking are not to be equated. Where focus marking is
unclear or ambiguous in my examples, I make use of small capitals to indicate
the marking of focus by intonation.
An outline of the project
I discuss these two approaches and the analyses that result from them in
Section 2. Essentially, justification for the first approach comes from the
truth-conditional equivalence between it-clefts and simple subject-predicate
sentences. From this perspective, it-clefts are viewed primarily as a means
of marking focus syntactically. The second approach, on the other hand,
builds on the fact that the it-cleft is a copular construction with a specificational meaning. So which is the better approach? The answer to this question depends upon what we think is the primary function of it-clefts and
asking which perspective can best explain the range of properties that itclefts display.
There are also different varieties of it-cleft which are sometimes
regarded as separate structures. One domain of variability involves the
category of elements that can occur in the postcopular (focal) position.
Although, most frequently, the focal element is a noun phrase, it-clefts
allow a range of other constituents to occur as the complement of be, such
as the prepositional phrase in (6). So should these examples be analysed in
the same way as those with nominal foci or do they require a separate
analysis? Can the sentence-final clause still be analysed as a restrictive
relative if the immediate antecedent is not nominal?
(6)
Its in October that hes leaving
Another domain of variation relates to the information status of the

sentence-final clause. In it-clefts, the clausal component is typically associated with expressing discourse-old information, as in (7). In this example,
we know from the prior discourse that a letter has been written and so the
open proposition x wrote it is given information. However, in other cases,
such as (8), the information expressed by the sentence-final clause is not
given by the previous discourse and the proposition, that someone once
said Laws are silent at times of war, does not even have to be known (or
familiar) to the intended audience. Do these examples represent two different types of it-cleft? And if so, how are they related? Is one more basic, or
prototypical, than the other?
(7)
A: Did Max write the letter?

B: No. It was Walter who wrote it.
(8)
(Start of lecture)
It was Cicero who once said, Laws are silent at times of war
A further domain of variation concerns the relationship between specificational and predicational tokens. It-clefts are usually identified as having a
specificational meaning (see above). However, superficially similar proverbial sentences, such as (9), have a predicational meaning. For instance, (9)
is most closely paraphrased by the predicational copular sentence in (10).
In both sentences, the postcopular element describes, rather than identifies,
the referent (the road that has no turning) as long.
(9)
It is a long road that has no turning
(10)
The road that has no turning is a long one
How does this structure relate to the specificational it-cleft, if at all? Can
proverbial sentences, such as (9), really be called clefts? It-clefts also seem
to resemble extraposed sentences, such as (12). For instance, on the surface,
the it-cleft in (11) differs only in that it contains a gap in the sentence-final
clause. Do it-clefts share more than just an apparent likeness with this
extraposed structure?
(11)
It was the Colonel [that __ survived]
(12)
It is a miracle [that he survived]
In this book, I provide answers to these questions by examining it-clefts

within the framework of construction grammar (cf. Croft 2001; Goldberg
1995, 2006; Lakoff 1987). Construction grammar was developed with a
view to providing full and explanatory accounts of specialized linguistic
patterns as well as broad generalizations. On this model, constructions are
not considered the epiphenomenal byproducts of a combination of componential meaning and highly general rules. Instead, aspects of form and
meaning can be encoded by the construction itself. Since much of the itclefts structure and use cannot be predicted from general patterns of correspondence, it is well-suited to treatment within a constructional approach. I
come back to this issue in Section 3.
In addition to providing a synchronic account of the English it-cleft, I
also examine the constructions diachronic development. Relevant questions here include: What is the origin of the it-cleft construction? How has
the it-cleft (and its relationship to other constructions) changed over time?
How did the different varieties of it-cleft emerge? This study contributes to
the recent literature on diachronic construction grammar. As a complex and
specialized linguistic pattern, the English it-cleft provides an example of
An overview of the literature on cleft sentences
how larger (multi-word) constructions undergo change. I discuss this aspect

of my investigation in Section 4.
In this section, I have provided a very brief outline of the different types
of phenomena discussed and the kinds of questions addressed throughout
this book. In the next section, I provide an introductory background into the
literature on cleft sentences. Section 3 asks why construction grammar is
helpful in the treatment of it-clefts. Here, I present an overview of the
particular kind of cleft analysis argued for in this book and I compare it to
other constructional accounts proposed in the literature. The historical facts
of the English it-cleft, along with my diachronic construction grammar account, are sketched in brief in Section 4. I explain how I use this diachronic
evidence to both support and inform my synchronic analysis. The methodology employed in this study is discussed in Section 5.
2.
As I noted in Section 1, authors tend to view it-clefts either from the

perspective of their relationship to truth-conditionally equivalent subjectpredicate sentences or from the perspective of their relationship to other
specificational copular constructions. In this section, I outline these two
approaches and the analyses that result from them. Although the individual
proposals differ, these opposing viewpoints lead to two different kinds of
analysis: those that treat the postcopular phrase as the preposed argument of
the proposition expressed in the sentence-final clause and those that consider the sentence-final clause to be associated in some way with the initial
element it. The purpose of this section is not to provide an exhaustive and
comprehensive review of the literature, but simply to highlight intellectual
trends in the history of the analysis of the construction.
2.1.
The expletive approach
For many authors, it-clefts are considered primarily as information structure variants of syntactically more basic sentences. From this perspective,
it-clefts do not differ dramatically in their semantic content from canonical
subject-predicate sentences, but are marked by the way that this informational content is presented (Ward, Birner, and Huddleston 2002). Unlike
their canonical counterparts, it-clefts have a fixed information structure: the
information that is to be foregrounded is placed in the postcopular position
while the remaining semantic content is backgrounded into a sentence-final

clause, as shown in (13).
(13)
It was [[Frank]i [that ____i complained]]
(14)
Frank complained
[it-cleft]
[canonical]
The analyses resulting from this approach assume that the focal element in
it-clefts enters into a predication relationship with the information in the
sentence-final clause; this accounts for their truth-conditional equivalence
with simple subject-predicate sentences. From this, it follows that the initial
element it and (in most accounts) the copular verb be are semantically
empty, serving only to introduce, or foreground the postcopular element.
As a result, in the cleft literature, these analyses are referred to cumulatively as the expletive approach; common to all such accounts is the
assumption that the initial pronoun it does not play an essential role in the
interpretation of the sentence.
An early example of an expletive account is detailed by Jespersen
(1937: 8389). He suggests that it-clefts are syntactically identical to their
noncopular counterparts except for the addition of a lesser subject and
verb and a connective word. So, for instance, the elements it, be and that
in (15) are semantically empty, with Frank and complained entering into a
predication relationship. The example is formalized using Jespersens notation.2
(15)
[It was] Frank [that] complained

[sv]
S [3c]
V
According to Jespersen, this analysis explains why it-clefts are used as a

means of marking focus syntactically. He notes, A cleaving of a sentence
by means of it isserves to single out one particular element of the
sentence and very often, by directing attention to it and bringing it, as it
were, into focus, to mark a contrast (Jespersen 1949: 147).
A number of similar analyses were developed within the generative
tradition of the 1980s. The details of these analyses differ. For example,
Rochemont (1986) suggests that the postcopular element is situated within
2. The use of square brackets in Jespersens (1937: 86) formalism indicates that
this information is extraposed relative to the sentence proper. Lower case s
and v indicate the lesser (i.e. expletive) subject and verb and 3 c represents a
tertiary connective.
the sentence-final clause at deep structure, while for Williams (1980) and
Heggie (1988) these components are coindexed at the level of surface
structure by a predication rule. Chomsky (1977) claims that the it-cleft is a
type of topicalization construction involving wh-movement. The relationship between the postcopular element and the sentence-final clause is therefore akin to that between topic and comment. For Delahunty (1982, 1984)
on the other hand, these components enter into a predication relation at the
level of logical form. Delahunty converts the sentence-final clause into a
function (via lambda abstraction) which takes the postcopular element as its
argument; after a reduction operation, the Logical Structure of cleft sentences is equivalent to that of their noncopular counterparts.
Although they use different mechanisms to achieve it, these authors
assume that it-clefts and truth-conditionally synonymous sentences share a
level of representation. Common to all of these accounts then is the treatment of the initial pronoun it as an expletive element, the analysis of the
sentence-final clause as being in some way related to the postcopular
element, and the understanding that the primary function of it-clefts is as a
focusing device.
2.2.
The extraposition approach
For others, it-clefts are considered foremost as specificational copular

sentences. From this perspective, it-clefts are analysed in relation to corresponding pseudocleft sentences and sometimes to noncleft copular constructions with an identifying function. The term pseudocleft is commonly
used to encompass both wh-clefts, which are introduced by wh-words, and
th-clefts, which are introduced by the definite article and one of a small
number of semantically general head nouns such as the one or the thing.3
(16)
Its grape soda that I like best
[it-cleft]
(17)
What I like best is grape soda
[wh-cleft]
(18)
The thing that I like best is grape soda
(19)
My favourite drink is grape soda
[th-cleft]
[noncleft NP be NP]
3. My use of the term th-cleft is from Collins (1991a, 1991b). This term is used in
a different sense in Ball (1977) and Hedberg (1990, 2000) to refer to cleft sentences introduced by demonstratives, such as this or that.
Each of these examples has the function of identifying (or specifying) the
postcopular element grape soda as matching a certain description. Like the
it-cleft in (16), pseudoclefts also contain clausal elements. However, in (17)
and (18) these clauses are in subject position. This suggests that the it-cleft
is an extraposition construction: the sentence-final clause is not connected
to the focal element; instead, it is related in some way to the initial it.
Again, an early example of such an approach is provided by Jespersen
(1927). Prior to his (1937) account, outlined above, he proposed a transposition analysis of it-clefts. In the following passage, Jespersen suggests
that it-clefts are paraphrased most closely by other specificational copular
constructions. Here the sentence-final clause is analysed as a restrictive
relative, modifying the constituent it. He notes,
it is not really the antecedent (or what looks like the antecedent) that is
restricted by a relative clause. When we say it is the wife that decides or
it was the Colonel I was looking for what we mean is really the wife is
the deciding person and the Colonel was the man I was looking for: the
relative clause thus might be said to belong rather to it than to the
predicative following after it is
(Jespersen 1927: 88)
Other accounts that view it-clefts in relation to specificational copular

sentences can be found in the transformational analyses of the 1970s, which
derive it-clefts from pseudoclefts (see Akmajian 1970; Gundel 1977), or
from the same source as pseudoclefts (see Wirth 1978). For these authors,
the clausal element in subject position is extraposed, leading to the manifestation of it as either a placemarker or a pronominal copy. Again, the
details of these analyses differ. For example, Akmajian and Wirth derive itclefts via extraposition rules that are particular to cleft sentences, whereas
Gundel suggests that this process is an instance of ordinary right-dislocation. For Gundel, the initial it is a pronominal copy of the right-dislocated
clause, whereas for Akmajian, it seems to be an expletive element.
Bolinger (1972) takes an approach that is more in line with Jespersens
(1927) original proposal, in which the relative clause restrictively modifies
the constituent it. He suggests that analytic compound relatives, as in (20),
provide an ideal source for it-clefts (Bolinger 1972: 110). Such sentences
can undergo what he labels inversion (extraposition-from-NP), whereby
the restrictive clause is extraposed but the nominal head remains in situ, as
in (21). For it-clefts however, this inversion is obligatory rather than
optional. On this analysis then, the constituent it is neither a placemarker,
nor a pronominal copy, but a restrictively modified pronoun.
A constructional approach to it-clefts
(20)
That which he stole was money
[analytic compound]
(21)
That was money which (that) he stole
[inverted compound]
(22)
It was money which (that) he stole

[it-cleft]
(examples from Bolinger 1972: 109)
Consequently, for these authors, there is little consistency as to the exact

role of it or how the relationship between it-clefts and other specificational
sentences works. Nevertheless, what these extraposition analyses share is a
concern for recognizing it-clefts primarily as specificational copular sentences.
3.
So which of these approaches is the better one? From the perspective of

construction grammar, the choice is straightforward. In this section, I explain why construction grammar is a useful framework for representing and
accounting for the unusual properties of it-clefts and why an approach that
examines clefts in relation to specificational sentences is more compatible
with the principles of construction grammar. I go on to provide an overview
of my account of it-clefts before showing how this improves on the previous constructional analyses proposed in the literature.
In construction grammar, larger linguistic patterns are represented as
symbolic pairings of form and meaning, much like individual lexical items.
These complex constructions are made up of smaller units, which are also
form-meaning pairs. However, since these correspondences are internal to
the larger construction, such compositional meanings may nevertheless be
construction-specific. Furthermore, on this model, constructions can sometimes encode meanings which are not compositional; that is, which cannot
be attributed to its individual components. In recognizing the construction
as a theoretical symbolic object, construction grammar therefore anticipates
the existence of (and can represent) idiosyncratic grammatical information
which cannot be predicted on the basis of highly general grammatical rules.
Within this framework, all aspects of form and meaning (including aspects
of use) are listed inside the construction and so form part of the speakers
grammatical knowledge. The theory of construction grammar is therefore
ideally suited to the task of providing full and comprehensive accounts of
the properties of more specialized linguistic patterns, such as the English itcleft construction.
10
As Goldberg (2003: 120121) notes, the explanatory power of constructional accounts comes from the requirement that each construction must be
motivated; that is, there must be some reason as to why this particular
construction should exist in the language. For the most part, the motivation
for a construction comes from within the grammar. On a usage-based constructional theory, a speakers grammatical knowledge is represented as a
network of constructions (form-meaning pairs). Specialized linguistic patterns inherit properties from more general patterns. The more properties a
construction inherits, the more it can be said to be motivated by the
language system. Constructions that are related to one another are shown to
inherit properties from the same general pattern, forming a family of
constructions. A more detailed introduction to construction grammar is
provided in Chapter 2.
The way that grammatical knowledge is organized in construction grammar suggests that analyses of it-clefts which are based on their relationship
to structurally simple noncopular sentences will result in a less satisfactory
account than an approach which views it-clefts in relation to other specificational copular sentences. On this model, inheritance links are posited
between constructions that are both formally and functionally related, with
an emphasis placed on similarities of surface form (Goldberg 2006: 23) and
aspects of meaning that go beyond truth-conditional synonymy (Goldberg
1995: 103). Goldberg (1995: 108) says, The intuition is that the existence
of a given form with a particular meaning in no way motivates the existence of a different form with a closely related meaning. Therefore, while
noncopular subject-predicate sentences can often be used to paraphrase itclefts, their truth-conditional synonymy is not necessarily expressed (as
closely) in the grammatical system. As we might expect, accounts which
view it-clefts in relation to structurally less complex sentences leave a
number of questions unresolved: Why should focus be marked using this
particular structure? Why do it-clefts have so many semantically empty
elements? Where do the existential presuppositions and the property of
exhaustiveness come from?
I view the it-cleft foremost as a member of the family of specificational
copular constructions. It-clefts, wh-clefts, th-clefts, all-clefts and certain
noncleft copular sentences all inherit properties from a more general, schematic, specificational copular construction. But what is a specificational
copular construction? And where does specificational meaning come from?
The answers to these questions are not obvious and a number of different
analyses have been proposed in the literature. In order to understand the
larger schema, or category, of copular constructions, I examine the nature
11
of these sentences in Chapter 3. I argue that in specificational copular sentences, specificational meaning results from a class-membership predication relation associated with the concept of definiteness. I show that this
type of analysis is able to account for data which has eluded alternative
approaches to specificational sentences.
On this account, many of the it-clefts properties are shown to be motivated; that is, they are simply inherited from the more basic specificational
copular construction. Following the extraposition accounts of Jespersen
(1927) and Bolinger (1972), I analyse the sentence-final clause as a restrictive relative, modifying the initial it. In particular, I argue in Chapter 4 that
it and the relative clause together operate like a discontinuous definite
description (see also Hedberg 1990, 2000; Percus 1997; Han and Hedberg
2008). Since definite descriptions exhibit existential presuppositions and
are associated with exhaustiveness (or inclusiveness), this analysis explains
why these properties are found in it-clefts too. It also reduces the number of
semantically dummy elements. For instance, on this account, the initial it
is not expletive and is instead shown to perform an important function.
Where my account advances the current literature is in providing a reason
as to why definite descriptions are a fundamental component of it-clefts
(and other specificational constructions). Furthermore, as I explain in Chapter 5, this particular analysis allows for a more straightforward account of
the relationship between specificational and predicational/proverbial itclefts.
Alternative constructional accounts of it-clefts have been put forward by
Lambrecht (2001) and Davidse (2000). However, while both authors extend
their accounts to other types of cleft sentence, neither makes use of a
system of inheritance. Consequently, they treat the it-cleft as a highly idiosyncratic construction. For example, Lambrecht (2001) views it-clefts in
relation to simple noncopular sentences, and so presents an expletive analysis. Thus, while his account is able to accommodate the it-clefts unusual
properties (through invoking the concept of the construction), it cannot
identify how they come about. Davidse (2000), on the other hand, analyses
the it-cleft as a highly complex structure involving two clauses (one of
which is unique to cleft constructions) which enter into different semantic
relationships with the postcopular element. Again, it is not clear how this
structure is motivated by the language system.
The constructional approach outlined here is therefore, in some ways, an
improvement on those of Lambrecht (2001) and Davidse (2000) since it
makes full use of the tools employed in construction grammar for making
generalizations. By examining it-clefts in relation to the taxonomy of
12
specificational sentences and exploiting an appropriate inheritance hierarchy, the motivation for this construction is maximized. Only after examining it-clefts in relation to the rest of the grammar are the exceptional or
truly construction-specific characteristics isolated. As Goldberg (2003:
118) observes, a given construction often shares a great deal with other
constructions that exist in a language; only certain aspects of its form and
function are unaccounted for by other constructions.
4.
A diachronic approach to it-clefts
We have seen then that construction grammar tolerates (although nevertheless seeks to limit) idiosyncrasies in the language system. However, ideally,
even exceptional properties should be provided with an explanation of
some sort. According to Goldberg (2003: 121), in such cases, motivation
can be provided by factors external to the grammar. In this section, I ask
whether historical evidence can provide motivation for, and so account for,
some of the construction-specific properties of the it-cleft.
From my synchronic analysis of the it-cleft as a type of specificational
copular construction, certain structural aspects remain a puzzle, such as the
modification of it by a restrictive relative clause and the extraposition of the
relative clause. By examining the it-clefts structural idiosyncrasies in relation to the language system of earlier periods of English, I show in Chapter
6 that although these properties are no longer motivated by the language
system, they are likely to have been inherited from formally related constructions existing at earlier periods of the language. In this way, the it-cleft
shows how the retention or entrenchment of once-motivated form-function
pairings can lead to construction-specific properties which are no longer
productive in other areas of the grammar.
For most types of it-cleft then, their seemingly idiosyncratic properties
become much less mysterious when examined in relation to the grammar of
earlier periods of English. However, there are subtypes of it-cleft which
exhibit properties that cannot be attributed to inheritance at any period of
the language. In Section 1, I introduced two varieties of it-cleft which are
sometimes treated as separate constructions from the it-cleft proper: those
with non-nominal foci and those with new information in the sentence-final
clause. The particular range of elements found to occur in the postcopular
position of the it-cleft is not shared by other specificational copular constructions. Likewise, the it-cleft seems to be the only kind of specificational
sentence to express brand-new information in the presuppositional clause
A diachronic approach to it-clefts
13
(see Prince 1978; Collins 1991a). As a result, the range of non-nominal foci
and the ability to express hearer-new information are properties which are
not inherited from the wider specificational construction (see Chapter 5).
This begs the question, where did these more idiosyncratic properties come
from?
In Chapter 7, I conduct a diachronic investigation, using data (from four
historical English parsed corpora) which spans from Old English to Modern
English. I find that the it-cleft occurs with an increasingly wide range of
foci and appears in a greater variety of discourse contexts over time. These
idiosyncratic properties of the it-cleft are therefore shown to be an outcome
of the constructions historical development. In Chapter 8, I ask how and
why the it-cleft construction has developed in this particular way. I interpret the changes to the function and use of the specificational it-cleft as an
example of schematization. On this account, novel instances are formed by
extension from the prototype, overriding inheritance from more basic
patterns. As these new types of instance become more conventional, the itcleft, in turn, becomes a more abstract and schematic construction. The itclefts construction-specific development is therefore shaped by its prototype, which differs in subtle ways from that of other specificational copular
constructions. I exemplify this with a short comparison of it-clefts and whclefts.
The historical evidence therefore demonstrates that the it-cleft was once
fully motivated by inheritance from the language system. Over time, the
construction has acquired a range of idiosyncratic properties via conventional pathways of change both fossilization and schematization. Therefore, while these construction-specific characteristics are not inherited from
more general patterns of correspondence, they are nevertheless shown to be
motivated by general principles of language change. A more comprehensive overview of usage-based approaches to constructional change is provided in Chapter 2.
With the inclusion of a substantial diachronic component, the present
study is able to contribute to the somewhat limited literature on the history
of the English it-cleft. Until very recently, Balls (1991, 1994a) work has
dominated the literature on this topic (see Filppula 2009; C. Johansson
2008; Los 2009; Los and Komen forthcoming; Patten 2010; Prez-Guerra
1999, forthcoming for more recent contributions). However, in Chapters 6,
7 and 8, I present a very different account of the specificational it-clefts
origin and subsequent diachronic development. The approach outlined here
is unique in that it reviews the historical data in light of an extraposition
(from-NP) account of it-clefts. As I explain, the historical evidence actually
14
provides considerable support for this particular synchronic approach.

Finally, the present study contributes to the recent literature on diachronic
construction grammar, in that it provides an example of how more complex
(multi-word and partially schematic) constructions undergo change.
5.
Methodology
This book examines the English it-cleft from both a synchronic and diachronic perspective. In the synchronic part, I rely largely on examples that
are either invented or taken from the literature. I have chosen to exemplify
my discussion in this way for several reasons. First, I am engaging with a
literature where the use of invented examples is common practice. Second,
since the issues surrounding the data are often complex, I have made an
effort to keep examples brief and to choose examples that highlight the
relevant features without requiring unnecessary explication. Furthermore,
in these chapters, the focus of my discussion is on the prototypical it-cleft
subtype. As a result, I am not always interested in detailing qualitative differences between individual instances.
However, where my concern is to highlight variation in the it-clefts
structure or to demonstrate aspects of use, I provide attested examples from
the British component of the International Corpus of English (ICE-GB),
extracted using the corpus utility program ICECUP 3.1. A search on the
CLEFTIT annotation produces an output of 430 hits within 422 text units.
From this data, I discounted incomplete tokens, tag questions and truncated
clefts (see Chapter 4, Section 1.1 for an analysis of such structures), as well
as a handful of instances which appear to have been mistakenly tagged as
it-clefts.4 This amounts to a data-set of 404 tokens, from which frequency
counts and proportions are measured. For more comprehensive quantitative
studies of present-day it-clefts in the ICE-GB see Gmez-Gonzlez (2004,
2007), Hasselgrd (2004), and Nelson (1997).
4. For example, in the utterance below, work is not specified as the thing she most
enjoys; instead, working full-time as a nursing auxiliary (referred to by it) is
described as enjoyable work. This is not an it-cleft, despite its mark-up in the
ICE-GB.
(i)
At the time of the accident she was thirty-nine years old, married, with
children, and working full-time as a nursing auxiliary at the Pembury
Hospital near Tunbridge Wells. It was work which she much enjoyed a
and to which she was fully committed. (S2A-062 008, Legal Presentation)
Methodology
15
The diachronic investigation makes use of data from four independent,

yet related, historical English corpora: the York-Toronto-Helsinki Corpus of
Old English Prose (YCOE), the Penn-Helsinki Parsed Corpus of Middle
English, second edition (PPCME2), the Penn-Helsinki Parsed Corpus of
Early Modern English (PPCEME) and the Penn Parsed Corpus of Modern
British English (PPCMBE). These four corpora form part of the same series
of syntactically annotated historical English corpora from the University of
Pennsylvania and the University of York. As a result, they all use the same
system of syntactic annotation and are accessed by the same search engine:
CorpusSearch2. Together, the corpora comprise over 5 million words of
running text and span the entire history of British English up until 1914.
This amounts to over 500 clear it-cleft tokens, dating from the mid-tenth to
the early-twentieth century. I discuss the method of extraction and selection
in detail in Chapter 7, Section 2.
The study undertaken here is corpus-based rather than corpus-driven;
that is, my cleft analysis informs my corpus investigation and, in turn, the
empirical evidence serves to support my constructional account of it-clefts.
Unlike in corpus-driven studies then, I approach the data with already established hypotheses. One reason for adopting this approach is that it-clefts
are notoriously difficult to identify and separate from superficially similar
but structurally distinct sentence-types, such as those containing extraposed
subject clauses (see Calude 2008a; Haugland 1993). Furthermore, since
there are different ways of analysing it-clefts, linguists may differ over
which examples count as clefts. It therefore seems preferable to have a
clear understanding of the criteria used to extract the relevant data, so that
at least the approach is consistent. It also means that I can address the issue
of how well my theory accounts for the data.5
5. See Tognini-Bonelli (2001) for an outline of the differences between corpusbased and corpus-driven approaches to corpus study. While I argue for a
corpus-based approach in this instance, Tognini-Bonelli (2001: Chapter 5) discusses the merits of a corpus-driven approach.
Chapter 2
A model of language structure and language change
1.
Some basic assumptions
This book examines the synchronic structure and diachronic development

of the English it-cleft within the framework of construction grammar. In
particular, I adopt the fundamental principles of usage-based theories of
construction grammar, and make use of concepts from Cognitive Grammar
(Langacker 1987, 1991), Cognitive Construction Grammar (Lakoff 1987;
Goldberg 1995, 2006) and Radical Construction Grammar (Croft 2001). In
this chapter, I outline some of the basic claims underlying the present study
and introduce some of the machinery that I make use of in my analysis of
cleft and copular sentences. In Section 2, I explain how grammatical knowledge is organized on a usage-based, constructional theory and discuss the
rationale behind this model of language structure. In Section 3, I show how,
together with usage-based assumptions, the conceptualization of language
as a hierarchical network makes a number of predictions regarding the
diachronic development of constructions which are compatible with wellattested pathways of change, such as grammaticalization. In Section 4, I go
on to provide a (very brief) indication of how I make use of these various
concepts in my analysis of the English it-cleft construction.
It should be noted that not all versions of construction grammar adopt
this same set of principles. The Construction Grammar of Kay and Fillmore
(1999), for example, is not usage-based. As a result, it differs from other
construction grammars in that it adopts a complete (rather than a default)
mode of inheritance, which licenses a non-redundant system of grammar.
As Goldberg (2006: 214, 216) notes, this version of construction grammar
has developed somewhat separately, in that it more closely resembles the
formalist theory Head-driven Phrase Structure Grammar (HPSG). In what
follows, I indicate where this version of construction grammar differs from
the other relevant theories and, as a result, from the fundamental assumptions of the present study.
A constructional model of language structure
2.
17
Construction grammar was developed with a view to providing full and

explanatory accounts of specialized linguistic patterns. Such configurations
are problematic for the componential model of grammatical knowledge that
underlies generative theories of grammar (see Croft 2007). On a componential model, each type of linguistic knowledge (syntax, semantics and so
on) makes up a separate component, with aspects of meaning and form
being mapped on to one another through general linking rules. The only
idiosyncratic and item-specific mappings between these components are
found in lexical items, which are both arbitrary and conventional. On this
model then, complex structures are built out of discrete, atomic elements in
accordance with the combinatorial rules specific to each component. This
reductionist perspective lends itself to the stronger hypothesis that constructions (more complex grammatical structures) are purely epiphenomenal. As Chomsky (2000: 8) says, grammatical constructions are taken
to be taxonomic artifacts, useful for informal description perhaps but with
no theoretical standing.
The componential model works well for regular syntactic expressions,
since, once we know the meanings typically associated with each of the
lexical items, we can determine the meaning of the construction as a whole.
However, the meaning and function of more specialized linguistic patterns
cannot be determined from the general rules of semantic interpretation for
their constituents as they exist outside of the particular construction; that is,
the meaning of specialized linguistic patterns is conventional in the sense
that it must be learned. Such structures are also sometimes noncompositional; that is, they express meanings which cannot be entirely broken
down into components and attributed to their individual formal elements.1
These, less regular expressions are therefore problematic for a strictly componential model of language structure, since aspects of their meaning
and/or use cannot be generated from the application of highly general linking rules.
Such phenomena represent strong evidence for the need to recognize the
construction as having an independent theoretical status. On this model of
grammar, constructions are symbolic pairings of form and meaning, much
like lexical items. The syntactic elements and semantic components that are
particular to each complex construction are related via symbolic links that
1. See Nunberg, Wasow, and Sag (1994) for a detailed discussion of the differences between the concepts of conventionality and noncompositionality.
18
are internal to that construction. This allows the representation of meanings

that are at once compositional (attributable to individual elements of form)
and conventional (construction-specific or irregular). In addition to housing
these smaller pairings of form and meaning, complex constructions contain
a further symbolic link which relates the entirety of the constructions form
to the constructions conventional meaning. This allows the representation
of noncompositional meanings, which are associated with the construction
as a whole, but cannot be attributed to any of its individual parts.
A model of grammar which recognizes the construction as a theoretical
object is therefore able to handle the existence of idiosyncratic grammatical
information. However, construction grammar takes this a step further, proposing that, in fact, all grammatical knowledge, including the predictable or
regular patterns of the language, should be represented as constructions. As
Fillmore, Kay, and OConnor (1988: 534) suggest, the machinery needed to
describe more peripheral constructions can be generalized to more familiar
structures.
Indeed, this follows directly from the usage-based assumptions of most
construction grammars. On a usage-based model, humans are not innately
programmed with linguistic knowledge; instead, all of language (and not
just the periphery) is learned inductively from the input, or rather, from the
speakers linguistic experiences. It follows from this that all grammatical
knowledge (both specialized linguistic patterns and broad generalizations)
should be given a uniform representation. Therefore, in usage-based constructional theories, the entire language system is made up of constructions
(form-meaning pairs). As long as a linguistic pattern is sufficiently frequent
that the speaker is likely to induce an abstract mental schema, then it may
be stored as a symbolic unit. Thus, on a usage-based framework, constructions are simply conventionalized chunks of linguistic knowledge (Goldberg 2006: 05).
These constructions form a structured inventory which makes up the
speakers knowledge of the language. This inventory is represented as a
taxonomic network of constructions with each construction constituting a
separate node (Croft and Cruse 2004: 262). The network is hierarchical,
showing that some constructions are more basic or general than others.
Lower-level constructions inherit attributes from higher-level constructions.
That is to say, more specialized and substantive (lexically filled) constructs
are instances of more general and schematic (lexically open) constructions.
Usage-based construction grammars adopt the default mode of inheritance.
On the complete inheritance model, an inheriting construction inherits all of
the information which is specific to a dominating construction. This means
19
that information only has to be represented once in the hierarchy: inherited

information is stored only in the dominating construction, at the highest
level possible. This amounts to a non-redundant system of linguistic knowledge.
It follows from the complete inheritance model that while a lower-level
construction may contain information which is not present in a dominating
construction, it may not contain information that conflicts with information
presented at a higher level, without resulting in ill-formedness (Goldberg
1995: 7374). Categories formed on the basis of these inheritance hierarchies are therefore classical; that is, a construction is a member of a larger
constructional category if and only if it inherits all of the grammatical
structure of the parent construction. It is this mode of inheritance that is
adopted in Kay and Fillmores (1999) Construction Grammar. On the
default inheritance model, however, conflict between the information specified in the inheriting construction and that specified in the dominating construction is permitted. This means that all of the attributes of a dominating
higher-level construction will be inherited by the lower-level construction
unless there is conflict. In this case, the more specific construction wins and
inheritance is limited to only non-conflicting information.2
Mismatch phenomena can therefore be accounted for easily, assuming a
default inheritance model. Mismatch describes mappings between form and
meaning which do not conform to more general patterns of correspondence.
Francis and Michaelis (2003) identify two different kinds of mismatch.
Complexity mismatch occurs when there is not a one-to-one relationship
between semantic components and formal elements. For instance, they note
that extraposition constructions, such as (1), involve an expletive it which
is present in syntax but semantically unspecified (Francis and Michaelis
2003: 4). Content mismatch, on the other hand, involves incongruous formmeaning mappings. Francis and Michaelis provide the example of predicate
nominals. They note that since noun phrases typically function as referring
expressions, their use in sentences like (2) involves a category mismatch, in
2. Hudsons Word Grammar (1990, 2007) also adopts a default inheritance model
of grammatical knowledge. Word Grammar shares much with constructional
theories in that it assumes that the syntax of language is comprised of a very
large number of constructions, each with its own peculiar interactions with
other constructions and with lexical items (Hudson 2007: 153). The difference
is that Word Grammar focuses on the (more or less specific types of) dependency relations between the words which make up these constructions (Hudson
2007: 156).
20
which the typical formal properties of one lexical category...are associated

with the typical semantic properties of another category (Francis and
Michaelis 2003: 5).
(1)
It is a miracle that he survived
(2)
John is a doctor
Since mapping between form and meaning is internal to the construction,

mismatch constructions are simply presented as containing information that
overrides inheritance from more general patterns. As Francis and Michaelis
(2003: 24) note, mismatch effects therefore provide evidence for symbolic
constructions and for inheritance hierarchies.
In the default mode of inheritance, information is stored redundantly. A
redundant system represents information not only on the highest possible
node, but at all levels in the hierarchy. Goldberg (1995: 98) argues in
favour of a redundant, or full entry, model since without it there is no
way of resolving conflict that may arise in cases of multiple inheritance.
For example, if a lower-level construction inherits information from two
different dominating constructions, but the information specified for these
constructions conflicts, how do we know which of the parent constructions
wins? In a case such as this, we can only know which information will be
inherited from which parent construction if this is specified redundantly in
the daughter construction.
A redundant system can also be argued for on psychological grounds.
While a non-redundant system is more economical, it requires maximum
on-line processing; that is, for a speaker to classify an utterance as an
instance of a lower-level construction, they must search all the way up the
inheritance tree in order to determine what the specifications of this construction are and whether the utterance matches them. In contrast, Goldberg
(1995: 74) says that the inheritance mechanism of our system is not an online process, but rather a static relation defined by shared information. The
evidence from cognitive psychology supports the latter system; it suggests
that Concepts and properties in human knowledge are organized with little
concern for elegance and parsimony (Barsalou 1992: 180).
By highlighting shared information, a redundant system also shows how
the constructional taxonomy supports itself. As Lakoff (1987: 538) notes,
the more redundancy a construction exhibits, the better it fits into the language network. In other words, the more properties that are inherited from
other constructions, the more we can say that the construction is motivated,
A constructional model of language change
21
or supported, by the language system. This means that we have an explanation as to why this particular construction with these particular formmeaning correspondences should be likely to exist in the language. Goldberg (2003: 120) notes that the requirement that every construction must be
motivated is What imbues a constructional approach with explanatory
adequacy. Therefore, while the symbolic nature of the construction is
designed to capture idiosyncratic information, the organization of grammatical knowledge in construction grammar nevertheless encourages and
facilitates the identification of generalizations (whether at more local or
more general levels in the taxonomy). In construction grammar then,
Exceptions are allowed to exist, but only at a cost to the overall system
(Goldberg 1995: 119).
Thus, a default inheritance model allows for partial generalizations to be
recognized. Some constructions (or instances of constructions) are better
(or more motivated) members of the constructional category than others. As
a consequence of this, the categories defined by the inheritance hierarchies
of this model are non-classical, with each category containing prototypical
members and non-prototypical members.3
3.
On a usage-based model, language change originates in language use, at the

level of the instance (or construct), and is dependent upon general cognitive
processes, such as categorization (Bybee 1985, 1995, 2006). As a result,
change is not confined to the acquisition process on this model; a speakers
grammar can change throughout their lifetime. Assuming a usage-based
approach, both language learning and language change involve the speaker
inductively generalizing over actually occurring instances to form mental
schemas (or constructions) which are represented in the language system.
The storage and organization of the speakers grammatical knowledge is
dependent upon, and can change according to, patterns of activation. Token
frequency (the number of times a given instance is activated) results in the
entrenchment of the instance as a unit in the language system. Furthermore,
the repeated use of an instance which is already stored as a conventional
unit in the speakers grammar activates (and so strengthens) that construct,
making it more entrenched. In contrast, type frequency (the frequency or
3. For a recent critique discussing the limits of constructional inheritance, see Sag,
Boas, and Kay (2012).
22
activation of different types of instance) results in the entrenchment of a

more basic schema. That is, the speaker abstracts over the different types of
instance to form a schema which stipulates only those characteristics shared
by all instances. The more types of instance a schema sanctions, the more
productive and schematic (or general) that construction will become (see
Bardal 2008).
Along with the non-classical categories of the default inheritance model,
these usage-based assumptions make some testable hypotheses regarding
the diachronic development of constructions. In particular, they suggest that
there are two different type of constructional change: one which is brought
about by token frequency and one which is dependent upon type frequency.
Although these changes have consequences for constructions at different
levels in the hierarchical network, they nevertheless involve the same process of conventionalization (or the entrenchment of schemas).
As shown above, type frequency results in the entrenchment of more
schematic and higher-level constructions. However, token frequency results
in the entrenchment only of a single, substantive, lower-level construct. As
Bybee (1985: 132134) suggests, the repeated use and activation of a construction which represents only one type of instance will only strengthen
the status of that particular schema; it will not serve to reinforce any
superordinate constructions. Consequently, the entrenchment of individual
instances is often found occurring alongside a concomitant loss (or weakening) of the overarching schema, which is no longer type productive. This
kind of constructional change can therefore lead to the fossilization of
entrenched patterns which are no longer properly integrated into the constructional taxonomy. For instance, Fillmore, Kay, and OConnor (1988)
note that in the comparative conditional the X-er the Y-er construction
(such as the more you practice, the easier it will get), the is not an instance
of the definite article, and is instead a relic of the Old English instrumental
demonstrative y. In many ways, fossilization depends upon a redundant
system of grammar, in which information inherited from the parent construction is also specified (or stored) in the daughter construction.
While token frequency accounts for historical retention, type frequency
accounts for novelty in the language system. On a usage-based model, new
types of instance are formed by extension from the prototype. Langacker
(1991: 295) argues that such extensions occur because of the pressure of
adapting a limited inventory of conventional units to the unending, evervarying parade of situations requiring linguistic expression. The new instances therefore represent non-prototypical members of the constructional
category, overriding inheritance from the overarching construction. As the
23
new and existing types of instance coexist, the speaker generalizes over
them to create a new level of abstraction, which resolves the former conflict
between the category and its new membership. This can, in turn, have repercussions for yet higher-order constructions (or categories) in the taxonomy, as existing schemas become more abstract in order to accommodate
(or sanction) the new lower-level constructions. This type of constructional
change therefore proceeds upwards throughout the hierarchy, leading to the
creation of new constructions and the reconfiguration of existing ones. As
Goldberg (2006: 62) observes, we constantly parcel out meaning, form
abstractions, and generalize over the instances we hear. This kind of constructional change is thus both gradual (proceeding in incremental steps)
and directional.
For historical linguists who accept a constructional model of language
structure, this type of constructional change constitutes a reimagining of the
well-attested directional change of grammaticalization. Although this label
traditionally applies to simple, atomic elements as they become more grammatical over time, such items are also regarded as constructions on this
model, since they are also symbolic units of form and meaning. From this,
it follows that changes which apply to substantive, simple constructions
should also affect more schematic and complex constructions. In other
words, if lexical items can grammaticalize, larger, less substantive constructions should also be subject to grammaticalization (Trousdale 2008a:
3334).
Assuming a constructional model of language structure, the lexicon and
the inventory of grammatical constructions are not separate components;
instead, they exist along a syntax-lexicon continuum, which ranges from the
most schematic and general constructions to fully substantive, unproductive
patterns. Thus, on this model, the cline from lexical to grammatical status is
re-envisaged as a hierarchy from more substantive to more schematic constructions. Within the framework of construction grammar, grammaticalization is therefore a process of schematization, in which the construction
becomes a more abstract, higher-level category and its internal composition
becomes less fixed. As Trousdale (2008b: 170171) comments, The more
schematic the construction, the more productive it will be (thus such constructions become aligned with what is usually called syntax and productive morphology).
In many ways, usage-based construction grammars are ideally suited for
providing a theory from which to study grammaticalization phenomena. As
we have seen, the hierarchical network is able to both represent and account
for these directional changes (see also Traugott 2007; Trousdale 2008a;
24
Fried 2008). Furthermore, the notion of a gradient continuum from lexicon

to syntax accords with the gradual nature of grammaticalization, in which
lexical items develop more grammatical functions through a series of small,
incremental stages (see Diewald 2006). As Langacker (1990: 16) observes,
a grammaticalizing element moves along this continuum rather than jumping from one discrete component to another. Finally, a focus on constructional change fits in with more recent approaches to the grammaticalization
of atomic elements as being context-dependent. For example, Himmelmann
(2004: 31) considers the grammaticalization of the individual element to be
dependent upon the schematicity of the surrounding construction; he argues
that the unit to which grammaticization properly applies are constructions,
not isolated lexical items (emphasis original); see also Bybee (2003).
This description of the grammaticalization of constructions (or rather,
grammatical constructionalization) also offers a way of accounting for the
existence of mismatch phenomena. On this model, grammaticalization is
akin to schematization or expansion. As such, the construction is extended
to accommodate items with which they are not prototypically associated.
Michaelis (2003: 263) uses the terms coercion and coercion effect for the
enriched interpretations which result from this procedure. In construction
grammar, coercion is accounted for by a combination of default inheritance
and the symbolic nature of the construction; that is, if the constructions
conventional meaning conflicts with the meaning typically associated with
a superimposed lexical item, the constructional requirements win out and
the lexical item conforms to them. Michaelis (2003: 268) refers to this as
the Override Principle.
An example of such template-based coercion is provided by Goldberg
(1995: 158), who identifies mismatch in instances of the English causedmotion construction between the semantics of the verb and the semantics
designated by the construction. She notes that in examples like Joe kicked
the dog into the bathroom, motion is coded by the verb and the preposition
into. However, in examples such as Sam squeezed the rubber ball inside the
jar and Sam urged Bill outside of the house neither the verbs squeeze or
urge nor the prepositions inside or outside independently code motion
(Goldberg 1995: 158; italics original). In such instances, the construction
coerces the locative term into a directional reading (Goldberg 1995: 159).
According to Michaelis (2003: 266), construction grammar is ideally suited
to representing these coercion effects, since it uses a single combinatory
mechanism, the construction, to account for both coerced and syntactically
transparent interpretations.
The application to it-clefts and copular constructions
25
An understanding of coercion also helps to explain the gradual nature of

constructional change. As Goldberg (1995: 159) comments, coercion is
governed by the extent to which there is a relationship between the inherent
meaning of the coerced item and the interpretation which it is given by the
construction. This explains why some items can be successfully coerced
into the construction, while others result in ungrammatically. For example,
in the case of the caused-motion construction above, the relationship
between the prepositional meaning and the constructional meaning is
straightforward: the location given by the preposition is interpreted as the
endpoint of a path to that location (Goldberg 1995: 159). On this account,
then, the grammaticalization of constructions is governed by the degree of
similarity between the new instances and the prototype; that is, between the
coerced item and the coerced interpretation.4
4.
The application to it-clefts and copular constructions
In this section, I briefly summarize how I make use of the various concepts
of usage-based construction grammar (introduced above) in my analysis of
the English it-cleft construction. I go on to highlight two important ways in
which my approach to cleft and copular constructions differs from much of
the relevant literature on this topic.
As we saw in Chapter 1, the it-cleft is a specialized linguistic pattern, in
that much of its structure and use cannot be predicted from highly general
rules. It is therefore well-suited to treatment within a constructional framework. Nevertheless, in the following pages, I show that the it-cleft is also a
motivated construction. By placing the it-cleft in a hierarchy of copular
constructions, I show that many of the it-clefts properties are inherited
from a larger specificational copular construction, which is itself a mismatch structure. In this way, the constructional framework helps us to identify generalizations, albeit often at a much more local level in the hierarchy.
I then go on to explain the it-clefts remaining idiosyncratic properties
by appealing to historical evidence. I show that the diachronic development
of the it-cleft construction involves both fossilization and schematization;
that is, the it-cleft preserves properties which were originally inherited from
4. However, this is not to say that coercion is predictive or deterministic in any
strict sense. Traugott (2007: 524) makes the point that coercion construed as a
strictly formal device is problematic for a theory of constructions as dynamic,
partially-productive, and contingent, not deterministic (Goldberg 2006: 217).
26
once productive patterns and has acquired new types of instance over time.
As the coerced interpretations found in these new mismatch constructs are
conventionalized, the it-cleft gradually develops a range of new functions
and ultimately becomes a more schematic and productive construction.
At a more fundamental level, my analysis is shaped by the assumptions
of construction grammar in two further ways. First, since the form-function
mapping is internal to the construction on this model, construction grammar differs from most other theories in that it encourages accounts which
discuss form and meaning together (rather than structure alone). As a result,
my analysis often focuses on symbolic and semantic descriptions of constructs typically viewed from a syntactic perspective (see Croft 2007: 490).
This is especially true of my analysis of specificational copular sentences in
the following chapter.
Second, construction grammar is a monostratal model of language; that
is, it does not recognize separate levels (or components) of syntactic structure. Therefore, while I often make use of linguistic terms which imply a
derivational process, such as clefted, extraposed, reverse and inverse, I do
not assume that these constructions are derived from more basic, underlying structures. Instead, my use of this terminology is intended only as a
way of engaging with a larger literature developed from within alternative
theories of grammar.
Chapter 3
Specificational copular constructions
1.
Different and competing analyses
In Chapter 1, I argued for an account of it-clefts which treats them foremost

as specificational sentences. In this chapter, I outline a particular analysis of
specificational copular constructions which focuses on defining the concept
of specificational meaning. The account is developed in relation to noncleft
NP be NP sentences before it is extended to other kinds of copular sentence, including th-clefts, wh-clefts and all-clefts. The purpose of this
chapter is to establish a unified account of the family of specificational
copular constructions which can be usefully employed in the analysis of itclefts developed in Chapter 4.
My analysis of specificational copular sentences begins with an examination of sentences of the type NP be NP, as in (1).
(1)
The thoracic surgeon is John McIntyre
How we analyse specificational sentences like (1) is dependent on whether

we choose to relate them to superficially similar sentences such as (2), and,
at a more fundamental level, on how we interpret the relationship between
sentences containing postcopular definite and indefinite noun phrases, as in
(2) and (3).
(2)
John McIntyre is the thoracic surgeon
(3)
John McIntyre is a surgeon
Most authors would agree that in example (3), which contains a postcopular
indefinite NP, John McIntyre is classified as a member of the set (or category) of surgeons and is therefore ascribed the property of being a surgeon.
The sentence therefore expresses a relation of set-membership (or class
inclusion). However, the analysis of sentences containing postcopular definite NPs, as in (2), is more controversial and corresponds, in part, to two
different and competing accounts of specificational NP be NP sentences,
such as (1). These are commonly referred to as the equative approach and
the inverse approach.
28
In what follows, I outline the main tenets of each type of analysis before
going on, in the remaining sections of this chapter, to advance a particular
account of specificational copular sentences which draws mainly from the
inverse tradition. I approach the literature and outline my analysis with the
assumptions of a set-theoretic account of nominal predication, defined as a
semantic relation between members and sets. Later in the chapter, I situate
my analysis within constructional and cognitive frameworks that take a different approach to the concept of nominal predication.
1.1.
The equative approach
For some authors, sentences with postcopular definite noun phrases, such
as (2) above, are semantically equative. On this analysis, the thoracic surgeon functions as a referring expression, picking out a particular individual
that is equivalent to John McIntyre. The sentence in (2) therefore tells us
that the individuals John McIntyre and the thoracic surgeon are one and the
same. Since the specificational sentence in (1) contains the same two noun
phrases, albeit in the reverse order, it can be accounted for in the exact
same way (see for example, Evans and Green 2006: 599).
On this account then, specificational copular sentences are provided
with the same analysis as identity statements, such as (4). This makes sense
because such sentences are also reversible, as shown in (5). It also helps to
explain why specificational sentences like (1) are felt to have an identifying
meaning.1
(4)
Trapper John is John McIntyre
(5)
John McIntyre is Trapper John
The account is further supported by the behaviour of sentences containing

postcopular indefinite NPs, as in (3). Such sentences, which have an ascriptive rather than identifying function, do not seem to share this characteristic reversibility, as shown in (6).
(6)
#A surgeon is John McIntyre
1. However, as Declerck (1988: 3) observes, the meaning relation involved in specificational sentences is not quite the same as that in identity statements; he
says, They are identifying in the sense that they reveal the identity of some entity not in the sense that they state a relation of identity between two entities.
29
Other proponents of the equative approach to specificational sentences

assume that sentences containing postcopular definite NPs, such as (2), are
instances of predication rather than equation. Here then, as in (3), the postcopular NP is a set-denoting predicate, of type <e,t>, which combines with
a referring expression, of type e.2 However, on the basis of evidence such
as (6), these authors argue that it is not possible to treat any constituent
appearing in [subject] position as predicated of a postcopular argument
(Heycock and Kroch 1991: 380; see also Rothstein 2001). As a result, specificational sentences like (1) must instead equate two phrases of the same
semantic type: in this case, two (type e) referring expressions. The apparent
inversion relationship between pairs of copular sentences such as (1) and
(2) is therefore illusory.
Since two referring expressions cannot combine directly (in that one is
not the argument of the other), most proponents of this approach claim that
in specificational sentences (and identity statements), the copula is a special
be of identity. An exception to this is Heycock and Kroch (1999), who suggest that the identity relation originates instead from the null functional
head of an equative small clause. Either way, common to all such accounts is the claim that the two NPs of a specificational NP be NP sentence
are of the same semantic type, which can therefore be equated or identified
(and which can also be reversed).
However, a problem with the equative approach is that the NPs in
specificational sentences are not as equal as those of identity statements.
For instance, while it seems reasonable to analyse the two proper names in
(4) as type e expressions referring to individuals, it is not at all obvious that
the definite NP the thoracic surgeon in (1) is being used in this way, at
least not to the same extent as the postcopular NP John McIntyre. Heycock
and Kroch (1999: 381) concede that the assimilation of specificational
sentences to equatives runs afoul of the intuition that the former are asymmetric in interpretation in a way that true equatives are not; they go on to
attribute this to a difference in information structure, rather than semantic
type.
2. In type-theoretic semantics, e defines an individual constant; that is, an argument expression which has constant reference to an individual. One-place predicates have the complex type <e,t>. This indicates that the expression combines with an individual expression (of type e), which results in a formula (an
expression of type t) (see Cann 1993: 8384). I make use of types throughout
this chapter as a way of engaging with the formal semantics literature on copular sentences and NPs (see especially Partee [1986] 2004a, [1987] 2004b).
30
Despite this problem, the equative approach to specificational sentences

is very popular, especially among those offering a semantic explanation for
syntactic connectedness (discussed in Section 4.3.2). More interestingly for
our purposes, this approach is also common among authors who, like me,
analyse it-clefts in relation to specificational copular sentences (see, Han
and Hedberg 2008; Percus 1997; Reeve 2012). I return to this issue in
Section 3 of Chapter 4.
1.2.
The inverse approach
The inverse approach to specificational sentences depends upon the initial

claim that sentences with postcopular indefinite and definite noun phrases
are both predicational (rather than equative). Sentence (2), repeated here as
(7), therefore contains a precopular referring expression (John McIntyre)
and a postcopular set-denoting NP (the thoracic surgeon), which ascribes a
property to the referent. On an inverse account, the specificational sentence
in (1), repeated here as (8), has the reverse (or inverse) configuration, with
the type <e,t> predicative NP appearing in initial position.
(7)
[predicational]
(8)
[specificational]
On some versions of this account, specificational and predicational copular

sentences are actually derived from the same small clause structure, in
which the referential noun phrase always precedes the predicative noun
phrase. The argument goes that if the referential noun phrase is raised into
the subject position, a predicational copular sentence is obtained. In contrast, specificational copular sentences result from a movement operation
which raises the predicative noun phrase (see Moro 1997; Mikkelsen 2005).
An advantage to the inverse approach is that it captures the intuition that
specificational sentences are semantically asymmetrical. Indeed, there is
good evidence for the claim that the initial NP in specificational examples
like (8) is non-referring. By experimenting with environments involving
left-dislocation structures, question-answer pairs and tag questions, Mikkelsen (2002, 2005) shows that while in predicational and equative constructions, anaphoric pronouns agree with the subject in terms of gender,
number and animacy, specificational copular sentences allow the pronoun it
to be anaphoric to gender-specific subjects, as shown in (10).
31
(9)
The tallest girl in the class is Swedish, isnt she?
(10)
The tallest girl in the class is Molly, isnt it?
(11)
[Pointing to a player on the field]

SHE is Molly Jacobson, isnt she?
[equative]
(examples from Mikkelsen 2005: 72)
[predicational]
[specificational]
These results indicate that there is a clear semantic difference between the
subjects of specificational and equative copular sentences. As Heycock and
Kroch (2002: 106) concede, This behaviour is quite unexpected under an
equative analysis. On an inverse account, however, the pronominalization
data gains a simple explanation; as Mikkelsen (2005: 66) observes, it is often anaphoric to predicative elements, as shown in (12) (see also Chapter 4,
Section 1.1).
(12)
LBJ is the President of the United States. He has been {it/*him}

since 1963.
Proponents of the inverse approach include (to name a few) den Dikken
(2006), Heggie (1988), Mikkelsen (2005), Moro (1997), Partee (2004a) and
Williams (1983). However, despite a large literature developing and supporting inverse accounts, the approach leaves a number of questions unresolved. For instance, a potential problem with this type of analysis is that
some predicational copular sentences resist inversion. We have already
seen, in (6) above, that indefinite NP predicates cannot typically occur in
precopular position. In fact, as I go on to explain in Section 3, some indefinite NPs, in certain contexts, can function as the initial NP of a specificational copular sentence. However, the exact criteria on which this distinction is based remain elusive.
Authors supporting an equative account argue that indefinite NP subjects will only be permissible if they allow a specific reading, on which
they refer to a particular individual (see Heycock and Kroch 2002: 112). In
contrast, those arguing for an inverse approach have sometimes appealed to
information structure conditions as an explanation for why indefinite NP
predicates are often barred from moving into precopular position (for example, see Mikkelsen 2005). In Section 3.2, I show that information structure alone cannot satisfactorily account for the data. It therefore remains a
challenge for the inverse approach to provide an adequate description of, as
well as an explanation for, the restrictions on indefinite NP predicates.
32
A further difficulty with the inverse approach is that it does not address
the question of where specificational meaning comes from. These works
stipulate that when inversion occurs, the result is a specificational copular
sentence. Mikkelsen (2005: 1) suggests that predicational copular clauses
tell us something about the referent of the subject, while a specificational
clause says who or what the referent is (emphasis original). However,
specificational meaning does not follow as a direct result of movement and
is not tied to a particular word order. For instance, noninverted NP be NP
sentences with focal subjects, such as (13), can perform the same specificational function of identifying rather than describing the referent as their
inverse counterparts (see Declerck 1988: 93). Such examples are sometimes labelled reverse specificational copular sentences.
(13)
A. Who is the winner? B. JOHN is the winner, isnt he?
(14)
A. Who is the winner? B. The winner is John, isnt it?

This poses the following, as yet unanswered, questions: what is the relationship between specificational and predicational meaning and how does
specificational meaning come about?
1.3.
A less formal approach
Higgins (1979) presents a less formal account of specificational copular

sentences which focuses on characterizing the nature of specificational
meaning as distinct from both identity and predication. For Higgins,
specificational sentences function like lists; that is, the initial NP acts as the
heading of the list and the postcopular elements serve as items on that list.
On this interpretation, the specificational NP be NP sentence in (15) can be
paraphrased as the list given in (16) (see Higgins 1979: 154).
(15)
The runners up were Max and Henry
(16)
Runners up: Max, Henry
Higgins suggests that specificational sentences involve a value-variable

relation. He notes that the heading of a list provides a variable, thereby
33
delimiting a certain domain, to which the items on the list conform as

values of that variable (Higgins 1979: 155).3
Higgins (1979: 214) maintains that this relation is not the expression of
some kind of identity and goes on to argue against the equative approach
to specificational sentences. However, Higgins does not align his analysis
with those involving predication either. He notes that The whole notion of
being about something is alien to a list; a list is neither about the
heading of the list nor about the items on the list (Higgins 1979: 214).
To cement this distinction, Higgins chooses to discuss specificational
sentences, not in terms of reference or predication, but using his own
terminology. For Higgins, the precopular NP is superscriptional and the
postcopular element is specificational.
Unfortunately, when it comes to characterizing the superscriptional and
specificational components, it is not very clear how Higgins terminology
relates to more well-defined concepts. Nevertheless, in what follows, I
suggest that Higgins characterization of specificational meaning is not
actually at odds with analyses involving predication. Certainly, Higgins
recognizes a semantic asymmetry between the two components of a specificational sentence and highlights the importance of the parallelism between pairs of predicational and specificational sentences, such as (17) and
(18).
(17)
That he hasnt come is a problem
(18)
The problem is that he hasnt come

(examples from Higgins 1979: 274)
Higgins (1979: 275) suggests that while these examples are presumably
not directly related,the existence of the parallelism clearly means something. While Higgins seems doubtful of the validity of movement-based
3. Declerck (1988) also engages with the concept of specificational meaning and
formulates an account based upon the same value-variable relation as Higgins.
However, by expanding on this idea, Declerck provides an extremely broad
definition of specificational meaning as pertaining to any sentence that gives
the answer to a wh-question, including ones with a predicational focus, as in (i).
This effectively reduces the concept of specificational meaning to contrastive
focus. Since such examples have a predicational or ascriptive meaning, they are
not considered specificational copular sentences here.
(i)
Q. What is John like? A. John is SILLY. (adapted from Declerck 1988: 39)
34
accounts here, I show that his observations square neatly with the particular
type of non-derivational inverse analysis that I propose.
2.
Specification as (the inverse of) nominal predication
In this section, I build up an account of specificational copular sentences

which adopts many of the same assumptions as the inverse approach. That
is, I assume that such sentences involve predication (rather than equation)
and that the subject NP can function as the semantic predicate. However,
since I assume a monostratal model of language, it does not follow on my
account that specificational sentences are derived via movement operations
(see Chapter 2, Section 4).4 Instead, I develop constructional schemas for
such sentences, which highlight their symbolic and semantic properties. In
particular, my focus is on explaining the relationship between specificational and predicational meaning. That is, I examine how and why specificational meaning arises in some sentences containing predicative NPs and
not others. Making use of Hawkins (1991) characterization of definiteness,
I explain why definite NP predicates are especially well-suited to the specifying function and go on to suggest some reasons why specification is so
often expressed via an inversion construction. Finally, I show that while the
account of nominal predication that I adopt here is not customary in constructional theories, it is not incompatible with these frameworks.
2.1.
Specification and definite NP predicates
To begin then, let us assume that the sentences in (19) and (20) are both
predicational, containing a precopular referring expression (John) and a
postcopular predicative NP which denotes the set of individuals that have
the relevant property (of surgeon or best surgeon).
(19)
John is a surgeon
[predicational]
(20)
John is the best surgeon
[predicational]
4. I follow Birner (1994, 1996), in using the terms inverse and inversion without
supporting a multistratal, movement-based analysis, but rather to conform to
traditional terminology for clarity (Birner 1994: 235).
35
This establishes the familiar view of nominal predication as an expression

of class or set membership. In (19) then, the referent John is categorized as
a member of the set of surgeons. In other words, John is ascribed the
property of being a surgeon. Similarly, for (20), we can say that the best
surgeon describes, or ascribes a property to, John.
However, the sentence in (20) allows an additional interpretation. If the
referent John is focused, the sentence acquires a specificational reading on
which John is identified as matching the description the best surgeon. In
contrast, the sentence in (19) does not invite this interpretation; even when
the referent John is focused, as in (21), there is still the sense that we are
ascribing a property to John.
(21)
JOHN is a surgeon
(22)
JOHN is the best surgeon
[predicational]
[specificational]
We can account for this difference by examining the nature of the classmembership relation in each sentence. In (20), the postcopular NP denotes
a set with only one member; there can only be one best surgeon. Therefore,
by classifying John as a member of this set, we are saying something about
John (describing him as being the best surgeon) and we are listing the
complete membership of the set best surgeon. The latter, specificational
reading is brought about if the referent (or membership) is focused, shown
in (22). In contrast, the predicate nominal in (19) denotes a set with many
members. By classifying John as a single member of the set of surgeons,
we are ascribing a property to John. However, since John does not come
close to making up the membership of this unrestricted set, a specificational
reading for (21) is not forthcoming.
This simple observation goes a long way to explaining why definite
noun phrases are so common in specificational NP be NP sentences. We
saw that in (20) above, the superlative best indicates a single member set.
However, in most specificational sentences, the set is restricted simply as a
result of it being marked as definite. For example, the sentence in (23)
allows a specificational interpretation despite the fact that there are other
thoracic surgeons in existence.
(23)
JOHN is the thoracic surgeon
[specificational]
Here, it is the presence of the definite article that marks this set as restricted, such that John represents an exhaustive list of its members.
36
According to Hawkins (1991), the definite article has three main properties. First, it carries with it the conventional implicature that the noun
phrase will be understood in relation to a shared set or pragmatic (P-) set
which is manifest in the speech participants mutual cognitive environment
(a notion borrowed from Sperber and Wilson (1986)). For example, in (23),
the thoracic surgeon can be understood in relation to the surrounding context (or larger situation set) expressed overtly in (24).5
(24)
JOHN is the thoracic surgeon (in this outfit)
Secondly, the definite article is associated with existence entailments. Accompanied by the implication of P-membership, it follows from (23) that
there is a thoracic surgeon in this outfit. Finally, the definite article is associated with uniqueness. This means that the descriptive content contained
within the NP applies uniquely within the P-set. For example, since the
definite noun phrase in (23) contains a singular noun, it can only be used to
describe one entity (and no more). Thus, it follows from (23) that there is
only one thoracic surgeon in this outfit. Hawkins extends the uniqueness
generalization to definite NPs containing plural nouns by suggesting that in
all cases, the definite NP applies to the total or maximal set of entities that
satisfy the description within the P-set. Hawkins refers to this regularity as
inclusiveness (see also Hawkins 1978).
Assuming this account of definiteness, we can now explain why definite
NP predicates allow a specificational interpretation. From the conventional
implicature of P-membership, it follows that a definite description denotes
a set of entities that exists within a shared environment. In other words, the
predicate denotes a set that is always understood to be restricted in some
way. For instance, the set of thoracic surgeons in this outfit is more specific
(has fewer members) than the set of all thoracic surgeons. If specificational
meaning involves listing all of the members of a described set, it follows
that this set is likely to have only a small membership. The predicative NPs
that allow a specificational interpretation will therefore denote restricted
sets.
In addition, it follows from the property of inclusiveness that when a
definite NP is predicated of a referring expression, the description will
5. This is not to say that the incomplete description in (23) contains additional
restrictive material at some underlying level of syntactic representation. Such
an analysis is dubbed the syntactic ellipsis approach by Stanley and Szab
(2000), and is strongly criticized therein.
37
characterize the individual(s) referred to uniquely (see Declerck 1986: 30).

Consequently, the subject NP will be understood to comprise a complete
list of entities that make up the restricted set. For example, the sentence in
(23) suggests that John is the only thoracic surgeon in this outfit. The same
is true of examples containing plural definite NPs, such as (25).
(25)
MAX and HENRY were the runners up
In this sentence, the definite NP predicate applies inclusively to all entities

that satisfy this description. Max and Henry are therefore taken to constitute
an exhaustive list of the members that make up the pragmatically restricted
set of runners up (in a particular race or competition). In other words, we
assume that there were no other individuals, besides Max and Henry, who
were runners up on that occasion.
In the account sketched here, specificational meaning is conceived of in
much the same way as by Higgins (1979). As Higgins suggests, specification involves the listing of entities. He elaborates by saying that the
heading of a list delimits a domain and each item on that list identifies a
particular member of that domain (Higgins 1979: 213). Here, Higgins gets
very close to the concept of specification as a class-membership relation,
despite his choice to discuss such sentences in terms not involving predication (see Section 1.3). In this section, I have viewed specificational meaning as a reimagining of nominal predication. On this interpretation, a class
inclusion relation is used to identify the membership of a set, rather than to
say something about an individual (as having the property of being a member of some set).
I have explained that definite NP predicates are especially well-suited to
the specifying function as a result of their unique semantic and pragmatic
properties. In particular, I have claimed that a specificational interpretation
is facilitated if the predicative NP denotes a restricted set (for which it is
possible to list all of its members) and if the individuals referred to can be
taken as an exhaustive list of members within the described set (on the
principle that a well-formed list should ideally be complete). As I go on to
explain in Section 3.1, indefinite NP predicates do not necessarily share
these properties. For instance, as we saw in (19) above, indefinite NPs can
denote very general sets, which are not understood in relation to a given Pset (see Hawkins 1991).
Finally, I have suggested that in predicational NP be NP sentences,
specificational meaning is accompanied by an information structure re-
38
quirement; in order to interpret a class-membership relation as identifying

the membership of some set, the members (or referents) should be in focus.
2.2.
Specification and inversion
In Section 2.1, I outlined an account of how specificational meaning arises

in certain cases of nominal predication. As a result, our focus so far has
been on so-called reverse specificational sentences, i.e. non-inverted NP be
NP sentences with a specificational meaning.6 However, while sentences
with postcopular predicative NPs can allow a specificational interpretation,
inverse NP be NP sentences consistently have a specificational, rather than
a predicational (or ascriptive), meaning. In this section, I examine why specification is associated with inversion and why a specificational inversion
construction should exist alongside its non-inverted counterpart.
On the account sketched so far, sentences with postcopular definite NPs,
such as (26), contain precopular referring expressions and postcopular predicative NPs. Following the inverse approach to specificational sentences
(outlined in Section 1.2), I assume that (27) involves the inverse alignment,
with the semantic predicate occurring in precopular position.
(26)
(27)
However, unlike most inverse accounts, I do not assume that the latter is
derived from the former, or that these sentences stem from the same
underlying structure. Rather, I suggest that NP be NP sentences with precopular predicative NPs make up a distinct construction that shares properties with other kinds of inversion construction, including VP inversion,
PP inversion and AP inversion, as shown in (28), (29), (30) and (31).
(28)
The winning couple was Pierce and Able
[NP inversion]
(29)
Sponsoring the event was Jonathan Tuttle
[VP inversion]
6. As den Dikken (2005) notes, there is confusion in the literature over the terms
reverse and inverse, since they have come to refer to opposing sentences. Their
use depends on which configuration is considered the more basic or canonical.
While I do not assume a derivational account, in what follows, I continue to use
terminology associated with the inverse approach.
39
(30)
Behind the scenes was Sidney Freedman
[PP inversion]
(31)
Particularly impressive were Leslies legwarmers
[AP inversion]
Although these sentences may not comprise a syntactically uniform category,7 Birner (1994: 235) observes that in all the different kinds of inversion, the logical subject [of which something is being predicated] appears
in postverbal position while some other, canonically postverbal, constituent
appears in clause-initial position. Furthermore, sentences (28) to (31) all
involve inversion around be and the clause-initial constituent is semantically predicative. Thus, specificational sentences like (28) might be said to
belong to the family of be inversion (or predicative inversion) constructions
(see Dorgeloh 1997: 84).8
The question then is why is specificational meaning often expressed via
an inversion construction? I suggest that there are two main reasons. The
first relates to information structure. In her corpus study of English inversion constructions, Birner (1994, 1996) finds that such sentences are subject to a robust pragmatic constraint. She notes that the preposed element
in an inversion must not be newer in the discourse than the postposed element (Birner 1996: 90). For Birner (1994: 234), the function of inversion
sentences is to allow the presentation of relatively familiar information
before a comparatively unfamiliar logical subject. As we saw in Section
2.1, specificational meaning is also tied to a particular information structure, arising in examples where the referring expression is in focus. Specification is therefore well-suited to the information structure requirements
of inversion, since the relatively informative membership information will
be situated in postverbal position, after the set description.
7. While the precopular NP in specificational sentences is clearly the grammatical

subject, the clause-initial elements of other types of inversion sentence display
mixed evidence for subject status (see Bresnan 1994; Mikkelsen 2005). For example, in the specificational (28), the verb agrees with the precopular NP; however, in the AP inversion in (31), it is the postverbal NP that determines agreement. As Mikkelsen (2005: 138) suggests, there may be a historical explanation
for this morphosyntactic distinction (see also Chapter 6, Section 5).
8. PP inversion commonly occurs with verbs other than be, with the clause-initial
constituent providing locative information. As a result, most authors recognize
two types of inversion: locative inversion, which occurs with verbs that take
locative complements, and the inversion of non-locative predicatives, which is
generally restricted to the verb be. Although the distinction is not clear-cut, it is
the latter that is most relevant to inverse accounts of specificational sentences.
40
Secondly, inversion provides an unambiguous way of distinguishing

between specificational and predicational meaning. We have seen that, for
sentences with postcopular definite NPs, the class-membership relation
allows both a predicational (ascriptive) and a specificational interpretation.
However, since inversion constructions have a fixed information structure,
the examples which allow a specificational interpretation will always be
interpreted as specificational. Conceivably, this situation results in specificational meaning becoming associated with NP inversion. As a result, this
particular type of inversion construction is constrained not only by discourse considerations, but also by the criteria which sanction specificational meaning. I come back to this issue in Section 3.3.
This accounts for the association between specificational meaning and
inversion. However, it also goes a long way to explaining why this rather
idiosyncratic construction should exist alongside the non-inverted predicate
nominal construction. I suggest that the specificational NP inversion construction is motivated (or supported) in three different ways.
First, this sentence type is part of a family of be inversion constructions
in which a predicative constituent appears in clause-initial position and the
argument of which it is predicated occurs postverbally. This mismatch
construction, which overrides more general patterns of correspondence
between form (syntax) and function (semantics), is therefore reinforced at a
local level in the inheritance hierarchy (see Chapter 2, Section 2). Second,
as we saw above, the NP inversion construction provides a syntactic way of
marking specification as distinct from predication. Finally, since all inversion constructions have the same fixed information structure, specificational inversion sentences conform to the generalization that given information should be presented prior to new information (see Ward, Birner, and
Huddleston 2002: 1372). In other words, this sentence type is motivated by
inheritance from a highly general information structure construction.
As a final question, we might ask, how is predication (or class inclusion)
actually achieved in the specificational inversion construction? In construction grammar, predication is interpreted as valency or relationality.
Verbs are understood to be inherently relational. For instance, the act of
sleeping requires a sleeper. In the sentence Henry sleeps then, the semantic
structure of the predicate sleeps contains a schematic substructure which is
elaborated by the non-relational argument Henry (Croft and Cruse 2004:
281). However, prototypical nouns (which denote objects), are not inherently relational, that is, they do not imply the existence of any other entity
(having a valency of zero). As a result, when nouns occur in a predicative
position, some semantic relation between the subject and the noun must be
41
coerced or construed. Thus, the noun is forced to behave as if it takes a

single semantic argument, having a valency of one (Croft 1991: 69). This
explains how indefinite NPs, which typically denote individuals, come to
denote sets (or categories) in the predicate nominal construction: they are
incorporated into a class inclusion relation (see Chapter 2, Section 2).
From this, NPs are not expected to have a relational meaning when they
occur outside of the predicative position; that is, we would not expect a
class inclusion relation to be forthcoming independently of the predicate
nominal construction. However, there is reason to suggest that definite
descriptions may be, in some sense, relational.
Hawkins (1991) characterization of definiteness is consistent with the
treatment of the as a form of universal quantifier, akin to all (see Hawkins
1978). However, while sentences with all make a claim about all of the
individual members of some set, sentences with the seem to make a claim
about a set as a whole. For instance, it is not necessary for all of the individual paving stones to be cracked in order for There are cracks in the
paving stones to be true (see Hawkins 1991: 409; Burton-Roberts 1981).
Thus, definite noun phrases appear to denote sets, even though their use is
often synonymous with reference to the members of this set.9
On this account, the definite article serves to delimit and isolate sets of
entities rather than individuals. Hawkins (1991: 410) says that there is
some unique maximal set of entities within a P-set to which expressions
such as the bridesmaids refer. For a singular definite NP, this set comprises
just a single member. In some sense then, definite NPs are relational, in
that the existence of a set implies the existence of its members. Declerck
(1987: 20) suggests that, Reference to a set implicates reference to all its
members (emphasis added).
In construction grammar, predication (or valence) is relative; as Croft
and Cruse (2004: 281) note, predicate and argument status depend on what
two semantic structures are being compared. Thus, we might suggest that
when a set-denoting definite description is compared with an individual
expression, a relation of class-membership can ensue. Here, the definite NP
9. Examples in which reference to a set as whole does not correspond to reference
to all of the members of a set therefore represent an argument against the claim
that the is a universal quantifier (contra Hawkins 1978; Neale 1990). See also
Graff (2001), who shows that quantifier phrases ranging over individuals differ
from definite NPs in that they cannot occur in predicative position (if they are
to be treated as semantic units), and Mikkelsen (2005: 159), who finds that NPs
introduced by all and every do not form acceptable specificational subjects.
42
is not used to refer, but to describe. On this account, specificational inversion sentences could be said to have the same syntactic structure as equative sentences, but a different semantic relation of class inclusion rather
than identity. In other words, the postcopular referring expression is understood to comprise the complete membership of a (pragmatically) restricted set. This may account for some of the similarities which have been
observed in the syntactic behaviour of specificational inversion sentences
and equatives.10
In addition, this scenario would further account for the relation between
specification and inversion. While definite NPs (which consistently allow a
specificational interpretation) invite a class-membership relation, indefinite
NPs must be coerced into it. I examine instances in which indefinite NPs
are coerced into the specificational inversion construction in Section 3.3.
2.3.
Capturing this account in cognitive and constructional frameworks
So far, I have proposed an analysis of specificational copular sentences

which explains how specificational meaning comes about in certain cases
of nominal predication and why it is often expressed via an inversion construction. However, at this point, it is important to note that the account of
nominal predication upon which this analysis is based (as involving a semantic relation between members and sets) is not standard in constructional
frameworks. In what follows, I argue in favour of a class-membership
account of both definite and indefinite nominal predication and show how
this can be expressed using the machinery of cognitive semantics.
On many constructional accounts, sentences with postcopular definite
noun phrases are understood in terms of an identity relation. Croft (1991:
69) suggests that class inclusion and identity are two of several possible
10. For instance, while the subject-predicate strings of predicate nominal sentences
(including non-inverted specificational sentences) can occur as the complement
of consider, shown in (ii) and (iii), specificational inversion and equative structures cannot, shown in (iv) and (v) (see also Heycock and Kroch 1999).
(ii) I consider [John the best doctor]
[predicate nominal]
(iii) I consider [JOHN the best doctor]
[non-inverted specificational]
(iv) *I consider [the best doctor John]

(v)
*I consider [John McIntyre Trapper John]
[specificational inversion]
[equative]
43
semantic relations involving predicate nominals. While most sentences

containing postcopular indefinite NPs are analysed as involving the relation of token to its subsuming type, postcopular definite NPs are said to be
coerced into a relation of token-token identity. In both cases, the copular
verb plays an important role in establishing the appropriate semantic relation. While the former involves a classifying be, the latter requires an
equational be or a be of identity.
However, Langacker (1991) outlines two objections to this type of analysis (which he refers to as a standard approach). First, he takes issue with
the claim that be necessarily expresses a semantic relation in such sentences, since comparable relationships are found in a variety of constructions without be (Langacker 1991: 65). For instance, (32) and (33) involve
the same nominal predication relation as their copular counterparts.
(32)
I considered John a good surgeon
(33)
That would make John the best surgeon
Langacker (1991: 65) concludes that the identity or inclusion relation is

independent of be, arguing that instead be is a meaningful element whose
primary function is temporal and aspectual.11
Secondly, Langacker claims that sentences with definite and indefinite
predicate nominals require parallel analyses. For Langacker (1991: 67),
such sentences are equivalent in all essential respects and differ only to
the extent that definiteness differs from indefiniteness. Langackers analysis of nominal predication therefore develops from the same basic assumptions as my account of specificational sentences. He notes that copular
sentences containing definite and indefinite descriptions are instances of the
same predicate nominative construction and exhibit the same semantic
relation. However, while I have assumed that such sentences involve class
inclusion, Langacker claims that they all predicate identity between two
instances (or individuals).
11. While specificational inversion structures cannot be embedded under consider
(see footnote 10), they can occur as the complement of make, as in (vi). This
suggests that the meaning relation involved in specificational sentences is also
independent of be. See also Higgins (1979: 161), who finds that certain other
verbs permit a meaning relation similar to specification (emphasis added).
(vi) But if what you say is true, that would make the real murderer John!
(Heycock and Kroch 1999: 381)
44
Langackers argument runs as follows. Since the copula be serves only

to temporalize stative predication, it must be the predicate nominatives
themselves that are relational. As such, they express a relation between a
trajector and a landmark, in which a more prominent entity (tr) is understood in relation to a less prominent one (lm). Langacker (1991: 66) posits a
derivational pattern, in which a nominal (profiling a thing X) is construed
as predicative (see Figure 1). The entity (X) becomes the landmark to
which the, as yet schematic, trajector is related via a profiled identity relation (represented by a dotted line).
Figure 1. The semantic derivation of predicate nominatives (Langacker 1991: 66)
In predicational NP be NP sentences then, the subject NP provides the

semantic characterization of the trajector. On this account, (34) specifies
the identity of two individuals: Auschlander (the trajector) and the Chief of
Services (the landmark).
(34)
Auschlander is the Chief of Services
Langacker (1991: 68) extends this analysis to sentences with indefinite NP

predicates, such as (35). Here, the landmark is an arbitrary (or imaginary)
member of the doctor category which is conjured up in the minds of the
speaker and hearer solely for the purpose of enabling an identity relation.
(35)
Auschlander is a doctor
Langacker posits the following semantic structure for the predicate nominative construction (see Figure 2). Here, the landmark is an instance (ti) of a
type specification (T), such as the category doctor or Chief of Services. A
relation of identity holds between this instance and the trajector (Auschlander) within the domain of instantiation.
45
Figure 2. The predicate nominative construction (Langacker 1991: 68)
On this account then, even sentences with indefinite NP predicates

involve a relation of identity, rather than class-membership. According to
Langacker (1991: 68), a class inclusion relation is only implied in such
sentences. For instance, in (35), the inclusion of Auschlander in the doctor
class is specified indirectly as a result of the identification of Auschlander
with an arbitrary, non-specific instance of the class doctor. Langacker
(1991: 67) justifies this account by claiming that, if the predicative NP
simply denoted a class, the presence of the indefinite article would be
unexpected; he says, It is not just a noun but a full nominal, so presumably
it does not represent a type specification but rather an instance of that
type.
However, J. Taylor (2002: 362) is not entirely satisfied with this analysis; he notes that, despite the presence of the indefinite article, which
seems to be an idiosyncratic feature of English, predicative nouns are like
adjectives, predicating a property of the subject referent. This is consistent
with findings from within formal theories of semantics. According to Graff
(2001), Partee (2004a) and Williams (1983), predicative NPs can conjoin
with other properties of type <e,t>, such as adjective phrases, and can
occupy (predicative) positions in which individuals (type e) and phrases
which quantify over individuals (<<e,t>,t>) are restricted from occurring.
As a result, most authors analyse indefinite predicative NPs as having an
(<e,t>) interpretation equivalent to a bare common noun interpretation.12
12. In formal semantics, common nouns are interpreted as <e,t> because they
denote sets of (or properties of) entities, rather than referring to individuals.
46
Langacker (1991: 69) provides this type of analysis for predicate nominatives containing simple, articleless nouns, as in the French example (36).
(36)
Auschlander est mdecin
Auschlander is (a) doctor
For such sentences, the predicate noun describes a type specification (T),
rather than an instance of a type specification (ti). The semantic characterization of the instance is instead provided by the subject nominal. For (36)
then, the landmark is the category doctor and the trajector is the instance
Auschlander. As shown in Figure 3, the profiled relationship (represented
by a dotted line) is thus a correspondence between a type and an instance.
This structure represents a genuine, or direct, instantiation relation between
the trajector instance and the type specification. Here, there is only one instance, provided by the subject nominal. Consequently, the class inclusion
relation is not mediated by an intervening identity relation between two
instances (cf. Figure 2).
Figure 3. Bare noun predicate nominatives (Langacker 1991: 69)
I suggest that this, much simpler, structure can be usefully extended to

English sentences with indefinite and definite NP predicates. On this account, predicate nominals designate types (i.e. classes or categories). They
are also relational, characterizing an instantiation relation between the landmark, which is semantically specified, and the trajector, which is specified
only schematically (to be elaborated by the subject nominal). From this, it
follows that the copula be is not necessary to establish a classifying relation
(contra Croft 1991). Instead, be functions only to express the continuation
of this instantiation relation over time.
Quantified NPs are therefore given a type <<e,t>,t> interpretation because they
quantify over and in relation to sets of entities. See also footnote 2.
Accounting for the behaviour of indefinite NPs
47
The nature of this instantiation relation is dependent upon the semantics

of definiteness and indefiniteness. For sentences involving definite NP predicates, instantiation relates a type specification with all possible instances
of this type. This follows from the property of inclusiveness associated with
the definite article (see Section 2.1). In contrast, sentences containing indefinite predicative NPs relate a type specification to just one of its instances
(see Section 3.1). The distinction between definiteness and indefiniteness
also impacts on the elaboration of the type (T). We saw in Section 2.1 that
while indefinite NP predicates can sometimes denote very general sets,
definite NP predicates are understood to denote restricted sets. In Cognitive
Grammar, the process of restricting or narrowing a type is referred to,
fittingly, as specification. In other words, definite noun phrases typically
designate highly specified types, with fewer instances.
Although this account of nominal predication is not consistent either
with Crofts (1991) constructional take on the standard approach or with
Langackers (1991) Cognitive Grammar analysis, it nevertheless makes use
of existing constructs in cognitive semantics, with one important proviso.
On most cognitive accounts, definite NPs invariably designate instances of
types (rather than types). The definite article indicates that this instance is
grounded; that is, it can be located within the speech event. As J. Taylor
(2002: 346) notes, the grounding of an instance is therefore tantamount to
referring to that instance. However, on an account that recognizes a predicative or non-referring use of definite NPs, the distinction between grounding and specification becomes blurred. By interpreting a type in relation to
the speech event (or mutual cognitive environment), we are necessarily
restricting that type, narrowing the number of its instances (see Section
2.1). This is therefore akin to the process of specification, which J. Taylor
(2002: 344) refers to as the distinctive function of modifiers and complements.
3.
The account outlined in Section 2 treats specificational NP be NP sentences

as predicate nominal constructs involving a class-membership (or class
inclusion) relation. Specificational meaning is understood to be an interpretation of this predication relation, on which the referring expression lists
the membership of some set (characterized by the predicate nominal). Definite NP predicates are shown to be well-suited to the specifying function:
the property of inclusiveness means that the entities referred to by the
48
argument expression will be taken to represent a complete list of members,

and the implication that definite NPs are understood in relation to a shared
environment means that the set is restricted (or the type is specified) so that
the number of possible members (or instances) is small enough to be usefully listed.
In this section, I examine whether this characterization of specificational
meaning can help to explain why indefinite NPs are often restricted from
occurring as the subjects of specificational inversion sentences and whether
it can account for particular examples in which indefinite NP predicates do
allow a specificational interpretation.
3.1.
Specification and indefinite NP predicates
Following Hawkins (1991) account, indefiniteness can be defined in the

following way. While definite NPs are associated with existence and uniqueness entailments, the use of the indefinite article entails only existence
and conversationally implicates (but does not entail) exclusiveness, or nonuniqueness. For instance, (37) entails that there is a surgeon and implies
that there is more than one surgeon (or that John is not the only surgeon).
(37)
John is a surgeon
Furthermore, membership of a P-set does not always hold with indefinites.

For instance, as shown in Section 2.1, we can interpret (37) as including
John in the very general set of surgeons. Hawkins (1991: 419) notes that
indefinite NPs can be interpreted in relation to a shared set (or mutual cognitive environment) on the condition that exclusiveness is satisfied. For example, if (37) were uttered inside a hospital, the likely interpretation would
be that John is a surgeon in this hospital. However, since this implication is
easily cancelled, Hawkins suggests that P-membership is conversationally
implicated in indefinites (and conventionally implicated in definite NPs).
Hawkins account provides an explanation for why sentences with
indefinite NP predicates rarely allow a specificational interpretation. As we
have seen, indefinite NPs sometimes denote very general sets (such as surgeons). Furthermore, even when the set is understood in relation to a shared
environment (as in surgeons in this hospital), the entity referred to by the
argument expression will normally be taken to form an incomplete list of
the entities within this set (see also Declerck 1986: 30). This follows from
the exclusiveness (or non-uniqueness) implicature of indefinites. Sentences
49
with indefinite NP predicates are therefore unlikely to provide meaningful

information about a sets membership and so cannot enable a specificational interpretation.
On this account, the fact that indefinite NPs rarely occur as the subjects
of specificational inversion sentences is somewhat predictable. First, as we
saw in Section 2.2, indefinite NPs typically denote individuals (or a single
member of a set), and so are not inherently relational. Second, even when
they are construed as predicates, indefinite descriptions are unlikely to be
interpreted as uniquely characterizing the referent (see Declerck 1986: 30)
such that the entity referred to is taken to represent the complete membership of the described set. Since the NP inversion construction is associated
with specificational meaning, this explains the unacceptability of inversion
sentences, such as (38).
(38)
#A surgeon is John
However, as mentioned in Section 1.2, some indefinite descriptions can,

in certain contexts, function as the initial NP of a specificational inversion
sentence. Mikkelsen (2005) provides the following example.
(39)
A philosopher who seems to share the Kiparskys intuitions on

some factive predicates is Unger (1972), who argues
(Mikkelsen 2005: 155)
So what is it about these particular indefinite NPs that allows them to meet
the requirements of this inversion construction and enables a specificational
meaning to occur? From the analysis sketched so far, we might expect that
the kinds of indefinite NP predicate to occur in the specificational inversion
construction would be those that share most in common with definite NPs.
In what follows, I show that while these indefinite descriptions certainly do
display characteristics associated with definiteness, they also perform a
useful, additional function which is not found in inversion sentences with
definite NP subjects. I begin by discussing a previous attempt at characterizing this subset of indefinite NPs from within the inverse tradition.
3.2.
An account based on discourse requirements
As I explained in Section 1.2, the question of why only a selection of

indefinite NP predicates can occur as the subjects of specificational inver-
50
sion sentences presents an important obstacle for inverse analyses based on

movement. Mikkelsen (2005) attempts to overcome this by putting together
an account based upon discourse requirements. Mikkelsen claims that a
predicative NP will only be raised to subject position if it contains discourse-old information. She notes that, in such cases, the predicative NP is
interpreted as the topic of the sentence; the preference for the topic to be
in subject position overrides the canonical word order, resulting in inversion (Mikkelsen 2005: 163).
On this basis, Mikkelsen explains the lack of indefinite NP predicates
that qualify for inversion as a consequence of the association between indefinite NPs and new information. For instance, while the definite article is
often said to have an anaphoric function, indicating that the entity can be
located within the previous discourse, the indefinite article is instead used a
means of introducing new entities into the discourse, shown in (40).
(40)
Mary saw a movie last week. The movie was not very interesting.
(Abbott 2010: 133)
As a result, indefinite NP predicates rarely meet the criterion for a verified

topic, and so do not qualify for the role of subject.
Mikkelsens account predicts that all indefinite NPs which are found in
the initial position of a specificational inversion sentence will contain
discourse-old information. This prediction is borne out. For instance, Mikkelsen notes that (39) is an extract from a text discussing the judgments of
Kiparsky and Kiparsky. The discourse-old element, the Kiparskys intuitions, therefore links the indefinite NP to the preceding discourse, allowing
it to function as topic. Mikkelsen (2005: 158) suggest that unmodified
indefinites, shown in (38) above, cannot meet this criterion because they do
not contain any material that could provide this kind of link. She notes that
the only way for unmodified NPs to function as topic is for the entire NP to
be discourse-old; in such cases, the indefinite article would be infelicitous
and the NP would have to be marked as definite.
On Mikkelsens account then, indefinite subjects of specificational inversion sentences have to meet two competing requirements: they must
contain enough discourse-old material to qualify for the role of topic while
still satisfying the Novelty Condition associated with indefinites (cf.
Heim 1982). For instance, while the indefinite subject in (39) contains discourse-old information, no philosopher who shares the Kiparskys intuitions on some factive predicates has been mentioned before and so it can
be marked as indefinite (Mikkelsen 2005: 155).
51
Mikkelsens (2005) account makes a lot of sense, since it reflects Birners (1994, 1996) finding that inversion is sensitive to discourse status (see
Section 2.2). However, discourse factors alone cannot fully explain the
restrictions on indefinite specificational subjects. For instance, Mikkelsen
provides examples, such as (41), where the predicative NP is discourse-old
and yet is felicitously marked as indefinite. Since these sentences meet
Mikkelsens criteria, they should qualify for inversion. However, as shown
in (42), this is not the case. Mikkelsen (2005: 159) therefore concludes that
her account, as it stands, cannot be the whole story.
(41)
Bill is a doctor. John is a doctor (too).
(42)
Bill is a doctor. #A doctor is John (too).

While I agree with Mikkelsen (2005) that discourse considerations are

certainly important for inversion constructions, I suggest that the particular
discourse status of indefinite specificational subjects follows from their
peculiar function in specificational sentences. I argue that these indefinite
NPs assert the existence of a non-exclusive, restricted set of entities as relevant to the shared discourse environment. In specificational inversion sentences then, indefinite NPs share many of the same properties of definite
NPs while still providing a unique discourse function. In what follows, I
advance this account, making use of Hawkins (1978, 1991) analysis of
definiteness and indefiniteness as well as Princes (1981, 1992) taxonomy
of discourse familiarity (also adopted by Birner 1994, 1996).
3.3.
An account based on definiteness
When appearing in the specificational inversion construction, definite

NP subjects are typically discourse-old. For instance, the use of the definite
article in (43) suggests that both the speaker and the hearer already know
that some entity (or singleton set of entities) exists which satisfies the description psychologist. This information may be explicitly evoked, involving the previous mention of a psychologist among the pragmatic set of
people talked about, or it may be inferable from the discourse context.
(43)
The psychologist is Dr. Hugh Beale.
52
For instance, if (43) were part of a discourse about a particular hospital, the
existence of a psychologist within this environment could be inferred from
our shared script or frame of hospitals i.e. from what we know about
hospitals. As Birner (1996: 95) notes, inferable information patterns with
evoked information in inversion sentences and can therefore be collapsed
into a single category (i.e. discourse-old information).
In contrast, indefinite specificational subjects will not be felicitous if
they are discourse-old; that is, if the information they express is already
present in the context of a particular speech event. To illustrate this point,
consider the example in (44). Given the right context, this could be an
acceptable specificational sentence. However, if the indefinite NP is explicitly evoked in the previous discourse, (44) becomes unacceptable, shown
in (45) and (46).
(44)
A psychologist who works at St. Eligius is Dr. Hugh Beale.
(45)
There is a psychologist at St. Eligius. #A psychologist who works at

St. Eligius is Dr. Hugh Beale.
(46)
There are several psychologists at St. Eligius. #A psychologist who

works at St. Eligius is Dr. Hugh Beale.
In the discourse context of (45), where there is no implication that there is

more than one psychologist at St. Eligius, the definite article would be
preferred, shown in (47).13 Similarly, while the exclusiveness condition of
indefinites is satisfied in (46), the numeral one is preferred, shown in (48).
(47)
There is a psychologist at St. Eligius. The psychologist is Dr. Hugh

Beale.
(48)
There are several psychologists at St. Eligius. One (of the) psychologist(s) working at St. Eligius is Dr. Hugh Beale.
In (48), one performs a partitive function in relation to the evoked set of

psychologists at St. Eligius. Since the existence of several psychologists entails the existence of one psychologist, this NP denotes a smaller subset of a
previously evoked set, the existence of which is entirely inferable from the
13. Although the first sentence of (45) and (47) contains an indefinite NP, it does
not imply exclusiveness (more than one). As Hawkins (1978: 222223) comments, the exclusiveness condition of indefinite noun phrases disappears when
they occur in there be sentences.
53
discourse context (see Prince 1981: 236). As shown in (46), the indefinite
article a does not have this function in specificational sentences; it is not
used to provide an already familiar or presupposed description or to classify
an entity as a single member of some previously evoked set of entities.
In the specificational inversion construction, descriptions introduced by
a therefore perform a different function from those introduced by the or
even one. Rather than indicating that the existence of a set of entities that
satisfy the description is already known or knowable from the discourse
context, the indefinite article is used to establish the existence of this set
and to position this information in relation to the discourse context. For
instance, as was shown above, the initial NP in (39), repeated here as (49),
provides a description which is brand-new to the discourse, but which is
also explicitly related to some element within the discourse context (in this
case, the Kiparskys). Assuming Princes (1981) taxonomy then, these
indefinite specificational subjects are brand-new anchored.
(49)
A philosopher who seems to share the Kiparskys intuitions on

some factive predicates is Unger (1972), who argues
In (49), the relative clause provides the anchor, or link, to the preceding
discourse. For Hawkins (1978, 1991), this is an establishing relative clause.
It adds to, or extends, the mutual cognitive environment in a way that is
relevant to and compatible with the interlocutors existing knowledge
(Hawkins 1991: 411). Fox and Thompson (1990) discuss this as a form of
grounding; the NP is grounded in the conversational space via its relationship to a referent which is clearly given in the immediate context.
From the account outlined in Section 2 above, we might expect that
indefinite NP predicates which contain establishing relative clauses would
be better suited to the specificational function, since they exhibit many of
the same properties as definite NPs. As we have seen, these indefinites are
explicitly anchored to the speech event. This is what enables the indefinite
NP in (49) to meet the discourse requirement of inversion, such that the
postcopular NP is newer to the discourse than the precopular NP (see
Birner 1994, 1996). Furthermore, since anchoring takes place via restrictive
modification, these indefinite NPs denote highly restricted sets; in other
words, the type is specified to a greater degree, limiting the number of
possible instances of that type (see Section 2.3).
Finally, as Hawkins (1978: 225) points out, the exclusiveness condition
is no longer operative in indefinite NPs with establishing relative clauses.
54
The indefinite NP in (49) is therefore noncommittal with respect to inclusiveness and exclusiveness (which are associated with definiteness and
indefiniteness, respectively (see Section 2.1)). In (49) then, type specification and non-exclusiveness result in the suggestion that the number of philosophers who share the Kiparskys intuitions on factive predicates is limited and there is no implication that there is in fact more than one such
philosopher. It is therefore possible that the description philosopher who
seems to share the Kiparskys intuitions on factive predicates characterizes
Unger uniquely. This indefinite noun phrase therefore enables a specificational reading for (49), such that the postcopular Unger can be taken as
representing the sets (potentially complete) membership.
Indefinite NPs with establishing relatives are therefore able to occur as
the subjects of specificational inversion sentences because they bridge the
gap between definiteness and indefiniteness. The account of specificational
meaning outlined in Section 2.1 is therefore able to explain what lies behind Declercks (1988: 19) observation that indefinite specificational subjects invariably express new information, but involve modifiers expressing
old information. In particular, it suggests that the fundamental property of
these particular indefinite NPs is that they do not rule out an inclusiveness
interpretation. While I have argued that, in this construction, indefinite NPs
introduced by the indefinite article often function differently from those
introduced by one, Declerck (1988: 20) does not make this distinction; he
claims that all types of indefinite specificational subject indicate exclusiveness and that their use is motivated by the semantic difference between
these NPs and the corresponding definites (emphasis added).
Of course, it is not just indefinite NPs with establishing relative clauses
that can occur in the specificational inversion construction, and in fact it is
not always the modifying information that contains discourse-old material.
For instance, the specificational sentence in (50) has an indefinite NP subject displaying adjectival modification. Here, the head noun psychologist
provides the link to the previous discourse and the adjective phrase especially talented expresses new information.
(50)
There are several psychologists at St. Eligius. An especially

talented psychologist is Dr. Hugh Beale.
As in (49) then, the indefinite NP an especially talented psychologist has

the information status brand-new anchored. We have seen above, in (48),
that the existence of several psychologists entails the existence of a (or
55
one) psychologist. However, the existence of an especially talented psychologist cannot be inferred from the preceding discourse.
The indefinite article therefore functions not to indicate that the existence of some described set of entities is already known or knowable from
the discourse context, but to actually assert the existence of this set and to
position this information in relation to the discourse context. While in (49),
the establishing relative also serves to extend the discourse environment,
the indefinite NP in (50) carries with it the implicature of P-membership,
such that the set psychologist is understood in relation to the discourse-old
set of psychologists at St. Eligius (see above and Hawkins 1991). This set is
further restricted, or specified, by adjectival modification to the extent that
it does not rule out a uniqueness (or inclusiveness) interpretation such that
Dr. Hugh Beale is the most talented psychologist at St. Eligius.
This account of indefinite specificational subjects seems to be on the
right lines, especially when we consider some less felicitous examples. For
instance, the indefinite NP in (51) involves adjectival modification and has
the same discourse status as that in (50). Nevertheless, there is the sense
that the specificational sentence in (51) is not as good as that in (50).
(51)
There are several psychologists at St. Eligius. ?A talented psychologist is Dr. Hugh Beale.
This is to be expected from the account sketched here. As we saw above,

the adverb especially, meaning above all, creates a uniqueness interpretation for (50). Without this adverb, it is more difficult to obtain a nonexclusive reading. Since talented is a gradable adjective, an individual can
be talented to a greater or lesser degree and the dividing line between what
counts as talented and what does not is difficult to determine. Therefore,
we might expect there to be more than one talented psychologist at St.
Eligius (even if some are considered to be less talented than others).
On this account then, we anticipate that indefinite NPs will be better
suited to the specifying function if they contain lots of modifying information (providing a more restricted type specification) and if they contain
modifiers which lexically imply uniqueness. Gradient acceptability judgments are therefore built into this account, since felicitousness is dependent
upon the degree to which the indefinite NP can be considered as having
properties associated more with definiteness than indefiniteness. As Mikkelsen (2005: 159) notes, a purely discourse-based account cannot explain
56
why the content and form of modification should be a factor in determining the felicity of indefinite specificational subjects.14
4.
Summarizing and extending the account
In this section, I summarize the account of specificational NP be NP

sentences developed so far and consider how this draws from, and advances
on, the existing literature. I then go on to examine the implications for other
types of specificational copular construction, building up a family of related
constructions. In particular, I outline the analyses of th-clefts, wh-clefts and
all-clefts that are most consistent with the account of specificational meaning developed here, before going on to advance an analysis of the it-cleft as
a specificational copular construction in Chapter 4.
4.1.
An overview of specificational NP be NP sentences
On the account developed so far, specificational meaning in NP be NP

sentences is quite literally the inverse of predication; that is, it derives from
14. Since additional material is often said to facilitate a specific reading for indefinite noun phrases (see Fodor and Sag 1982), one could argue that the felicitousness of indefinite specificational subjects is dependent on a specificity condition similar to that found in sentences containing individual-level predicates
(see Heycock and Kroch 2002: 112). However, the kinds of indefinite NPs that
make acceptable specificational subjects do not readily occur with predicates
expressing permanent properties without inducing a generic reading, as shown
in (viii) compared with (ix).
(vii) A doctor who finds cures for rare diseases is Jonathan Tuttle.
(viii) A doctor who finds cures for rare diseases is intelligent. (generic reading)
(ix) A certain doctor we both know is intelligent.
(specific reading)
Furthermore, indefinite specificational subjects do not pronominalize with gender-specific tag questions, as we would expect on a referential reading, shown
in (x) and (xi).
(x)
A certain doctor we both know is intelligent, isnt he?
(xi) *A doctor who finds cures for rare diseases is Jonathan Tuttle, isnt he?
57
the same nominal predication relation of class inclusion but involves interpreting this relation from the opposite perspective, as listing the membership of a set rather than attributing a property to a referent. In Section 2.1,
definite NP predicates were shown to be especially well-suited to enabling
a specificational interpretation because they are always understood in relation to the speech event (and so denote restricted sets) and because they are
associated with inclusiveness. For sentences with definite NP predicates
then, the referring expression is understood to provide a complete, exhaustive list of the members which constitute the restricted set.
The specificational interpretation also relies on a particular information
structure, in which the referring expression is in focus. From this, it follows
that specificational sentences will invariably meet the discourse requirements of inversion, such that the precopular element is not newer to the
discourse than the postcopular element (which itself conforms to highly
general information structural tendencies). This, along with the fact that the
NP inversion construction provides an unambiguous, structural way of distinguishing between specificational and predicational meaning, helps to explain the association between specification and inversion (see Section 2.2).
In Section 2.3, this account of specificational NP be NP sentences was
situated within cognitive and constructional frameworks. Here, it was argued that the class inclusion relation of nominal predication does not arise
via a special use of classifying be (contra Croft 1991), nor is it implied
indirectly as a result of an identity relation between an instance and an arbitrary, nonspecific instance of a type (contra Langacker 1991). Instead,
predicate nominals (definite and indefinite) were analysed as designating a
type and as profiling a relation between that type and the schematic instance of that type, which is then elaborated by an argument expression. On
this account, predicate nominals are properly relational, expressing an
instantiation (or class inclusion) relation between an instance and a type.
In Section 3, we saw that by focusing on the question of what constitutes specificational meaning, rather than on the conditions affecting syntactic movement operations, we can provide something of an explanation
for the behaviour of, and restrictions on, indefinite specificational subjects.
As was shown in Section 3.2, these issues have not been satisfactorily
addressed in movement-based inverse accounts. It was argued, in Section
3.3, that the indefinite article performs a useful function in specificational
inversion sentences and so provides a contrast with the use of the definite
article. First, it allows the speaker to introduce a brand-new set description
into the communication space, with the proviso that it is anchored to the
discourse in some way. Second, it means that the speaker can be noncom-
58
mittal or evasive with respect to inclusiveness. In other words, the referring

expression can plausibly be taken as representing a sets membership even
though it may not in actuality constitute an exhaustive list of members.
Despite providing this unique and useful function, we saw that the
felicitousness of indefinite specificational subjects is nevertheless governed
by their similarity to definite NPs, which we have said contain properties
well-suited to specificational meaning. In particular, the indefinite NP
predicate must exhibit, or at least not be at odds with, some characteristics
associated more with definiteness than indefiniteness. For instance, indefinite NPs will be deemed more acceptable as the subjects of specificational
inversion sentences if they are understood in relation to the communication
space and if they contain modifying information which specifies the type to
such a degree that exclusiveness is not implied.
4.2.
Positioning this account in relation to the literature
The account of specificational NP be NP sentences built up throughout this

chapter is based on the premise that such sentences involve nominal predication (or class inclusion) rather than equation and so captures the same
intuitions as the inverse analyses of specificational sentences outlined in
Section 1.2. However, while the inverse approach focuses on the derivational relationship between specificational inversion sentences and their
predicational (or non-inverted) counterparts, the account developed here
assumes a non-derivational, monostratal model of language. Thus, pairs of
inverse and non-inverted NP be NP sentences are not derived from a common underlying structure, but are instances of two separate constructions
which are related by their position within the hierarchical taxonomy of
grammatical knowledge.
As we saw in Section 2.1, certain instances of the predicate nominal
construction allow a specificational interpretation. All such sentences share
the same semantic relation (of class inclusion) and exhibit the same syntactic structure (see also footnote 10). I suggest that specificational sentences
with postcopular predicative NPs (commonly referred to as reverse specificational copular sentences) form a subtype of the predicate nominal construction, labelled the specificational non-inversion construction in Figure
4. The psychological reality of this sub-construction can be argued for on
the grounds that such sentences have a fixed information structure which is
not found in the more general predicate nominal construction; that is, the
subject must be in focus. Furthermore, these sentences have a distinct,
59
specifying function which requires that the predicate denotes a restricted set
or designates a highly specified type a characteristic which falls out from
the semantics and pragmatics of definite NP predicates (see Section 2.1).
Specificational inversion sentences, on the other hand, involve the same
class inclusion relation as sentences with postcopular predicate nominals,
but have different structural properties (see Section 2.2).15 We could argue
that speakers recognize the parallelism between the specificational inversion construction and the predicate nominal construction, and so form an
abstraction, shown as predicate nominal semantics in Figure 4. Although
this construction mostly comprises semantic information (outlined in Section 2.3) and does not specify a particular linear order, predicate nominal
semantics still constitutes a form-meaning pairing because the type specification must be provided by a nominal element. As I explain in Chapter 5,
the same is not true for the element characterizing the instance of this type,
as a variety of phrasal categories can be used to perform a referring function. The construction therefore has a highly abstract form, specifying for
two phrasal elements (XP and NP) mediated by the copular verb be.
The specificational inversion construction further stipulates that the type
must be specified (or the set restricted) to the extent that a specificational
interpretation can be achieved. The construction involves mismatch, in that
the grammatical subject is the semantic predicate and the semantic argument (a referring expression) is in the complement position. In addition, the
specificational inversion construction has a fixed information structure,
with focus placed invariably on the postcopular XP, which characterizes the
instantiation of the type (or the membership of the set). As we saw in
Section 2.2, the specificational inversion construction shares some of these
properties with other kinds of be inversion construction. Despite differences in their syntactic behaviour, these inversion sentences exhibit a similar mismatch between form and function (with the logical subject occurring
in postverbal position and a predicative element appearing in clause-initial
position) in addition to sharing the same discourse requirements. The family of be inversion constructions may therefore provide local support (and
motivation) for the symbolic and information structural properties of the
specificational inversion construction, as shown in Figure 4.
15. As noted in Section 2.2, it is conceivable that the specificational inversion construction shares the same syntactic (subject be complement) structure as that
found in equative sentences one which is underspecified for syntactic predication (see footnote 10). However, since I have not explored the syntax of these
constructions in any detail, this is not represented in Figure 4.
60
Predicate nominal semantics
be inversion construction
Specificational
inversion construction
Predicate nominal
construction
Specificational noninversion construction
Figure 4. A constructional taxonomy of predicate nominal inheritance relations
As a result of its theoretical differences from other inverse analyses, the

account developed over this chapter focuses not on identifying a system of
rules for governing movement operations but on characterizing the relationship between specificational and predicational meaning. By recognizing the
role of definiteness in enabling the specificational interpretation of NP be
NP sentences, this account advances the semantic (as opposed to the structural) side of the inverse approach. In doing so, it owes much to Higgins
(1979) characterization of specificational sentences as functioning like lists
(see Section 1.3). As I explained in Section 2.1, Higgins sometimes gets
very close to depicting specificational meaning as a class inclusion relation.
For instance, he notes that in specificational (inversion) sentences, the postcopular NP says what constitutes or makes up the object referred to by the
subject noun phrase (Higgins 1979: 150). In other words, a specificational
interpretation involves listing the membership of a set or class, rather than
using class inclusion to attribute a property to a referent.
The analysis of specificational meaning developed here also has a number of points in common with the account of Blom and Daalder (1977).
These authors suggest (as I do) that specificational sentences involve a
class inclusion relation. However, for Blom and Daalder, this is tantamount
to a hyponymy relation. Their argument runs as follows. Predicational NP
be NP sentences involve a class-membership relation in which the subject
referent is a hyponym of the superordinate category described by the predicate nominal. Specificational (inversion) sentences have the reverse alignment, with the subject denoting a more general concept than the postcopular NP.
Declerck (1988: 92) criticizes this account, claiming that although predicational NP be NP sentences often do express a class-membership relation,
61
there are subtypes of predicational sentences where no idea of classinclusion (or class-membership) appears to be present. In evidence of this,
he cites examples with definite NP predicates, such as (52).
(52)
John is the best musician in town
(Declerck 1988: 92)
Declerck (1988: 93) also argues against Blom and Daalders claim that
specificational copular sentences involve a hyponymy relation; he observes
that There is no difference in generality between the variable NP and the
value NP. Declerck suggests that this is the reason why the two NPs are
reversible because they exhibit no difference in specificity.16
The account of specificational meaning developed here provides a solution to the discrepancy between these two arguments. On the one hand, I
assume (with Blom and Daalder 1977) that specificational inversion sentences involve class inclusion relations and are therefore, in some sense, the
reverse of predicational NP be NP sentences. However, I also agree with
Declerck (1988: 93) that specificational sentences are not characterized by
a difference in generality. For sentences with definite NP predicates, such
as (52) above, there is, as Declerck comments, no difference in the generality of the two NPs, since both are inherently singular. However, what Declerck does not acknowledge is that the postcopular NP can nevertheless be
interpreted as denoting a (highly restricted, all-inclusive) set or class. As we
have seen, this property of definite NPs is what enables these sentences to
16. For Declerck (1988: 47), the asymmetry of specificational sentences instead
stems from the fact that whereas the value NP is strongly referring, the variable NP is weakly referring or attributive. In other words, the variable NP is
used to refer to an unknown individual whoever or whatever is the so-andso (Donnellan 1966: 285). However, Mikkelsen (2005: 89) shows that in tag
questions, subjects with unknown reference differ from the subjects of specificational inversion sentences in that they do not pronominalize as it, but with
the gendered pronouns she/he, as shown in (xii) (see also Declerck 1988: 56).
(xii) The murderer (whoever he is) must be insane, mustnt he/*it?
(xiii) The murderer is Andy Schroeder, isnt it?
Furthermore, as Higgins (1979: 269) observes, the concept attributive does not
accurately characterize the function of specificational sentences. On an attributive use, the speaker does not know the identity of the referent; however, specificational sentences are instead used when the speaker does know (and intends
to inform the hearer of) the referent to which the description applies.
62
acquire both predicational (x is a member of the set Y) and specificational

(x makes up the complete membership of the set Y) interpretations.
4.3.
Other specificational copular constructions
This account of specificational meaning can be extended to other kinds of

copular sentence. In this section, I outline comparable analyses of various
different types of pseudocleft construction before going on to examine the
implications for it-clefts in Chapter 4. Pseudocleft constructions include thclefts, wh-clefts, and all-clefts, shown in (53), (54) and (55).
(53)
The thing that I like best is grape soda
[th-cleft]
(54)
[wh-cleft]
(55)
All he drinks is grape soda
[all-cleft]
The term pseudocleft stands alongside the term cleft (used as a synonym for
it-cleft) to indicate that while it-clefts can almost always be paraphrased by
noncopular sentences, pseudoclefts usually, but do not always, have simple
sentence counterparts (see Ward, Birner and Huddleston 2002: 1423). For
instance, (53) and (54) correspond in meaning to I like grape soda best and
(55) can be paraphrased as he only drinks grape soda. However, there is no
corresponding noncleft for (56) because *I like about it that its so sweet is
ungrammatical.
(56)
What I like about it is that its so sweet
[wh-cleft]
The relationship between pseudoclefts and simple noncopular sentences

is therefore less systematic than that existing between it-clefts and their
more basic counterparts. As a result, pseudoclefts are not as well-suited to a
derivational (movement-based) account. Nevertheless, Ward, Birner, and
Huddleston (2002: 1423) maintain that clefts and pseudoclefts share the
same information packaging function of dividing the message into two
parts. Consequently, pseudoclefts are often discussed in relation to more
basic sentences. For example, Lambrecht (2001: 469) analyses their introductory elements (the thing that, what) and the copula as semantically inert,
suggesting that while pseudoclefts are not derived, they nevertheless have
the same semantic structure as simple noncopular sentences. In what
follows, I present analyses which examine pseudoclefts not in relation to
63
their noncopular counterparts but as fully-fledged copular constructions.

The aim is that by the end of this chapter we will have established a unified
analysis of the family of specificational copular sentences which can be
usefully employed in the analysis of it-clefts in Chapter 4.
4.3.1.
Th-clefts as specificational copular sentences
Collins (1991b: 483) defines th-clefts as pseudoclefts introduced by the in

conjunction with the proform equivalents of the English interrogatives
(thing, one, place, time, reason, way). However, despite this formal distinction, th-clefts are essentially just specificational NP be NP sentences
introduced by definite NPs. As a result, this sentence type can be accounted
for using the same analysis as that proposed in Section 2 above. On this
account, the th-cleft in (57) is an instance of the specificational inversion
construction, containing a definite NP predicate in subject position. Therefore, the only thing that separates this th-cleft from the NP be NP sentence
in (58) is the form of the initial NP.
(57)
The one who wants to be a concert pianist is Charles
(58)
The aspiring pianist is Charles
The main difference between (57) and (58) is that the th-cleft in (57)
contains an obligatory restrictive relative clause. As shown in (59), the sentence becomes unacceptable when this restrictive information is removed
(see Lambrecht 2001: 469). Of course, the initial NP in (58) can also be
restrictively modified, shown in (60). However, here, the reduced relative
provides the P-set (this outfit), within which the existence and uniqueness
of the singleton set aspiring pianist holds (see Hawkins 1991). Since this Pset is already accessible from the immediate context (as indicated by the
use of the definite article) it does not have to be expressed overtly, shown
in (58).
(59)
#The one is Charles
(60)
The aspiring pianist in this outfit is Charles
In contrast, the relative clause in (57) provides the singleton set information
(wannabe concert pianist) that distinguishes Charles from the other entities
that make up the larger P-set. Here, the head noun one performs a partitive
64
function in relation to the contextual P-set, such that Charles is specified as

the one (out of the people in this outfit) who wants to be a concert pianist.
In (59) then, the initial NP does not contain any information that uniquely
describes Charles as distinct from the other members of the P-set and so
does not constitute an acceptable specificational sentence.
The obligatory presence of the restrictive relative explains why these
particular NP be NP sentences are characterized as clefts (or pseudoclefts).
For many authors, the concept cleft is tied to the correspondence of such
sentences to noncopular paraphrases. In (57) above, the relative clause
information x wants to be a concert pianist forms the basis of a noncopular
counterpart, for which the postcopular Charles functions as the missing
subject. However, in (60), it is the head noun, rather than the relative
clause, that provides the distinguishing information. As a result, Charles is
in this outfit does not adequately capture the meaning of (60) and so this
NP be NP sentence cannot be characterized as cleft.
What distinguishes th-clefts from other NP be NP sentences then, is how
the informational content of the initial predicative NP is distributed; the
correspondence of th-clefts to noncopular sentences is secondary to this and
is dependent upon it. However, it is not clear to me whether this small
formal distinction is recognized by speakers; that is, whether th-clefts make
up a separate construction from noncleft NP be NP sentences. The answer
depends on the psychological reality of the concept (or rather, the construction) cleft. For Collins (1991b: 484) th-clefts include only those examples
that can be analysed as in any sense a cleaving of a simple sentence.
Since I view cleft sentences as fully-fledged copular constructions, the term
cleft has no theoretical significance on my account. Therefore, it may be the
case that the th-cleft examples are simply instances of the specificational
inversion construction and can be accounted for in the exact same way as
other specificational NP be NP sentences.
4.3.2.
Wh-clefts as specificational copular sentences
The analysis of specificational NP be NP sentences outlined in Section 2

can also be extended to incorporate wh-clefts. On this account, the precopular constituent of a wh-cleft is a fused relative. Fused relatives function as
noun phrases in which the head is incorporated into the relative lexeme.
Therefore, what I like best in (61) corresponds to a definite NP with an
integrated relative clause, such as the x [I like x best] or the x such that I
like x best (see Ward, Birner, and Huddleston 2002: 1420).
(61)
65
Much of the literature on wh-clefts involves the discussion of connectivity

(for a comprehensive overview, see den Dikken 2005). In what follows, I
show that while some of the connectivity evidence seems to support the
treatment of wh-clefts as specificational NP be NP sentences, other connectivity data is problematic for this type of analysis.
Most connectivity effects are found not only in wh-clefts, but in all
types of specificational sentence, and so support a unified analysis of such
constructions. For example, both the wh-cleft in (62) and the NP be NP
sentence in (63) exhibit reflexive connectivity. Here, the postcopular reflexive yourself is governed by the pronoun you embedded inside the restrictive
relative clause.
(62)
What you should try instead is shaving yourself in the evenings
(63)
The approach you should try instead is shaving yourself in the

evenings
(examples from Higgins 1979: 57)
In generative and transformational theories of grammar, binding effects

such as this represent clause-internal structural relations between syntactic
elements. Consequently, connectivity evidence has sometimes been used to
support analyses of wh-clefts in which the postcopular constituent is either
extracted from within, or coindexed with a gap inside, the embedded relative clause. On such accounts, binding in wh-clefts works in much the same
way as it does in simple noncopular sentences, such as (64) (see Hankamer
1974).
(64)
You should try shaving yourself in the evenings
However, since noncleft NP be NP sentences like (63) cannot be derived

from simple noncopular sentences, Higgins (1979) suggests that connectivity in specificational sentences requires an explanation which is independent from syntactic c-command. Higgins (1979) argues instead for an
interpretive account, suggesting that connectivity effects are related to, and
in fact result from, the meaning of specification. More developed analyses
taking a semantic approach to connectivity are found in Jacobson (1994),
Sharvit (1999, 2003) and Heller (2002, 2005). For these authors, binding is
66
a consequence of semantic composition and depends upon the semantic

properties of what/the and be.17
It is beyond the scope of this book to develop an account of connectivity
in specificational sentences. For our purposes, it is sufficient to note that
the shared behaviour of wh-clefts and other specificational sentences with
respect to connectivity supports a unified analysis of these constructions.
However, there are some types of connectivity that are specific to the whcleft, such as NPI connectivity, shown in (65). Since the licensing of the
negative polarity item any breaks down in the reverse wh-cleft of (66),
such examples can be analysed as specificational NP be NP sentences with
postcopular fused relatives. However, den Dikken, Meinunger and Wilder
(2000) argue that canonical wh-clefts like (65) require a separate analysis.
(65)
What John didnt buy was any wine
(66)
* Any wine was what John didnt buy
These authors argue for an ellipsis approach where the licensing negation ccommands the NPI at S-structure within an elliptical IP, shown in (67).
They go on to suggest that this type of wh-cleft patterns with questionanswer pairs, and analyse the wh-clause as an interrogative rather than a
fused relative.
(67)
What John didnt buy was [he didnt buy any wine]
(den Dikken, Meinunger and Wilder 2000: 50)
For den Dikken, Meinunger and Wilder (2000) then, not all types of whcleft can be reduced to specificational NP be NP sentences. However, the
question-in-disguise theory of wh-clefts is not without its problems. For
one thing, it stipulates enormous amounts of elided material that are not
always supported by agreement factors. In addition, NPI connectivity can
sometimes feature in th-clefts, which contain what are undisputably DPs
containing headed relative clauses in their pre-copular position (den Dikken, Meinunger and Wilder 2000: 81).
(68)
?The only thing John didnt buy was any wine
17. However, since many of these accounts depend upon a be-of-identity, they
are somewhat at odds both with analyses that view specificational sentences as
instances of nominal predication and with evidence that specificational meaning (whether class inclusion or identity) is actually independent of be.
67
The only way for these authors to account for examples like (68) is to
suggest that the precopular DP is a concealed question, whether this is
realized syntactically at an underlying level (requiring further ellipsis) or
purely in terms of semantic function (see also Schlenker 2003).
Exactly how NPI connectivity works in specificational sentences is not
clear to me. However, I am skeptical of an approach that provides superficially similar and functionally analogous sentence types with such different
and unrelated underlying structures. As Schlenker (2003: 212) observes, it
is not entirely impossible for a negative polarity item to be licensed in a
reverse wh-cleft.
(69)
To buy any wine was what he refused to do (Schlenker 2003: 212)
As a result, I maintain a fused relative analysis for all types of wh-cleft and
will assume that the gradient acceptability of foci containing NPIs is evidence that wh-clefts and reverse wh-clefts form distinct constructions that
have developed their own idiosyncratic properties (see also Chapter 8).18
As shown in Figure 5, the wh-cleft construction is an instance of, and
inherits from, the specificational inversion construction. Thereverse whcleft construction is a subtype of the specificational non-inversion construction. Like other reverse specificational sentences, reverse wh-clefts are
instances of the predicate nominal construction. The wh-cleft and the
reverse wh-cleft are positioned at a lower level in the constructional
hierarchy, which reflects the fact that they are less schematic (or general)
than the higher level constructions. For instance, in addition to their construction-specific behaviour, wh-clefts and reverse wh-clefts also have a
more substantive (fixed) form. While the specificational inversion and noninversion schemas involve nominal predication, the wh-cleft constructions
further specify for a fused relative (selecting from a limited range of whrelatives) in the semantic predicative position.
18. I suspect that the restrictions on any and (to a lesser extent) some may be
related to their extreme non-specificity. For instance, unlike in (66), the NPI of
(69) is embedded inside a specific action to buy any wine. Likewise, (68)
becomes less acceptable without the adverb only, which establishes a contrast
between wine (as a specific foodstuff) and the other produce that John did buy.
Therefore, the wh-cleft may differ from other specificational sentences only in
that it is unusually well-suited to occurring with non-specific, abstract foci. I
provide more evidence to this effect in Chapter 8, where I compare it-clefts to
wh-clefts.
68
Predicate nominal
construction
Specificational
Wh-cleft
Reverse wh-cleft
Figure 5. Incorporating wh-clefts into a family of specificational constructions
4.3.3.
All-clefts as specificational copular sentences
Like th-clefts and wh-clefts, all-clefts, such as (70), are a variety of pseudocleft (see Collins 1991b: 483). As with th-clefts, the precopular constituent
here is clearly nominal and so can be easily incorporated into the analysis
of specificational NP be NP sentences outlined in Section 2.
(70)
All that he asked for was a grape soda
However, as Traugott (2008) notes, all-clefts have an additional function,

producing a below expectation reading which is also found in th-clefts
containing the adverb only, as in (71).
(71)
The only thing he asked for was a grape soda
The all of all-clefts therefore differs from the ordinary use of this word. In
(72), all has the opposite effect: that of emphasizing the number of things
hed asked for. While all in (72) corresponds in meaning to the quantifier
everything, we cannot replace the all of (70) with this word, shown in (73).
(72)
And after all that hed asked for
(73)
*Everything that he asked for was a grape soda
69
Therefore, rather than quantifying over individuals (all things), the all of
all-clefts seems to be used to emphasize the totality of the set, as inclusive
of all and only those entities to which the description applies.
Since all-clefts contain this unvarying substantive element, which furthermore fulfils a somewhat idiosyncratic function in these sentences, they
can be treated as instances of a distinct all-cleft construction, which is itself
a subtype of the specificational inversion construction (see Figure 6). Like
other kinds of specificational copular sentence, the all-cleft has a counterpart which does not involve inversion, shown in (74). These reverse allclefts form a subcategory of specificational non-inversion sentence. For a
more detailed analysis of all-clefts, see Tognini-Bonelli (1992).
(74)
And that is all Im going to say about the matter

Predicate nominal
construction
Specificational
Wh-cleft
All-cleft

Reverse
wh-cleft
Reverse
all-cleft
Figure 6. Incorporating pseudoclefts into a family of specificational constructions
4.3.4.
A family of specificational copular sentences
Throughout Section 4.3, I have outlined accounts of th-clefts, wh-clefts and

all-clefts which are consistent with their treatment as specificational copular sentences involving nominal predication. In the case of th-clefts, these
examples can be properly integrated as instances of the specificational
(non-)inversion construction (see Section 4.3.1). In contrast, wh-clefts and
all-clefts have been shown to form less abstract generalizations at an
intermediary position in the inheritance hierarchy, shown in Figure 6.
Regardless, it is noticeable that although these sentence types are referred
70
to cumulatively as pseudoclefts, no overarching pseudocleft construction is

posited in our taxonomy of specificational copular sentences. As I explained above, this transformational terminology refers to the relationship
between such sentences and their noncopular paraphrases and so has no
theoretical significance on an account which treats them as fully-fledged
copular constructions. From this perspective, the only commonality unique
to these sentence-types is a predicative noun phrase containing a semantically general head noun and (therefore) an obligatory relative clause. It is
not clear to me that this is a sufficient basis for the claim that the pseudocleft category has any psychological reality.
Chapter 4
It-clefts as specificational copular sentences
1.
The English it-cleft
In the introductory chapter, it was noted that English it-clefts are comprised
of four main elements: an introductory it, a form of the copular verb be, a
postcopular phrasal element and a sentence-final clause. However, our understanding of how these components function and relate to one another
differs from one analysis to the other. Consequently, it is common in the
cleft literature for these structural subparts to be labelled using construction-specific, theory neutral terminology. Following Hedberg (1990), I
make use of the terms shown in (1) throughout my discussion.
(1)
Cleft pronoun + copula + clefted constituent + cleft clause
Also in Chapter 1, we saw that there are two main types of analysis that
have been proposed to account for the it-clefts unusual structure. One of
these hinges on the correspondence between it-clefts and simple noncopular
sentences. On this account, the cleft clause is predicated of, or is in some
other way related to, the clefted constituent, while the cleft pronoun is
semantically expletive. The other type of analysis stems from the treatment
of it-clefts as specificational copular sentences. From this perspective, the
cleft clause is in an extraposed position and is related in some way to the
cleft pronoun, which may play an important role in the interpretation of the
sentence.
In what follows, I outline my own version of the latter type of analysis,
based upon the account of specificational copular sentences developed in
Chapter 3. I then go on, in the remaining sections of this chapter, to discuss
my analysis in relation to the cleft literature. In Section 2, I pit my analysis
against an expletive account, before comparing it to other versions of the
extraposition analysis in Section 3. Finally, in Section 4, I show how my
analysis of it-clefts differs considerably from the alternative constructional
accounts proposed in the literature.
72
1.1.
A discontinuous constituent account of it-clefts
In Chapter 3, I advanced an analysis of specificational copular sentences as

involving nominal predication and showed that definiteness is an important
concept in the creation of specificational meaning. Since it-clefts are specificational copular constructions, it follows that they too should contain a
nominal, predicative element which will likely exhibit some of the characteristics associated with definiteness.1 This suggests a discontinuous constituent analysis of the it-cleft, on which the cleft pronoun and the cleft
clause function together as a definite NP predicate. On this type of account,
the it-cleft in (2) is effectively reduced to a specificational NP be NP sentence, corresponding to the th-cleft in (3).
(2)
It was Frank that complained
(3)
The one that complained was Frank
From these examples, we can see that the highlighted cleft clause in (2)
corresponds to the relative clause in (3). As a result, I suggest that the cleft
clause is in fact a restrictive relative, albeit in an extraposed position. From
this, it follows that the cleft clause must be relative to the initial it, providing us with a semantic NP be NP structure.
Like other personal pronouns, it is definite. Functioning as a pronominal
form of the definite article, it therefore imbues the discontinuous NP with
characteristics associated with definiteness. The cleft pronoun also acts as
the head noun of the description, performing a role equivalent to the one or
the thing in th-cleft sentences (cf. Postals 1970 transformational analysis
of personal pronouns). However, as is the case for th-clefts, it is the relative
clause that provides the distinguishing classifying information. For instance, in (2) and (3), the act of complaining uniquely characterizes Frank
as distinct from the other members of the relevant background set (such as,
the people in this outfit).
On this account then, it-clefts are specificational inversion sentences in
which the semantic predicate is a discontinuous definite NP. It-clefts therefore involve an additional mismatch between the syntax and the semantics:
while a nouns modifiers typically appear inside the NP, in it-clefts and
1. However, this does not follow on all accounts that treat it-clefts as specificational copular sentences. Declerck and Seki (1990: 31) say, As is well-known,
it-clefts are the only type of specificational copular structure in which the variable need not be a nominal.
73
other sentences containing extraposed relative clauses, the semantic modifier is situated at the end of the clause (Culicover and Jackendoff 2005:
163). However, unlike other sentences containing extraposed relatives, the
it-cleft has a relatively fixed structure: the cleft clause must be extraposed
and cannot occur in a position adjacent to the cleft pronoun, as shown in
(4). This also accounts for the unacceptability of sentence (5), in which the
complete definite description is in postcopular position. Unlike NP be NP
or pseudocleft sentences then, the it-cleft does not have a corresponding
non-inversion construction.
(4)
*It that complained was Frank
(5)
*Frank was it that complained
The ungrammaticality of sentences like (4) and (5) has been used as an
argument against analyses which claim that the cleft clause forms a semantic constituent with the initial it (see Jespersen 1937: 8485). Left unexplained, this data is certainly problematic for the account sketched here and
I return to the issue of why these structures represent unacceptable alternatives in Chapter 6. For now, however, it is sufficient to note that the itclefts extraposed structure conforms to, and is supported by, a variety of
information structure generalizations (or constructions).
Like other specificational inversion constructions, it-clefts have a fixed
information structure, in which the focus is located in clause-final position.
Thus, all such constructions conform to the principle that the relatively new
information should not occur in subject position. However, it-clefts also
conform to the information structure tendency for heavier constituents to
occur near the end of the clause (see Hawkins 1994, 2004; Wasow 1997,
2002). As Prince (1978: 886) notes, the cleft clause in it-clefts is on average
nearly twice as long as the clefted constituent (see also Collins 1991a). The
it-cleft configuration is therefore motivated by some highly general, and often competing, information structure principles. As Lambrecht (2001: 488)
comments, cleft formation is a way for a language to have its cake and eat
it too.
In part, this explains why the it-cleft configuration exists as a useful and
motivated construction in our taxonomy of copular sentences. As shown in
Figure 7, the it-cleft is a less schematic subtype of specificational inversion
construction. It is more specific, in that it contains the substantive element
it and it specifies for an extraposed restrictive relative clause. Together, the
cleft pronoun and the cleft clause function as the definite NP predicate,
74
which describes the entities referred to by the clefted constituent. It-clefts

therefore exhibit the same semantic structure of class inclusion as other
specificational copular sentences.
Of course, it is not typical for personal pronouns to be restrictively
modified in this way and so the it-cleft configuration is relatively idiosyncratic. This represents a good argument for its treatment as a symbolic construction with internal form-meaning mappings, rather than as a product of
componential meaning and general combinatorial rules (see Chapter 2,
Section 2). However, while construction grammar tolerates and anticipates
idiosyncratic information, it still requires that such properties are motivated. In Chapter 6, I look to the language system of earlier periods of
English to explain why the cleft pronoun differs from other, more typical,
uses of it.
Specificational
It-cleft
Wh-cleft
All-cleft
Figure 7. Incorporating the it-cleft into a family of specificational constructions
There are also other types of specificational sentence which share the itclefts irregular configuration. Aside from an initial it, cleft structure can be
introduced by other elements, such as the demonstrative pronouns this and
that (see Hedberg 2000). These demonstrative clefts are used to indicate
temporal or spatial deixis within the immediate discourse context.2
2. I use the term demonstrative cleft to refer to sentences which have the extraposed structure of it-clefts, but which are introduced by demonstrative pronouns. The term is used in a different sense in Calude (2007, 2008b) to refer to
a subset of reverse pseudoclefts with demonstrative subjects, such as thats
what I thought.
(6)
This is Oliver London were talking about, isnt it?
(7)
That was Oliver London who just phoned
75
For instance, this in (6) indicates that the description person were talking
about is dependent upon the current discourse situation. On the other hand,
(7) is tied to an event in the recent past. Here, that indicates that the
description person who just phoned is only slightly removed from the
immediate context and still depends upon this context for its interpretation.
There-clefts, such as (9), are also related structurally to the it-cleft. They
differ from it-clefts in that they do not carry an exhaustiveness effect. For
instance, we assume from the it-cleft in (8) that Mark is the only person (in
this department, for instance) who is working Saturday. However, the
there-cleft in (9) is used to indicate that Mark and Oliver may or may not
comprise a complete list of people available to work Saturday.
(8)
Its Mark thats working Saturday
(9)
Well, theres Mark and Oliver that are available to work Saturday
Therefore, while the same structural configuration is shared by it-clefts,

demonstrative clefts and there-clefts, we have seen that the choice of using
it, this, that or there to introduce this structure has implications for sentence
meaning. This supports the claim that the cleft pronoun is a meaningful
element which provides an important semantic contribution. For example,
the non-exhaustive reading of there-clefts suggests that the property of exhaustiveness in it-clefts can be attributed to the initial it (see Section 1.2.3).
The demonstrative cleft and the there-cleft can be incorporated into our
taxonomy of specificational copular sentences, part of which is shown in
Figure 8. Since, as we have seen, it-clefts, demonstrative clefts and thereclefts share the same structural configuration, we might wonder whether
they form a distinct category, shown here as the overarching cleft schema.
It is difficult to say whether speakers actually recognize this overarching
category and whether this abstraction is realized in the language system.
Consequently, the cleft schema is depicted in Figure 8 using a series of
dashed lines, with the arrows pointing upwards to demonstrate that this abstraction may not be independently motivated. Nevertheless, what is clear
is that the it-cleft, the demonstrative cleft and the there-cleft are structurally
related in some way. I build up a historical account to this effect in Chapter
6.
76
Specificational
Cleft schema
Wh-cleft
There-cleft
Demonstrative
cleft
All-cleft
It-cleft
Figure 8. Incorporating cleft varieties into a family of specificational constructions
The treatment of it-clefts as specificational inversion sentences also has

implications for the analysis of so-called truncated clefts, as in (10). As the
name suggests, many authors have viewed such sentences as reduced or
elliptical it-clefts, including Bring (1998), Declerck (1983a, 1988), and
Declerck and Seki (1990). On this type of account, the cleft clause is missing from the second sentence in (10) because it is immediately recoverable
from the discourse, as shown in (11).
(10)
Was it Oliver who came to work Saturday? No, it was Mark.
(11)
Was it Oliver who came to work Saturday? No, it was Mark who
came to work Saturday.
However, this account is not consistent with the extraposition-from-NP

analysis of it-clefts argued for here. For instance, I suggested above that the
cleft pronoun it performs a role equivalent to the one in th-clefts. From this,
it follows that the truncated it-cleft in (11) patterns with the truncated thcleft in (12). However, as I explained in Chapter 3, Section 4.3.1, examples
such as (12) do not constitute acceptable specificational sentences. Here, it
is the relative clause information that uniquely characterizes Mark as
distinct from his workmates and so this constituent cannot be felicitously
removed.
(12)
77
Was it Oliver who came to work Saturday? #No, the one who came
to work Saturday was Mark.
This suggests that it was Mark in (10) cannot be analysed as a truncated

it-cleft. Rather than forming part of a discontinuous description (along with
an elided cleft clause), I suggest that the initial it in such sentences is a full,
non-restricted pronoun which is anaphoric to a complete predicative NP.
On this account then, the pronoun it in (10) is anaphoric to the discontinuous definite description it (the one)...who came to work Saturday. Examples
of truncated clefts are therefore really specificational inversion sentences
with pronominal subjects. The same can be said for instances with that,
shown in (13). Here, the pronoun is in bold and its antecedent is underlined.
(13)
Was it Oliver who said hed work Saturday? No, that was Mark.
A similar analysis is put forward by Mikkelsen (2007), who builds up a

case for analysing the initial it and that of such sentences as property
denoting anaphoric pronouns. She notes that, when used referentially, it and
that can only denote non-human, type e individuals. However, in contexts
involving specificational inversion sentences, it and that can nevertheless
be anaphoric to subjects that describe human individuals, shown in (14) and
(15). Mikkelsen (2007: 54) concludes that, in such cases, it and that are in
fact property denoting anaphors (extensionally, type <e,t>).
(14)
The best baker is Beverly, isnt it?
(15)
The best baker, that/it is Beverly.

[left-dislocation structure]
[tag question]
This analysis is supported by the use of it and that as property anaphors in

other (non-specificational) contexts. For instance, in (16) and (17), the antecedents to it and that are non-referring, predicative expressions.
(16)
Beverly is employee of the month. It/That is a prestigious position.
(17)
They said that Beverly was accomplished and she certainly is that.
Mikkelsen goes on to extend this analysis to truncated clefts, which she

analyses as specificational clauses containing pronominal subjects. Using
the example in (18), Mikkelsen notes that the truncated cleft in A1 corresponds to the specificational NP be NP sentence in A2. The only difference
78
is that in A1, the initial NP depends upon the context for its interpretation.
In this sentence then, Beverly is specified as the individual who bears a
certain contextually salient property of being the best baker (Mikkelsen
2007: 49).
(18)
Q: Who do you think the best baker is?

A1: It/Thats Beverly.
A2: The best baker is Beverly.
Therefore, in terms of our constructional taxonomy, truncated clefts are

instances of the specificational inversion construction. Since such examples
are effectively NP be NP sentences and the use of it and that as property
denoting anaphors is independently attested, it is not clear whether these
sentences form a distinct truncated cleft construction. Either way, the
important thing to note is that, since such sentences do not contain (elided)
extraposed relative clauses, they do not inherit the same cleft schema as itclefts, demonstrative clefts or there-clefts (see Figure 8).
1.2.
Explaining the it-clefts pragmatic properties
In Section 1.1, we saw that the account of specificational copular sentences developed in Chapter 3 calls for a discontinuous constituent analysis
of it-clefts, in which the cleft pronoun and the cleft clause (analysed here an
extraposed restrictive relative) function together as a definite NP predicate.
On this account, it-clefts are understood to involve the same nominal
predication relation and exhibit the same characteristics of definiteness as
other copular sentences with a specificational meaning. In what follows, I
show that many of the it-clefts functional and structural properties simply
fall out from this analysis. In particular, what are often considered to be
idiosyncratic characteristics of the it-cleft are shown to be inherited either
from the wider specificational inversion construction or from other more
general form-meaning correspondences.
In this section, I discuss the it-clefts semantic-pragmatic features before
going on to examine how the constructions structural characteristics relate
to an extraposition-from-NP account in Section 1.3. Here, I show that the
pragmatic properties of focus, presupposition, exhaustiveness, and contrast
can be explained either as characteristics of definiteness more generally or
as products of specificational meaning. While other authors have certainly
recognized such correspondences between it-clefts and definite NPs (see
79
Section 3), I provide an explanation for them in relation to my own account

of specificational sentences, which makes use of Hawkins (1978, 1991)
conceptualization of definiteness.
1.2.1.
Focus
The it-cleft is widely regarded as a device for marking focus syntactically.

Indeed, on accounts that treat it-clefts as structural variants of noncopular
sentences, this is understood to be the constructions primary function. For
instance, Lambrecht (1994, 2001) suggests that it-clefts provide a way of
avoiding the ambiguities associated with prosodic focus marking. For example, in the noncopular (19), nuclear stress is placed on the final element.
(19)
John kissed MARGARET
This utterance can be interpreted in two ways, shown in (20) and (21). The
context in (20) results in an argument-focus reading, where the object
Margaret constitutes the focus phrase, while in (21) the entire predicate is
in focus.
(20)
Who did John kiss? John kissed [MARGARET]
(21)
What did John do? John [kissed MARGARET]
In contrast, the corresponding it-cleft, shown in (22), does not allow both of
these readings. Here, the focused element (Margaret) is placed in a syntactically marked postcopular position while the remaining content of (19) is
located in the cleft clause. This provides an unambiguous argument-focus
reading; as Lambrecht (2001: 489) notes, it-clefts are used to prevent
unintended predicate-focus construal of a proposition.
(22)
It was [Margaret] that John kissed
However, on the assumption that it-clefts are specificational copular

sentences, the constructions primary function is not to provide a structural
alternative to prosodic focus marking in noncopular sentences, but to
express a specifying class inclusion relation. On this account, the rigid
argument-focus structure that Lambrecht (2001) observes falls out directly
from the it-clefts specifying function. As I explained in Chapter 3, specifi-
80
cational meaning in copular constructions involves a reinterpretation of a

nominal predication relation. Rather than ascribing a property to a referent,
specificational sentences specify the membership of a set (i.e. the instances
of a type specification).
The difference between predicational (or ascriptive) and specificational
meaning is therefore partly information structural. In non-inverted predicate
nominal sentences, a specificational interpretation relies on the prosodic
focus marking of the subject argument, shown in (23). On the other hand,
the inversion construction shown in (24) has a fixed information structure,
containing a postcopular focal element. This provides an unambiguous
specificational reading, since the membership (rather than the set) information, is always in focus.
(23)
JOHN is the thoracic surgeon
[specificational non-inversion]
(24)
The thoracic surgeon is John
As a subtype of the specificational inversion construction, the it-cleft inherits this fixed focus structure. Therefore, in (22), the focal element Margaret is not the argument of the predicate kissed, as in (19), but of the predicate nominal it (the one)who John kissed. On this analysis, it-clefts are
not information structural variants of simple noncopular sentences. Instead,
they are specificational copular sentences, and so exhibit the focus structure
associated with specifying the membership of a set.
1.2.2.
Presupposition
Focus is a relational concept, often defined as the unit of information structure where the assertion differs from the presupposition (see Lambrecht
1994; Vallduv and Engdahl 1996; Erteschik-Shir 2007). For instance, in
(20) above, the information John kissed x is already known. As the only
unpredictable element, Margaret is the focus of the sentence, providing the
value for the variable x in the open proposition. For authors that view the itcleft primarily as an information packaging construction, the cleft configuration performs two roles: it foregrounds the focal element while at the
same time backgrounding the remainder of the sentence into a separate,
subordinate clause.
On this account, the it-cleft functions not only as a focusing device, but
as a way of marking presupposed information (see Ward, Birner, and
81
Huddleston 2002: 1415). For instance, in (22), the proposition John kissed
somebody is taken as a precondition to the assertion that John kissed Margaret. Information that is presupposed is characteristically preserved under
negation. For instance, (25) asserts that John didnt kiss Margaret but the
presupposition that John kissed somebody (else) remains in force (see
Halvorsen 1978).
(25)
It wasnt Margaret [that John kissed]
The presuppositional nature of it-clefts follows naturally from an extraposition-from-NP account. In Section 1.1, I claimed that the cleft clause
restrictively modifies the cleft pronoun, forming a discontinuous definite
description. As we saw in Chapter 3, Section 2.1, the definite article is
associated with existence entailments. For instance, (24) above entails that
the singleton set thoracic surgeon exists (or is not empty). More specifically, this sentence entails a presupposition which remains constant under
negation, shown in (26). This sentence presupposes, but does not assert,
that there is a thoracic surgeon.3
(26)
The thoracic surgeon isnt John
On the account of it-clefts argued for here, it follows that the discontinuous
constituent should also carry existential presuppositions. For instance, both
(22) and (25) presuppose that that the individual(s) that John kissed exist(s), while (22) also entails that the individual that John kissed exists.
As in it-clefts then, the negation of the copular verb in specificational
sentences like (26) does not affect the existential presuppositions of definite
descriptions; here, neither the existence of the set of thoracic surgeons nor
that of the individual John is denied. Instead, what is negated is the existence of an instantiation or class inclusion relation between these elements.
Tellingly, however, both (25) and (26) carry the assumption that some other
3. The presuppositional nature of definite descriptions in it-clefts and other specificational copular sentences is reinforced by their fixed information structure,
since topics have a tendency to be presupposed (see Horn 1986). As Abbott
(2006: 13) notes, When definition descriptions are functioning as predicate
nominals, and thus as focus, they lose their presuppositional character to a great
extent. However, since in specificational (non-)inversion sentences, the referring expression is always in focus, these existential presuppositions remain in
force (see Chapter 3, Section 2).
82
individual matches the description (or instantiates the type). This supports
the suggestion, made in Chapter 3, Sections 2.2 and 2.3, that the definite
NP subject of specificational inversion sentences is a relational element. In
other words, the existence of the set requires the existence of its membership or instantiation. The purpose of a specificational sentence then is to
elaborate the (as yet) schematic trajector role of instance. Since (25) and
(26) only identify an individual that does not perform this function, the act
of specification has not yet been achieved. Such sentences therefore suggest
that some other entity exists that can complete this instantiation relation.
On the account proposed here, the presuppositional nature of it-clefts
does not follow from the backgrounded position of the cleft clause and does
not warrant independent explanation in relation to the cleft configuration.
Instead, it follows from the semantic contribution of the cleft it. As we have
seen, the it-cleft shares this perfectly regular pragmatic property with other
definite descriptions and (with respect to negation, for instance) exhibits the
same behaviour as other specificational copular sentences.
1.2.3.
Exhaustiveness
In addition to its fixed focus structure and existential presuppositions, the

use of the it-cleft also signals exhaustiveness. For instance, the simple noncopular sentence in (27) does not preclude the possibility that John kissed
additional people besides Margaret on the occasion in question. However,
from the corresponding it-cleft in (28), we assume that Margaret was the
only person that John kissed.
(27)
John kissed Margaret
(28)
It was Margaret that John kissed
Again this pragmatic property falls out neatly once we treat the it-cleft as a
specificational copular sentence containing a (discontinuous) definite NP
predicate. On Hawkins (1978, 1991) account, the definite article is associated with inclusiveness, such that the NP applies to the complete set of
entities that satisfies this description (within a pragmatically restricted context). As I explained in Chapter 3, this characteristic of inclusiveness brings
about an exhaustiveness interpretation for specificational sentences containing definite NP predicates. In such cases, the entities predicated of by
83
the definite NP will be understood as comprising a complete list (or the

totality) of individuals that constitute the described set (or category).
Therefore, in (28), the description it...that John kissed necessarily
applies to all of the individuals that John kissed on a particular occasion,
simply by virtue of being marked as definite (by the cleft pronoun it). From
this, it follows that the individual Margaret is understood to uniquely satisfy this description, representing an exhaustive list of people John kissed.
In Chapter 3, I suggested that this pragmatic property is extremely compatible with the specifying function, since in such sentences the referential
NP must be taken to represent (or at least give a satisfactory indication of)
the membership of a set. Ideally then, specificational sentences offer a
complete or exhaustive list of members, inclusive of all individuals that
comprise the described set.
A somewhat similar suggestion is made by Declerck (1988: 30), who
views specification as a value-variable relation. He claims that it-clefts such
as (28) carry an exhaustiveness implicature stemming from the Maxim of
Quantity (make your contribution as informative as required). Declerck
notes that, if the speaker adheres to the Maxim of Quantity when formulating this specificational it-cleft, they will necessarily provide an exhaustive
list of the values that satisfy the variable described by the cleft clause that
John kissed. Declerck (1988: 30) concludes, It is clear, then, that exhaustiveness follows directly from the act of specification itself. Exhaustiveness
is nothing else than exhaustive listing. See also Horn (1981) for a (generalized) conversational implicature analysis of exhaustiveness in it-clefts.
I agree with Declerck that exhaustiveness is a concept strongly related
to that of specificational meaning, in that it conforms to our intuitions of
what constitutes a well-formed list. Nevertheless, specificational meaning
does not create and cannot explain exhaustiveness in it-clefts. Rather, this
pragmatic characteristic stems ultimately from the concept of definiteness
(specifically the property of inclusiveness). This, among other things, offers
a reason as to why definite NPs are well-suited to occurring in specificational copular constructions and provides support for the analysis of itclefts as containing discontinuous definite NPs.4
4. Declerck (1988) does go on to relate the property of exhaustiveness to the
concept of definiteness. However, it is not clear how this relationship transpires
in Declercks analysis of it-clefts. For Declerck (1988: 31), the cleft clause is
not a restrictive relative (modifying it) and instead represents an incomplete
variable expression which functions like a definite nominal. However, since
the cleft clause does not contain a definite element, it is difficult to see where
84
This account also explains why certain types of specificational sentence

consistently indicate exhaustiveness while others do not. As was shown in
Section 1.1, there-clefts do not (and cannot) rule out a non-exhaustive interpretation; it is entirely possible, given (29), repeated from (9), that there are
other colleagues, besides Mark and Oliver, available to work Saturday.
(29)
Well, theres Mark and Oliver that are available to work Saturday
Since it-clefts and there-clefts share the same cleft structure, the differences
between them must rest on the nature of the cleft pronoun (see Section 1.1).
On the account developed here, the cleft it performs the same function as
the definite article. Since definiteness entails uniqueness, it follows that the
act of specification in it-clefts will necessarily be interpreted as exhaustive.
However, there-clefts seem to pattern more closely with specificational
sentences introduced by indefinite NPs in this regard.
Indefinite NPs are not associated with inclusiveness but, as shown in
Chapter 3, they can still occur as the subjects of specificational inversion
sentences as long as they do not implicate exclusiveness/non-uniqueness
(therefore explaining the preference for restrictively modified indefinites).
In specificational sentences then, indefinite NPs are noncommittal with
respect to inclusiveness and exclusiveness. From this, it follows that the
entities predicated of by the indefinite NP could conceivably, (but may in
reality not) comprise a complete list of entities that satisfy the description.
There-clefts also allow the speaker to be evasive about whether the act of
specification is exhaustive or not. For instance, it is possible, following
(29), that Mark and Oliver are the only colleagues available to work
Saturday. Indeed, this state of affairs most likely represents the speakers
(perhaps limited) knowledge on this matter. However, by choosing a therecleft (rather than an it-cleft), the speaker indicates that they are unwilling to
commit to the stronger proposition.
Assuming a discontinuous constituent account then, the property of exhaustiveness in it-clefts can be explained as a direct consequence of the
property of inclusiveness (or uniqueness) associated with definiteness.
While exhaustiveness is a concept which is well-suited to the specifying
function, the act of specification does not, in and of itself, bring about an
definiteness comes from on this account. Furthermore, as I explain below, the
non-exhaustive there-cleft data suggests that definiteness (and, from this,
exhaustiveness) must be attributed to the pronoun it, rather than to the cleft
clause.
85
exhaustive interpretation.5 Non-exhaustiveness is tolerated in specificational sentences as long as the list of entities can be taken as representative
(or at least indicative) of the membership of the described set.
1.2.4.
Contrast
It-clefts are also said to have a contrastive function. For instance, (30)
contrasts Charles with other people who do not share his ambition. More
specifically, it invites a contrast between Charles and the other people who
make up some contextually relevant set of individuals. For example, if used
to specify the only aspiring pianist in this outfit, (30) will necessarily invite
a contrast between Charles and the other people in this outfit. This sense of
contrast is even more pronounced in negative it-clefts. For instance, in (31)
Charles is contrasted with the individual who satisfies this contextually
relevant description. As Declerck (1992: 214) observes, contrastiveness is
a relative characteristic: the sense of contrast becomes greater if the set of
potential values is limited and clearly defined.
(30)
Its Charles that wants to be a concert pianist
(31)
Its not Charles that wants to be a concert pianist
5. Nevertheless, for some authors, exhaustiveness cannot occur without an act of

specification. From this perspective, negative it-clefts (which do not specify a
sets instantiation) are not really exhaustive. For example, although only one
entity is listed in (i), it does not follow that John kissed only one person. As
shown in (ii), we can qualify this negative statement by listing two (or more)
entities which satisfy the description given.
(i)
It wasnt Margaret that John kissed
(ii) It wasnt Margaret that John kissed it was Kellye and Ginger
Such examples have led Atlas and Levinson (1981) and Declerck (1988) to
treat positive and negative it-clefts differently, with only the former containing
exhaustiveness entailments/implicatures. However, once we relate exhaustiveness to inclusiveness, both types acquire a uniform interpretation. On this account, what is at issue in (i) is the totality of the set of people John kissed. The
sentence in (ii) is therefore unproblematic; it simply asserts that the complete
set of people John kissed is not made up of Margaret, but of Kellye and Ginger.
Therefore, exhaustiveness, or the fact that we are talking about the totality of
the set, is a prominent feature of both positive and negative it-clefts.
86
Yet again, the contrastive nature of it-clefts can be explained as a consequence of the characteristics of definiteness. As we have seen, definite
descriptions are always understood in relation to some mutual cognitive
environment, or pragmatic set. They are also associated with inclusiveness,
such that they denote a set of entities which exists uniquely within the
relevant context (see Hawkins 1991). Therefore, assuming that it-clefts
contain definite NP predicates, it follows that Charles is the only aspiring
concert pianist within some mutually understood context. It is because of
this property of exhaustiveness that (30) invites a contrast between Charles
and the other entities within the relevant pragmatic set that do not match
this description (or which fall outside of this more restricted set).
For the negative it-cleft in (31), contrast follows both from the existential presuppositions of definite NPs and from the (conventional) implicature
of P-membership. As we saw in Section 1.2.2, presuppositions remain
constant under negation. The it-cleft in (31) therefore presupposes the
existence of some individual that wants to be a concert pianist and further
implicates that this entity exists within some relevant pragmatic set. Like
other negative specificational sentences, negative it-clefts do not elaborate
an instantiation relation because they do not provide the membership of a
set. However, since the existence of a set implies the existence of its members, the negative it-cleft in (31) invites a contrast between Charles and the
as yet unidentified individual within the P-set that satisfies this description.
It-clefts therefore share the same contrastive function as other specificational copular sentences containing definite, predicative NPs. However, as
with other sentence-types characterized as cleft, it-clefts exhibit less variation in how this contrastive information is presented. For example, the
noncleft NP be NP sentence in (32) has two possible readings. Depending
on the context, (32) can be used to contrast Dr. Hugh Beale with other nonpsychologists working at the hospital or with some other group of psychologists who dont work at St. Eligius hospital.
(32)
The psychologist who works at St. Eligius is Dr. Hugh Beale.
In the former, the restrictive relative clause provides the shared environment (or P-set information) of hospital workers from which the sole psychologist is identified. In the latter, the relative clause tells us what distinguishes Dr. Hugh Beale from some shared set of psychologists. As shown
in (33), this latter reading is not possible if the relative clause is omitted.
Here, the head noun must be interpreted as containing the distinguishing
information and the relevant P-set information is determined by the context.
(33)
87
The psychologist is Dr. Hugh Beale.
In it-clefts, however, it is always the relative (or cleft) clause that

provides the distinguishing classifying information. For instance, in a discourse about a group of psychologists, (34) specifies Dr. Hugh Beale as the
only one working at St. Eligius. In order to capture the second reading of
(32) above, we must alter the information in the cleft clause, shown in (35).
Uttered in a discourse about St. Eligius, (35) contrasts Dr. Hugh Beale with
the non-psychologists working at the hospital.
(34)
Its Dr. Hugh Beale who works at St. Eligius.
(35)
Its Dr. Hugh Beale thats the psychologist.
It-clefts differ from noncleft NP be NP sentences in that the head noun of

the definite description is semantically underspecified and invariably relates to the contextual P-set information (see also the discussion of th-clefts
in Chapter 3, Section 4.3.1). For instance, in (34) and (35) respectively, the
cleft pronoun isolates a singleton set from the relevant P-set of psychologists and hospital workers, to which the descriptive content of the restrictive relative clause applies uniquely.
The contrastive function of it-clefts therefore follows from the analysis
of the cleft pronoun and cleft clause as a discontinuous constituent carrying
the existential and uniqueness entailments associated with definiteness.
Since definite descriptions are always understood in relation to a mutual
cognitive environment (or shared, pragmatic set), they inevitably bring
about a contrast with the other entities within that environment. Furthermore, the particular way that contrast works in it-clefts (and other sentence
types characterized as cleft) follows from the fixed, substantive structure of
the (discontinuous) description as containing a semantically underspecified
head noun and a contentful relative clause.
1.3.
Explaining the it-clefts structural properties
We have seen then that the constructions pragmatic properties acquire a

straightforward analysis once we assume a discontinuous constituent
account. In this section, I examine the it-clefts structural characteristics,
including the behaviour of the cleft clause, some difficult constituency evidence, and the complicated facts surrounding agreement. At first glance,
88
these outwardly complex and irregular properties seem to highlight various

problems with the extraposition-from-NP account argued for here. However, in what follows, I show that the cleft clause and the constituency data
are not incompatible with, and actually exhibit behaviour which is to be
expected from, an account which analyses the clausal component as an
extraposed relative, restrictively modifying the cleft pronoun. I go on to
show that while the agreement evidence presents more of a challenge for
my account, and warrants a supplementary explanation, the analysis of specification as a type of class inclusion relation provides a useful perspective
from which to examine this data. I conclude that the agreement facts gain at
least a plausible explanation on the account of it-clefts presented here.
1.3.1.
The behaviour of the cleft clause
The cleft clause exhibits a variety of different and seemingly contradictory

properties, making it very difficult to conclusively classify this construct as
an instance of one or another recognized clause type. As Lambrecht (2001:
468) notes, its category membership is sometimes debatable or unclear.
In some ways, the cleft clause is undeniably analogous to restrictive
relative clauses. For instance, the cleft clause occurs with the same range of
introductory elements as restrictive relatives. Most tellingly, the cleft clause
is often introduced by that and sometimes occurs without an overt relative
pronoun, as in (36). While both of these options are possible for restrictive
relatives, shown in (37), they are not found in other, nonrestrictive clause
types. For instance, nonrestrictive relatives require an overt wh-relative
pronoun, shown by the ungrammaticality of (38).
(36)
It was Oliver London that/ I used to work with
(37)
The guy that/ I used to work with lives there
(38)
*Oliver London, that/ I used to work with, lives there
While nonrestrictive relatives are separated from the rest of the sentence by
commas or pauses, as in (38), the cleft clause does not share this property.
This suggests that, like the restrictive relative clause in (37), the cleft clause
is interpreted as forming a constituent with some other element in the sentence.
However, additional evidence suggests that, unlike restrictive relatives,
the cleft clause does not form a constituent with its immediate antecedent.
89
As shown in (36), the clefted constituent in it-clefts can be a proper name.

Nonrestrictive relatives, which provide additional or ancillary information,
can also attach to these phrasal elements, as in (39). In contrast, proper
names cannot normally be modified by restrictive relatives, shown in (40).
(39)
Oliver London, who I used to work with, lives there
(40)
*Oliver that I used to work with lives there
As Langacker (1991: 432) notes, restrictive relatives function to modify or

restrict the type specification of the antecedent noun. In (37), the relative
clause aids identification by providing a more detailed specification, since
guy I used to work with is more specific than the type designated by guy.
However, in (40), the proper name Oliver is a complete noun phrase used to
refer to a specific, uniquely identifiable, individual. As such, it cannot be
further modified by a restrictive relative.6
Restrictive relative clauses and their antecedent nouns form a constituent, and therefore an intonation unit, within the noun phrase, with nuclear
stress falling on the tone-final element, shown in (41). However, since itclefts are typically spoken with a fall-rise tone, shown in (42), the cleft
clause does not form an intonation unit with the clefted constituent (see
Halliday 1967: 237; Davidse 2000: 1103). This suggests that, like nonrestrictive relatives, the cleft clause functions as a separate constituent from
its phrasal antecedent.
(41)
[The doctor who phned me] lives there
(42)
[It was the dctor] [who phned me]
An additional problem with a restrictive relative analysis of the cleft

clause is that, while both configurations can occur without an overt relative
pronoun, this phenomenon is far less constrained in it-clefts (see Sornicola
1988: 346). For instance, the zero realization of the relative pronoun is not
found in restrictive relatives where the antecedent acts as the subject of the
6. As Langacker (1991: 59) points out, a restrictive relative clause can modify a
proper noun only if it is preceded by a determiner, as in (iii). Here, Oliver is
used not to refer to a specific entity, but as a common noun denoting a more
general type person with the name Oliver.
(iii) I just bumped into the Oliver you used to work with
90
proposition in the relative clause. This is evidenced by the ungrammaticality of (43) in comparison to the grammatical (37), where the guy functions
as the complement of a preposition. In contrast, the cleft clause can occur
without an overt relative marker even when the relative element functions
as the subject, shown in (44).
(43)
*The man paid for that is over there
(44)
It was your husband paid for that
(Lambrecht 2001: 470)
This evidence has led many authors suggest that the cleft clause is a
structurally unique type of clause, specific to the cleft configuration. While
the cleft clause exhibits the same internal structure as a restrictive relative,
it does not restrictively modify its immediate antecedent. Consequently,
some have labelled the cleft clause as a type of nonrestrictive relative
clause, despite a lack of correspondence between the cleft clause and other
nonrestrictive relatives (see for instance, Lambrecht 2001). As Huddleston
(1984: 462) comments, this type of analysis is very largely ad hoc the
relative clause is of a kind that is sui generis, unique to this construction.
However, once we assume an extraposition-from-NP analysis of itclefts, the seemingly contradictory behaviour of the cleft clause becomes
straightforward and is shown to be consistent with a restrictive relative
analysis. As Jespersen (1937: 83) notes, when talking about his earlier
transposition theory, the cleft clause is felt to be, and is treated like, a
relative clause, though it does not logically restrict the word with which it
is connected. This observation is unproblematic for an extraposition-fromNP account, since here the cleft clause restrictively modifies the initial it.
Therefore, the antecedent of the cleft clause is not the immediately
preceding clefted constituent, but the cleft pronoun.
As shown in (45), the prosodic structure of it-clefts supports the claim
that the cleft clause forms a discontinuous constituent, and an intonation
unit, with the initial (unstressed) it. As Huddleston (1984: 461) comments,
this type of analysis also allows us to explain the difference in meaning
between superficially similar sentences such as (45) and (46). On this
account, the sentence in (46) is not an it-cleft, but a specificational NP be
NP sentence containing a restrictively modified referring expression. Here,
the NP the doctor who phned me comprises a single intonation unit, with
the noun doctor functioning as the antecedent of the restrictive relative.
What characterizes (45) as an it-cleft then, is the restrictive modification of
the pronoun it.
(45)
Q: Who phoned? A: [It] was the dctor [who phned me]
(46)
Q: Who was at the door? A: It was [the doctor who phned me]
91
Once we treat the cleft pronoun as the antecedent of the cleft clause, we
can also speculate as to why it-clefts permit zero realization of the relative
marker in more contexts than ordinary restrictive relatives. As was noted
above, restrictive relatives cannot normally occur without an overt relative
marker when the relative element functions as the subject of the embedded
clause. For instance, in (47), the relative proposition contains a gap in
subject position which corresponds to the antecedent noun (__phoned me).
(47)
*The doctor phoned me is at the door
Presumably, the problem with such sentences is that they create an unintended garden-path reading in which the modified NP is taken to be a
complete sentence. For instance, in (47) the doctor phoned me is a wellformed proposition. Here, the VP may not be interpreted as a restrictive
relative ( phoned me) but as predicated of the NP the doctor. In contrast,
for relative clauses with object gaps, as in (48), zero realization of the
relative marker is unproblematic. Since the doctor I phoned is not a wellformed sentence, it must, in all contexts, be interpreted as a restrictively
modified NP.
(48)
The doctor I phoned is at the door
In (47) then, miscommunication occurs because the doctor is incorrectly

analysed as an unmodified NP. However, on an extraposition-from-NP
account of it-clefts, phoned me does not modify the doctor in (49) and is
instead relative to it. Consequently, the zero realization of the relative
marker does not create the same parsing ambiguity as in (47).
(49)
It was the doctor phoned me
Nevertheless, it is worth pointing out that while it-clefts can certainly

tolerate zero markers in clauses with subject gaps, this is by no means the
preferred option. Out of 31 it-clefts without overt relative markers in the
ICE-GB corpus, only 12 contain subject gaps in the cleft clause and 10 of
these are reduced cleft clauses. Considering that 67% of all it-clefts in the
ICE-GB exhibit subject gaps (270 out of 404), we would expect there to be
92
twice as many zero relativizers functioning as subjects compared with nonsubjects. Instead, zero markers functioning as objects and adjuncts are 1.6
times more frequent (9.5 times more if we factor out the reduced relatives).
Tellingly, the zero relatives with subject gaps are restricted to spoken communication. For instance, example (50), along with 8 out of the 10 it-clefts
containing reduced relative clauses, occurs in spoken sports commentaries.
(50)
And its the grey Altaya still still has it from Young Pretender in
the yellow
(S2A-006 176, Channel 4 Racing)
The spontaneous commentary sub-genre of the ICE-GB has been noted for
its high concentration of it-clefts and shows a strong preference for it-clefts
that deviate from the conventional pattern (see Nelson 1997). For instance,
the majority of it-clefts with initial focus, such as (51), are also found in
spoken sports commentaries.
(51)
Simeone it is who chips the ball in

(S2A-010 174, BBC Radio 5 International Soccer Extra)
This suggests that the occurrence of it-clefts with zero relatives containing subject gaps is exceptional and that, in general, it-clefts exhibit the
same preference as other restrictive relative clauses for zero realization in
clauses with non-subject gaps.7 This is exemplified by (52), which contains
two cleft clauses performing different functions. While the first of these
(youd spoken to__) has an object gap and a zero relativizer, the second
clause (who__ spoke to you) has a gap in subject position and seems to
require the occurrence of an overt relative marker.8
(52)
Was there some doubt in your mind as to whether it was a lady or

gentleman youd spoken to or who spoke to you
(S1B-006 108, legal cross-examination)
7. As Curme (1931: 186) notes, the that should not be omitted if it is needed to
keep the thought clear, i.e., to indicate the oneness of the words in the subject
[cleft] clause and to maintain the integrity of the group as a distinct grammatical element in contradistinction to other elements in the sentence.
8. The it-cleft in (52) is perhaps predicational rather than specificational, with the
focal element ascribing a property to an individual (see Chapter 5). However,
since I will claim that both varieties of it-cleft exhibit the same discontinuous,
extraposition-from-NP structure, they are equally relevant to the discussion surrounding the realization of the relative marker.
93
In sum, then, the extraposition-from-NP analysis of it-clefts allows for

the cleft clause to be unequivocally classified as a restrictive relative. With
this analysis in place, much of the seemingly incongruous behaviour of the
cleft clause becomes predictable from what we know about how restrictive
relative clauses operate in the rest of the grammar. Of course, on this
account, the cleft clause cannot be said to be a typical restrictive relative,
since it is extraposed relative to a pronominal element. Outside of the itcleft construction, the pronoun it functions as a full NP, and so would not
normally qualify as the antecedent of a restrictive relative (see the discussion of proper names above). Importantly, however, the restrictive modification of pronouns is not unprecedented. I return to this issue in Chapter 6,
where I show that an extraposition-from-NP account of it-clefts is supported by empirical, historical evidence.
1.3.2.
The evidence for VP constituency
A potential problem for the extraposition-from-NP account of it-clefts is

the apparent evidence that the postcopular element and the cleft clause form
a constituent. Delahunty (1982) invokes five tests for VP constituency,
including VP-deletion, VP-conjunction, right-node-raising, parenthetical
formation, and VP-fronting. With respect to each of these environments,
shown in examples (53) to (57), Delahunty claims that the clefted constituent and the cleft clause operate as a syntactic unit (in bold). This is unexpected on the account proposed here, which treats the cleft clause as any
other extraposed relative.
(53)
I said that it should have been Bill who negotiated the new
contract, and it should have been
[VP-deletion]
(54)
It must have been Fred that kissed Mary but Bill that left with her
[VP-conjunction]
(55)
It could have been and it should have been Bill who negotiated
the new contract
[right-node-raising]
(56)
It must have been, in my opinion, the cyanide that did it

[parenthetical formation]
(57)
?I said that it was Bill that argued the case, and Bill who argued
the case it was
[VP-preposing]
(examples from Hedberg 1990: 98, her judgments)
94
However, in what follows, I show that not all of this data is problematic for
a straightforward extraposition-from-NP analysis of the cleft clause and
that not all of these configurations are reliable tests for constituency.
First, it can be argued that examples of VP-deletion, such as (53), relate
to a type of specificational construction different from the it-cleft. In Section 1.1, I argued that specificational sentences of the type it be NP are not
truncated it-clefts, in which the cleft clause has been elided or deleted.
Instead, they are simply NP be NP sentences with pronominal subjects.
Assuming an inverse analysis of specificational sentences, the subject it is
therefore a property denoting anaphor which refers back to a predicative
expression (see also Mikkelsen 2007). From this, it follows that the second
instance of it should have been in (53) is not an incomplete it-cleft but an
incomplete it be NP sentence, corresponding to it should have been Bill.
Here, the pronoun it is anaphoric to the description given in the preceding
it-cleft, itwho negotiated the new contract. On this account, (53) involves
the deletion only of the NP Bill rather than the string Bill who negotiated
the new contract. Consequently, (53) cannot be used as evidence for analysing the clefted constituent and the cleft clause of it-clefts as forming a
syntactic constituent.
Secondly, we can set aside Delahuntys constituency evidence on the
basis that such tests are notoriously unreliable. For instance, Croft (2001)
comments that coordination cannot be depended upon to uncover constituency. He provides the following examples of English coordinate constructions which support constituents that are not supported by other criteria
(Croft 2001: 189).
(58)
[Jenny makes]?? and [Randy sells]?? the prints
(59)
Jenny gave [the books to Randy] ?? and [the magazines to Bill]??

(examples from Croft 2001: 189)
As these examples show, it is not only constituents that can conjoin. Croft
suggests that coordinate constructions combine structures which are perceived to have sufficient points of commonality. Coordination therefore
demonstrates evidence of conceptual grouping rather than constituency (see
also Wierzbicka 1980, Ch.7). Consequently, the conjoining of Fred that
kissed Mary with Bill that left with her in (54) is not necessarily indicative
of VP-internal constituency.
Likewise, right-node-raising and parenthetical formation tests can give
false results for constituency. For instance, Hedberg (1990: 98) observes
95
that these constructions do not entirely exclude extraposed relative clauses

from appearing in the position filled by the cleft clause in clefts. She
provides the following examples.
(60)
Nobody would and nobody could drink instant coffee who

knew anything about espresso
[right-node-raising]
(61)
Nobody would ever, in my opinion, drink instant coffee, who knew

anything about espresso
[parenthetical formation]
(examples from Hedberg 1990: 99)
In each case, the extraposed relative clause who knew anything about
espresso clearly restricts, and is dependent upon, nobody. Nevertheless, on
the assumption that these constructions are reliable tests for constituency,
(60) and (61) indicate that the relative clause is internal to the VP (in bold).
These ordinary extraposed relatives therefore exhibit the same pattern of
behaviour as the it-clefts in (55) and (56) above. Consequently, right-noderaising and parenthetical formation cannot be used as evidence against an
analysis of it-clefts involving extraposition-from-NP.
This leaves us with VP-fronting (or VP-preposing). As Hedberg (1990)
observes, extraposed relative clauses cannot occur in the VP-fronting construction. On the basis of (62), Hedberg (1990: 99) concludes that the
relation which holds between the cleft pronoun and the cleft clause is not
identical to the relation which holds between an extraposed relative clause
and its NP head.
(62)
*I said a candidate would win who had charisma, and win who
had charisma, a candidate did
(Hedberg 1990: 99)
However, VP-fronting in it-clefts is also unacceptable to many speakers,

leading Delahunty (1982) to disqualify this evidence, treating examples like
(57) as ungrammatical. Nevertheless, in a later paper, Delahunty (1984: 71)
claims that the low acceptability of such sentences appears to be a matter
of stylistic awkwardness rather than ungrammaticality. Hedberg (1990,
2000) agrees, arguing that while some examples are certainly more unacceptable than others, including (63), preposing can be used as evidence for
VP-internal constituency in it-clefts. She presents the it-cleft in (64) as an
example with an equally acceptable status as the straightforward case of
VP-preposing in (65) (see also Reeve 2010: 50).
96
(63)
*I said it is John thats an interesting guy, and John thats an

interesting guy it is
(Hedberg 2000: 916)
(64)
I said it would be a conservative whod win and a conservative

who won it certainly was
(Hedberg 1990: 98, her judgment)
(65)
I said that I would finish by September, and finish by September I

did
(Hedberg 1990: 98)
However, I disagree with Hedbergs (1990) judgments. The it-cleft in (64)

is clearly less acceptable and is certainly less natural than the noncleft
example in (65). In conclusion then, examples of VP-preposing are at best
unnatural and at worst ungrammatical. Consequently, I do not view this
data as strong evidence for VP-internal constituency (see also Delin 1989).
In sum, Delahuntys (1982) constituency evidence does not undermine
the extraposition-from-NP analysis of it-clefts argued for here. As shown
above, we can account for this data fairly neatly (where grammatical) by
comparing it-clefts to other sentences containing extraposed relatives. In
addition, it is worth noting that we can invoke similar constituency tests as
evidence that the cleft clause does not form a constituent with the postcopular element. For instance, Han and Hedberg (2008: 359) provide the
example in (66), in which a parenthetical adverbial phrase comes between
these two components.
(66)
It was Kim, in my opinion, who won the race

(Han and Hedberg 2008: 359)
1.3.3.
The evidence from agreement
Agreement evidence is often called upon to argue against extraposition

accounts of it-clefts. On the extraposition-from-NP account argued for here,
we anticipate that, when the cleft clause contains a gap in subject position,
the verb inside the cleft clause will agree with the antecedent pronoun it.
Data which demonstrate the existence of a subject-verb agreement relation
between the verb in the cleft clause and the clefted constituent are therefore
extremely problematic for this analysis. For instance, Jespersen (1937: 84)
suggests that agreement in it-clefts represents overwhelming evidence
against an extraposition account, claiming that there is almost universal
agreement with regard to person and number of the verb in the relative
clause with the immediate antecedent.
97
However, in what follows, I show that that the agreement patterns of itclefts present a far muddier picture than Jespersen portrays here. In truth,
the facts surrounding agreement are very complicated and are difficult to
account for on any one analysis of it-clefts. While the person agreement
evidence is suggestive of an extraposition account (contra Jespersen 1937),
number agreement is more problematic, requiring an ancillary explanation.
According to Jespersen (1937: 84), the pattern shown in (67) represents
the ordinary system of person agreement in it-clefts. Here, the verb in the
cleft clause agrees with the postcopular first person pronoun and the clefted
constituent is in the nominative case. This pattern would suggest that the
clefted element acts as the preposed argument of the proposition expressed
in the cleft clause, I am responsible.
(67)
It is I who am responsible
(Akmajian 1970: 153)
However, Akmajian (1970: 150) finds that the most common agreement
pattern is actually that which is represented by (68). Here, the postcopular
pronoun is in the objective case and the verb in the cleft clause is systematically third person, regardless of the clefted constituent.
(68)
It is me who is responsible
Akmajian (1970) finds that the majority of speakers consistently adopt the
pattern in (68). Furthermore, all speakers who display the pattern in (67)
nevertheless allow (68) as an acceptable alternative. The remaining speakers in Akmajians data set display the pattern in (69). As in (68), verbal
agreement is consistently third person, but these speakers mark the clefted
element as nominative when no surface subject intervenes between it and
the verb of the clause. As Akmajian (1970: 152) notes, this strategy often
gives the misleading impression of case agreement between the clefted
constituent and the relative element.
(69)
It is I who is responsible
So, in all of the different dialects that Akmajian (1970) observes, third
person marking of the embedded verb is customary (and for many speakers
is routine). In only one of the less common dialects is it possible for person
marking to be influenced by agreement with the clefted constituent (contra
Jespersen 1937). Indeed, among it-clefts in the ICE-GB corpus, there are no
examples in which the person marking of the cleft clause has clearly been
98
influenced by that of the clefted constituent; the verb embedded in the cleft
clause allows a third person interpretation in all 404 it-clefts. In contrast,
there are 4 instances in which there is an observable discrepancy between
the person marking of the clefted constituent and that of the cleft clause.
For instance, in (70), the postcopular me is a first person pronoun while the
embedded verb has is an unambiguously third person singular form. Case
aside, this lack of agreement produces the ungrammatical unclefted string
*I has ended up as the criminal.
(70)
The irony is, as Linda pointed out, its me thats ended up as the
criminal.
(W1B-007 099, social letter)
The evidence from person agreement therefore seems to support our

discontinuous constituent account, on which the cleft clause is extraposed
relative to the third person pronoun it. As Akmajian (1970: 151) notes, itclefts show the same pattern of agreement as pseudocleft sentences, such as
the th-cleft in (71). In these sentences, the embedded verb is systematically
marked as third person following agreement with a third person head noun
(in this case, one).
(71)
The one who is responsible is me
On this type of analysis, we anticipate that, as in other specificational inversion constructions, the postcopular pronoun in it-clefts will be assigned
objective case as a result of its postverbal position. However, in the ICEGB, we find only 5 instances where the focal pronoun is clearly objective
despite there being a subject gap in the cleft clause, as in (70). In contrast,
there are 9 tokens containing postcopular pronouns in nominative case,
including (72).
(72)
it was he who recommended that CPRE should be contacted.

(W1B-020 117, business letter)
Akmajian (1970) suggests that such deviations from the expected pattern may result from the use of nominative case as a prestige form. This is
certainly supported by the ICE-GB data. While all 5 tokens with objective
pronouns occur in relatively informal, unplanned modes of communication,
including direct conversation and social letters, nominative case marking is
found in more formal and planned discourses, such as scripted broadcasts,
published works and business letters (see Gmez-Gonzlez and Gonzlvez-
99
Garca 2005: 165). However we choose to explain them, the case agreement facts do not represent a strong argument against an extraposition account of it-clefts; the fact remains that the person agreement data supports a
discontinuous constituent account, such as the one argued for here.
In contrast, number agreement is certainly problematic for this account.
As shown in (73), the embedded verb does not agree in number with the
singular pronoun it, but with the clefted constituent. In the corresponding
th-cleft, on the other hand, the embedded verb also allows plural marking,
but this is in agreement with both the clefted constituent and the plural head
noun ones, shown in (74).
(73)
Its John and Margaret who are responsible
(74)
The ones who are responsible are John and Margaret
It-clefts and pseudoclefts therefore have different patterns of number agreement (see also Huddleston 1984: 461). Of course, this is unexpected on a
discontinuous constituent account, which views it-clefts in relation to other
specificational inversion sentences. Nevertheless, in what follows, I show
that we can provide something of an explanation for number agreement in
it-clefts once we treat this discontinuous description as the semantic predicate.
In Chapter 3, I argued for a class inclusion analysis of specificational
inversion sentences, in which the precopular NP denotes a set (or type) and
the postcopular NP specifies the individuals (or instances) that make it up.
In (74) then, the precopular NP the ones who are responsible denotes a
single set but, with regard to number, is conceived of in terms of its multiple entities or instances. The predicative NP of a specificational sentence
is therefore interpreted in the same way as any other type of collective entity. As Croft (2001: 128) notes a collective entity is conceptually both
singular and plural; it is singular because it functions as a singular unit, but
it is also plural in that it is made up of a multiplicity of individuals. Alternative conceptualizations sometimes lead to variation and change. Croft
(2001: 127) discusses the following example of a phenomenon which has
now become a convention of the language. Here, a noun which refers to a
collective entity is marked as singular, but is followed by a plural verb.
(75)
Section 278 of the Highways Act 1980, therefore, provides that if a

highway authority are satisfied that it would be of benefit to the
public
(Croft 2001: 127)
100
It is possible then that the mixed pattern of agreement in it-clefts with

plural foci is somehow related to the different ways of conceptualizing the
discontinuous definite description. As shown in (73), while the verb in the
relative clause is marked as plural (are), the copular verb of the matrix
clause is consistently singular (s). We might argue then that while the
matrix copula shows agreement with the singular set (marked as singular by
it), the embedded verb agrees in number with the members of this set. This
might sound opportunistic, but it is worth noting that a similar state of
affairs is found in NP of NP constructions. For instance, in (76), the matrix
verb has agrees with the singular noun group, while the verb are in the
relative clause shows agreement with antecedent noun physicians. On this
analysis then, (73) informs us that the set (of individuals who are responsible) is comprised of John and Margaret.
(76)
The group of physicians who are responsible for this mix up has
been ordered to disband.
In essence then, the cleft pronoun is underspecified for number, characterizing sets with singular and plural membership in the exact same way.
The extraposed position of the cleft clause therefore provides the it-cleft
with a useful function: it allows the speaker to withhold information about
number until the act of specification is complete. For instance, in (77), the
speaker contrasts plural referents (the tadpoles) with a singular entity (the
frog), suggesting that both could potentially comprise the membership of
the underspecified set. This becomes much more difficult when the speaker
is forced to choose between a singular and plural head noun, show in (78).
(77)
Its not the frog but the tadpoles which tell us the truth about our
class system
(S2B-036 044, BBC Radio 4 debate)
(78)
??The things which tell us the truth about our class system arent
the frog but the tadpoles
In addition, it is worth noting that there is variation in the it-clefts number

agreement pattern. Although not attested in the ICE-GB, some speakers
allow the verb in the relative clause to be marked as singular (agreeing with
it and the matrix copula) even when the clefted constituent is plural. For
101
instance, in my dialect, (79) is an equally acceptable alternative to (73)

above.9
(79)
Its John and Margaret thats responsible for this
In conclusion then, number agreement is not an insurmountable obstacle

to analysing the cleft clause as extraposed, relative to it. Once we treat the
cleft it as a morphosyntactically singular, but conceptually underspecified
pronoun, number agreement is understood to be a semantic phenomenon,
with the verb in the cleft clause showing agreement with the membership of
the described set. As I go on to show in Chapters 5 and 6, the analysis
sketched here successfully accounts for the different agreement patterns of
specificational and predicational it-clefts and is furthermore supported by
historical evidence.
1.4.
Interim summary
Throughout Section 1, I have advanced and supported a discontinuous constituent (extraposition-from-NP) analysis of it-clefts on which the cleft pronoun and the cleft clause function together as a definite NP predicate. After
demonstrating how this particular analysis follows neatly from the account
of specificational copular sentences as involving a nominal predication
relation in which the semantics and pragmatics of definiteness play an
important part (outlined in Chapter 3), I went on to show that many of the
9. This is the pattern most often found in wh-clefts. Like the cleft it, what is inherently singular, but can be used to denote a set with plural membership. Typically, the verb in the cleft clause and the matrix copula agree with the singular
what, as in (iv). However, other times the matrix copula shows agreement with
the postcopular NP, as in (v). This suggests that, for both wh-clefts and it-clefts,
the verb adjacent to what or it will be consistently marked as singular, while the
verb which appears later in the sentence is subject to influence from the sets
plural membership (which is made explicit by the postcopular NP).
(iv) What has made this especially upsetting has been their constant
reassurances and their lack of contrition
(v)
What makes something a pencil are superficial characteristics such as a

certain form and function
(Partee 1998, citing Higgins p.c.)
102
it-clefts functional and structural properties fall out from this analysis
straightforwardly.
In Section 1.2, I showed that a discontinuous constituent analysis of itclefts is very successful at both accounting for and explaining the constructions pragmatic properties. From this perspective, many of the it-clefts
pragmatic characteristics are inherited from the wider specificational inversion construction and conform to the general patterns of correspondence
found in definite NPs. For instance, the fixed focus-structure of it-clefts
was shown to follow from the analysis of specificational meaning as a reinterpretation (or information structural variant) of a class-inclusion predication relation. Likewise, once we invoke Hawkins (1978, 1991) account of
definiteness, the properties of presupposition, exhaustiveness and contrast
are explained as products of the existential, inclusiveness and familiarity
conditions associated with definite NPs.
Section 1.3 examined the it-clefts structural characteristics in relation to
an extraposition-from-NP analysis. It was shown that the behaviour of the
cleft clause and the it-clefts person agreement patterns are consistent with,
and acquire an explanation on, this type of account. Furthermore, the
examination of so-called constituency tests revealed that it-clefts can be
manipulated in many of the same ways as other sentences involving
extraposed relatives. Finally, the number agreement data was identified as
problematic for a discontinuous constituent account. Nevertheless, as
shown in Section 1.3.3, we can account for inconsistencies in the way that
number is marked in it-clefts once we adopt a predicational analysis of
specificational sentences. On this type of analysis, the discontinuous
description is semantically predicative, denoting a set of entities which can
be conceptualized as both singular (in relation to the set) and plural (in
relation to its members). As I explain in Chapter 5, this approach provides
an illuminating way of accounting for differences between specificational
and predicational it-clefts. First, however, in the remaining sections of this
chapter, I situate my account in relation to the current cleft literature.
2.
A comparison with expletive accounts of it-clefts
As I explained in Chapter 1, expletive is the name given to it-cleft analyses

which assume that the cleft clause is directly predicated of (or is in some
other way related to) the clefted constituent. This accounts for the truthconditional equivalence between it-clefts, such as (80), and simple subjectpredicate sentences, such as (81).
(80)
It was Frank that complained
(81)
Frank complained
103
From this, it follows that the cleft pronoun and (in most accounts) the copular verb do not play an important role in the interpretation of the sentence.
On expletive accounts then, the cleft it is invariably a semantically expletive element. It is this characteristic that gives this type of analysis its name.
While all expletive accounts adopt this same set of basic assumptions,
authors differ as to how they perceive the relationship between the cleft
clause and the postcopular element and how they think this relationship
comes about. Nevertheless, whether it is achieved underlyingly at deep
structure (Rochemont 1986), via indexing at surface structure (Chomsky
1977; Heggie 1988; Williams 1980), as a result of lambda conversion at the
level of logical form (Delahunty 1982, 1984), or through some combination
of the above (. Kiss 1998), these authors all assume that it-clefts share a
level of representation with simple subject-predicate sentences.
There is of course some evidence for the claim that the clefted element
and the cleft clause form a constituent, including number agreement (see
Section 1.3.3), VP-constituency tests (see Section 1.3.2) and possibly also
syntactic connectedness (see Chapter 3, Section 4.3.2). However, in many
ways, expletive accounts are not as successful as discontinuous constituent
analyses at accounting for the it-clefts characteristics, leaving a number of
questions unresolved.
Expletive accounts tend to take a syntax-centred approach to it-clefts.
As a result, they rarely tackle the issue of how this construction acquires its
pragmatic properties. In Section 1.2, I showed that the characteristics of
existential presupposition, exhaustiveness and contrast result from the
semantic contribution of the cleft it, which performs the same function as
the definite article. However, the cleft pronoun is semantically empty on
expletive accounts. Since the only contentful elements are those found in
the corresponding noncopular sentence, no elements remain that could be
said to generate these pragmatic properties. As a result, it is difficult, on an
expletive account, to explain exactly how the it-clefts meaning differs
from that of its simple sentence paraphrase.
For example, . Kiss (1998) attempts to account for exhaustiveness in
the syntax, as resulting from cleft structure. She claims that exhaustive
identification is a function of structural focus (. Kiss 1998: 251). Identificational focus (which expresses exhaustive identification) has a designated
syntactic position, occupying the specifier slot of the focus phrase. In .
104
Kiss analysis, the clefted constituent occupies this scope position, thereby
performing two important roles. Syntactically, the clefted constituent
functions as an operator which marks the sentence part following it and ccommanded by it as the scope of exhaustive identification (. Kiss 1998:
253). Semantically, the clefted element expresses the complete set of
entities for which the predicate in the cleft clause holds. In this way, . Kiss
provides a syntactic explanation for the fact that it-clefts are associated with
exhaustiveness.
However, there are several problems with this account. For one thing, it
is highly idiosyncratic. . Kiss (1998: 258) assumes that exhaustiveness is
dependent upon a syntactic structure which in English is particular to the itcleft configuration. However, we have seen, in Section 1.2.3 above, that exhaustiveness is found in other specificational copular constructions (which
do not exhibit the cleft structure). Consequently, the account is unable to
capture existing generalizations in the language. Secondly, not all types of
cleft sentence are exhaustive (see Section 1.2.3). Non-exhaustive thereclefts represent an important exception to the claim that exhaustive identification is tied to cleft structure. Instead, they provide support for theories in
which exhaustiveness is signalled by the cleft it.
As shown in Section 1.1, the existence of non-exhaustive there-clefts as
well as deictic demonstrative clefts represents strong evidence against the
claim that the cleft it is semantically expletive, since a change in the cleft
pronoun results in a concomitant change in meaning (see Hedberg 2000).
Without recognizing the semantic contribution of the initial it (and its relation to the definite article), expletive analyses are unable to account for the
it-clefts pragmatic properties.
The expletive analyses also struggle to account for some of the it-clefts
structural properties. As I explained in Section 1.3.1, the behaviour of the
cleft clause is largely consistent with a restrictive relative analysis, if we
assume that the cleft clause is relative to it. However, if we take the clefted
constituent to be the antecedent, the cleft clause seems to exhibit a range of
contradictory properties which make classification difficult. For some authors, the cleft clause shares only a superficial similarity to relative clauses;
Delahunty (1982), Rochemont (1986) and Heggie (1988) argue that relative
pronouns, like who, only occur in the cleft clause as a result of analogy
with relative clauses. Others, including Chomsky (1977) and Williams
(1980), suggest that the cleft clause has the same internal structure as a
restrictive relative but, since the cleft clause does not restrictively modify
its immediate antecedent, it is said to perform a different function from
other restrictive relatives.
105
On expletive accounts then, the cleft clause cannot be classed with any
known clause type. Although the cleft clause looks and behaves in many
ways like a restrictive relative, it is really a non-modifying sentential
predicate. From this, it follows that the cleft clause is a structurally unique
type of clause, specific to the cleft configuration. However, the expletive
accounts seem to be missing an important generalization here. If the cleft
clause is structured internally like a restrictive relative, then the chances are
it also has a modifying function. If this clause does not modify the clefted
constituent then we have to ask, what does it modify? The discontinuous
constituent account provides us with a suitable answer: the cleft clause
restrictively modifies the cleft pronoun it.
Person agreement data is also problematic for expletive accounts. As I
explained in Section 1.3.3, the verb in the cleft clause typically shows
agreement with the third person pronoun it, rather than with the clefted
constituent, shown in (82), repeated from (68). Such data is especially
awkward for expletive accounts which invoke movement to extract the
clefted constituent out of the cleft clause, including Rochemont (1986),
since these examples do not have corresponding noncopular paraphrases,
demonstrated in (83).
(82)
It is me who is responsible
(83)
*Me is responsible
In order to accommodate this data, . Kiss (1998: 259), whose main

argument requires movement, suggests that the clefted constituent can also
be base-generated in the postcopular position and coindexed with the whpronoun in the cleft clause. Nevertheless, while expletive analyses invoking
the base-generation strategy can cope with the person agreement data, this
is still suggestive of an alternative account.
In sum, it seems that expletive accounts have less explanatory potential
than the extraposition-from-NP account argued for in Section 1.1. Once we
treat the cleft clause and the cleft pronoun as forming a discontinuous
definite description, the it-clefts pragmatic and structural characteristics
become less mysterious and are found to be consistent with what we know
about other types of specificational construction as well as definite noun
phrases and restrictive relative clauses, more generally. By maximizing correspondence between it-clefts and their noncopular paraphrases, expletive
accounts cannot capitalize on these generalizations. From this perspective,
106
it-clefts cannot be easily incorporated into a unified analysis of specificational copular constructions.
Heggies (1988) work is exceptional in that it does attempt to integrate
an expletive account of it-clefts into a general theory of copular sentences.
Assuming an inverse analysis of specificational sentences, Heggie claims
that all copular constructions involve predication of some sort. Since most
expletive accounts assume that the cleft clause is predicated of the clefted
element, Heggie maintains that it-clefts are like other copular sentences.
For Heggie, the copula is a raising verb which takes a small clause complement. It functions as a verbal operator, creating a predicate out of any
phrasal category via coindexing. This index then spreads to the subject of
the small clause via predication (following Williams 1980). In the case of
the it-cleft then, the cleft clause becomes the predicate and the clefted
constituent functions as the subject of the small clause.
However, there is an important difficulty with Heggies analysis. On the
inverse account, specificational copular sentences always involve nominal
predication. If it-clefts are a subtype of specificational copular sentence, it
follows that these too should involve nominal predication (see Section 1.1).
Assuming an expletive account, however, it-clefts involve a unique form of
predication. Consequently, Heggie (1988: 183) makes the verbal operator
function so unrestricted that it allows be to create small clause structures
which do not exist in any other context, such as the CP-small clause found
in it-clefts. For Heggie then, it-clefts are like other specificational copular
sentences in that they involve predication. However, since they do not share
the same type of predication, it is difficult to see what is gained from this
unified analysis.
Without the assumption that nominal predication generates specificational meaning in all such constructions, Heggies account does not benefit
from the key generalization that it-clefts contain definite descriptions. Thus,
despite Heggies (1988) attempt, expletive analyses are not well-suited to
being incorporated into a general theory of specificational copular sentences. As I see it then, the main problem with the expletive approach is
that it does not prioritize the relation between it-clefts and other specificational constructions. While these authors may recognize that it-clefts are a
subtype of specificational copular sentence, this fact does not form the
basis of their analyses.10 By examining it-clefts in relation to structurally
10. Den Dikkens recent work may represent an exception to this claim. He treats
it-clefts as specificational inversion sentences (as do I). On his account, the
initial it is a (meaningless) pro-predicate (which has undergone inversion) and
A comparison with extraposition accounts of it-clefts
107
less complex sentences, expletive accounts overlook the useful generalizations that exist at a local level in the hierarchy. As Hedberg (1990: 35)
observes, nothing is what it seems on an expletive approach.
3.
A comparison with other extraposition accounts of it-clefts
The discontinuous constituent account argued for in Section 1.1 is therefore

in many ways more successful than expletive accounts at providing explanations for the it-clefts characteristics. But how does this particular analysis measure up to more similar accounts? In what follows, I show how this
analysis compares to other accounts which assume that the cleft clause is
related to the cleft pronoun in some way. I begin by outlining some of the
early extraposition accounts of the 1970s before focusing on more recent
discontinuous constituent accounts. I show that while my analysis draws
heavily from this literature, capturing many of the same insights, it provides a novel perspective from which to explore the crucial role that definite descriptions play in it-clefts.
3.1.
The early extraposition accounts of the 1970s
The extraposition approach to it-clefts became especially popular during

the transformational tradition of the 1970s. Within this decade, a number of
analyses were developed which derived it-clefts from pseudocleft sources
via extraposition. For example, Akmajian (1970) derives it-clefts from their
corresponding wh-clefts, shown in (84) and (85). For Akmajian, wh-clefts
differ from th-clefts like (86), in that the wh-clause is a headless relative,
and so does not function as a full NP (see the discussion in Bolinger 1972).
(84)
It was a car that John bought
the clefted constituent is the underlying subject. The clefted constituent is

structurally related to the cleft clause via asyndetic specification (not predication). On this analysis, the cleft clause is a base-generated, right-dislocated, free
relative containing a radically empty null head which enters into a concord relationship with the clefted element. Den Dikken (2009: 22) concludes that all
specificational copular sentences have a predicational source. Nevertheless,
since on this account, the it-clefts specificational meaning does not arise via
nominal predication, den Dikken does not consider the role of definiteness in
such sentences.
108
(85)
What John bought was a car
(86)
The thing that John bought was a car

(examples from Akmajian 1970)
Akmajian claims that these wh-clauses are generated in the structure given
as (87) below (where what is associated with something). He then invokes a
construction-specific Cleft-Extraposition Rule which extraposes the initial
clause of the wh-cleft to the end of the sentence, leaving the it in subject
position (Akmajian 1970: 161). In some cases, such as (84) above, the whrelative must also be replaced by that (Akmajian 1970: 164).
(87)
[it [John bought something]] ( what John bought)

As I see it, Akmajians account misses a crucial generalization: the cleft

it, what and the thing all function as the article and head noun of definite
descriptions. On this analysis, the extraposition of the wh-clause of a whcleft results in a right-dislocated wh-cleft rather than an it-cleft. Gundel
(1977) agrees, and provides an analysis in which it-clefts are derived from
right-dislocated structures. In right-dislocated wh-clefts, it acts as a pronominal reference to the right-dislocated fused relative, shown in (89). For
Gundel, it-clefts are derived by an optional, construction-specific Variable
Head Deletion rule.11 This rule deletes the head element of the compound
pronoun what as well as the intervening sentence boundary between the
dislocated clause and the postcopular element. The result, Gundel claims, is
an it-cleft containing a sentence-final restrictive relative and a referential
subject (it) that acts as a pronominal copy for the dislocated constituent,
shown in (90).
(88)
What you heard was an explosion
(89)
It was an explosion, what you heard
(90)
It was an explosion that you heard

(examples from Gundel 1977: 543)
11. Gundel (1977: 557) acknowledges that her Variable Head Deletion rule has no
independent motivation in English, and the fact that it can apply only in ID
[specificational] structures is admittedly suspicious.
109
For both Akmajian (1970) and Gundel (1977) then, the cleft clause is a
headless relative clause and the cleft pronoun is a semantically redundant
placemarker or pronominal copy for the extraposed (or dislocated) clause.
However, there are two other transformational analyses which argue (as I
do) that the cleft clause forms a definite description with the cleft it. For
example, Wirth (1978) derives it-clefts (and wh-clefts) from copular sentences whose subjects contain relative clauses with the head nouns one or
thing, as in (91). These (th-cleft) structures undergo extraposition, shown in
(92), before restricted nominal reduction applies, which deletes the head
noun. Since the cannot occur without a following noun, shown in (93), it
must be inserted instead. Wirth (1978: 60) claims that it is a syntactic
variant of the, appearing when there is no head noun in the NP.
(91)
The thing which (that) struck the house was lightening
(92)
The thing was lightening which (that) struck the house
(93)
It(*The) was lightening which (that) struck the house

(examples adapted from Wirth 1978)
Wirth (1978) therefore recognizes the relationship between the cleft it

and the definite article and invokes a rule of extraposition-from-NP (rather
than extraposition of the NP). From this, it follows that the cleft clause
forms a discontinuous description with the subject element. However,
Wirth also makes use of a similar head deletion rule to Gundel (1977). On
Wirths account then, the cleft it does not really function as the antecedent
noun of the cleft clause. In contrast, Bolinger (1972) suggests (as do I) that
the cleft pronoun restrictively modifies the pronoun it. For Bolinger, itclefts are derived from the same rule of inversion (essentially extraposition-from-NP) that creates the structure in (94). This rule inverts the
restrictive relative clause of the analytic compound relative in (95), leaving
the nominal head in initial position.
(94)
That was money which (that) he stole
(95)
That which he stole was money
From this, it follows that the it-cleft in (96) is derived from the structure in
(97). Here, it functions as the nominal head of the restrictive relative that he
stole. While inversion is optional for (95), it applies obligatorily when the
nominal of the main clause is it, as in (97) (Bolinger 1972: 110).
110
(96)
It was money that he stole
(97)
*It that he stole was money
3.2.
The more recent discontinuous constituent accounts
(examples from Bolinger 1972: 109)
Bolingers (1972) and Wirths (1978) transformational analyses therefore

represent forerunners of the discontinuous constituent account of it-clefts.12
This type of analysis has gained renewed interest in recent years, resulting
in a number of different versions of this approach. For example, Percus
(1997) derives it-clefts from an underlying structure containing a definite
determiner and a null head, shown in (98b). The relative clause of this
structure is then extraposed, before the DP (containing the definite article
and CP trace) is realized as it at morphological spell-out, shown in (98d).
On this account then, it is not restrictively modified by the cleft clause;
instead it substitutes for a complete DP.
(98)
a. It is [JOHN] F that Mary saw.

b. [IP [DP the 0 [CP OPi that Mary saw ti]]j [VP tj is John]]
c. Extraposition
[[IP [DP the 0 tk]j [VP tj is John]] [CP OPi that Mary saw ti]k]

d.
It
is John
that Mary saw
Spell-Out: [DP the 0 tk] It
(Percus 1997: 338)
Percus provides support for his analysis by identifying several properties

which it-clefts share with specificational sentences containing definite NPs,
including existential presuppositions and exhaustiveness (see Sections 1.2.2
and 1.2.3). He claims that all such sentences are semantically equative,
containing a different copula from that occurring in predicational sentences.
Percus (1997: 376) notes that while the latter relates an individual and a
predicate and does not allow inversion, the former allows inversion,
selecting two arguments of the type of individuals with the restriction
12. Other early analyses which foreshadow later discontinuous constituent accounts
include Jespersens (1927) transposition theory (see Chapter 1, Section 2.2)
and also Fowler and Fowlers (1908) discussion of itthat. For instance,
Fowler and Fowler analyse the it-cleft It is money that I want in relation to the
sentence It (the thing) that I want is money, where the antecedent it followed
by a defining relative clause with that...gives us the subject of a predication.
111
that one of its syntactic arguments be a Du [unknown individual] (cf.

Declerck 1988; Chapter 3, footnote 16).
Hedbergs (2000: 894) analysis is based on the claim that definite NPs
are comprised of two parts: an INDEXICAL component, which is expressed
by the determiner head and determines the relation of the referent to the
context, and a DESCRIPTIVE component, which is expressed by its nominal
complement and describes the referent (emphasis original). For Hedberg
(2000: 898), it is analyzed as an allomorph of the. In it-clefts then, the
cleft pronoun functions as the definite article head (the indexical component) of the DP, while the cleft clause comprises the descriptive component. This is shown in (99) and (100), where the descriptive component is
highlighted in bold.
(99)
[DP [D the] [NP [NP dog] [PP next door]]]
(100)
[DP [D it][CP who won]]
(examples from Hedberg 2000: 898)
Hedberg (2000: 898) claims that the definite article is realized as it as a

consequence of the extraposed position of the cleft clause; that is, when no
descriptive content (NP or CP) immediately follows.13 On Hedbergs
account, the complement clause of the subject DP is extraposed into a
position internal to another DP inside the VP, shown in (101). The cleft
clause is therefore syntactically adjoined to the clefted constituent in the
position of a nonrestrictive relative clause (Hedberg 2000: 915).14
(101)
[IP [DP [D it]]i [I' [I wask] [VP [DP [DP Max] [CPi [whoj] [C' tj won]]]]]]
For Hedberg (2000) then, the cleft clause is semantically related to the
cleft pronoun, but forms a syntactic constituent with the clefted element,
inside the VP. This allows Hedberg to capture (most of) the generalizations
afforded by a discontinuous constituent account, without coming up against
Delahuntys (1982) VP constituency evidence. However, as was shown in
Section 1.3.2, it-clefts behave like other sentences containing extraposed
13. This aspect of Hedbergs analysis is not supported by the historical evidence. In
earlier periods of the language, the pronoun it sometimes occurs as the immediate antecedent of a restrictive relative (see Chapter 6).
14. Hedberg (1990) offers a different syntactic analysis, in which the cleft clause
(analysed as a restrictive relative) adjoins to an embedded VP rather than a DP.
Nevertheless, on both accounts, the cleft clause forms a constituent with the
clefted element within VP, following Delahunty (1982).
112
relatives in relation to most of these constituency tests (see also Hedberg

1990). Consequently, they do not seem to function as a pressing reason for
positing a construction-specific and very complex type of extraposition (see
Hedberg 2000: 913). As a result of her syntactic analysis, Hedberg analyses
the cleft clause as nonrestrictive, relative to the clefted constituent. I argued
against this position in Section 1.3.1.
Hedberg offers a similarly manifold analysis of the it-clefts semantics.
She suggests (as I do) that while the clefted constituent is a type e referring
expression denoting individuals, the discontinuous constituent is of a higher
type, which is consistent with its analysis as either a predicate (type <e,t>)
or as a generalized quantifier (type <e,t>,t>) (Hedberg 2000: 917). However, elsewhere in her paper, Hedberg (2000: 898) describes the combination of cleft pronoun and cleft clause as a discontinuous referring expression (emphasis added) and she makes use of a separate identificational
copula to obtain a specificational, rather than a predicational, interpretation (Hedberg 2000: 916). As I explained in Chapter 3, Section 1.1, the
two be approach is typical of equative rather than inverse accounts of
specificational sentences.
In subsequent work, Han and Hedberg (2008) propose an analysis of itclefts within the framework of Tree Adjoining Grammar. Like Hedberg
(2000), their aim is to capture the best of both the extraposition and
expletive accounts (Han and Hedberg 2008: 345). In order to do this, these
authors exploit the machinery of Tree Adjoining Grammar, which provides
structures with both a derived tree, which represents surface constituency,
and a derivation tree, which represents semantic composition and syntactic
dependencies (Han and Hedberg 2008: 355). Han and Hedberg suggest that
the cleft clause is syntactically adjoined to the clefted constituent in the
derived tree, while the cleft clause and the cleft pronoun form a syntactic
and semantic unit in the derivation tree (represented by placing elementary
trees for them in a multi-component set).
Han and Hedberg suggest that this analysis reflects the parallelism
between it-clefts and copular sentences with definite descriptions, while
accounting for evidence that favours an expletive account, such as VP constituency and number agreement. However, although their grammatical
model allows it, the fact remains that the syntactic and the semantic structure of it-clefts are very much at odds with one another on this account.
Han and Hedberg (2008) also adopt an equative account of specificational
it-clefts, in which an equative copula takes two DP arguments, shown in
(102). I return to Han and Hedbergs analysis in Chapter 5, where I show
113
that an equative account is not actually capable of explaining the it-clefts

number agreement properties.
(102)
THEz [won(z)] [z = Ohno] (It was Ohno who won)

(Han and Hedberg 2008: 349)
Reeve (2010, 2012) has also proposed a two-tiered analysis of it-clefts.

He suggests that the cleft clause is a restrictive relative which semantically
modifies the cleft pronoun but syntactically modifies the clefted element.
Reeve presents a syntactic analysis which is essentially parallel to Hedberg
(2000); that is, the cleft clause is adjoined to the clefted element within DP.
For Reeve, however, there is no relation of extraposition between the cleft
clause and it. In support of this analysis, Reeve claims that the cleft clause
behaves more like other relatives modifying VP-internal DPs than relatives
modifying subject DPs.15 This begs the question, how can the cleft clause
ever be interpreted as restricting the reference of it (Reeve 2011: 170).
15. For instance, Reeve (2010: 5657) suggests that it-clefts with reduced relatives,
such as (vi) pattern with object relatives rather than subject relatives. While
many authors have observed that extraposed relatives cannot normally be
reduced, shown in (vii), Reeve claims that this is only true for subject relatives,
providing the grammatical example (of a reduced extraposed object relative) in
(viii) (see also Reeve 2012: 39).
(vi)
It was John [sitting outside]
(vii) *A man came in to the room [now sitting outside]

(viii) I saw a man yesterday [now sitting outside]
(examples from Reeve 2010: 57)
However, this data is not straightforward, since the intervening element in (viii)
is not comparable to that in (vii). Once we put a concrete NP between the
extraposed clause and its antecedent, as in (ix), the reduced object relative
becomes less grammatical. Furthermore, extraposed subject relatives become
more acceptable in contexts without this intervening NP, as in (x). This suggests that the restrictions on extraposed reduced relatives are based on gardenpath ambiguities, which the it-cleft is not subject to (see also Section 1.3.1).
(ix)
??I saw a man in the room [now sitting outside]
(x)
?A man came in (yesterday) [now sitting outside]
114
As a solution to this problem, Reeve claims that relative clauses have

two licensing conditions, one thematic and one syntactic, which in it-clefts
are licensed by different DPs (the cleft it and the clefted constituent, respectively). In order to satisfy both conditions, the cleft clause undergoes
obligatory extraposition to VP and the two licensers are related through
semantic equation. Reeve (2012: 4) suggests that specificational sentences
contain a functional head,...Eq(uative), in the extended verbal projection
which semantically equates the denotations of the two DPs. On this
account, the cleft pronoun and the clefted element are of the same semantic
type (e.g. type e). It therefore follows that while the cleft clause is said to
semantically modify the initial it, the pronoun alone comprises a complete,
referring DP. For Reeve then, the it-cleft is a noncompositional configuration; there exists a mismatch between the semantics and the syntax which is
difficult to reconcile.
3.3.
A different extraposition account?
In this section, we have seen that the account of it-clefts developed in this
chapter follows a long tradition of treating the cleft clause as being in some
way related to the cleft pronoun. However, not all of these accounts assume
(as I do) that the cleft clause and the cleft pronoun form a discontinuous
definite description, including Akmajian (1970) and Gundel (1977). Of the
ones that do, not all claim (as I do) that the initial it is restrictively modified
by the cleft clause, including Hedberg (2000), Percus (1997) and Wirth
(1978). Furthermore, many of the more recent accounts, such as Han and
Hedberg (2008) Hedberg (2000) and Reeve (2010), claim that the cleft
clause also forms a syntactic constituent with the clefted element. For these
authors then, the it-cleft configuration is not a straightforward case of extraposition-from-NP (as I have argued). Consequently, the analysis presented
here is perhaps most similar to Bolingers (1972) transformational account
(see Section 3.1).
As I explained in Section 3.2, the most recent discontinuous constituent
accounts combine elements of both extraposition and expletive analyses in
order to account for the it-clefts contradictory properties. However, it is
often difficult to see how the two disparate aspects of these analyses come
together under a cohesive account. As we have seen, the best of both
approach results in some very complicated analyses, which invoke a host of
construction-specific devices to sanction the it-clefts distinctive behaviour.
Consequently, it might be better to account for some of these phenomena in
A comparison with constructional accounts of it-clefts
115
different ways. In this book, I maintain a fairly uncomplicated analysis of

the it-cleft as an extraposition-from-NP construction. I provide new evidence in support of this analysis from a variety of sources and show that
much of the more challenging data is not actually inconsistent with this account, once viewed from a different perspective (see especially Chapter 6).
It is also interesting to observe that most of the discontinuous constituent accounts have assumed that it-clefts are semantically equative (Bolinger
1972; Han and Hedberg 2008; Percus 1997; Reeve 2010, 2012). Instead, I
have incorporated it-clefts into an inverse analysis of specificational sentences which suggests that the concept of definiteness plays a fundamental
role in the creation of specificational meaning. This account of specificational copular sentences provides strong support for analysing it-clefts as
containing discontinuous definite NP predicates. Thus, while many authors
have identified points of similarity between it-clefts and specificational
sentences containing definite NPs, the analysis of specificational meaning
as a class inclusion relation (along with Hawkins 1978, 1991 conceptualization of definiteness), gives an account of how these shared properties
come about (see Section 1 and also Chapter 3).
4.
A comparison with other constructional accounts of it-clefts
The analysis developed throughout this chapter is presented within the

framework of construction grammar. As I explained in Chapter 1, the itcleft is especially well-suited to this treatment, since construction grammar
was designed to account for specialized linguistic patterns which do not
conform to highly general grammatical rules. A constructional approach
encourages us to find generalizations at a more local level. For instance,
when examined in relation to a small family of related constructions, the itcleft is shown to inherit properties from more basic schemas (specifically,
the specificational inversion construction) and to conform to the behaviour
of definite NPs and restrictive relatives, more generally. A constructional
approach is also useful in that it anticipates that the grammar contains some
measure of idiosyncratic information (see Chapter 2). As shown in Section
1.1, once we (momentarily) factor out the non-standard use of it and the
restrictive modification of a detached, pronominal element as constructionspecific properties, our analysis of English it-clefts becomes far more
straightforward, allowing the relevant generalizations to be made.
However, there are other constructional accounts of it-clefts in the literature which are very different from that argued for here. In what follows,
116
I show that while each of these accounts makes reference to some of the
basic claims and concepts that underlie the constructional framework, they
overlook some of the explanatory objectives which the theory was designed
to meet.
For example, Lambrecht (2001) puts forward an expletive account of itclefts, which is largely based on Jespersens (1937) analysis. Specifically,
Lambrecht (2001: 472) agrees with Jespersen that the sequence it isand
the connective pronoun or marker (when present) do not enter fully into the
semantic composition of the sentence. For Lambrecht (2001: 470) then,
the it-clefts main function is as a syntactic focusing device, with it and be
providing the pragmatic role of focus marker (see also Section 1.2.1). On
his account, the gap in the cleft clause is coindexed with the clefted constituent and the cleft clause is a kind of nonrestrictive relative, in that it has
the internal structure of a restrictive relative clause but does not modify its
antecedent (see also Section 1.3.1). In this way, Lambrecht (2001: 473)
simply extends the taxonomy of relative clauses to encompass the specific
type of clause found in it-clefts.
Lambrecht (2001: 466) claims that his approach is constructional, in that
the it-cleft is treated as a noncompositional configuration, with properties
that require independent explanation. For instance, on his account, the itcleft involves a complexity mismatch, expressing a simple proposition via
biclausal syntax (Lambrecht 2001: 466), it contains a construction-specific
type of relative clause, and it has pragmatic properties which are associated
with the construction as a whole (rather than attributed to any individual
component). Lambrechts analysis is therefore constructional in the sense
that he believes, as I do, that the concept and architecture of the construction is extremely useful for the study of specialized linguistic patterns, such
as the it-cleft.
However, while Lambrecht extends his analysis to pseudoclefts (arguing
that initial what and the one are also semantically inert dummy elements,
see Chapter 3, Section 4.3), he does not situate the it-cleft within a hierarchical network of constructions and does not exploit a system of inheritance. As Goldberg notes, constructional taxonomies are organized in a way
that prioritizes similarities of surface form (Goldberg 2006: 23) and aspects
of meaning that go beyond truth-conditional synonymy (Goldberg 1995:
103). By examining it-clefts in relation to truth-conditional paraphrases
which are both formally and functionally less complex, Lambrechts account does not identify (what I believe are) the key generalizations shared
among the family of copular constructions, and instead emphasizes the itclefts idiosyncratic and irregular properties.
117
Unlike Lambrecht, Davidse (2000) does not assume that it-clefts are
semantically equivalent to their noncopular counterparts and instead claims
that both the matrix clause and cleft clause contribute to the constructions
representational semantics. For Davidse, the matrix clause is an identifying
clause; in other words, it functions as a specificational copular sentence.
Using Hallidays (1967) terminology, Davidse labels the pronoun it as the
identified and the clefted constituent as the identifier. In many ways, this is
akin to a value-variable relation. As Davidse (2000: 1121) comments, The
identified can be likened to the unknown x in a mathematical equation
and the identifier to its actual value in that equation.
Davidse (2000: 1125) suggests that the definite pronoun it also performs
a quantifying role, imposing an exhaustively specifying value onto the
complement of be via equation. The relative clause then takes this complement (which has been exhaustively quantified) as its antecedent, incorporating it into a value-variable relation, as defined by Declerck (1988). In
other words, the clefted element constitutes the specific value, or total set
of instances, for the variable or entity involved in the situation designated
by the relative clause (Davidse 2000: 1125). Davidse (2000: 1128) provides evidence that the cleft clause is neither a restrictive nor a nonrestrictive relative and concludes that the relationship between the cleft clause and
its antecedent has a special status.
Davidses constructional account is therefore much more compositional
than Lambrechts (2001), but is in many ways just as idiosyncratic. Like
Lambrecht, she argues that it-clefts can and should be analyzed as
constructions in their own right (Davidse 2000: 1124). This leads Davidse
to claim that it-clefts contain a construction-specific type of relative clause
and that they are properly biclausal, with each clause entering into a
different semantic relationship with the clefted constituent. Davidse (2000:
1127) concludes that it-clefts do not simply express specification - they
SPECIFYINSTANCES AS VALUES CORRESPONDING TO A VARIABLE
(emphasis original). On Davidses account then, specificational meaning is
marked twice: once by the identifying matrix clause and once by the valuevariable relationship expressed by the relative clause. However, aside from
being uneconomical, this is also unnecessary: it increases the complexity of
the it-cleft structure in a way that is not supported by inheritance from the
language system.
Paveys (2004) Role and Reference Grammar analysis also makes use of
the construction as a unit of grammatical knowledge. Following Declerck
(1988), Pavey suggests that it-clefts involve a value-variable relationship,
in which the postcopular element provides the value for the variable
118
described by the cleft clause. She notes that the relationship between the
cleft clause and the rest of the it-cleft construction results in a sens global
particulier that will be characterized differently from both restrictive and
non-restrictive relative clauses (Pavey 2004: 201). For the matrix clause,
Pavey adopts a simple predicational structure, in which the postcopular
referring expression acts as a kind of pragmatic predicate (a concept
borrowed from Lambrecht (1994: 231)). She claims that that the clefted
constituent value modifies the variable expressed by the cleft clause
[and]assists the hearer in making a full identification of the underspecified argument in the cleft clause (Pavey 2004: 200). The syntactic
argument of this postcopular predicate is the non-referring cleft pronoun,
which, for Pavey, serves only to reflect the main specificational function of
the sentence.
On Paveys (2004) account then, the it-cleft is a highly idiosyncratic
construction. Not only is the syntactic predicate a referring expression and
the syntactic argument a non-referring pronoun, but, the clefted constituent
is also a pragmatic predicate, involved in a relation of identification (rather
than predication) with the presupposed cleft clause. For Pavey (2003: 12)
then, the structure, meaning and function of it-clefts is particularly complex since there is no straightforwardly isomorphic link between the various elements at the syntactic, semantic and the pragmatic level. On top of
this, the cleft clause is related to the rest of the sentence in a way that is
both syntactically and semantically idiosyncratic and the cleft it has a
construction-specific meaning. Pavey (2004: 254) makes use of the concept
of the construction primarily as a way of housing these idiosyncratic
properties; she notes that These idiosyncrasies of the construction require
special conditions and these are stated in the constructional template.
Paveys reliance on construction-specific phenomena is to some extent
unexpected, since she views the it-cleft, as I do, as one of a family of
copular constructions, and as a specificational construction in particular
(Pavey 2004: 189). Nevertheless, Pavey (2004: 187) notes that her it-cleft
analysis is not consistent with either equative or inverse accounts of specificational sentences. Furthermore, the it-cleft is syntactically distinct, since
it does not contain two noun phrases connected by the copula (Pavey
2004: 185).16 For Pavey then, the only thing that it-clefts share with other
16. This syntactic difference is also present in Declercks (1988) original analysis.
He notes that while the subject NP of a specificational NP be NP sentence
contains a variable x, shown as the head noun in (xi), the x variable is not
represented in the it-clefts syntactic form. In such sentences, the clefted ele-
119
specificational sentences is a pragmatic (rather than semantic) predicational

(or rather identifying) function (see above).
The accounts of Lambrecht (2001), Davidse (2000) and Pavey (2004)
are therefore constructional in the sense that the it-cleft is recognized as
having the status of a construction; that is, it is treated as a conventional
unit of linguistic knowledge containing construction-internal mappings
between form and meaning (see Chapter 2). However, since these authors
do not invoke inheritance, or look for significant pattern resemblances with
more basic constructions, the it-cleft is not properly situated in the language
network. This leads Lambrecht, Davidse and Pavey to stipulate a lot of
idiosyncratic information, without showing how the it-cleft construction is
motivated. As Goldberg (2003: 118) comments, a given construction often
shares a great deal with other constructions that exist in a language; only
certain aspects of its form and function are unaccounted for by other constructions. Therefore, despite popular misconceptions, the constructional
framework neither supports nor encourages highly idiosyncratic analyses of
extremely noncompositional structures.
In accordance with the principles of construction grammar, the approach
undertaken in this work seeks to maximize inheritance from the language
system, emphasizing both formal and functional correspondences between
it-clefts and other, more basic constructions. The more properties a construction inherits, the more we can say that the construction is motivated
(see Chapter 2, Section 2). In other words, the it-cleft is properly integrated
into, and is therefore supported by, a network of related constructions. Only
after examining it-clefts in relation to the rest of the grammar should the
exceptional or truly construction-specific characteristics be isolated and
even these should ideally be provided with an explanation of some sort (see
Goldberg 2003: 121). Following this model trajectory, we begin, in the
remaining chapters, to examine some aspects of the it-cleft construction
that require explanation (or rather motivation) from outside of the family of
specificational copular sentences.
ment provides the value for the variable described by the cleft clause, shown in
(xii) (Declerck 1988: 185).
(xi) The one(x) that complained was Frank
(xii) It was Frank that complained
Chapter 5
Other varieties of it-cleft
1.
Beyond the archetypal it-cleft
In Chapter 4, the it-cleft construction was integrated into an inverse account

of specificational copular sentences, outlined in Chapter 3. On this analysis,
the cleft clause is an extraposed restrictive relative which modifies the definite pronoun it to form a definite description; this discontinuous constituent
is predicated of the postcopular referring expression, which (due to a fixed
information structure) is interpreted as constituting the membership of the
described set. With this analysis in place, many of the it-clefts properties
were shown to be inherited from the specificational inversion construction
and to conform to the behaviour of definite NPs more generally. However,
there are some varieties of it-cleft that have fewer properties in common
with the family of specificational copular constructions. In what follows, I
show how the analysis developed so far can be extended to accommodate
these different varieties of it-cleft, before going on to ask how they are
motivated, if not by inheritance from the larger specificational schema.
As I explained in Chapter 1, not all of the tokens which appear to be itclefts have a specificational meaning some have a predicational meaning
instead. In these tokens, known as predicational and/or proverbial it-clefts,
the clefted constituent provides descriptive rather than specifying information. For instance, (1) describes the image presented in the poem as a beautiful image. Likewise, the proverbial token in (2) advises that roads which
have no turnings are long roads. As a proverb, it provides reassurance that
a life without some surprises (good or bad) is very monotonous; in other
words, we can expect to experience difficult times if we want to live a full
life.
(1)
He asks God to disperse these mists, which blot and fill My

perspective still as they pass; Or else remove me hence unto that
hill Where I shall need no glass. It is a beautiful image which is
presented here, it is one of longing.
(W1A-018 061, student examination script)
(2)
Beyond the archetypal it-cleft
121
Such examples have often been regarded as forming a structurally distinct

sentence type, which is only superficially related to the specificational itcleft. However, in Section 2, we will see that accounts which examine itclefts in relation to other copular constructions anticipate the existence of
predicational it-clefts and that a discontinuous constituent account lends
itself well to both the specificational and predicational varieties of the construction.
Another domain of variation concerns the range of elements that can occur in the postcopular position of specificational it-clefts. So far, our focus
has been exclusively on sentences of the type NP be NP. However, while
the predicative expression of a specificational copular sentence is always
nominal, the referring expression does not have to be. All specificational
constructions, including it-clefts, allow a variety of phrasal categories to
occur in this position, as demonstrated by the PP-focus it-cleft in (3).
(3)
It was to the Corkscrew that I directed my steps

(W2F-011 009, fiction)
However, since the same range of potential focal elements is not shared by
the different specificational constructions, this property cannot be attributed
to inheritance. Nevertheless, in Section 3, I show that an inverse account is
very good at explaining why some elements make better it-cleft foci than
others and I suggest that historical evidence may be useful in showing how
a construction-specific category of it-cleft foci has emerged.
There is further variation in the discourse-status of the cleft clause. Typically, this component expresses discourse-old information. A very clear
example is given in (4). Here, the fact that someone communicates about
the threat of war is given in the immediate discourse.
(4)
Who communicates about the threat and for what purposes? The
answer is extremely simple. It is domestic politicians who communicate about the threat in order to mobilise public support for their
own policy and power base.
(W2A-017 030, academic writing)
However, in some cases, the cleft clause contains brand-new information.

For instance, the it-cleft in (5), from Birner, Kaplan, and Ward (2007: 323),
occurs as the first sentence of a newspaper article andcould also felicitously begin a history lecture. Here, the information in the relative clause
is not given in the previous discourse and the proposition, that Lewis and
Clark reached the Pacific Ocean, may not even be known to the intended
122
audience. Instead, such sentences often function to inform the header/reader

of a brand-new, yet presupposed, proposition.
(5)
It was 200 years ago this month that Lewis and Clark reached the
Pacific Ocean
(Chicago Tribune, 23/11/05 [Birner, Kaplan, and Ward 2007: 323])
I discuss these examples, labelled informative-presupposition it-clefts by

Prince (1978), in Section 4. Here, I suggest that other specificational constructions do not occur with brand-new information in the predicative NP
and conclude that this construction-specific property cannot be inherited
from the specificational inversion construction a claim which is somewhat at odds with the views of Gundel, Hedberg and Zacharski (2001).
Instead, I suggest that, to explain how IP it-clefts are motivated, we need to
conduct a diachronic investigation.
2.
Predicational (and proverbial) it-clefts1
Perhaps because they are less common than their specificational counterparts, predicational (and proverbial) it-clefts are rarely discussed in the cleft
literature and many authors do not recognize them. For example, Gundel
(1977: 547) claims that the specificational-predicational distinction is not
applicable to it-clefts, which can only have an ID [identifying] interpretation. Furthermore, authors who do discuss these predicational tokens often
treat them as forming a distinct and unrelated sentence-type. For example,
Ball (1977: 68) claims that predicational (including proverbial) it-clefts are
superficially similar but syntactically unrelated to specificational it-clefts.
Likewise, Lambrecht (2001: 502503), who discusses only the proverbial
examples, suggests that they cannot be analysed as true cleft sentences.
Declerck (1983b, 1988) provides a more complicated analysis, in that he
treats proverbial and predicational it-clefts as separate constructions. While
the proverbial examples are said to represent a type of their own, different
from genuine clefts (Declerck 1983b: 16), Declerck integrates the nonproverbial predicational tokens into his analysis of specificational it-clefts,
all the time maintaining that the it-cleft is essentially specificational in
nature (Declerck 1983b: 18).
1. The analysis and arguments put forward in this section were originally presented as a conference paper at the Cleft08 workshop, Berlin (see Patten 2008).
Predicational (and proverbial) it-clefts
123
In what follows, I claim that this treatment of the predicational examples

as distinct from the it-cleft proper (or as somehow partly specificational),
stems from the fact that these authors (Ball 1977; Declerck 1983b, 1988;
Lambrecht 2001) adopt an expletive approach, which struggles to account
for the predicational it-cleft data. I explain that a discontinuous constituent
analysis is, in contrast, well-suited to the treatment of predicational it-clefts
and show that the existence of a predicational variety is expected on (and so
supports) an account which views the it-cleft foremost as a copular construction.
2.1.
An expletive approach to predicational it-clefts
As was shown in Chapter 4, Section 2, expletive accounts typically

examine it-clefts in relation to their noncopular counterparts, and so assume
that the cleft clause is predicated of (or is in some other way related to) the
clefted constituent. From this perspective, the cleft clause exhibits a peculiar mix of properties which makes classification difficult (see also Chapter
4, Section 1.3.1). For instance, Declerck (1983b: 12) claims that the cleft
clause is not a genuine restrictive relative, Lambrecht (2001: 473) suggests
that it is a type of nonrestrictive relative which is unique to cleft sentences
and Ball (1977: 58) maintains that while its internal structure is like a
restrictive relative, it has no apparent head. Regardless of the particulars,
these authors all agree that the cleft clause is not an ordinary restrictive
relative clause.
However, proverbial it-clefts do not show this same correspondence to
noncopular sentences. Instead, authors agree that they are paraphrased most
closely by predicational NP be NP sentences containing restrictive relative
clauses. For example, the meaning of (2) above, repeated here as (6), is not
that, in general, long roads do not have turnings, as in (7), but that roads
without turnings are long roads, as in (8).
(6)
[proverbial it-cleft]
(7)
A long road has no turning
(8)
A/The road that has no turning is a long road

[predicational paraphrase]
[noncopular paraphrase]
On this basis, Lambrecht (2001: 503) suggests that the construction has a
noncleft status, since, for him, the concept cleft is inextricably tied to the
124
semantic equivalence of such examples to noncopular sentences.2 Likewise,

Declerck (1983b: 14) concludes that they are not really cleft sentences but
represent a type of sentence that is homophonous with clefts. They differ
from clefts in that they involve a true restrictive relative clause. Instead,
these authors analyse proverbial it-clefts as derived from, or as instances of,
either extraposition-from-NP (see Ball 1977; Declerck 1983b: 15) or rightdislocation, shown in (9) (see Lambrecht 2001; Declerck 1983b: 16).
(9)
It is a long road, a road that has no turning
[right-dislocation]
We have seen then that the expletive approach cannot be applied to the
proverbial it-cleft. As Lambrecht (2001: 502) observes, it is a construction
whose subsumption under our cleft definition poses serious problems. The
proponents of this approach are therefore forced to examine specificational
and proverbial tokens in relation to two different sentence-types, only one
of which has an NP be NP structure. Unsurprisingly, this results in two different analyses, involving differences in the status of the cleft clause. In a
circular line of argumentation, this is then used as evidence that the proverbial examples should be treated as forming a structurally distinct sentence
type from the specificational it-cleft.
The situation becomes even less clear when we consider non-proverbial
predicational tokens, such as (1) above, reproduced in part here as (10). As
Ball (1977: 61) notes, these examples pattern with proverbial it-clefts in
that they also correspond to predicational copular sentences, shown in (11).
Ball therefore analyses all it-clefts with a predicational meaning in the same
way, as forming a sentence-type distinct from the it-cleft proper.
(10)
It is a beautiful image which is presented here

(11)
The image which is presented here is a beautiful image
However, while the separate treatment of proverbial it-clefts can perhaps be

justified on the basis that they speak of hypothetical circumstances and, as
proverbial expressions, are possibly instances of an older linguistic pattern
(see Declerck 1983b: 1516), there is less reason to assume that the nonproverbial predicational tokens are distinct from the specificational it-cleft.
2. Ball (1991: 149) employs similar criteria; she notes that the predicational cleft
is a construction distinct from the specificational cleft, which can be uncleft.
125
As a result, Declerck tries to incorporate these examples into his expletive

account.
For Declerck (1988), specificational sentences involve a value-variable
relation: in it-clefts, the clefted element specifies the value for the variable
described by the cleft clause. On this analysis, the it-cleft is an inherently
specificational structure. Therefore, while Declerck (1983b, 1988) acknowledges that predicational it-clefts have an ascriptive meaning and
identifies several characteristics that they share with other predicational
sentences, his objective is to account for and explain how predicational
meaning ensues from what is an essentially specificational configuration.
As a result, Declerck (1983b: 18) suggests that tokens like (10) are really
borderline cases: although the information they convey is predicational,
they are formally cleft sentences, and cleft sentences are essentially specificational in nature.
Declerck claims that in such sentences, the modifying adjective (in this
case beautiful) is the only predicational element; the head noun (image) is
specificational, in that it constitutes the value for a variable. He notes that
If we leave out the predicational modifier, what remains must be a good
specificational cleft, shown in (12) (Declerck 1983b: 31).
(12)
It is an image which is presented here
Declerck suggests that the reason why examples like (10) are felt to be
predicational is that the adjective constitutes the only new information: the
rest of the NP is presupposed, representing information that is given or
known. On this account, the predicational meaning of these specificational
it-clefts comes about more or less accidentally (Declerck 1983b: 38).
Thus, from Declercks analysis of these predicational tokens, the claim
that it-clefts are fundamentally specificational in nature can be maintained.
However, Declerck concedes that there are some examples which must be
analysed as purely, rather than partly, predicational it-clefts. For example,
in (13), the head noun issue represents old information: what is new is that
this issue has now become important. However, the fact that become is
used instead of be makes it impossible to maintain that the focal NP as a
whole is specificational (Declerck 1983b: 38). Likewise, the presence of
no in (14) indicates that the entire NP, including the head noun idiot, is exclusively predicational. Declerck (1983b: 18) notes that the substitution
of no for not a in a postcopular NP is possible only if the NP in question is
predicational (more specifically, if it expresses a property).
126
(13)
After the recent troubles it has become an extremely important

issue that the ministers must discuss next week
(14)
It certainly was no idiot who planned this

(examples from Declerck 1983b: 3839)
Declerck treats such examples as exceptional cases. He notes, Purely

predicational cleftsare possible only because a specificational interpretation is explicitly excluded by the lexical material of the head clause, such
as by the use of become or the presence of no (Declerck 1983b: 41). There
is also the suggestion that these examples may be the product of language
change. Declerck (1983b: 38, 39) says that this type of cleft can hardly be
recognized as specificational any more; in fact, it is no longer specificational at all (emphasis added). For Declerck then, it-clefts are only recognized as non-specificational, or as exclusively predicational, as a last
resort.
However, there are some it-clefts which express a predicational meaning
but which do not contain adjectival modifiers or other lexical material said
to suppress a specificational interpretation. For instance, in (15), the it-cleft
is used not to identify the entity who made the suggestion originally (a specific woman), but to inform us that the individual who made the suggestion
originally was female. In this context then, the unmodified clefted constituent a woman denotes a class or property, rather than referring to an individual: it is clearly a predicational NP (see also Hedberg 1990: 57). This is at
odds with Declercks (1983b: 31) claim that such structures are necessarily
good specificational cleft[s].
(15)
So dont debar women from being able to contribute to cricket

because after all women did invent overarm bowling in eighteen
hundred and five and then it was adopted officially sixty years later
by which time everybody had forgotten that it was a woman who
made the suggestion originally
(S1B-021 084, BBC Radio 4 Sport on Four)
Therefore, while Declerck attempts to integrate predicational clefts into

his account, they are ultimately treated as aberrations of the specificational
it-cleft to a greater or lesser degree. However, as shown in (15), this cannot
account for all such examples. Instead, the predicational tokens seem to
form a distinct, and yet structurally-related class of it-cleft. Furthermore, by
recognizing that true predicational (i.e. non-specificational) it-clefts exist,
127
the status of the proverbial it-cleft is called into question. If predicational itclefts are possible, why should the proverbial examples, which also have a
predicational meaning, require a separate analysis? In conclusion then, the
expletive approach is not well-suited to the treatment of predicational
varieties of it-cleft, including the proverbial tokens. It encourages authors
either to overstate the structural differences between predicational and
specificational it-clefts, or to overlook their differences in meaning.
2.2.
Predicational it-clefts and the inverse approach
However, as Hedberg (1990, 2000) and Han and Hedberg (2008) observe,
the existence of predicational it-clefts becomes predictable on an account
which views it-clefts as copular constructions, since, as we saw in Chapter
3, noncleft NP be NP sentences also come in both specificational and predicational varieties, shown in (16) and (17) below.
(16)
(17)
The thoracic surgeon is a heavy drinker
[specificational]
[predicational]
In Chapter 3, I analysed specificational meaning as a reinterpretation of the

nominal predication relation associated with definite NP predicates. On this
account, specificational copular sentences, such as (16), are the inverse of
predicate nominal sentences; that is, the postcopular element John McIntyre
is a referring expression and the precopular element the thoracic surgeon is
a predicative NP. However, this definite NP can also function as a referring
expression when predicated of by a postcopular indefinite NP, as shown in
(17). In (16) then, John McIntyre is classified as the thoracic surgeon; in
(17), the thoracic surgeon is classified as a heavy drinker. Since other types
of specificational copular sentence have also been shown to contain definite
NP predicates (see Chapter 3, Section 4.3), it follows that they too should
correspond to predicational NP be NP sentences in which the definite
element functions as a referring expression.
For instance, in the wh-cleft in (18), the postcopular clause that he could
use a drink is predicated of by the fused relative what (the thing that) John
said. However, in (19), this same phrase functions as a referring expression,
where what John said is described, or classified, as a secret.
(18)
What John said was that he could use a drink
[specificational]
128
(19)
What John said was a secret
[predicational]
As Williams (1983: 430) comments, on an inverse account, the distinction

between predicative and referential NPs corresponds with two types of
pseudoclefts: predicative free relatives occur in the specificational pseudocleft, and referential free relatives occur in the predicational pseudocleft.
Likewise, on the assumption that it-clefts contain discontinuous definite
noun phrases (see Chapter 4), we would expect this element to allow both
predicative and referential readings. For instance, in the specificational itcleft in (20), the discontinuous constituent (it + relative clause) functions as
a predicative NP, denoting a set of which the referent Margaret is the sole
member, an interpretation which follows from the uniqueness entailments
of definite NPs (see Chapter 3, Section 2). However, in the predicational itcleft in (21), repeated from (15) above, this same NP us used to refer to an
individual, while the postcopular element functions as the predicative NP.
Here, it (the one) who made the suggestion originally is classified as a
member of the set of women. Since the predicate is indefinite in this
instance, it carries an exclusiveness interpretation, such that the individual
in question is not the only woman in existence (see Chapter 3, Section 3.1).
(20)
It was Margaret who made the suggestion originally
(21)
It was a woman who made the suggestion originally

(S1B-021 084, BBC Radio 4 Sport on Four)
The existence of a predicational it-cleft variety therefore follows neatly

from an account which assumes both an inverse analysis of specificational
copular sentences and a discontinuous constituent analysis of it-clefts. As a
result, they can be accounted for easily. While specificational it-clefts are
instances of the specificational inversion construction, in which a (discontinuous) predicative NP precedes a referring expression, predicational itclefts inherit from the predicate nominal construction. As we might expect,
the roles are reversed or inverted in such sentences, with a (discontinuous) referring expression preceding a postcopular predicative NP.
On this account then, the predicational examples form a separate construction from the specificational it-cleft, but they are nevertheless structurally related, sharing the same discontinuous (extraposition-from-NP)
structure. From this, it follows that they are sub-constructions of the same
it-cleft schema, shown in Figure 9. While the it-cleft schema is not fully
motivated in present-day English (despite support from a variety of basic
129
information structure generalizations), we can assume that speakers recognize this fixed and irregular structure and so form the more abstract mental
representation (see also Chapter 4, Section 1.1). In Chapter 6, we will see
that the it-cleft schema obtains its formal and functional properties from
configurations existing outside of the present-day language system.
Specificational
It-cleft schema
Specificational it-cleft
schema
Predicate nominal
construction
Predicational it-cleft
schema
Proverbial it-cleft
schema
Figure 9.
Incorporating predicational and proverbial it-clefts into a taxonomy of

constructions involving nominal predication
In addition to their formal resemblance (which brings with it a semantically underspecified definite NP), the two types of it-cleft exhibit the same
predicate nominal semantics (also shown in Figure 9). However, since
specificational it-clefts inherit from the specificational inversion construction and predicational it-clefts inherit from the predicate nominal construction, they differ primarily in how their formal elements and meaning components correspond to one another. While in the specificational examples,
the initial discontinuous definite NP functions as the semantic predicate, in
predicational it-clefts, this is treated as an act of reference.3
3. Further evidence that there are two different types of it-cleft (predicational and
specificational) becomes apparent when embedding under consider. Specificational it-clefts, like other specificational inversion sentences, require the overt
presence of to be when occurring in this environment, shown in (ii) (see also
Chapter 3, footnote 10). However, predicational it-clefts pattern with other pre-
130
On this account, the clefted constituent is either a referring expression,

resulting in a specificational it-cleft, or it is a predicative expression, resulting in a predicational it-cleft. From this perspective, it does not make sense
to speak of the specificational and predicational elements within the focal
NP, contra Declerck (1983b, 1988). Declercks borderline cases are
therefore properly predicational; in (10), repeated here as (22), it is not only
the adjective beautiful, but the entire NP a beautiful image that is semantically predicative. In such tokens, the postcopular head noun (in this case,
image) provides a value for the underspecified head noun contained within
the cleft pronoun it, as shown in the NP be NP paraphrase in (23). As a
result, the postcopular noun is interpreted as given information, while the
attributive adjective bears the main (focal) stress.
(22)
It is a beautiful image which is presented here

(23)
The image which is presented here is a beautiful image
The proverbial examples can also be accounted for straightforwardly on

this approach. As shown in Figure 9, proverbial it-clefts are also instances
of the predicational it-cleft construction. In such tokens, the discontinuous
definite NP has generic reference; that is, it refers to an abstract individual
a hypothetical entity which exhibits properties shared by all entities that
satisfy this description. For example, in the proverbial it-cleft in (24) and its
predicational NP be NP paraphrase in (25), it (the mother) that has such
children (that is, the hypothetical mother that has children which behave as
well as these ones do) is described, or classed, as being necessarily a happy
mother.
dicational NP be NP sentences in that the subject-predicate string can occur as

the complement of consider without to be, shown in (iv). Declerck (1983b: 30)
notes this parallel, but nevertheless insists on the essentially specificational
nature of clefts.
(i)
I consider his best friend *(to be) John
(ii) I considered it *(to be) John who is his best friend

(iii) I consider John (to be) his best friend
[predicational NP be NP]
(iv) I consider it (to be) an interesting subject that they are discussing tonight
(examples from Declerck 1983b: 30)
(24)
It is a happy mother that has such children
(25)
The mother that has such children is a happy mother
131
Because such sentences involve a particular type of reference and because

they have specialized stylistic and metaphorical functions, we can recognize a distinct proverbial it-cleft construction, shown in Figure 9. Furthermore, since each proverb is a learned and invariable saying, the instances of
the proverbial it-cleft construction should also occur within our taxonomy
as highly entrenched and substantive constructs (see Chapter 2, Section 3).
For reasons of space, this is not represented in Figure 9.
Therefore, the discontinuous constituent account of it-clefts argued for
here allows us to provide proverbial and non-proverbial predicational itclefts with a unified analysis, while at the same time acknowledging their
familial relationship with specificational it-clefts, contra Ball (1977) and
Declerck (1983b, 1988). In turn, the proverbial examples could provide us
with a clue to finding out more about the it-clefts structural configuration.
As Declerck (1983b: 15) observes, the type of sentence under consideration is essentially proverbial and, like most proverbs, reminiscent of an
older stage in the English language. As we have seen, for Declerck, this
counts as an argument for analysing the proverbial examples as forming a
distinct sentence type from the modern-day specificational it-cleft. However, on the assumption that all types of it-cleft (including the proverbial
tokens) share the same structural configuration, it follows that the it-cleft
schema might well be a relic from an earlier period of the language. I
explore this hypothesis in Chapter 6, where I research the history of the itcleft construction.
2.3.
Predicational it-clefts and the equative approach
So far, then, we have seen that predicational (and proverbial) it-clefts can
be successfully accounted for assuming the analysis of it-clefts and copular
constructions developed over Chapters 3 and 4. This represents an improvement on the expletive approach of Ball (1977), Declerck (1983b, 1988) and
Lambrecht (2001), where the specificational-predicational distinction in itclefts is eliminated rather than explained. However, as we saw in Chapter 4,
Section 3.2, there are alternative discontinuous constituent accounts in the
cleft literature, of which some have been argued to be able to accommodate
predicational it-clefts (see especially Han and Hedberg 2008). Where they
132
differ from the current account is in assuming that specificational it-clefts

are semantically equative. In what follows, I show that the inverse approach
adopted here is in a better position to explain the predicational it-cleft data.
On Han and Hedbergs (2008) account, specificational it-clefts involve a
two-place equative predicate taking two referential arguments: the discontinuous constituent and the clefted constituent. This is represented using the
formula in (26). In contrast, we have seen that predicational it-clefts contain
a postcopular predicate with a single argument: the discontinuous constituent. This is shown in (27).
(26)
THEz [won(z)] [z = Ohno] (It was Ohno who won)
(27)
THEz [beat(z, John)] [kid(z)] (It was a kid who beat John)
(examples from Han and Hedberg 2008: 349)
The specificational-predicational distinction here is not quite as tidy as on

the account outlined in Section 2.2 above. There, it was suggested that specificational and predicational it-clefts exhibit the same predicate nominal
semantics; they differ mainly in terms of which elements perform which
function. For Han and Hedberg, on the other hand, these two varieties of itcleft have very different semantic structures, each involving a distinct
copular verb and argument structure. More problematically, however, their
analysis is unable to account for an observable discrepancy between specificational and predicational it-clefts.
As we saw in Chapter 4, Section 1.3.3, specificational it-clefts referring
to plural entities are nevertheless introduced by a singular cleft pronoun.
For instance, while the postcopular NP in (28) refers to two individuals, the
discontinuous NP cannot be introduced by the plural pronoun they. However, as Ball (1977) observes, predicational it-clefts with postcopular plural
NPs are introduced by plural pronouns, rather than the singular it, as in
(29).
(28)
It is (*They are) John and Margaret who are responsible
(29)
They are (*It is) nice boots youre wearing
This data is indicative of a semantic difference between the discontinuous

constituents of specificational and predicational it-clefts. On my inverse
account, the discontinuous description of a predicational it-cleft refers to
individuals. As such, it is expected that this constituent will be marked for
133
number.4 However, the discontinuous description of specificational it-clefts

is a predicative, set-denoting NP. In Chapter 4, Section 1.3.3, I suggested
that in examples such as (28), this NP is conceptually both singular and
plural, in that it denotes a singular set made up of plural members. While
the matrix copula is shows agreement with the singular set (marked as singular by it), the embedded verb are agrees with the membership of this set.
However, for Han and Hedberg (2008), the discontinuous constituents
of both specificational and predicational it-clefts are semantically referring
expressions. Consequently, these authors do not anticipate any differences
in the behaviour of these NPs. While, on their account, the plural marking
of the predicational it-cleft in (29) is to be expected, they cannot begin to
explain why specificational it-clefts like (28) do not share this property.
They note, Why equative clefts require singular cleft pronouns when they
contain a plural clefted constituent does not follow from our theory and
remains a puzzle (Han and Hedberg 2008: 372).
In contrast, it is the plural marking of predicational it-clefts that is problematic for Reeves (2010) equative account (see Chapter 4, Section 3.2).
Based upon the specificational data, Reeve concludes that clefts only allow
neuter singular pronominal subjects, regardless of the gender and number
of the clefted constituent, and builds a syntactic explanation for this into his
analysis. Consequently, predicational it-clefts with plural subjects are unaccounted for. Reeve (2010: 113) suggests that such examples do not involve a true cleft structure and provides these false predicational clefts
with an analysis similar to that argued for here for all cleft types.
It therefore seems that a discontinuous constituent approach is more
successful at accounting for the predicational it-cleft data if we assume an
inverse analysis, on which the precopular NPs of specificational and predicational sentences are semantically distinct. In turn, the data provides
further evidence against an equative analysis of specificational copular
sentences (see also Chapter 3, Section 1).5
4. Den Dikken (2009) also uses this evidence to show that while the cleft pronoun
of specificational it-clefts is (for him) a pro-predicate, the initial it of predicational it-clefts is a referential pronoun.
5. However, Birner, Kaplan and Ward (2007) propose a particular type of equative analysis which they argue can account for such data. Focusing on demonstrative clefts (in our terminology), they analyse the cleft pronoun that and the
clefted constituent as semantically referring expressions. The authors suggest
that clefts with plural foci can only occur with singular subject pronouns if
there is an open proposition (OP) in the discourse. They claim that in such sentences, that refers to the instantiation of the variable in the OP, rather than to
134
In sum, the account of specificational meaning and specificational itclefts developed throughout Chapters 3 and 4 lends itself well to the treatment of predicational and proverbial it-clefts, and is better at accounting for
this data than either expletive or equative analyses. In particular, I have
suggested that while predicational it-clefts are structurally and semantically
related to specificational it-clefts, they inherit from (and are thus motivated
by) the predicate nominal construction, rather than the specificational inversion construction (see Figure 9 above). It follows from this that while
the discontinuous definite NP in specificational it-clefts denotes a set of
entities, this same constituent in predicational it-clefts is used to refer to a
member, or to some of the members, of a set.
3.
It-clefts with non-nominal foci
So far, the account developed here has focused on the level of correspondence between it-clefts (of both specificational and predicational varieties)
and sentences of the type NP be NP. As a result, the discussion has centred
on it-clefts with noun phrases in the postcopular position. However, while
such examples are known to represent the most common variety, specificational it-clefts can occur with a range of clefted elements, such as prepositional phrases, clauses and adverb phrases, shown in (30), (31) and (32),
respectively.
(30)
It was to the Corkscrew that I directed my steps

(W2F-011 009, fiction)
(31)
Its only if you have one of them and your parents have a licence
that youre covered under that (S1A-078 085, direct conversation)
(32)
And it was only then that it i it uh the wound opened

(S1B-066 025, legal cross-examination)
the plural entities in the discourse or situational context; this variable is singular
regardless of the cardinality of its instantiation (Birner, Kaplan and Ward
2007: 330). However, the existence of a salient OP is not as reliable as the
specificational-predicational distinction in predicting the form of the subject
pronoun in clefts with plural foci: specificational tokens with plural foci occur
with singular subject pronouns even in discourse initial position (despite the absence of a salient OP). Nevertheless, some aspects of this analysis come close
to the account of number agreement argued for here; on both accounts, the cleft
pronoun relates to a singular concept rather than referring to plural individuals.
135
Predicational it-clefts, on the other hand, do not share this property; in

these sentences, the clefted constituent is always nominal. This state of
affairs provides strong support for the analysis argued for here, in which itclefts are situated within a family of constructions that share the same
predicate nominal semantics. Since, as we have seen, the clefted constituent
in predicational it-clefts is semantically predicative, we would not expect
anything other than an NP to occur in this position. Given that the precopular referring expression is also nominal (since the discontinuous definite
NP structure represents a substantive feature of the it-cleft schema), the
predicational it-cleft is a true NP be NP construction.6 In specificational itclefts, on the other hand, the discontinuous predicative expression is inherently nominal, but the clefted referring expression does not have to be.
It-clefts7 with non-nominal foci are therefore anticipated on this account
since predicative NPs can sometimes be predicated of non-nominal items,
such as the PP subject under the bed in (33). Like other kinds of it-cleft
then, those with non-NP foci contain a discontinuous definite description
which is made up of the pronominal head it and a restrictive relative clause.
For instance, the PP-focus it-cleft in (34) corresponds to the th-cleft in (35)
and the wh-cleft in (36).
(33)
Under the bed is a great place to hide
6. For Reeve (2010: 54), the impossibility of predicational it-clefts with adjectival
predicates supports his claim that the cleft clause syntactically modifies the
clefted constituent; he suggests that (v) is ruled out because the clefted AP
feline does not correspond to the gap in the cleft clause, shown in (vi). He
notes that for other discontinuous constituent accounts, on which (v) corresponds to the predicational copular sentence in (vii), the judgment of ungrammaticality is mysterious (see also Reeve 2012: 3537). However, on the
account argued for here, in which it-clefts always involve nominal predication,
the impossibility of (v) is explained easily.
(v)
*It is feline that I am pointing at
(vi) *I am pointing at feline

(vii) The thing that I am pointing at is feline
(examples from Reeve 2010: 52, 54)
7. In keeping with the cleft literature, I continue to use the term it-cleft as a convenient shorthand for specificational it-cleft. In the remainder of this work, I
explicitly indicate where the discussion turns to predicational it-clefts or relates
to the it-cleft schema which encompasses both of these subtypes.
136
(34)
It was with a knife that I cut it
(35)
The way that I cut it was with a knife
(36)
How I cut it was with a knife
(Fowler and Fowler 1908)
That such examples can be so easily integrated into a discontinuous constituent account represents a further advantage of this type of analysis. In
contrast, on some expletive accounts, it-clefts with non-nominal foci represent a structurally distinct sentence type from those containing postcopular
NPs. As we have seen, on this type of account, the cleft clause is related in
some way to the clefted constituent. For Delin (1989), when the clefted
constituent is nominal, the cleft clause can be analysed as a restrictive relative.8 However, since relative clauses cannot modify non-nominal antecedents, she argues that it-clefts with non-NP foci instead contain a sentential
complement (see also Ball 1994a and the discussion in Chapter 8, Section
3.1).
Nevertheless, while the discontinuous constituent analysis argued for
here can straightforwardly accommodate examples with non-nominal foci,
it is difficult to properly explain, or even describe, which elements can and
cannot occur as the clefted constituent. For one thing, the range of possible
it-cleft foci differs from that of other specificational copular sentences. For
example, it-clefts can occur with certain prepositional phrase foci that are
not found in other constructions, shown in (37) and (38).
(37)
It is with great pride that I accept this award
[it-cleft]
(38)
*How I accept this award is with great pride

[wh-cleft]
(examples from Gundel 1977: 548)
Since they cannot be derived from equally acceptable pseudocleft sources,

such examples present a particular problem for transformational analyses
(see Chapter 4, Section 3.1). On this basis, Pinkham and Hankamer (1975)
conclude that it-clefts with non-NP foci are derivationally distinct from
those with NPs in postcopular position. They suggest that only the latter
can be derived by extraposition (forming deep-clefts); all other it-clefts
8. As Delin notes (1989: 60), a restrictive relative analysis for it-clefts with
nominal foci is not uncontroversial on the expletive approach, since relative
clauses cannot ordinarily modify full NPs (see also Chapter 4, Section 1.3.1).
Likewise, on the account argued for here, the restrictive modification of pronominal it awaits further justification (see Chapter 6).
137
must be derived from their corresponding noncopular sentences via the

extraction of the clefted element (forming shallow clefts).9
On a non-derivational extraposition account, such as the one argued for
here, it-clefts like (37) do not present this same problem. However, what
they do show is that the acceptable it-cleft foci comprise a constructionspecific category, governed by factors peculiar to this construction. Since
the range of it-cleft foci is not shared among the family of specificational
copular constructions, it cannot be accounted for by inheritance and instead
requires an independent explanation.
However, it is difficult even to identify the boundaries of this category,
since some XPs make for acceptable foci only in certain circumstances. For
this reason, we cannot rely on a rule-based, syntactic account to explain
which items can and cannot occur in postcopular position. For instance, Delahunty (1982, 1984) devises a syntactic account to sanction the full range
of it-cleft foci (including NPs, PPs, AdvPs, particles, APs and quantifier
phrases), while excluding all other constituents. For Delahunty, the clefted
constituent and the cleft clause are syntactic sisters within VP (see Chapter
4, Section 1.3.2). As a result, he makes use of Jackendoffs (1977) rules for
the projection of V to explain why only certain syntactic categories occur in
postcopular position. For example, the nonexistence of a phrase structure
rule like VP V VP S accounts for the absence of it-clefts with VP foci,
such as (39).
(39)
*It was complain that Frank did
However, because these phrase structure rules apply wholesale to each

phrasal category, they do not predict that some members of a category can
make for better (or more acceptable) it-cleft foci than others, or that their
relative acceptability can depend on a variety of factors. For instance, while
adverbial foci are sanctioned by Delahuntys rules, the AdvP-focus it-cleft
in (40) is questionable. As shown in (41), the focal adverb slowly becomes
more acceptable when placed in a contrast set with other manner adverbs.
To account for this data, Delahunty (1984: 80) appeals to the it-clefts pragmatic properties; he suggests that the unacceptability of (40) perhaps has
to do with the uniqueness and exhaustiveness implicatures associated with
the form. For Delahunty, (41) is more acceptable than (40) because the itclefts implicatures are negated.
9. For other attempts to account for it-clefts without pseudocleft sources from
within a transformational analysis, see Emonds (1976) and Gundel (1977).
138
(40)
?It was slowly that Mary dressed to go out
(41)
It wasnt only slowly that Mary dressed, but carefully too

(examples from Delahunty 1984: 80)
Since Delahunty ultimately relies on a pragmatic explanation to account

for the differing acceptability of permissible it-cleft foci, the rationale for
positing rule-based, syntactic restrictions is called into question. As we
have seen, this construction-specific category cannot be satisfactorily
defined in terms of general syntactic rules or major syntactic categories.
Perhaps then, the range of it-cleft foci is governed purely by function rather
than form. Assuming the inverse analysis of specificational sentences developed over Chapter 3, the specificational it-cleft contains a discontinuous
predicative NP and a postcopular referring expression. We might expect
then that only those elements that can be given a referential interpretation
will be able to occur in the postcopular, focal position. While this approach
cannot explain why it-clefts have a different range of foci from other specificational copular constructions, it nevertheless helps us to understand why
some items are more likely than others to occur in this position.
Since NPs denote objects, they are well-suited to the referring function
(see Croft 1990). Therefore, we anticipate that NPs will make up the most
common class of it-cleft foci (by my calculation, accounting for 311 (77%)
of the 404 it-clefts in the ICE-GB). Indeed, some authors, including
Davidse (2000: 1116), claim that the postcopular position is a strictly
nominal slot: all non-NP elements occurring in this position are rankshifted into the category of NPs via nominalization. Similarly, for Partee
(2004a), non-nominal focal elements are nominalized via the typeshifting
operation nom. However, it isnt really true to say that non-nominal it-cleft
foci are reclassified as noun phrases. For instance, as Partee observes, the
non-NP foci of cleft sentences do not display the same morphological
markings as other nominalized forms. In discussing the specificational
adjective phrase focus wh-cleft in (42), Partee (2004a: 198) asks why is
the form here unusual rather than unusualness?
(42)
What John is is unusual
(Partee 2004a: 190)
Partee (2004a: 199) chooses to leave the question of what licenses the
proposed zero-morphology of unusual on the specificational reading of
(42) unanswered. However, the evidence suggests that nominalization is
139
not necessarily involved in sanctioning the non-NP foci of specificational

inversion sentences.
Instead, these non-NP items appear to be unchanged with regard to both
meaning and form. Only their function is altered, with characteristically
modifying or predicative elements being employed as referring expressions.
A similar account is provided by . Kiss (1998). She suggests that the focal
element in identificational (our specificational) sentences must denote a
type e individual. Since VPs and other predicative elements do not denote
individuals, they do not ordinarily make for acceptable clefted constituents,
shown in (43). Nevertheless, . Kiss (1998: 262) follows Szabolcsi (1983)
in assuming that non-individuals can sometimes be individualized, that is
presented as discrete entities. For example, (44) involves the listing of
predicative APs, in which a two-member set of properties (including tired
and sick) is established (. Kiss 1998: 262, italic original).
(43)
*Its sick that he was
(44)
Its not sick that he was but tired

(examples from . Kiss 1998: 262)
On the account presented here, we might say that by contrasting sick with
tired, the latter is conceived of as a distinct property (situated among a
restricted background set of related properties), which uniquely matches the
description it (the thing) that he was, and so comprises the membership of
this set.
Likewise, focusing adverbs often serve to make characteristically nonreferring foci more acceptable. For instance, the adverb only improves the
acceptability of the AdvP-focus it-cleft in (45) by placing the manner
adverb gradually in relation to some excluded alternatives (see also (41)).
(45)
And its only gradually that people with the qualities required for
success in a command economy can develop the qualities that are
required to survive in a free market economy
(S2B-047 068, non-broadcast, scripted speech)
As Borkin (1984: 127) notes, other adverbs without a contrastive concern

for alternatives, including just, precisely and exactly, can also function to
underscore the correctness of identification. These various coercion strategies produce the same effect: the clefted element is presented as a discrete
entity with clear-cut boundaries.
140
By providing a functional (rather than formal) explanation for the range

of it-cleft foci, we can account for the fact that this construction-specific
category is not comprised of complete classes of phrasal items. We are also
able to explain why some property-denoting elements have to be coerced
into the postcopular slot in order to qualify as acceptable foci. However,
our account cannot yet explain why the it-cleft seems to have different
criteria for measuring what constitutes acceptable foci from other specificational copular constructions. For instance, as I go on to show in Chapter
8, the wh-cleft construction is in many ways more relaxed than the it-cleft
about what can enter into the referential slot, allowing a greater variety of
property-denoting items (including APs and VPs) to occur in postcopular
position, shown also in (42). I suggest that a possible way to account for
these differences is to examine how these constructions have developed
over time (see also Chapter 7).
4.
Informative-presupposition (IP) it-clefts
As I explained in Chapter 4, the it-cleft has a fixed information structure.

Like other specificational inversion sentences, the construction contains a
postcopular focal element. On this focus structure, which is associated with
specificational meaning, the membership (rather than the set) information is
at issue (see Chapter 4, Section 1.2.1). Like other sentences containing
definite NPs, the it-cleft is also presuppositional. The discontinuous definite NP carries an existential presupposition such that the entities which
satisfy the description are assumed to exist (see Chapter 4, Section 1.2.2).
This information is treated as a precondition to the main assertion. From
this, it follows that it-clefts, like other specificational inversion sentences
with definite predicative NPs, are associated with expressing new information in the postcopular XP and old information in the precopular (or discontinuous) NP. In terms of information structure then, the archetypal itcleft behaves in ways that are consistent with its treatment as a specificational inversion construction.
However, it is well known that the relationship between focal and new
information, and between presupposed and old information, is imperfect.
Entities which are already known to us, and which have been mentioned in
the discourse, can nevertheless be in focus. For instance, (46) could occur
towards the end of a discussion about John and the other members of the
surgical team. What is new or unpredictable in (46) is not John, but the fact
that John (as opposed to any other entity) instantiates the type best surgeon.
(46)
141
The best surgeons got to be JOHN
Likewise, information which is presupposed is not always manifestly old.

Presupposed information need not be present in the prior discourse or even
part of shared knowledge. Instead, it could be inferable from, or simply
consistent with, the speech participants mutual cognitive environment. For
example, (47) could form part of a discourse about an impending operation.
In this instance, the presence of an anaesthesiologist may not be already
known to the hearer and may not have been mentioned in the preceding
discourse. Nevertheless, this information is inferable from the shared script
or frame of operations i.e. from experiential or cultural knowledge about
operations in general (see also Chapter 3, Section 3.3).
(47)
The anaesthesiologist is Dr. Vijay Kochar
To some extent then, we expect that it-clefts, like other specificational

inversion sentences, will deviate from the archetypal old-NP be new-XP
configuration. However, what is less clear is how our analysis can account
for instances such as (48), which contain brand-new information in the cleft
clause. Here, the existence of an individual who once said Laws are silent
at times of war does not form part of a previous assertion and is not
necessarily shared or inferable knowledge.
(48)
(Start of lecture)
(Patten 2010: 222)
Are these examples, labelled informative-presupposition it-clefts by Prince

(1978), also predictable from the behaviour of specificational inversion
sentences and definite NPs more generally?
For Hedberg (2000) and Gundel, Hedberg and Zacharski (2001), such
examples fall out straightforwardly assuming an analysis on which it-clefts
contain discontinuous definite descriptions. Using Gundel, Hedberg and
Zacharskis (1993) Givenness Hierarchy, these authors note that the entities
referred to by definite NPs are very often unfamiliar to the addressee. For
instance, Gundel, Hedberg and Zacharski (2001) find that for roughly 44%
of the tokens in their corpus, the addressee could not be expected to have an
existing representation of the entity in memory. On this basis, the authors
conclude that the definite article in English is actually unspecified for
familiarity and that it only requires the addressee to be able to assign a
142
unique representation to the phrase by the time it has been processed

(Gundel, Hedberg and Zacharski 2001: 288).
They go on to claim that this account can be naturally extended to itclefts, on the assumption that the cleft it and the cleft clause form a discontinuous definite description. Citing the corpus studies of both Delin (1989)
and Hedberg (1990), Gundel, Hedberg and Zacharski (2001: 292) find that
the proportion of it-clefts containing nonfamiliar cleft clauses is relatively
high, ranging between 49% and 72%, respectively. They conclude that such
examples should not be treated as exceptional cases requiring independent
explanation. As Hedberg (2000: 903) states, informative presupposition itclefts are analogous to definite descriptions whose referent is uniquely
identifiable but not yet familiar. Thus, a notable advantage to this
approach is that it allows for a unified account of all types of specificational
it-cleft. In other words, the informative-presupposition examples are shown
to behave in ways that are consistent with the behaviour of definite NPs
more generally.
However, there are several reasons to suspect that the facts are not quite
as simple as Hedberg (2000) and Gundel, Hedberg and Zacharski (2001)
suggest. For one thing, these authors group inferable and new information
together as forming a single nonfamiliar category. They reason that new
and inferable entities are not already in memory (at least not necessarily
so) at the point when the phrase that refers to them is encountered
(Gundel, Hedberg and Zacharski 2001: 280). However, it is not always
clear where the boundary lies between old and new information and the
correct classification of inferables is both problematic and debatable (see
Prince 1981: 252). For many authors, including Hedberg and Fadden
(2007) and Prince (1992), inferables share the same information status as
the entities from which they are inferable. On this approach, most inferables are classified with old, rather than with new, information. The evidence for this association comes from Birner (1994, 1996). In her study of
English inversion sentences, she finds that inferrables are treated exactly
as though they had been explicitly evoked. That is, inferrables are treated
as no less given in this sense than explicitly evoked elements (Birner
1996: 102, emphasis original).
Therefore, it seems that Gundel, Hedberg and Zacharskis (2001) claims
regarding the proportion of unfamiliar entities referred to by definite NPs
may be somewhat overblown. Of course, this does not undermine the
observation that definite NPs do not always refer to familiar entities. For
instance, Hawkins (1978) mentions several unfamiliar uses of the definite
article in noun phrases containing explanatory modifiers. In all such cases,
143
the modifier itself takes over the role of previous discourse, and enables
the hearer to identify some set of objects within which he is to locate the
referent (Hawkins 1978: 149).10 For instance, while (49) does not require
that the hearer has prior knowledge of the woman in question, (50), without
the relative clause, does presuppose familiarity on the hearers part.
(49)
Whats wrong with Bill? Oh, the woman he went out with last
night was nasty to him.
(50)
The woman was nasty to him. (examples from Hawkins 1978: 131)
For Hawkins (1991: 411), the reason for this difference is that (49) contains
an establishing relative clause which extends the relevant P-set information
(the facts and entities contingently associated with Bill) in a way that is
compatible with the hearers knowledge. In order to accomplish this, the
hearer-new entity (the woman) must be explicitly related or anchored to
an individual who IS mutually known in this case to Bill (Hawkins 1991:
411, emphasis original).
However (and here we come to the second complication), definite NPs
in specificational sentences do not appear to share this same range of unfamiliar uses. For instance, it would undoubtedly seem strange to utter (51) if
the referent that satisfies the description woman he went out with last night
were not at issue; that is, if the information that Bill went out with a woman
last night were not already known.
(51)
The woman he went out with last night was Cathy Martin
In specificational sentences then, definite descriptions are associated with

familiar information. As we saw in Chapter 3, Section 3, brand-new, anchored information can be incorporated into the precopular predicate if the
indefinite article is used. For instance, in (52), the indefinite article indicates that the description woman he went out with last night is hearer-new
information and informs us that there is a woman who Bill went out with
last night. Here, it is still possible that this description characterizes Cathy
10. For Abbott (2004: 137), this represents a further difficulty with Gundel, Hedberg and Zacharskis (1993, 2001) model. On their hierarchy of cognitive statuses, it is the article which encodes degree of accessibility. However, Abbott
notes that this approach does not satisfactorily explain why more accessible
referents require less descriptive information in the NP and vice versa.
144
Martin uniquely; in other words, that Cathy Martin represents the membership of the (potentially singleton) set of women he went out with last night.
(52)
A woman he went out with last night was Cathy Martin
These indefinite NP subjects therefore seem to fill a gap in the function

of definite NPs. Like their definite counterparts, indefinite specificational
subjects are anchored to the conversation space, they denote restricted
(rather than general) sets, and exclusiveness is not implied (see Chapter 3,
Section 3). This observation ties in well with Hawkins (1991: 417) claim
that the and a provide a grammatically, and psycholinguistically real contrast set, in which the is the logically stronger member of the pair. He
notes that the interpretation of a depends on whether the could not have
been used instead, and from the precise interpretation of the when it could
have been used (Hawkins 1991: 417). In other words, indefinite NPs will
display characteristics associated with definite NPs, only when there is no
longer any competition between the and a (Hawkins 1991: 422).
Moreover, the association between definite NPs and old (or salient)
information in specificational copular sentences does not seem to be reliant
on their preverbal position. For instance, it is difficult to imagine a context
for the specificational non-inversion sentence in (53) in which the identity
of the referent matching the description woman he went out with last night
is not at issue; in other words, where the existence of such a woman is not
already known.
(53)
CATHY MARTIN was the woman he went out with last night
This is not to say that sentences with postcopular definite NP predicates

cannot occur with unfamiliar information in the postcopular position. However, such examples are perhaps better classified as predicational, rather
than specificational copular sentences. For instance, (54) does not require
that the hearer already knows that Bill went out with a woman last night.
However, since the referent Cathy is discourse-old information, (54) seems
to be about this individual, describing, or ascribing a property to, Cathy.
(54)
This is Cathy. Cathys the woman he went out with last night.
The third and final complication is that informative-presupposition itclefts seem to perform a pragmatic function which is different from many
other unfamiliar uses of definite noun phrases. Given the it-clefts structural
145
configuration, we might expect the cleft clause to sometimes function as an

establishing relative clause. However, the cleft clause of informativepresupposition it-clefts does not have to be anchored to the prior discourse
and is not necessarily built on (forming an extension of) an aspect of the
hearers existing knowledge. For example, (48) above, repeated for convenience as (55), provides brand-new, unanchored and non-inferable
information.
(55)
(Start of lecture)
(Patten 2010: 222)
Such instances therefore seem to be rooted not in the hearers existing

knowledge but in the area of general knowledge, or established fact. As
Prince (1978: 899900) notes, the informative-presupposition it-cleft serves
TO MARK A PIECE OF INFORMATION AS FACT, known to some people
although not yet known to the intended hearer (emphasis original). In truth
then, the it-cleft is very much associated with known information, so much
so that it goes beyond consideration of what the hearer knows.
The situation, then, is more complicated than Hedberg (2000) and Gundel, Hedberg and Zacharski (2001) suggest. As we have seen, informativepresupposition it-clefts exhibit a discourse-pragmatic function which is not
shared by other specificational copular sentences, with or without inversion. For instance the pseudoclefts which correspond to (55) are not so
suited to discourse-initial position, shown in (56) and (57).
(56)
(Start of lecture)
#The one who once said, Laws are silent at times of war was
Cicero
[specificational inversion sentence]
(57)
(Start of lecture)
#CICERO was the one who once said, Laws are silent at times of
war
[specificational non-inversion sentence]
The suggestion here is that the audience should already know that someone
once said, Laws are silent at times of war. From this, it follows that the
informative-presupposition it-cleft cannot be explained as a straightforward
product of inheritance, nor does it seem to be governed purely by the word
order considerations which result from the extraposed position of the relative clause (contra Abbott 2000: 1429).
146
Furthermore, we have seen that the specialized function of the informative presupposition it-cleft differs from many of the recognized unfamiliar
uses of definite NPs in other contexts and represents a very extreme form of
unfamiliar use. In definite NPs, new information is typically presented as
an extension of the hearers existing knowledge; it is anchored to, or built
upon, what the hearer already knows. In this way, definite NPs in general
seem to be associated with hearer-old information. However, in IP it-clefts,
the presupposed information does not necessarily have to be (assumed to
be) consistent with, or in any way expected from, the hearers knowledge.
As Prince (1978: 898) comments, the whole point of these sentences is to
INFORM the hearer of that very information. Nevertheless, the new information is presented as non-debatable because it is marked as established
fact. It-clefts are therefore associated with factual information. To formalize this distinction, Prince (1978: 903) reserves the term given for information assumed to be in the hearers consciousness and known to refer to
information that the speaker represents as being factual. For Prince, it is
the latter notion which is pertinent to it-clefts.
In sum, the characteristics of these informative-presupposition examples
are not entirely predictable from analyses in which it-clefts contain discontinuous definite descriptions, particularly if they are analysed in the
context of a family of specificational copular constructions. Therefore,
contra Hedberg (2000) and Gundel, Hedberg and Zacharski (2001), the itclefts discourse-pragmatic properties do seem to require an independent
explanation. In Chapter 7, I provide evidence to suggest that the it-cleft has
acquired construction-specific properties over time. This accounts for some
much discussed differences in the discourse functions of it-clefts and whclefts (see Chapter 8).
5.
Summary
Throughout Chapter 4, I built up an analysis in which it-clefts are treated

foremost as specificational copular sentences. In particular, my aim was to
provide a constructional account which maximizes motivation for the itcleft by situating it in relation to a family of constructions as both an
instance and inheritor of an overarching specificational inversion construction. Throughout Chapter 5, I have considered three varieties of it-cleft
which cannot be accounted for in this way: predicational (and proverbial)
it-clefts, it-clefts with non-NP foci and informative-presupposition it-clefts.
I have shown that while my analysis extends naturally to incorporate predi-
Summary
147
cational and proverbial it-clefts (owing to inheritance from the predicate

nominal construction), there are some aspects of the specificational itclefts behaviour that remain a mystery.
As a result, it seems that we must look outside of the present-day
language system to explain the range of it-cleft foci and the informationstatus of the cleft clause. This is the focus of the diachronic investigation in
Chapter 7, which demonstrates how the specificational it-cleft has acquired
these construction-specific properties over time. First, however, we must
address the idiosyncratic structural properties shared by all types of it-cleft,
such as the restrictive modification of the pronoun it, the obligatory extraposition of the cleft clause, and the constructions unusual agreement patterns. In Chapter 6, I show that these properties are historical remnants of
an earlier period of English. The historical evidence provides both an
explanation for these idiosyncratic properties and demonstrable support for
a discontinuous constituent account of it-clefts.
Chapter 6
The it-cleft and earlier periods of English
1.
Beyond the present-day language system
At this point, we begin our investigation into the history and diachrony of
the English it-cleft. In the remaining chapters, I show how, once we look
beyond inheritance from the present-day language system, we can find an
explanation for some of the residual construction-specific properties which
remain unaccounted for on our analysis. Before investigating the development of, and changes to, the construction in Chapter 7, I begin, in this chapter, with a consideration of the properties that have stayed the same over
time. Here, I examine the it-clefts structural idiosyncrasies in relation to
the language system of earlier periods of English.
As I explained in Chapter 4, there are some aspects of the discontinuous
constituent account of it-clefts that remain a puzzle. On this analysis, the
cleft clause is a restrictive relative clause, which modifies the cleft pronoun
it. However, we have already seen (in Chapter 4, Section 1.3.1) that proper
names and pronouns cannot normally be modified by restrictive relatives.
How then do we explain this fundamental structural property, which has
played an instrumental role in our analysis so far? A further difficulty is the
extraposed position of the cleft clause. While extraposed relative clauses
are not uncommon, extraposition-from-NP is not normally obligatory. In itclefts, however, the cleft clause cannot appear in a position adjacent to its
pronominal antecedent. The success of a discontinuous constituent account
therefore depends on an answer to the question, why is the relative clause
always extraposed?
The final puzzle of interest to us here relates to agreement. As we have
seen in Chapter 4, Section 1.3.3, the person agreement data largely supports
a discontinuous constituent account, in that the verb in the cleft clause
typically agrees with the third person pronoun it. Number agreement, on the
other hand, seems to follow the opposite pattern, with the embedded verb
agreeing with plural foci, rather than with the inherently singular pronoun
it. In order to account for these agreement facts, I suggested that in it-clefts
with plural foci, the discontinuous description is interpreted as a kind of
collective entity that is, as having both singular and plural conceptualizations. In Chapter 5, Section 2, I showed that this analysis neatly accounts
The early history of the English it-cleft
149
for the fact that number agreement works differently in specificational and
predicational it-clefts. Nevertheless, we have yet to provide any real evidence for it. The question remains, why is number agreement in it-clefts not
straightforward?
In this chapter, I provide an answer for each of the above questions by
outlining the relevant historical evidence. In many ways, my approach here
follows from Declercks (1983b: 15) suggestion that the proverbial it-cleft
configuration is reminiscent of an older stage in the English language. On
the assumption that proverbial, predicational and specificational it-clefts
share the same discontinuous constituent structure (see Chapter 5, Section
2), it follows that the it-cleft schema may itself be characterized as a historical remnant, which better reflects the language system of earlier periods
of English.
The chapter is structured as follows. In Section 2, I outline the early
history of the it-cleft construction, discussing the various debates over its
first attestation. I then go on, in Sections 3, 4 and 5, to address each of the
three problem cases in turn: the restrictive modification of the pronoun it,
the obligatory extraposition of the cleft clause and the unusual pattern of
number agreement. In Section 6, I examine some new evidence which is
specific to the historical data that of gender agreement. We will see that
the it-clefts early gender agreement patterns represent very clear evidence
in support of a discontinuous constituent account. Finally, in Section 7, I
provide an account of the constructions origin and diachronic development
which is consistent with the historical evidence. I conclude that the it-cleft
structure has become increasingly idiosyncratic over time, not as a result of
any internal changes, but in relation to the surrounding language system.
2.
For many authors, the it-cleft is an Early Middle English development.

For example, Ball (1991), whose work constitutes the most comprehensive
diachronic treatment of the cleft construction to date, suggests that the first
specificational it-cleft does not arise in her data until the late thirteenth century; she provides the following example of an early focus-first it-cleft from
The South English Legendary.
(1)
A-bidez, quath is holie man: ore louerd is guod and freo.

e deuel it is at bringuth is wedur
150
Stay, said this holy man, our Lord is good and free. The devil it is
that brings this weather
(1280-90 South English Legendary [Ball 1991: 158])
Balls findings are supported by the observations of both Visser (1963) and
Mitchell (1985a). For instance, Visser (1963: 49) does not identify any Old
English cleft tokens with it (or hit); he notes that hit is omitted in Old
English, or its place is taken by t. Likewise, Mitchell (1985a: 622)
does not find any Old English examples corresponding to the modern use
of it to give emphasis, as in Its food that I want. He elaborates, This
is not surprising, since OE achieves the same emphasis by giving the noun
initial position, providing the example given as (2) below (Mitchell 1985a:
622). Visser (1963: 49) cites this same example as an OE cleft, without hit.
(2)
min fder is e me wuldra

my father is that me glorifies
It is my father that glorifies me
(lfric, Catholic Homilies [Ball 1991: 27])
In proposing her origin story for the Early Middle English it-cleft, it seems
that Ball is influenced by the accounts of Visser and Mitchell. She suggests
that the specificational it-cleft developed out of this Old English NP beon
relative clause configuration due to the rise of expletive subjects in Middle
English. She notes, the focused subject [of (2), min fder] becomes the
predicate complement and dummy hit appears in its place (Ball 1991: 68).
Recently, however, Filppula (2009) has reconsidered the Old English
evidence, extracting tokens from the York-Toronto-Helsinki Parsed Corpus
of Old English Prose (YCOE). He concludes that YCOE contains several
clear examples of clefts with hit that can be considered specificational in
meaning and hence genuine instances of it-clefts (Filppula 2009: 277).
Although I adopt a very different methodology from Filppula (see Chapter
7, Section 2), I agree with him on this point. I have identified seven tokens
in the YCOE that clearly warrant a specificational it-cleft analysis, including (3) below.
(3)
& cyde hit am Iudean. t hit wre se hlend e hyne hlde.

(cowsgosp,Jn_[WSCp]:5.15.6088)
[The man went away] and told the Jews that it was Jesus who had
healed him.
(Ball 1991: 40)
151
Interestingly, this token also appears in Balls data. However, she analyses
it (and other similar examples) as having a subtly different function as well
as a different structure from the it-cleft proper (see Ball 1991: 6367). I
discuss this aspect of Balls analysis, and identify several problems with it,
in Section 6 of this chapter.
The YCOE data also includes six examples of the predicational it-cleft
construction, including (4). In this token, Zosimus incorrectly classifies the
individual praying as a spirit rather than as a living being. A predicational
interpretation for (4) is confirmed when, later in the text, the individual in
question (who turns out to be St. Mary of Egypt) says ic nan gast ne eom
ac merge and axe and eall flsc (I am no spirit, but embers and ashes,
and all flesh (Skeat 1881a: 21)). Here, the NP spirit is clearly used to
express a property rather than to refer to an individual.
(4)
a a he on re eoran lg astreht a hwon hit gast wre t

r mid hwylcere hiwunga gebde hi.
(comary,LS_23_[MaryofEgypt]:278.187)
Whilst then he lay prostrate on the earth he [was troubled in his
mind, considering whether] at all it might be a spirit that, by some
strange appearance, was praying there
(Skeat 1881a: 19)
Ball (1991) agrees that such tokens represent early examples of the presentday predicational it-cleft. On Balls analysis then, the specificational it-cleft
is a more recent construction than the predicational it-cleft; for her, the two
types of it-cleft are therefore both structurally and historically distinct (see
Chapter 5, Section 2.1).
The present-day demonstrative cleft is also attested in Old English. For
example, Ball (1991) identifies several Old English tokens with the structure t/is NP beon relative clause, including (5) and (6) below. The data
is comprised of both specificational and predicational tokens. For instance,
(5) seems to me to be a clear occurrence of a specificational that-cleft.
Likewise, (6) is analysed by Ball (1991: 38) as an OE predicational thiscleft containing an indefinite NP predicate. Here, the vision revealed to the
young man is described as being a wonderful vision.
(5)
he a wolde yt cunnian, hwt t wre, t he r ehyrde.

he still wanted to verify what that/it was that he had heard.
(Gregorys Dialogues [Ball 1991: 36])
152
(6)
is is to soan sum wundorlic gesih e god lmihtig ysan

geongan men onwreogan hf.
This is truly a wonderful vision that God Almighty has revealed to
this young man
(The Seven Sleepers, LS 23 [Ball 1991: 38])
In conclusion then, it seems that the present-day it-cleft construction

dates back to the Old English period. Despite some disagreement regarding
the earliest attestation of the specificational it-cleft variety, the YCOE data
includes tokens which clearly have a specifying function and which can be
accounted for straightforwardly by assuming a discontinuous constituent
analysis of specificational it-clefts (see further, Section 6). What is more,
the Old English data exhibits the same family of constructions that we find
in present-day English, with specificational and predicational varieties of itand demonstrative clefts occurring side by side. This means that the structural configuration shared by all of these constructions (referred to as the
cleft schema in Chapter 4, Section 1.1) is not a recent development, but a
relic from earliest English. This provides support for the historical approach
adopted in this chapter, in which the (now) idiosyncratic properties of the
cleft configuration are reconsidered in relation to the language system of
earlier periods of English.
3.
A restrictively modified pronoun?
As noted above, in Section 1, the cleft structure is unusual in that the cleft
clause restrictively modifies the cleft pronoun. However, though atypical, it
should be noted that restrictively modified pronouns do occur in presentday English, albeit in restricted environments. For example, relative clauses
can restrict the very general type specification of indefinite pronouns,
shown in (7). Furthermore, for some speakers, definite object pronouns can
be followed by reduced relatives in expressions such as (8) and (9) below.
(7)
[someone [who lives here]]
(8)
[them [round the corner]]
(9)
[him [next door]]
In addition, restrictively modified pronouns occur in formulaic expressions,

such as (10). Here, the pronoun he and the following relative clause form a
153
definite description (akin to the one who laughs last), which is used to refer
to a hypothetical entity.
(10)
[He [who laughs last]] laughs longest
Although no longer a productive pattern, the example in (10) is an instance

of a historically much more regular construction involving determinative
pronouns; that is, pronouns modified by restrictive relatives.1 Perhaps then,
the present-day cleft structure is also an instance of the once productive
determinative pronoun construction.
There is considerable historical evidence in support of such a claim. For
instance, the early demonstrative clefts clearly warrant a determinative pronoun analysis. As Ball (1991: 5859) comments, t and is could both occur with restrictive relative clauses in Old English, shown in (11) and (12).
Note that in (12), the relative clause is adjacent to the demonstrative is, in
a position that is unacceptable in present-day English, shown in (13). On
the basis of such evidence, Ball analyses Old English t- and is-clefts,
including (5) and (6) above, as determinative pronoun constructions involving extraposition-from-NP.
(11)
Hi forbrndon a a burh 7 t e binnan hyre ws

They burnt then the city and that which was in it.
(The Heptateuch, Josh [Ball 1991: 59])
(12)
la, Petrus, is t ic e sde, m beon hrdlice ecyed.

lo Peter this that I you said may be quickly proved
What I have said, Peter, can be quickly proved.
(Gregorys Dialogues [Ball 1991: 58])
(13)
*This that I said to you can be quickly proved
(Ball 1991: 58)
There is also evidence for analysing the it-cleft as a (now conventionalized) instance of the determinative pronoun construction. Like that and this
1. I follow Ball (1991) and Declerck (1988) among others in describing pronouns
modified in this way as determinative. However, Curme (1931) uses the term in
a slightly different sense from other authors. He suggests that the clausal component in it-clefts and other determinative pronoun constructions is a subject
clause rather than a restrictive relative. On this account, the cleft it and other
determinatives serve as anticipatory subject[s], pointing to the following subject clause (Curme 1931: 188).
154
(or t and is), it could also, at one time, function as a determinative pronoun. In the Middle English examples below, (h)it occurs as the immediate
antecedent of a restrictive relative clause.
(14)
forhohe forte don hit t te unche uuel of & eil forte heren.
(c1220-25 CMHALI,142.261)
...unless they scorn to do what they think wrong and ill to hear of
(Cockayne 1866: 24)
(15)
is is it at setti ee in silence
This is what sets you in silence
(e15th The Book of Privy Counselling [Ball 1991: 59])
According to Ball (1991: 59), the determinative use of it does not appear
until Early Middle English; she states that Hit, to my knowledge, does not
occur in construction with a relative clause in OE. However, the YCOE
contains 16 tokens in which a relative clause is analysed as having the antecedent hit, all of which involve extraposed relatives, as in (16) and (17).
While these extraposed clauses could be independent from, yet coreferential with, hit, we cannot rule out a determinative interpretation, on which hit
is restrictively modified. At the very least, these tokens are relevant to an
extraposition analysis of it-clefts and suggest that a determinative use of it
may have originated in such extraposition structures (see also Section 4).
(16)
Ne t to nahte nyt ne bi t man godne mete ete oe t betste

win on gebeorscipe drince, gif t gelimpe t he hit eft
spiwende anforlte, t he r to blisse nam & to lichoman
nyttnesse.
(coblick,HomS_17_[BlHom_5]:57.35.701)
It is all to no purpose for a man to eat good meat, or at a feast to
drink the best wine, if it happeneth that he afterwards spews up and
loses that which he previously received for enjoyment, and for the
advantage of his body
(Morris 1880: 56)
(17)
Ne wicwee we be re eadigan Marian a ecan riste eah for

wrscipe gehealdenum geleafan us gedafena t we hit wenon
swior onne we unrdlice hit gesean. t e is uncu buton
lcere frcednysse.
(cocathom1,CHom_I,_30:431.80.5925)
Nor do we deny the eternal resurrection of the blessed Mary,
though for caution, preserving our belief, it befits us that we rather
hope it, than rather rashly assert what is unknown without any
danger
(Thorpe 1844: 441)
155
In addition to providing us with the relevant determinative structure, the

language system of earlier periods of English also helps to account for the
semantic interpretation of the early cleft tokens. As shown in (3), the OE
hit-clefts, like their present-day counterparts, can occur with human foci; in
this instance, the discontinuous NP characterizes a human referent. In
present-day English, this is difficult to account for, since it is associated
with reference to non-human entities. In Old English, however, the pronoun
it could also be used to refer to human individuals. For instance, in the OE
example given in (18), hit is used as the subject of a predicational sentence
to refer back to the old man. In present-day English, we would instead have
he as the subject.2
(18)
Hwt is es ealda man? Se engel him to cw, Hit is an biscop se

dyde mare yfel onne god
Who is this old man? The angel said to him: He is a bishop
who did more evil than good
(The Blickling Homilies [Ball 1991: 25])
The history of the pronoun it therefore offers a plausible explanation as to

how and why the discontinuous constituent of it-clefts can describe or refer
to human entities, in specificational and predicational varieties respectively.
This hypothesis is supported by the correspondence that exists between
the cleft it and the determinative uses of he and she in present-day English.
As shown in (19) below, he and she had a determinative function in Middle
English (see Ball 1991: 147). However, unlike determinative it and this, he
and she can still sometimes occur with adjacent relative clauses, as shown
in (10) above.
(19)
Bridgume iss he att hafe brid

The bridegroom is the one that has the bride
(c1175 The Ormulum [Ball 1991: 147])
2. Indeed, he replaces it in an Early Middle English version:

(i)
a escade paul to mihhal hwet e alde mon were. a cwe mihhal heh
angel he wes an biscop on eore liue
Then Paul asked Michael who the old man was. Then said Michael the
Archangel: He was a bishop in the other life
(1185-1225 Lambeth Homilies [Ball 1991: 25])
156
In support of a determinative pronoun analysis of proverbial it-clefts, Declerck (1988: 155) notes that when the discontinuous NP refers to a human
entity, the pronoun it can be replaced with the more regular determinative
pronouns he/she (emphasis added). Making this same point, Curme (1931:
189) discusses the following examples, in which the hypothetical human
who follows his own instructions can be introduced either with determinative it or he.
(20)
It is a good divine that follows his own instructions

(Shakespeare, Merchant of Venice [Curme 1931: 189])
(21)
He is a good divine who follows his own instructions

(Curme 1931: 189)
In conclusion then, the evidence from Old and Middle English provides
a suitable historical context for the discontinuous constituent analysis of itand demonstrative clefts argued for here. We have seen that the restrictive
modification of pronouns was once a much more productive pattern, with
it, that and this occurring in positions adjacent to modifying relatives. Furthermore, we have found that the pronoun it could at one time be used with
human reference. This explains just how it is possible for discontinuous
descriptions headed by neuter it to describe or refer to human entities in
specificational and predicational it-clefts respectively, as shown in (3) and
(20) above. Finally, we have found that it, like that and this, may have been
used determinatively since Old English, occurring in sentences with extraposed relatives, and that the more regular determinative pronouns he and
she can, like the cleft it, occur with extraposed relative clauses, as in (21).
This historical evidence provides support for a discontinuous constituent
(or extraposition-from-NP) analysis of both the earliest OE cleft tokens and
their present-day counterparts. Most authors agree that at least some types
of it-cleft require an analysis on which the pronoun it has a determinative
function. While Declerck (1983b, 1988) limits this analysis to proverbial itclefts and Ball (1977, 1991) applies it to her predicational tokens as well as
to all types of Old English demonstrative cleft, the historical facts provide
equal evidence in support of a determinative analysis for the specificational
it-cleft subtype. They show without doubt that the proposed cleft structure,
on which an extraposed relative clause restrictively modifies the cleft pronoun, was motivated by the language system of earlier periods of English,
inheriting properties from the then productive determinative pronoun
construction and conforming to historical uses of the pronoun it.
An obligatorily extraposed relative clause?
4.
157
Nevertheless, a question remains with this historical account of the presentday it-cleft. If the cleft it is truly a determinative pronoun, then how do we
account for the fact that the extraposition of the cleft clause is obligatory?
In other words, why can the cleft clause never occur in a position adjacent
to the pronoun it? While the fixed position of the cleft clause has been
regarded by many as a strong argument against an extraposition-from-NP
analysis (see Jespersen 1937: 8485, for example), I present evidence that
demonstrates that the present-day it-cleft configuration is better attuned to
the language system of Old English, when the extraposition of relative
clauses was very much the preferred option. I suggest that while the language system has changed, the it-cleft has retained the earlier pattern as an
entrenched (fixed) form.
Although extraposed relatives are not especially uncommon in presentday English (see Francis 2010: 65), they are clearly atypical, and on some
models of language are said to derive from their more basic, non-extraposed counterparts. In Old English, however, relative clauses are routinely
found sentence-finally. For instance, ONeil (1977: 199) goes so far as to
say that the most striking characteristic of the relative clause [in OE is] the
fact that such clauses are (almost) always at the margins of the main clause,
(almost) never flanked by material from the main clause, as in (22) below.
(22)
[a men common on East Engle [e onm anum scipe wre]]

the men came to East Anglia who on the one ship were
(ONeil 1977: 200)
On this basis, ONeil argues against a derivational account of such tokens,

on which the relative clause is underlyingly embedded under NP. He notes
that it is only by the Middle English period that relative clauses are regularly found in a position adjacent to their nominal antecedents.
More recently, Surez-Gmez (2006) has provided empirical support for
ONeils claim. She examines a database of 518 Late Old English and Early
Middle English relative clauses, which excludes clause-final tokens that do
not have the potential to show variation in their position with respect to the
main clause. Surez-Gmez (2006: 80) finds that In LOE extraposed relative clauses are almost twice as frequent as clause-internal relative clauses.
However in EME the frequency of integrated clauses increases importantly
and even slightly outnumbers extraposed relative clauses. She takes this as
evidence of a clear development whereby relative clauses, which initially
158
appear adjoined to the margins of the main clause in positions extraposed

from their antecedents, have become more integrated (or intraposed) within
the main clause (Surez-Gmez 2006: 127).
This evidence suggests that the extraposed structure of the early it-clefts
was consistent with the behaviour of OE relative clauses, more generally. It
could also account for the fact that determinative it does not appear with
adjacent relative clauses until Early Middle English (see Section 3). The
overall picture then, is of an originally well-motivated it-cleft construction
which conforms to wider generalizations involving relative clauses and
determinative pronouns. It seems then that while relative clauses gradually
became more integrated with the sentences to which they were attached
(Ball 1991: 60), the it-cleft construction simply retained the older pattern.3
This raises the following question: why didnt this change also affect the
relative clause in it-clefts? In other words, why has the it-cleft construction
been safeguarded against changes to the wider language system?
The reason must be that the OE it-cleft configuration is still motivated in
some way. In other words, while the obligatorily extraposed cleft clause is
no longer consistent with the behaviour of relative clauses more generally,
the cleft structure (as a whole) has come to be aligned with, and supported
by, other basic patterns and principles in the language system. As we saw
in Chapter 4, Section 1.1, the cleft configuration conforms to some highly
general information structure principles, such as the tendency for heavier
constituents to occur near the end of the clause (see Hawkins 1994, 2004;
Wasow 1997, 2002). This, accompanied by the fact that focal information
is placed within a cognitively-preferred clause-final position, makes the
cleft structure particularly well suited to linguistic generalizations brought
about by processing demands.
3. In addition to its extraposed position, the present-day cleft clause may also have
retained another trait of the early English relative clause. Fischer et al. (2000:
93) note that in Old and Middle English, zero relatives were most common
where the antecedent acts as the subject of the proposition in the relative clause,
as in (ii). While zero-subject restrictive relatives are no longer found in presentday English, Fischer et al. (2000: 94) note that the early tokens closely resemble the zero-type still acceptable in it-clefts and there be sentences (see Chapter 4, Section 1.3.1).
(ii) I know no knight in this contrey [] is able to macche hym
I know no knight in this country [who] is able to match him
(Malory, Works [Fischer et al. 2000: 93])
159
Indeed, Surez-Gmez (2006) finds evidence which suggests that the

principle of end-weight is a contributing factor in the position of relative
clauses in Early Middle English. While in Old English, extraposition is
preferred irrespective of length, as integrated clauses start to become more
frequent, it is only the longer relative clauses that are still associated with
extraposition (Surez-Gmez 2006: 103). Surez-Gmez does not measure
the influence of length relative to that of the main clause, which may have
more relevance for the preservation of cleft structure than absolute length.4
Nevertheless, the data supports a scenario in which the it-cleft was shielded
from the initial move towards embedded relative clauses in EME as a consequence of its information structure.
In addition, prosodic factors could help to explain why the extraposed
position of the cleft clause is obligatory. As we saw in Section 3 above, the
pronoun it begins to appear with adjacent relative clauses in Early Middle
English. However, in most of these examples, the determinative description
occurs in a postverbal position. Ball (1991: 263) claims that for whatever
reason, it + relative clause is not attested in first position. Although this
statement may be too strong for the Middle English corpus employed here,5
intraposed relatives adjacent to clause-initial it do seem to be dispreferred.
From this, we might conclude that the historical non-occurrence of it-clefts
with non-extraposed (or intraposed) relatives is part of a more general
phenomenon affecting the determinative it. Bolinger (1977: 76) offers a
possible explanation, attributing it to the prosody of English; he notes that
It is normally stressless, but in initial position followed by an obligatory
stressless that it would have to be stressed.6
4. See also Francis (2010), who finds that relative weight is a significant factor in
the extraposition of relative clauses in present-day English.
5. There is one example in the PPCME2 in which it + relative clause functions as
a left-dislocated subject, introducing a conjoined proposition, shown in (iii).
(iii) And hit at Saynt Nycholas hadde by hys holy prayer, hit was of such
fuson and plente, at hit fond all e pepull to ete and to sowe e erthe
aftyr
(a1415-1500 CMMIRK,13.363)
And that which Saint Nicholas received by his holy prayer, [it] was of
such abundance and plenty, that it provided [for] all the people to eat and
to sow the garden [there]after
6. Although it-clefts with intraposed relatives are not attested, we do find examples with determinative it that can be interpreted as reverse specificational sentences; that is as specificational inversion constructs. For example, in this token
160
Ball (1991: 60) suggests that there may also be interpretive reasons why
the cleft clause never came together with its antecedent, but remained in
fixed clause-final position. Discussing a determinative pronoun analysis of
the OE predicational tokens, she notes that, as in present-day English, the
focal noun is often interpreted as given information, forming part of the
discontinuous referring expression (see Chapter 5, Section 2). For instance,
(6) above, repeated here as (23), describes the vision revealed to the young
man. In contrast, in the non-extraposed version, translated in (24), this that
(or is e) is interpreted more generally as what. Ball (1991: 61) concludes
that the correct interpretation of the predicational cleft depends crucially
on the relative clause coming after the predicate nominal, so that there were
sound interpretive reasons for its staying put.
(23)
is is to soan sum wundorlic gesih e god lmihtig ysan

geongan men onwreogan hf.
This is truly a wonderful vision that God Almighty has revealed to
this young man
(The Seven Sleepers, LS 34 [Ball 1991: 38, 58])
(24)
*This that God Almighty has revealedis truly a wonderful

vision
(Ball 1991: 38, 58)
As a final possibility, we might speculate that, over time, it-clefts have

come to align less with (the now integrated) relative clauses, and more with
other kinds of extraposition construction. While we are used to thinking of
extraposed sentences as less basic structures, studies have repeatedly shown
that, for clausal subjects (such as that-clauses and to-infinitives), extraposition is more common than non-extraposition (see Erdmann 1987; Kaltenbck 2004); Calude (2008a: 13) concludes, extraposition appears to be the
norm rather than the exception. Thus, while it-clefts have a distinct structure, in that they do not involve the extraposition of a complete subject expression, the superficial similarity that exists between it-clefts and cases of
it-extraposition could well provide support for the it-clefts outward appearfrom Early Modern English, that (the act of unconditional friendship) is specified as the thing that allows us to benefit mankind, better ourselves and so
make ourselves dearer to God. Such examples show that, at earlier periods of
English, it-clefts (like other subtypes of specificational inversion sentence) had
corresponding non-inverted forms (cf. example (5) in Chapter 4, Section 1.1).
(iv) and indeed that is it that can make the difference; we must be friends to
all
(1662 JETAYLORMEAS-E3-P1,14.63)
161
ance, as shown in (25) and (26). Indeed, Prez-Guerra (1998) explores the
historical relationship between it-clefts and it-extraposition and finds that
they undergo a parallel increase in frequency from the Early Modern period
onwards.7
(25)
It was the Colonel that survived
(26)
It is a miracle that he survived
[it-cleft]
[extraposed that-clause]
We can conclude then that while the cleft clauses extraposed position is
no longer consistent with the behaviour of restrictive relative clauses more
generally, this was not always the case. At earlier stages of English, relative
clauses frequently occurred at the margins of the main clause. As the
external structure of relative clauses has undergone change, leading to more
integration between the relative and its antecedent, the it-cleft has remained
the same. Now, the it-clefts extraposed structure is supported largely by a
combination of interpretive, information structural and prosodic factors as
well as via analogy with other extraposition constructions. Therefore, while
the motivation for cleft structure is less direct in the system of present-day
English, the it-cleft can still be shown to conform to larger generalizations.
Furthermore, even though the fixed, discontinuous relationship between
the cleft clause and its pronominal antecedent is now specific to the cleft
schema, the internal structure of the cleft clause is nevertheless inherited
from the restrictive relative clause construction. Again, this is supported by
historical evidence. Undertaking a comparative study of the relative markers in NP-focus it-clefts and restrictive relative clauses, Ball (1994b) finds
that the cleft clause and other restrictive relatives have undergone parallel
diachronic developments.
Focusing on clauses in which the gap functions as the subject of the
embedded verb, Ball shows that in both it-clefts and restrictive relatives,
that is the preferred complementizer up until the eighteenth century, when
there is a sharp increase in the use of wh-pronouns. However, Ball notes
that while the pronoun which undergoes a rise in the eighteenth century, it
never achieves the same high frequency as who. Ball interprets this data as
7. Prez-Guerra suggests that this increase may be tied to a corresponding decrease in right-dislocation structures. While in right-dislocation, the dislocated
constituent is syntactically loose, in it-clefts and it-extraposition, the extraposed clause is integrated at sentence level (Prez-Guerra 1998: 15). From
this perspective, the cleft construction actually conforms to the tendency for
languages to move towards more integrated structures.
162
evidence of a paradigmatic shift in the system of relative markers (see also

Ball 1996). She notes that during the seventeenth century, who replaces
personal which, leading to a distinction in the restrictive relative paradigm
between personal and non-personal antecedents. This is followed by the assignment of that to the non-personal category. As a result of the decline in
that with personal antecedents, who has become the dominant form for cleft
clauses and restrictive relatives with personal subject antecedents, at least
in standard written British English. On the other hand, non-personal which
remains in competition with that and so is not subject to the same rate of
increase.
Ball argues that it-clefts and restrictive relative clauses undergo these
same changes at roughly the same time and at the same rate. She notes that
the rate of decrease in that with personal antecedents is not significantly
different in clefts and restrictive relatives, a finding which supports the
hypothesis that cleft complements and restrictive relative clauses share a
syntactic structure (Ball 1994b: 196). Therefore, the historical evidence
suggests that while its extraposed position is reminiscent of an earlier stage
of the language, the cleft clause is nevertheless a fully-fledged restrictive
relative which inherits its internal structure from the more productive construction.
5.
An unusual pattern of agreement?
Assuming a discontinuous constituent analysis of cleft structure, the specificational it-cleft displays an unusual pattern of number agreement, in that
the verb embedded inside the cleft clause does not formally agree with its
antecedent the singular pronoun it. For instance, in (27), the embedded
verb (are) is plural, in agreement with the focal NP (John and Margaret).
In contrast, the subject pronoun and the matrix copula are both marked as
singular (it is). So, we might ask, what is the history of this curious system
of number agreement?
(27)
Its John and Margaret who are responsible
On first appearances, the diachronic data does not seem overly helpful.
For instance, it is well known that throughout the it-clefts entire history,
the verb in the cleft clause shows agreement with the focal NP, rather than
with the initial it (see Ball 1991). Consequently, we cannot claim that this
pattern of number agreement has undergone change. However, what has
163
changed over time is the behaviour of the matrix copula. In what follows, I
show that the history of agreement in the matrix clause in fact provides us
with a way of understanding how present-day agreement works in the cleft
clause.
Throughout Old English and well into Middle English, the matrix verb
agrees in number with the clefted constituent (see Visser 1963: 49 and Ball
1991: 283288). For example, the text in (28) contains two it-clefts. In the
first, the form of the copula is agrees with the singular referent your own
spirit. However, in the second instance, the matrix verb and the postcopular
NP are both marked as plural.
(28)
when it is in owe spirite at speki ees iueles, or it ben ees

oer iuel spirites at speken hem in ee.
when it is your own spirit that speaks these evils, or it {are/is}
these other evil spirits that speak them in you
(e15th A Tretis of Discrescyon of Spirites [Ball 1991: 286])
It is not until the early fifteenth century that we begin to see the situation
that we have today, with the matrix copula being marked as singular, even
in cases where the clefted constituent refers to plural entities. Example (29),
which is from a slightly later manuscript of the same text as (28), shows the
more modern pattern. Here, the postcopular NP is plural but the matrix verb
is in the singular form.
(29)
or it is ees oer iuel spirites

or it is these other evil spirits
(mid-15th A Tretis of Discrescyon of Spirites [Ball 1991: 286])
So how should we account for the it-clefts early agreement pattern and
what can it tell us about the present-day it-cleft? Once again, the answer
lies in the history of the cleft pronoun it. Throughout Old and Early Middle
English, the pronoun it could sometimes be used with plural reference (see
Mitchell 1985a: 622). For example, in (30), OE hit is used to refer back to
three different events, classed here as signs.
(30)
And eac his riste of daee, and his upstige to heofenum, and ealle
a wundra e he worhte, ealle hit wron tacna
And, likewise, his resurrection from death, and his ascension to
heaven, and all the wonders which he wrought all these were
signs
(lfric, Catholic Homilies [Thorpe 1844: 146, 147])
164
Similarly, in the EME token in (31), hit is used alongside ha (they) to refer
to a plural referent, the clever and wise words.
(31)
witti ant wise w[o]rdes hit weren ef ha neren false

clever and wise words they would be if they werent false
(1200-25 Seinte Katerine [Ball 1991: 154])
From this, it follows that, in (28) above, the plural copula (ben) shows
agreement, not only with the plural postcopular constituent, but also with
the pronoun it. So, in the earlier it-cleft agreement pattern then, number
marking is actually consistent across the whole of the construction.
The historical facts therefore demonstrate that while the verb embedded
in the cleft clause has always been subject to plural marking (in agreement
with the clefted constituent), this was originally in agreement both with the
matrix copula and, most importantly, with its nominal antecedent the
determinative it. For example, in the Old English it-cleft in (32), taken from
the YCOE, the verb in the cleft clause (feallan) carries the plural present
ending -a, and so shows agreement with the plural marking (-an) of the
postcopular masculine noun steorra. However, since the matrix copula sind
is also formally plural, we may assume that the initial hit is also plural,
being used to describe a plural entity.
(32)
Ac hit ne sind na steorran t r fealla

But it is not stars that fall there
(cotempo,Temp:9.1.286)
For the early it-cleft tokens then, number agreement is accounted for easily
on a discontinuous constituent analysis: the cleft clause shows agreement
with its pronominal antecedent, which can be used to refer to or describe
both singular and plural entities. Therefore, the historical it-cleft tokens are
not problematic for this account in the same way as their present-day counterparts.
Thus, we have seen that the earlier number agreement pattern is
consistent with a discontinuous constituent analysis (and in fact with any
analysis) of it-clefts and that the more unusual, present-day pattern is a
development. So how can we account for this change? As we have already
seen, the pronoun it was once maximally underspecified; that is, it could
describe or refer to plural as well as singular individuals and to human as
well as non-human referents (see example (18) above). To an extent, the
specificational it-cleft has retained this historical flexibility, in that the
present-day it-cleft permits both human and plural foci. In these sentences,
165
a discontinuous NP headed by it can describe (or be predicated of) singular

or plural, human or non-human referents.
However, since in most contexts the pronoun it has come to be restricted
to denoting non-human, singular individuals, the cleft it is also perceived of
as a morphologically singular pronoun (by analogy). This is demonstrated
by the changes to subject-verb agreement in the matrix clause, as the matrix
copula now consistently takes the singular form. The development of it as
an invariably singular pronoun has therefore created a complicated pattern
of number agreement. Although the discontinuous NP still describes plural
entities and the verb in the cleft clause still takes plural marking, the cleft
pronoun and the matrix verb are formally singular. The historical evidence
therefore supports an analysis of the present-day it-cleft in which the cleft it
is semantically underspecified and, although formally marked as such, is
not limited to describing singular, non-human referents.
On such an account, number agreement in the cleft clause is semantic,
rather than morphosyntactic (see Chapter 4, Section 1.3.3). This is supported by those occasional examples in which the cleft clause does not
show agreement with a plural postcopular NP. For instance, out of a handful of Late Middle English tokens with plural foci, Ball (1991: 307) points
to the following example, also found in the PPCME2, in which the postcopular pronoun ye is plural but the embedded verb hath is marked as
singular. She notes that while ye is formally plural, in this example, it has
singular reference. Ball (1991: 307) concludes that the embedded verb
agrees in number with the referent of ye (emphasis added).
(33)
Truly, sir, she seyde, I trowe hit be nat ye that hath slayne my
husbonde, for he that dud that dede is sore wounded, and is never
lykly to be hole, that shall I ensure hym.
(a1470 CMMALORY, 202.3233)
it is not you that has slain my husband
(Ball 1991: 307)
Visser (1963: 49) discusses a similar example from Early Modern English;
he notes that The form makes in the following passage from Shakespeares
Alls Well (IV, ii, 21) is remarkable. Here, the embedded verb does not
agree with the postcopular plural NP the many oaths. Instead, makes shows
agreement with the referent for which this negative it-cleft holds true (the
plain single vow), even though this is not specified until later in the sentence. Similar examples occur in present-day English, where agreement in
the cleft clause seems to be governed by semantic factors, such as the act of
specification (see Chapter 4, Section 1.3.3).
166
(34)
Tis not the many oaths that makes the truth, But the plain single
vow that is vowd true
(Shakespeare, Alls Well That Ends Well [Visser 1963: 49])
However, while number agreement in specificational it-clefts therefore

seems to be semantic, the cleft clause in predicational tokens shows formal
agreement with the cleft pronoun. As we saw in Chapter 5, Section 2.3,
present-day predicational clefts with plural foci are introduced by plural
pronouns, such as they. Yet again, this agreement pattern seems to result
from changes to the pronoun it. According to Visser (1963: 50), they-clefts
date back to the sixteenth century. The development of this construction
therefore follows the reanalysis of it, occurring only after its matrix copula
is consistently marked as singular. Before this time, plural predicational
clefts are also introduced by the underspecified pronoun it, as shown in this
Early Middle English token from the PPCME2 (see also Ball 1991: 154).
(35)
Ah et me teone mare t ha tuket ure godes to balewe & to

bismere & seid hit beo deoflen t in ham dearie.
(c1200-25 CMKATHE,27.136)
But it grieves me yet more, that she brings our gods into contempt
and derision; and says that they are devils that lurk within them.
(Ball 1991: 154)
Interestingly, the changes to the pronoun it in Middle English are

followed by a period of variation, in which specificational clefts with plural
foci can also be introduced by plural pronouns. For example, in this Early
Modern English they-cleft from Locke, the collective entity our offspring is
contrasted with other species as the things we neglect to discipline when
young. In present-day English, this token would be introduced by it, which
(along with the adverb only) suggests that the example prefers a specificational reading.
(36)
We are generally wise enough to begin with them when they are
very young and discipline betimes those other creatures we would
make usefull to us. They are only our ofspring that we neglect in
this point and haveing made them ill children we foolishly expect
they should be good men.
(1685 LOCKE-E3-H,51.125)
While both the specificational and predicational varieties of it-cleft have

therefore been affected by the changing function of the pronoun it, this has
167
nevertheless resulted in the development of different patterns of agreement

for the two constructions. The fact that the specificational it-cleft did not
fully develop (or, at least, did not retain) a variant with they, suggests that
there is a semantic difference between the discontinuous constituents of
specificational and predicational it-clefts. This, in turn, supports an inverse
(rather than an equative) analysis of specificational meaning (see Chapter 5,
Section 2.3). In particular, I have suggested that the discontinuous NP of
the predicational tokens is marked for number because it refers to individuals. In contrast, the discontinuous NP of specificational it-clefts denotes a
set. In examples with plural foci then, the description is conceptually both
singular and plural, denoting a singular set with plural members. By not
marking number until the act of specification is complete, the construction
allows the speaker to contrast singular and plural entities as possible
instantiations of the same set (see Chapter 4, Section 1.3.3). There is therefore a practical advantage for retaining the original, underspecified use of it
in the specificational variety.
However, on the account sketched here, we might also expect to see a
difference in the introductory pronouns of specificational and predicational
clefts with human foci; that is, for he or she to replace it in the predicational
examples and for underspecified it to be found in the specificational tokens.
The existence of predicational it-clefts like (37) disproves this rule. As den
Dikken (2009: 6) notes, such examples have postcopular predicates that
should be predicated of humans; he continues, The use of referential it as
the subject here raises some questions.
(37)
Its no RELIABLE man that you hired, but a crook

(Declerck 1988: 160)
On the current analysis, such examples gain a historical explanation: the

use of it with human reference is a relic from an earlier time. Nevertheless,
there is variation in the introductory pronouns of predicational clefts which
is not found in the specificational variety. This conforms to the predictions
of the current account and provides additional evidence in support of it. As
Curme (1931: 189) notes,
We thus often use it even where we point to persons, provided the desire is
to identify; but when the desire is to describe, we may say with Shakespeare It is a good divine that follows his own instructions (Merchant of
Venice, I, II, 15); or more commonly we replace it here by a personal pro-
168
noun that indicates gender and number: He is a good divine who follows
his own instructions.8
In conclusion, the historical evidence has shown that number agreement

in specificational it-clefts hasnt always been a problem for a discontinuous
constituent analysis. In Old and Early Middle English, when it had variable
reference, it-clefts with plural foci were marked as plural throughout. It is
only with changes in the usage of it that the construction has developed its
unusual pattern of number agreement, with the formally singular cleft it
retaining its original semantic flexibility. The historical evidence therefore
supports the account of number agreement sketched in Chapter 4, Section
1.3.3, and provides a reason for its idiosyncratic nature. In addition, the fact
that the predicational it-cleft was more thoroughly affected by the changes
to the pronoun it supports the claim, outlined in Chapter 3, Section 2, that
specification is the inverse of predication, rather than an instance of equation. On this account, the discontinuous NP in specificational it-clefts is
semantically predicative, rather than referring.
However, it should be noted that while the history of number agreement
strengthens the particular analysis of it-clefts argued for here, the history of
person agreement seems to argue against such an account. We have already
observed that in present-day English, the verb embedded in the cleft clause
is typically in the third person and the postcopular NP is often found in the
objective case. This pattern is consistent with an extraposition-from-NP
analysis, on which the cleft clause modifies the third person pronoun it (see
Chapter 4, Section 1.3.3). However, when personal pronouns first appear in
postcopular position (in the Late Middle English period), they are always in
the nominative case and the verb embedded in the cleft clause sometimes
shows person agreement with the postcopular pronoun (see Ball 1991: 308
309). Ball (1991: 309) provides the example in (38). Here, the embedded
verb have agrees with the clefted constituent the first person, nominative
pronoun I.
(38)
Wherefore it is onely I that haue offenced.

(1531 Elyot Gouernour II [Ball 1991: 309])
8. Such variation does not seem to be limited to proverbial predicational tokens,

as shown in (v).
(v)
Hes a good man (that) youve got there
169
This earlier agreement pattern would seem instead to provide evidence

in favour of an expletive analysis of it-clefts, in which the postcopular NP
functions as the underlying subject to the verb embedded in the cleft clause.
However, there is additional evidence which argues against this conclusion.
For one thing, postcopular personal pronouns are found in the nominative
case even when they cannot function as the subject of the verb in the cleft
clause. For example, in (39), hee is nominative, despite its correspondence
to the object of the embedded proposition you love __ best.
(39)
Why, is it hee you love best, quoth the Parson?

(1597 DELONEY-E2-P1,16.235)
Furthermore, throughout the Middle English period, postcopular pronouns

in simple copular sentences were also in the nominative case. Smith (1906:
7881) cites several examples, including (40). Here, it is not possible to
analyse the nominative pronoun he as subject of some additional clause.
Instead, this pattern can be used in support of an inverse analysis of these
specificational sentences. Smith (1906: 81) suggests that such examples are
evidence of a transition period, during which time the postcopular pronoun changes from having subject to object status.
(40)
but by cause he knewe not his sheld he demed it was not he.
(1485 Malory, Le Morte dArthur (Cx) [Smith 1906: 80])
Discord between the person marking of the verb in the relative clause
and that of its pronominal head, shown in (38), also extends beyond the itcleft construction. For example, the verb have in the reverse pseudocleft of
(41) shows agreement with the subject of the matrix clause rather than with
the third person determinative he. As Ball (1999) notes, this agreement
pattern can sometimes penetrate even a relative clause with a noun head,
such as that given in (42).
(41)
Y am he that haue synned, and Y dide wickidli

I am the one that have sinned, and I acted wickedly
(a1440 Wycliffite Later Version [Ball 1991: 311; Ball 1999])
(42)
I am, said she, a gentylwoman that am disheryted, whiche was

somtyme the rychest woman of the world.
(1485 Malory, Le Morte dArthur (Cx) [Ball 1999])
170
The earlier pattern of person agreement in it-clefts is therefore indicative of

larger generalizations involving Late Middle English copular constructions
and relative clauses. In truth then, the historical data does not necessarily
support an expletive analysis of it-clefts and cannot be used as evidence
against a discontinuous constituent account.
6.
The evidence from Old English gender agreement
In addition to the evidence involving earlier patterns of number and person

agreement (which has helped to explain the agreement facts in present-day
it-clefts), a historical perspective brings with it some new agreement data,
concerning gender. In Old English, pronominal relatives inflect for gender,
number and case; that is, they are in agreement with the gender and number
of their antecedent nouns, while case is inflected according to the relative
clause. The Old English cleft data is therefore extremely instructive, since
we have an additional means with which to identify the structure of these
early cleft sentences. In what follows, I show that gender agreement in OE
it-clefts favours a discontinuous constituent analysis and is problematic for
expletive accounts (on which the antecedent of the cleft clause is the clefted
constituent, rather than the cleft pronoun). Indeed, the Old English tokens
have often been discounted from earlier historical accounts because they do
not conform to an expletive analysis, and so (on this thinking) cannot be
considered true clefts.
Old English has three different relativization strategies: relative clauses
can be introduced by the indeclinable particle e, by pronominal relatives
inflected for gender (se, so, t), or by some combination of the two, such
as the see relative (Mitchell and Robinson 2007: 7677). For examining
cleft structure, the relevant tokens are those in which the relative clause is
not introduced by the indeclinable e and where there is disagreement
between the neuter cleft pronoun and the gender marking of the clefted
constituent. In such cases, the form of the relative pronoun makes it possible for us to identify whether the antecedent is the neuter hit or the postcopular noun. There are 6 relevant tokens in the YCOE, out of a total of 13
it-clefts. In 5 of these 6 tokens, the relative pronoun is t.
While relative that is now invariable (which was already standard by the
thirteenth century), it is for the most part agreed that in Old English, neuter
singular t was still part of a well-preserved demonstrative/relative
declension (Mitchell 1985b: 108). Although we can identify precursors to
the modern use of that in OE, lack of agreement involving neuter singular
171
t was very rare. Indeed, Mitchell (1985b: 101) finds that most of the
apparent examples of relative t showing lack of concordcan be otherwise explained. Furthermore, OE t could function as the complement of
a pied-piped preposition (as in urh t by which), a characteristic unique
to pronominal relatives (Surez-Gmez 2006: 85). Therefore, the use of
t in the YCOE cleft tokens suggests that the antecedent of the relative
clause is the neuter pronoun hit.
Among them, these 5 YCOE tokens represent both specificational and
predicational varieties of it-cleft and contain both masculine and feminine
postcopular nouns. (43), for instance, is clearly a predicational it-cleft (see
the discussion of example (4) in Section 2). Here, the postcopular noun gast
is masculine and so is at odds with the relative t. As Ball (1991: 39)
notes, Bourcier (1977: 326) discusses this same example and takes hit to be
the antecedent of the relative clause.
(43)
a a he on re eoran lg astreht a hwon hit[n.] gast[m.]

wre t[n.] r mid hwylcere hiwunga gebde hi.
(comary,LS_23_[MaryofEgypt]:278.187)
Whilst then he lay prostrate on the earth he [was troubled in his
mind, considering whether] at all it might be a spirit that, by some
strange appearance, was praying there
(Skeat 1881a: 19)
Example (44) is also predicational. Here, Philosophy describes the grace

granted to the wicked as temporary rather than everlasting. The predicative
noun gifu is feminine, but the relative clause is introduced by the neuter t
rather than the feminine so.
(44)
a cw ic: Nu ic (on)gite t hit[n.] (n)is ecu gifu[f.] t[n.] he

gif m yflum, ac is hwilchwugu eldcung & andbid s hehstan
deman.
(coboeth,Bo:38.119.33.2388)
Now I understand that it is not an everlasting grace that He
granteth to the wicked, but a manner of delay and waiting for the
Highest Judge.
(Sedgefield 1900: XXXVIII)
Example (45), however, is clearly specificational. Here, the neuter t does

not agree with either Petrus or ngel, both of which are masculine. This
leads Mitchell (1985b: 102) and Ball (1991: 67) to conclude that the initial
pronoun hit is the antecedent of the relative clause. Mitchell (1985b: 102)
says, t refers back to hit.
172
(45)
a cwdon a geleafullan, Nis hit na Petrus t

r
not-is it-n. not Peter-m. REL-n. there
cnuca, ac is his ngel.
knocks but is his angel-m.
(cocathom1,CHom_I,_ 34:474.247.6867)
Then the faithful said: It isnt Peter who is knocking there, but his
angel.
(Ball 1991: 39, translation and gloss)
Similar examples are shown in (46) and (47). Here again, the postcopular
nouns steorra and lichoma are both masculine, while the cleft clause is
introduced by the neuter pronoun t.
(46)
Ac hit[n.] ne sind na steorran[m.] t[n.] r fealla

But it is not stars that fall there
(cotempo,Temp:9.1.286)
(47)
ond him tywde a wunda on his handum ond on his fotum ond a
gewundedan sidan, t hi y solicor ongeaton t hit[n.] ws
solice his agen lichoma[m.] t[n.] r of deae aras.
(comart3,Mart_5_[Kotzor]:Ma27,A.2.493)
and showed them the wounds on his hands and on his feet and his
wounded side, that they might understand with greater certainty
that it was truly his own body that had there arisen from death.
(Herzfeld 1900: 5153)
In only one example does the relative pronoun agree in gender with the
clefted constituent but not with the cleft pronoun. In (48), we have two see
relatives. In the first, the relative pronoun is in the nominative case (se e),
while in the second, it is in the accusative case (one e). The masculine
forms se and one are consistent with the postcopular Swithhun, rather than
with the neuter cleft hit.
(48)
a cw t wif him to. t hit[n.] wre Swyun[m.] se[m.] e

hine lrde mid re halgan lare. and one[m.] e he geseah on
re cyrcan swa fgerne.
(coaelive,LS_[Swithun]:388.4463)
Then said the woman to him, that it was Swithhun who had
instructed him in this holy lore, and whom he had seen so glorious
in the church.
(Skeat 1881b: 465)
Unlike the other examples, this token seems to favour an expletive account,
on which the antecedent to the relative clause is the postcopular constituent.
173
However, there is an alternative explanation for this agreement pattern.

Rather than forming a conjoined relative with e, the masculine pronoun se
could be functioning as the demonstrative head of the relative clause (see
Allen 1980: 108109). Ball (1991: 40) states that the pronoun in the second
conjunct (one) cannot be accounted for on this analysis because the accusative form clearly takes its case from the relative clause (he had seen x in
the church). Nevertheless, the first conjunct is properly ambiguous and can
be interpreted either as a restrictive relative clause, or as a right-dislocated
demonstrative-headed relative. On the latter reading, (48) is made up of a
simple copular sentence (it was Swithhun) followed by a complete definite
description containing two relative clauses (he who had instructed him in
this holy lore and whom he had seen so glorious in the church).
However, while it is certainly possible that the token in (48) is not a true
it-cleft, there is no solid reason to disqualify it other than that it supports an
expletive account. I therefore continue to count this token as a cleft, while
noting that, in general, gender agreement in the Old English data can only
be accounted for by a discontinuous constituent (or extraposition-from-NP)
analysis. While my data includes only it-clefts, Ball (1991) finds this same
agreement pattern in the Old English demonstrative clefts. For example, in
the predicational t-cleft in (49), the neuter relative (t) shows agreement with the initial pronoun rather than with the masculine postcopular
noun journey. As Ball (1991: 59) notes, The gender of the relative pronoun suggests that [the initial] t is the head of the relative clause.
(49)
t
ws geocor si,
t
se
DEM-n.n.s was grievous journey-m. REL-n.s. DEM-m.n.s
hearmscaa to Heorute ateah!
despoiler
to Heorot took
That was a painful journey that the loathsome despoiler had made
to Heorot.
(Beowulf [Ball 1991: 35])
On the basis of gender then, OE hit-, t- and is-clefts clearly warrant

an analysis involving extraposition-from-NP. Indeed, none of the relevant
authors have argued otherwise. For instance, Ball (1991: 63) follows Culicover and Rochemonts (1990) account of present-day English extraposed
relative clauses and claims that, in these Old English tokens, the relative
clause is base-generated in an extraposed position adjoined to IP and is
coindexed with the neuter subject pronoun (hit, t or is). What has been
called into question, however, is the status of these Old English tokens as
genuine clefts. In other words, because such examples do not conform to an
174
expletive analysis of cleft sentences, they have often been discounted as

representing early instances of the present-day it-cleft construction. This is
especially true of Balls (1991) work.
From the outset, Ball (1991) assumes an expletive analysis of presentday it-clefts, on which the cleft pronoun is a dummy element. From this
perspective, the OE gender agreement pattern is a surprising fact: if these
are true clefts, in which the focus is the logical antecedent of the relative
pronoun, we should expect gender agreement with the focus (Ball 1991:
3940). By way of a solution to this problem, Ball builds a case for analysing the OE tokens as instances of a separate construction. For the predicational examples, this is relatively straightforward. We have already seen
(in Chapter 5, Section 2) that predicational it-clefts are incompatible with
expletive accounts and so are treated by many as forming a structurally
distinct sentence type from the it-cleft proper (see Ball 1977). For Ball
then, as here, the predicational it-cleft has an extraposition-from-NP structure that has remained relatively unchanged since Old English. The only
substantial difference between our accounts is that, for Ball, this does not
represent the true cleft structure.
However, it is much more difficult to argue that the OE specificational
tokens form a separate construction from the present-day specificational itcleft. To do so, Ball claims that they are functionally (as well as formally)
distinct, invoking Higgins (1979) distinction between specificational and
identificational copular sentences. According to Higgins (1979: 265), the
sentence in (50) has two possible readings. The specificational reading tells
us that the individual Mary Gray matches the description the girl who helps
us on Fridays. However, on the identificational reading, we are told the
name (Mary Gray) of the individual referred to by the definite NP the girl
who helps us on Fridays.
(50)
The girl who helps us on Fridays is Mary GRAY

(Higgins 1979: 265)
Higgins class of identificational copular sentences and his use of the term
identificational are not widely accepted.9 Nevertheless, Ball uses this definition to argue that the Old English non-predicational tokens are actually
9. Indeed, Mikkelsen (2004, 2005) argues that Higgins (1979) identificational
class is not a semantically uniform category and goes on to show that many of
his examples can be accommodated into a tripartite taxonomy of predicational,
equative and specificational copular sentences.
175
identificational rather than specificational it-clefts. That is, she claims that
they do not identify a specific referent; instead, they identify only the name
of an already established individual.
For instance, after examining the context for (45) above, shown in (51),
Ball (1991: 64) explains that there is an otherwise established entity in the
context [the one knocking at the door] which lacks only a name.
(51)
Micel wurscipe is cristenra manna, t gewhilc hbbe fram his

acennednysse him bethtne engel to hyrdrdene, swa swa be am
apostle Petre awriten is, aa se engel hine of am cwearterne
geldde, and he to his geferum becom, and cnucigende inganges
bd. a cwdon a geleaffullan, Nis hit na Petrus t r
cnuca, ac is his ngel.
It is a great honour for Christian men, that each has from his birth
an angel assigned to him in fellowship, just as of the apostle Peter
it is written, when the angel led him from the prison, and he came
to his companions, and knocking prayed for admission, the faithful
said, It isnt Peter who is knocking there, its his angel.
However, this cannot be right. Here, the faithful mistake St. Peter, who they
believe to be in prison, for his angel; that is, they (wrongly) identify his
angel as the one who is knocking there. In (51) then, what is at issue is not
the name of the referent (as Peter or his angel) but the referent itself. This
example clearly conforms to our understanding of specificational meaning
as involving the specification (or identification) of the individuals that are
uniquely characterized by the definite description (i.e. the members that
make up the described set).
Likewise, Ball (1991: 64) claims that token (3), repeated here as (52),
occurs in the same kind of identificational context. In this example, Jesus
has healed a lame man. However, the man does not know the identity of the
one who healed him until he encounters Jesus again. Realizing who he is,
the man tells the Jews.
(52)
& cyde hit am Iudean. t hit wre se hlend e hyne hlde.

(cowsgosp,Jn_[WSCp]:5.15.6088)
and told the Jews that it was Jesus who had healed him.
(Ball 1991: 40)
176
For Ball (1991: 66), the central issue is the name of an entity which has
already been at least partially described, and which is salient. However,
the man provides much more information than a simple name or label here
(the Saviour, or literally, the healer). Instead, he refers to an individual,
pointing out Jesus for the benefit of the Jews. Indeed, Ball (1991: 66)
acknowledges that these Old English hit-clefts also allow a specificational
interpretation; she concludes, There is a fine line between identification
and specification.
In conclusion then, there is little justification for characterizing the Old
English cleft data as performing a naming or labelling function, rather than
a specificational one. There is therefore little evidence for suggesting that
the Old English hit-clefts are instances of a structurally distinct construction. Tellingly, Ball only distinguishes between specification and identification when discussing the Old and Early Middle English data. For subsequent periods, Ball finds it difficult to implement this distinction and
consequently abandons it. She notes, Specificational and identificational
it-clefts are classed together here because of the difficulty of reliably distinguishing themWhile the distinct semantics of the predicational cleft
continue to set it apart, the other two types are close both in interpretation
and function (Ball 1991: 220). This suggests that Balls main purpose in
employing this distinction for Old English is to allow her to reconcile the
existence of early tokens involving extraposition-from-NP with an expletive account of specificational it-clefts.
In sum, the Old English gender agreement data provides strong support
for extraposition-from-NP accounts, so much so that it forces proponents of
expletive analyses to discount and explain away this data as evidence of a
different kind of copular construction.10
10. Filppula (2009) is exceptional in that he assumes that it-clefts have an expletive
structure while maintaining that the YCOE contains genuine it-cleft tokens with
a specificational meaning. Tellingly, however, Filppula does not discuss the
gender agreement evidence in any detail, except to note that, for the example
given as (45) and (51) above, Mitchell [1985b: 102] takes hit and not the
focused noun, as in clefts, to be the antecedent of the relative pronoun here and
therefore rejects this as an example of a cleft sentence (Filppula 2009: 277).
The it-cleft as a relic from an earlier time
7.
177
Examining the it-clefts early history therefore provides us both with a new
dataset for analysing cleft structure and with an alternative language system
in which to look for generalizations that help to explain the constructions
(now) idiosyncratic properties. This allows us to address questions which
were left unresolved by a discontinuous constituent analysis of the presentday facts. For instance, we saw in Section 3 that the it-clefts discontinuous
(it + relative clause) NP is an instance of a once productive pattern in which
determinative pronouns are modified by restrictive relative clauses. In Section 4, we saw that the sentence-final position of the relative clause was the
norm in Old English, which suggests that the it-cleft represents a reflex of
the older pattern. Finally, in Section 5, we saw that the cleft pronoun it has
more in common semantically with the Old and Early Middle English lexeme it than with the present-day pronoun. With this observation made, we
can better understand the it-clefts unusual pattern of number agreement.
In this way, the it-clefts construction-specific formal properties have
been shown to have their sources in configurations which no longer exist
(or are no longer productive) in present-day English. In other words, these
now idiosyncratic properties were at one time inherited from more general
linguistic patterns, including the determinative pronoun construction, the
paratactic relative clause construction, which displays a loose degree of
integration (Surez-Gmez 2006: 52), and the OE/EME lexeme it. These
early inheritance relations are given in Figure 10 as dashed lines, indicating
that they are representative of an older system of English.
OE/EME
lexeme it
OE paratactic
relative clause
Determinative
pronoun
It-cleft schema
Figure 10. Incorporating historical motivation for the it-cleft configuration
The it-clefts syntactic configuration, as it is conceived on a discontinuous

constituent account, was therefore originally motivated by the language
178
system of earlier periods of English; that is, it was supported by a family of

constructions. On this account, the it-cleft schema is a relic from an earlier
time. While the structured inventory of the language has changed since Old
English, the it-cleft construction has been largely unaffected. From this, it
follows that the it-clefts historical development involves the entrenchment
(or fossilization) of once productive patterns.
In Chapter 2, Section 3, we saw that, on a usage-based model of change,
fossilization is a product of token frequency (that is, the number of times a
given instance is activated). While the it-cleft is a relatively low frequency
construction, it undergoes a gradual, yet substantial, increase over time (see
Chapter 7). This serves to cement the status of the it-cleft schema as a conventional unit in the speakers grammar. However, the strengthening of this
individual, lower-level construction does not result in the activation, or the
reinforcement, of its superordinate schemas (see Bybee 1985: 132134). Instead, it occurs alongside the weakening (or loss) of their productivity; that
is, the overarching constructions no longer sanction different types of instance. This usage-based account assumes a redundant system of grammar,
where information inherited from the dominating construction is also specified the inheriting construction.
This analysis explains how, as the dominating constructions have either
fallen out of use (such as the determinative pronoun construction) or undergone significant changes (such as the external syntax of restrictive relatives
and the restriction of it to non-human, singular individuals), their influence
has become entrenched within the it-cleft schema. The it-cleft has therefore
become increasingly idiosyncratic over time, not through any internal
changes, but in relation to the surrounding language system. This idea, that
irregularity is sometimes the relic of historical regularity, is well understood (see Chapter 2, Section 3). For example, irregular past tense forms
such as stand/stood are the entrenched relics of a once productive pattern of
strong verbs. Likewise, idioms such as kith and kin (meaning friends and
family) and with might and main (meaning with a lot of strength) contain
words which are no longer found outside of these formulaic expressions
(see Fillmore, Kay, and OConnor 1988).
However, an important, and as yet unresolved, question for such usagebased hypotheses is posed by Croft (2007: 504); he asks, How many
tokens is enough to entrench a linguistic unit? Certainly, in the case of the
it-cleft, it is unlikely that frequency alone can account for its entrenched
form. In Section 4, I suggested that the it-cleft has become less associated
with the constructions which inform the behaviour of its component parts
and instead aligns with other constructions which support the it-clefts
179
structure as a whole, such as information structure generalizations and

extraposition constructions. Thus, while the it-cleft has become more
entrenched over time, resulting in a fixed form which is resistant to the
changes affecting its component parts, it is not entirely unsupported. In this
instance then, fossilization is accompanied by pattern-alignment with more
general structures. Since the configuration is now motivated less directly,
its structural details are difficult to make sense of without considering the
historical evidence.
Contra Declerck (1983b: 15) then, it is not only the proverbial cleft that
is reminiscent of an older stage in the English language, but all varieties
of it-cleft and demonstrative cleft. The basic structure of present-day clefts
is therefore older than the individual proverbial cleft tokens, which have
emerged through a similar process of entrenchment (or fossilization). Ball
(1991: 148) records the first proverbial cleft as occurring in Early Middle
English, in The Owl and the Nightingale, shown in (53).
(53)
Nis no man for is bare songe

Lof ne wr not sue longe;
Vor at is a furwore man
at bute singe not ne can.
Not-is no man for his mere song / Esteemed not is not very long
For that is a worthless man / That but sing nought not can
No man for his mere song is esteemed for very long; for that is a
worthless man who can do nothing but sing
(1250-1300, The Owl and the Nightingale [Ball 1991: 148])
In such examples, the discontinuous definite NP has generic reference (see

Chapter 5, Section 2.2). That is, it evokes an abstract, hypothetical entity
whose type is given by the predicate NP (man) and whose basic attributes
are specified by the at-clause (here, the man who can do nothing but
sing) (Ball 1991: 149). Since such tokens are often used metaphorically,
and so can be applied in a number of similar situations, their frequency may
be relatively high. This leads to the entrenchment of the instance as a substantive construct, stored low in the constructional taxonomy as a fixed and
invariable saying.
However, while it is clear that the storage and retention of cleft structure
involves entrenchment at various levels in the constructional hierarchy, we
have yet to explain how this discontinuous structure first came into being.
Nevertheless, the taxonomic network of related constructions provides us
with a plausible hypothesis. As instances of either the specificational inver-
180
sion construction or the predicate nominal construction, we would expect

the Old English cleft tokens to have developed from simple NP be NP
sentences introduced by the neuter pronominal subjects, is, t and hit. As
Ball (1991: 24) notes, it is plausible that a language could have the simple
copular sentence without having the cleft.
For instance, (54) is a simple OE specificational inversion sentence, in
which the initial hit is a full anaphoric NP with the meaning the one that
was standing there.
(54)
sde t Petrus r stode. a geleaffullan cwdon t hit

said that Peter there stood the faithful
said
that it
nre
Petrus, ac wre his engel.
not-were Peter but were his angel
[Rhoda] said that Peter was standing there. The faithful said that it
wasnt Peter, but was his angel.
Once we add a paratactic relative clause (in accordance with the Old English determinative pronoun construction), we obtain a specificational it-cleft
as in the example in (45), repeated below as (55). Here, the initial hit is
restrictively modified by the extraposed clause and functions as the determinative head of the definite description it (the one) that is knocking there.
(55)
a cwdon a geleafullan, Nis hit na Petrus t r cnuca, ac is

his ngel.
(cocathom1,CHom_I_, 34:474.247.6867)
Then the faithful said: It isnt Peter who is knocking there, but his
angel.
(Ball 1991: 39)
Ball (1991: 67) provides a similar account of these OE tokens; she suggests
that the predicational and identificational hit-clefts are simple sentence[s]
expanded by the adjunction of a relative clause.
However, Ball provides a different origin story for the specificational itcleft (see Section 2). She suggests that this construction originated in Early
Middle English, with examples such as (56), repeated from (1).
(56)
A-bidez, quath is holie man: ore louerd is guod and freo.

e deuel it is at bringuth is wedur
[As St. Edmund was preaching, it became overcast, a terrible wind
began to blow, and it grew dark. People began to leave.] Stay, said
181
this holy man, our Lord is good and free. The devil it is that
brings this weather
(1280-90 South English Legendary [Ball 1991: 158])
Ball (1991: 158) claims that this example has a different function from the
OE hit-clefts, since It is not the case that there is a storm-bringing entity in
the context whose identity is at issue. Rather, the devil is contrasted with
God as the value of X in the OP [open proposition] X brings this weather.
More importantly, however, this EME token differs from the OE examples
in that it is actually amenable to an expletive account of the specificational
it-cleft, occurring as it does only after the system of gender agreement has
already broken down (see Section 6).
Searching for a viable source construction, Ball comes across an Old
English NP beon relative clause configuration, exemplified by (57), also
(2). She suggests that such examples acquired expletive subjects in Middle
English; The resulting structure is superficially similar to but structurally
distinct from the OE identificational hit-cleft (Ball 1991: 68).
(57)
min fder is e me wuldra

my father is that me glorifies
It is my father that glorifies me
However, while Ball (1991: 51) translates these examples as it-clefts, she
notes that they are closer in form to reverse pseudoclefts, since the initial,
focal NP is the grammatical subject. Furthermore, Ball acknowledges that
this source construction is very rare in Old English; it occurs in her data
only in translations of Latin headless relatives and is sometimes avoided
even then. For example, (57) is an English translation of the Latin original
est Pater meus, qui glorificat me. She concludes that in Late West
Saxon, at least, NP/PRO BEON REL-CLAUSE was not the preferred construction for marking focus and open proposition (Ball 1991: 52).
Balls (1991) account of the specificational it-clefts origin is therefore
shaped by her adherence to an expletive analysis. Instead, I suggest that the
historical data calls for a much simpler account, on which the specificational it-cleft develops in OE, alongside the predicational variety. In proposing two different origin stories for present-day it-clefts, Balls account is
in many ways parallel to Matsunamis (1961: 6) earlier claim that the identifying and disjunctive functions of it-clefts originally belonged to different
constructions,they were combined in one and the same formula in the
182
course of the syntactical development of English. Indeed, Ball makes use

of Matsunamis distinction between the identifying and emphasizing (or
contrastive) functions of it-clefts as a way of differentiating between identificational and specificational it-clefts.
For example, in support of a non-specificational analysis for the OE hitclefts in (51) and (52) above, Ball (1991: 66) notes that Matsunami finds
(52) to be identifying rather than emphasizing. She also cites Mitchell
(1985a: 622), who discards (51) as an example of the modern use of it to
give emphasis. Ball goes on to employ this distinction in her assessment
of (57) as properly specificational. Again she cites Mitchell (1985a: 622),
who says that OE achieves emphasisby giving the noun initial position. However, for Matsunami (1961: 14), example (57) has an identifying nature and so is also of the identifying pattern. This suggests that the
functions of identifying and emphasizing are simply not appropriate for
distinguishing between different kinds of copular sentence.
As we saw in Chapter 3, the purpose of a specificational sentence is to
correctly identify the entity which uniquely matches the description given
by the predicative NP. The identifying function then, neither characterizes,
nor properly corresponds to, Higgins (1979) definition of the identificational copular sentence as a naming or labelling device. That Ball conflates
Matsunamis identifying-emphasizing distinction with Higgins characterization of identificational and specificational copular sentences therefore
results in a misunderstanding of the nature of specificational meaning and
further undermines the basis on which the OE hit-clefts are separated from
(and discounted from) the later specificational it-cleft data.11
11. An example of this is Balls (1991: 156) treatment of the following token. Here,
St. Mary converses with someone who she thinks is the gardener. However, it
is really Jesus, who has risen from his tomb. Ball analyses (vi) as an Early
Middle English identificational it-cleft. She notes that On the specificational
reading, which is brought about by unclefting [shown in (vii)], Mary would
be strangely mistaken about the event taking placeWhereas on the identificational reading, she is only mistaken about the identity of her interlocutor, not
about what is actually happening (Ball 1991: 156157).
(vi) Heo wende hit were e leyhtunnward. at to hire spek.
She thought it was the gardener that was speaking to her
(1250-1300 Passion of Our Lord [Ball 1991: 156])
(vii) Heo wende e leyhtunnward spek to hire.
She thought the gardener was speaking to her
(Ball 1991: 157)
Summary
8.
183
Summary
In this chapter, I have examined it-clefts from earlier periods of English in

relation to the synchronic language system of the time. In doing so, I have
identified some illuminating generalizations which are not at all transparent
in present-day English and have posited inheritance relations which are no
longer productive in the constructional taxonomy. The historical evidence
therefore provides motivation (via default inheritance) for properties which
are specific to a discontinuous constituent (extraposition-from-NP) analysis
of it-clefts. We now have a plausible explanation as to what sanctions the
restrictive modification of the pronoun it, why the cleft clause is obligatorily placed in an extraposed position and which events have led to the itclefts unusual pattern of number agreement. In addition, while the evidence suggests that the it-cleft configuration has changed very little over
time, a historical perspective nevertheless provides us with the opportunity
to examine those structural properties which have undergone change, and
which are no longer observable in present-day English. In particular, I have
shown that Old English gender agreement patterns offer strong support for
a discontinuous constituent account.
The historical cleft data is therefore extremely instructive. It not only
supports the analysis of cleft structure argued for here, but also provides us
with a number of plausible explanations as to why the present-day it-cleft is
structured in this way. As Matsunami (1961: 1) observes, not one of the
various synchronic accounts of it-clefts has succeeded in explaining its
grammatical irregularity, for which only the historical consideration seems
to be equal to answer, but to which, to my mind, enough notice has never
been given.
However, this reasoning is problematic. While (vii) prefers a predicate-focus

reading, on which Mary could be mistaken about what is happening i.e. that he
spoke to her, this information is presupposed in the corresponding it-cleft and is
therefore not at issue. Even if we interpret specificational it-clefts as information-structure variants of noncopular sentences, they nevertheless provide an
unambiguous argument-focus reading (Lambrecht 2001: 485), on which only
the identity of St. Marys interlocutor is at issue (see Chapter 4, Section 1.2.1).
Chapter 7
The it-clefts development over time
1.
A diachronic investigation
In Chapter 6, we saw that a historical perspective provides motivation for

some of the it-clefts more idiosyncratic properties by revealing inheritance
relations that are internal to the structured inventory of an obsolete system
of grammar. In particular, the cleft structure has been shown to be a relic
from an earlier time, shaped by generalizations which are no longer true of
the present-day language system. In this sense then, the story of the it-cleft
is one involving an absence of change, or rather, a resistance to change. It is
this process of conventionalization that, above all else, creates the illusion
of idiosyncrasy. Nevertheless, the it-cleft construction has not been entirely
impervious to change, as we will see in this chapter. Most noticeably, the
frequency of the it-cleft has increased and its function has expanded over
time. In what follows, I chart the it-clefts diachronic progress all the way
from Old English to Modern English. I conclude that the construction has
gradually acquired some of its more idiosyncratic properties as a result of
language change. In this way, the diachronic evidence provides additional
motivation for the present-day it-cleft construction, via factors which are
external to the language system (see Goldberg 2003: 121).
The chapter is structured as follows. In Section 2, I discuss aspects of
the methodology, including the corpora used and the method of data extraction. Here, I detail and defend a much more rigorous selection process than
that employed in Filppulas (2009) recent study of Old and Middle English
it-clefts. I then go on to provide some basic frequency information in Section 3, before focusing in more detail on the it-clefts gradual acceptance of
a wider range of focal elements in Section 4 and its appearance in a greater
variety of discourse contexts in Section 5.
2.
The corpora, the search and the selection process
For this diachronic investigation, I make use of data from four independent,
yet related, historical English corpora: the York-Toronto-Helsinki Corpus of
Old English Prose (YCOE), the Penn-Helsinki Parsed Corpus of Middle
185
English, second edition (PPCME2), the Penn-Helsinki Parsed Corpus of

Early Modern English (PPCEME) and the Penn Parsed Corpus of Modern
British English (PPCMBE). These four corpora form part of the same series
of syntactically annotated historical English corpora from the University of
Pennsylvania and the University of York. As such, they are broadly (and as
much as possible) comparable in terms of the size of the corpus, the number and size of the text files and genre composition. All employ roughly the
same system of syntactic annotation, which permits the searching of
syntactic structure using the specially-designed CorpusSearch2 program.
Together, the YCOE, the PPCME2, the PPCEME and the PPCMBE comprise over 5 million words (totalling 5350956), spanning the entire history
of British English up until 1914.
The relevant tokens were retrieved by searching for the structure of the
cleft clause, labelled as CP-CLF. The annotation provides cleft clauses with
the same internal structure as that of relative clauses, but analyses them as
daughters of IP. This mark-up is not inconsistent with either an expletive or
an extraposition-from-NP account: it is intended only as an aid to searching, not as a linguistic statement. Nevertheless, there is some indication that
the annotation process has been influenced by the standard assumptions of
an expletive analysis. Given an extraposition-from-NP account, we would
assume that a search of cleft clauses would isolate not only it-clefts, but
also demonstrative clefts, introduced by this and that (see Chapter 4, Section 1). However, the PPCME2 and PPCEME annotation manual (Santorini
2010) associates the label CP-CLF exclusively with it-clefts, suggesting
that this is a structurally unique, construction-specific type of clause.
Since the focus here is on it-clefts, this assumption is not so problematic
for the purposes of data collection. However, in the YCOE specifically, the
adoption of an expletive account is much more overt and has an unwelcome
effect on the search output. The YCOE reference manual states that the CPCLF structure is not contained within its antecedent or coindexed to the
expletive subject, which may be empty or overt (A. Taylor 2003, emphasis
added). This statement includes the fundamental assumption of an expletive
analysis and is at odds with a discontinuous constituent account. It assumes
further that, since the cleft pronoun is a meaningless dummy element, it can
be realized as an empty subject. The search for CP-CLF in the YCOE therefore locates not only all it-clefts, but also tokens with the surface structure
be XP relative clause, which are given an empty expletive subject in the
syntactic annotation. The YCOE reference manual confirms that, When
the HIT subject of a cleft is missing an empty expletive subject is added
(A. Taylor 2003).
186
If, however, we acknowledge the evidence that the cleft pronoun is not
in fact expletive, and instead functions as a definite determinative pronoun,
the inclusion of sentences without determinative pronouns as instances of
cleft structure makes little sense. True to form, once we examine the discourse contexts of these hit-less tokens, they appear to be instances of an
OE presentational construction rather than a subtype of it-cleft.
2.1.
OE presentational/impersonal sentences
A search of the CP-CLF structure in the YCOE yields 94 cleft tokens. The
majority of these examples occur without the pronoun hit. This includes 50
tokens from a single text file, cobede.o2, taken from Bedes Ecclesiastical
History of the English People, such as (1) and (2) below.
(1)
a ws fter hire deae, t a broor oerra weorca swiur

gemdon & isse cirican timbro forlton soefon gear.
(cobede,Bede_3:6.176.8.1722)
Then after her death, the brethren were more occupied with other
works, and for seven years neglected the erection of the church
(Miller 1898: 177)
(2)
a ws fter elberhtes forfore, Eadbald his sunu feng to am

rice, ond he sona micle wonunge & werdlan ws re mrwan
cyrican weaxnisse.
(cobede,Bede_2:5.110.22.1042)
Then after thelberhts death his son Eadbald succeeded to the
throne and soon he was cause of great loss and injury to the growth
of the tender church
(Miller 1898: 111)
In these examples, the postcopular PP provides a temporal setting for the

event described in the following clause. Although parsed as clefts in the
YCOE, and labelled time-clefts in the reference manual (A. Taylor 2003),
these tokens have a very different function from the specificational it-cleft.
For one thing, the examples do not contain presupposed t-clauses. As we
have seen, in Chapter 4, Section 1.2.2, presupposition characterizes all different kinds of specificational copular sentence, by simple virtue of the fact
that they contain definite descriptions. Even in informative-presupposition
it-clefts, brand-new information in the cleft clause is marked as expected in
some sense or not at issue (see Chapter 5, Section 4).
187
Once we examine the discourse context of these hit-less OE tokens, it is

difficult to interpret them as cleft sentences. For example, (1) occurs as part
of a discourse which follows the story of the abbess, who starts to build a
church in her monastery but dies before she can finish it. Surprisingly, it is
a full seven years before the brethren recommence the building work and
carry out the abbess plans. That the brethren are so diverted from this task
is not only new to the reader, it is presented as the main informational content of the sentence; that is, as an unexpected turn of events. Similarly, in
(2), Eadbalds ascension to the throne is presented as a highly significant
event. In the prior context, Bede writes of thelberhts life and death, with
no mention of Eadbald. As a result, we cannot interpret (2) as simply providing us with the timing of Eadbalds ascension to the throne. Instead, the
event signals the beginning of a cruel and chaotic reign, which contrasts
starkly with the good deeds of thelberht.
The non-presuppositional YCOE tokens therefore lend themselves to a
presentational interpretation, in which actions or events are presented in a
neutral manner. Here, I follow Ball (1991) in referring to such sentences as
impersonals. On this account, the verb wesan is not the copula that we find
in clefts, but has a full meaning, similar to happen or come to pass (see
Visser 1963: 50). This is followed by a sentential complement (rather than
a relative clause), which provides the primary informational content. Such
sentences often occur with scene-setting adjuncts, such as the PPs after her
death and after thelberhts death in (1) and (2) above, which leads to a
superficial similarity with it-clefts containing prepositional foci. However,
in impersonals, this postcopular adjunct is optional; it is not an obligatory
part of the sentence.
As Ball (1991: 95) observes, Old English has a number of presentational
verbs of this type, including (ge)weoran, gelimpan, geslan, getimian,
getidian, and cuman. These happen-class verbs commonly occur with the
expletive subjects hit and t, shown in (3). However, when some other
element occupies the first position, hit and t are optional (Ball 1991: 93).
This explains why (1) and (2), which are introduced by the adverb a, contain empty expletive subjects.
(3)
t gelamp on sumre niht, t r com sum man to s halgan

weres sprce.
It happened one night that there came a certain man to speak to the
holy man
(Saint Guthlac [Ball 1991: 99])
188
Ball (1991) explicitly discusses Bedes Ecclesiastical History in relation

to this OE impersonal construction. She notes that the historical works,
with their focus on events, favour the occurrence of HAPPEN-impersonals
(Ball 1991: 98) and goes on to find a significant association between their
occurrence and the start of major segments in the narrative. An impersonal
analysis for the examples from Bede is further supported by Millers (1898)
translations, shown in (1) and (2) above. Miller chooses to translate these
passages, not as present-day it-clefts, but as simple sentences with initial
thematic temporal expressions. As Ball (1991: 95) observes, these tokens
cannot be rendered in any other way, because present-day English has no
verb which introduces an event into the discourse, in a neutral manner, via
a that-clause.
I have taken care to exclude those tokens which are plainly impersonals
(both with and without it) from my historical dataset and include only those
tokens which clearly prefer an it-cleft analysis. This has had an enormous
impact on my calculation of the frequency of YCOE it-clefts, leaving me
with only 13 relevant tokens. In contrast, Filppula (2009) is not as selective
in his recent approach to the YCOE data. He claims that the YCOE timeclefts, such as the tokens from cobede.o2, clearly belong to the class of
informative-presupposition clefts (Filppula 2009: 272). As a result, Filppula translates the OE hit-less impersonals, including (1) and (2) above, as
present-day it-clefts, shown in (4) and (5) below. In all other respects, Filppula accepts Millers (1898) translations. He notes, I have followed here
Millerstranslation but added clefting to the beginning of the sentence as
indicated by the parsing in YCOE (Filppula 2009: 274).
(4)
Then [it] was after her death that the brethren were more occupied
with other works, and for seven years neglected the erection of the
church
(Filppula 2009: 272)
(5)
Then it was after thelberhts death that his son Eadbald succeeded to the throne,
Filppula does not discuss the discourse context or the function of these
examples in any detail. However, he does comment that they do not contain
presupposed that-clauses, despite going on to classify them as informativepresupposition clefts (see Filppula 2009: 273). This suggests that Filppulas
figure of 89 YCOE clefts is, at least to some extent, overblown. Indeed,
Filppula (2009: 276) acknowledges that, some of the structures parsed as
189
clefts in YCOE, and especially the type of time-clefts found in Bedes

texts, may be open to differing interpretations.
While the hit-less impersonals are mostly confined to Old English, the
variant with expletive it continues to appear. For example, the sixteenth
century token in (6) does not allow a cleft analysis because we cannot
establish a plausible focus. Instead, the example prefers an impersonal
interpretation, akin to just as it happens that or just as it is the case that.
It is telling that in later versions of this text, such as (7) and (8), the passage
is rendered as a simple sentence with an initial adverbial expression, rather
than as an it-sequence.
(6)
And I do not speke nowe of the voluntarye mouynges of the soule,

that hath knowledge, but of the natural intencion of thynges, euen
as it is that we do digest meates, that we haue eten without
thynkyng thereon howe it is digested and as we do take wynde and
breathe in slepe, not knowyng thereof.
(1556 BOETHCO-E1-H,80.509)
(7)
As our meate we take without great study

(1593 BOETHEL-E2-H,68.223)
(8)
Thus we swallow our Meat without thinking of it,

(1695 BOETHPR-E3-H,145.236)
Likewise, examples containing the question word how, such as (9), often
favour an impersonal reading. These tokens seem to question how it is even
possible that the situation denoted by the that-clause has come about, rather
than asking in what manner it took place (see also Ball 1991: 104).
(9)
Then saith the woman of Samaria vnto him, How is it that thou,
being a Iewe, askest drinke of me, which am a woman of Samaria? For the Iewes haue no dealings with the Samaritanes.
(1611 AUTHNEW-E2-H,IV,1J.361)
Even in the Late Modern English period, we find similar examples containing the fixed phrases hence it is and therefore it is. As shown in (10)
and (11), the adverbs hence and therefore make unlikely cleft foci; there is
certainly no element of contrast here. Instead, these introductory phrases
serve a presentational function, which can be paraphrased as and so it happens or and so it is the case.
190
(10)
Hence it is, that the will, induced by powerful reasons, sometimes

chuses and embraces death, altho' nature dreads and abhors it;
(BOETHRI-1785,123.244)
(11)
and therefore it is that we have inserted some passages verbatim

from his work, Neuralgia and its Counterfeits
(POORE-1876,163.56)
2.2.
Existential sentences with it
In addition to the impersonal/presentational tokens, the CP-CLF output also

includes some existential sentences. Formerly, existential sentences could
occur with it as well as there. For instance, the rhetorical question below
asks what is there (or what exists) that a man can do to another man which
cannot be done, in turn, to him. The expected answer to this question is of
course nothing. The follow-up sentence offers an example of this truism
by reminding us that Busiris, who frequently killed his guests, was in turn
killed by one of his guests Hercules.
(12)
To go further, what is it that any man may do to another, which

another may not do again to him? We are told, that it was the
Custom of Busiris to kill his Guests, and himself at last was killed
by Hercules his Guest.
(1695 BOETHPR-E3-P1,76.509)
While these tokens are superficially similar to clefts, they are pragmatically
distinct. Existential sentences question rather than presuppose the existence
of the entity described in the that-clause (Ball 1991: 269). For this reason, I
have excluded these tokens from my data-set.
I have also eliminated 11 tokens of the type whase itt iss attfrom the
PPCME2 data. This construction, found in The Ormulum (cmorm.po.m1),
also seems to prefer an existential reading. For example, the first sentence
of (13) serves to establish a generic, indefinite entity whoever there is that
loves peace, which is then predicated of in the following sentence, shall
find Jesus.
(13)
& whase itt iss att lufe gri & follhe wi hiss herrte, att
mann shall findenn Jesu Crist To beon wi himm i blisse.
(c1200 CMORM,I,227.1888)
191
And whoever it is that loves peace And follows with his heart That
man shall find Jesus Christ To be with him in bliss.
(= Whoever loves peace...shall find Jesus Christ...)
(Ball 1991: 185)
As a result, there is very little difference in meaning between these tokens
and corresponding simple sentences introduced by indefinite wh-subjects,
as shown above. Ball (1991: 188) concludes that their appearance in The
Ormulum, which is written in verses of 15 syllables, may be due to the fact
that itt is att is metrically useful. Whatever governs its appearance, it is
clear that Orms construction is more existential than cleft. As Ball (1991:
186) notes, if they are it-clefts then they are of a type not yet seen and no
longer extant.
2.3.
The pattern I it am
11 tokens introduced by the pattern I it am have also been omitted from the
PPCME2 search results. Ten of these, shown below, are from a single passage of the same text: Julian of Norwichs Revelations of Divine Love.
(14)
ofte tymes oure lorde Ihesu sayde to me, I it am that is hiaste. I

it am that ou luffes. I it am that thowe lykes. I it am that owe
serves. I it am at ou langes. I it am that owe desyres. I it am
that thowe menes. I it am at is alle. I it am that haly kyrke preches
the and teches the. I it am that schewed me are to the.
often times our lord Jesus said to me, I am the one that is the
highest. I am the one that you love. I am the one that you like. I am
the one that you serve. I am the one that you long [for]. I am the
one that you desire. I am the one to whom your intention is
directed. I am the one that is all. I am the one that holy church
preaches to you and teaches you. I am the one that showed myself
to you before.
(c1400-50 CMJULNOR,59.256-265)
Although these tokens look like focus-first specificational it-clefts, they are
actually predicational. The function of these sentences is to tell us more
information about the subject (Jesus), rather than to identify the referent
described by the discontinuous definite NP. Here, the pronoun I is the
thematic subject, which shows agreement with the copular verb am. The
postcopular relative modifies the determinative it to form a predicative NP.
192
The configuration I it am is therefore equivalent to I am he (see Ball 1991:

71). Watson and Jenkins (2006: 206) note that the use of I it am rather than
I am he may be a way of deemphasizing the maleness of Jesus at a time
when the gender-neutral construction I am the one had not yet developed.
2.4.
Other constructions mistaken for clefts
The search for CP-CLF in the historical English parsed corpora produces
several other structures which are easily mistaken for it-clefts. For example,
although (15) below is annotated as a cleft by the YCOE (and is treated as
such by Filppula (2009: 274)), it is most likely a simple NP be NP specificational copular sentence followed by a nonrestrictive relative.
(15)
...and t hus afylde mid ormtum stence, t man eae mihte

witan t hit se deofol ws e hine dwelian wolde;
(coaelive,LS_[Martin]:770.6456)
...and that house filled with intense odour so that one could easily
know that it was the devil who wanted to mislead him;...
In this example, the devil pretends to be Christ. What is at issue here, then,
is the identity of this vision, as either Christ or the devil. If the vision is
truly Christ, then no deception can be said to have occurred. Therefore, the
information in the relative clause x wanted to mislead him cannot be presupposed. Instead, it provides additional information about the devil, explaining his purpose for appearing as Christ.
Likewise, I do not consider (16) to be an it-cleft, despite its mark-up in
the YCOE. Here, the relative clause is properly restrictive, but it modifies
the postcopular noun Veronica rather than the initial it. In this example,
Veronica is treated as a common noun, representing the type people named
Veronica. The initial hit is used with human reference.
(16)
...and eac hyt ws seo ylce Veronix, e s hlendes reafes

thran and wear urh t fram s blodes fleusan gehled.
(covinsal,VSal_1_[Cross]:18.1.129)
because [and also] she was the same Veronica who had touched
the Saviours garment, and was healed from the flow of blood.
(Cross 1996: 114)
193
A similar example is found in the output for the PPCMBE. In (17), the zero
relative modifies the noun time to form a postcopular NP (the first time I
have been quoted as an authority by an eminent outsider). The initial it is a
full anaphoric pronoun referring to the situation depicted in the previous
sentence.
(17)
I believe to-day there was in the Times what I may call the first
personal thorough recognition of my working life from an
educational quarterIt is the first time I have been quoted as an
authority by an eminent outsider,
(THRING-187X,216.28)
Finally, it is well-known that sentences with extraposed that-clauses are

difficult to distinguish from it-clefts (see Calude 2008a; Haugland 1993). In
the following example from the PPCMBE output, the that-clause (that she
did) is an extraposed subject, which is predicated of by the VP is in the
family. The intended meaning here is something like that she did say it and
think it is known in the family.
(18)
That poor child Carinthia Jane, when first she beheld Old
England's shores, tossing in the packet-boat on a wild Channel sea,
did say it and think it, for it is in the family that she did; and no
wonder that she should
(MEREDITH-1895,19,132.8-9)
2.5.
Interim summary
The CP-CLF annotation in the English historical parsed corpora is therefore

applied to several constructions which are not it-clefts. This is unsurprising,
since it is notoriously difficult to identify and isolate it-clefts from other,
superficially similar but structurally distinct, sentence types. In most cases,
however, a manual check of the discourse context of each token provides
enough evidence for a reliable classification. Other times, the judgements
are much more subtle and I discuss some of the difficult cases in Section 4.
After eliminating examples which prefer an impersonal reading, sentences
with existential it, and tokens where the clausal component has some other
relationship to the rest of the sentence, what remains is a dataset of specificational and predicational it-clefts spanning over 1000 years and comprising over 500 tokens (with 514 it-clefts in total).
194
3.
Frequency information
Frequency information, charting the development of the English it-cleft, is

provided in Table 1. These figures represent data from all four corpora: the
YCOE (1.5 million words), the PPCME2 (1.2 million words), the PPCEME
(1.7 million words) and the PPCMBE (1 million words). The data is mostly
separated into 70100 year intervals, using time periods recognized by the
English historical parsed corpora series.1 However, the Old English data
has not been divided up in this way, since it is difficult to separate the
YCOE text-files into discrete time periods. Table 1 indicates both raw numbers and frequencies which are normalized to a corpus of 500,000 words
(shown in parentheses). The numbers given here are quite conservative; I
have only included what I believe are clear cases of specificational and
predicational it-clefts.2
Table 1. The frequency of it-clefts throughout Old, Middle and Modern English
OE (1150)
ME 1 (11501250)
ME II (12501350)
ME III (13501420)
ME IV (14201500)
E I (15001569)
E II (15701639)
E III (16401710)
MBE I (17001769)
MBE II (17701839)
MBE III (18401914)
Specificational
7 (2.4)
3 (5.8)
3 (10.2)
12 (12.3)
17 (32.0)
26 (22.6)
40 (30.6)
103 (91.1)
82 (137.2)
105 (142.4)
95 (168.8)
Predicational
6 (2.1)
2 (3.9)
1 (3.4)
3 (3.1)
1 (1.9)
1 (0.9)
1 (0.8)
1 (0.9)
1 (1.7)
1 (1.4)
3 (5.3)
Total
13 (4.5)
5 (9.7)
4 (13.6)
15 (15.4)
18 (34.0)
27 (23.4)
41 (31.4)
104 (92.0)
83 (138.9)
106 (143.7)
98 (174.2)
These figures show a general increase in the occurrence of it-clefts over

time. The normalized frequencies, in particular, increase steadily at almost
every interval. The only exception relates to the period MEIV (14201500),
which contains a higher proportion of it-clefts than the following two subperiods of Early Modern English. The reason for this aberration seems to
1. The PPCME2 is divided into subcorpora indicating both composition date and
manuscript date. Here, I conflate these subcorpora to indicate only the time of
composition (where known).
2. Nevertheless, as a consequence of reexamining the data, the frequencies given
here are slightly higher than those given in Patten (2010) for Late Middle and
Early Modern English.
Frequency information
195
lie in the high frequency of it-clefts in two Late Middle English texts: The
Book of Margery Kempe and Malorys Morte Darthur. 11 of the 18 it-clefts
in MEIV appear in these two text-files (cmkempe.m4 and cmmalory.m4),
which contain several instances of hidden or mistaken identity. In (19), for
example, a woman is berated for not believing that the one who spoke to
her was truly God. Likewise, in (20), King Arthur fails to identify Merlin
(who is in disguise) as the man currently speaking to him.
(19)
& is chastisyng schal enduryn xij days tyl u wyl beleuyn at it is

God whech spekyth to e & no deuyl.
(c1450 CMKEMPE,146.3375)
and this chastising shall endure twelve days until you will believe
that it is God who speaks to you and no devil.
(20)
And Merlion was so disgysed that kynge Arthure knewe hym nat
Sir, seyde thes two knyghtes, 'hit ys Merlion that so spekith unto
you.
(a1470 CMMALORY,30.952)
In any case, Table 1 suggests that the it-cleft has steadily become a more
established construction over time, occurring just under 40 times more frequently in the late nineteenth century data than in Old English. What is
more, the figures clearly show that this increase is limited to the specificational variant of this construction. While the predicational it-cleft is present
throughout all periods of English, it maintains a relatively low frequency.
In Old English, the specificational it-cleft and the predicational it-cleft have
roughly the same rate of occurrence. In MEI, the frequency of the predicational it-cleft is 2/3 the size of that of the specificational it-cleft, which
drops to 1/3 at MEII and again to 1/4 MEIII. However, from the end of the
Late Middle English period onwards, the two constructions take different
paths and their frequencies are no longer comparable.
Interestingly, the predicational it-cleft is attested slightly earlier than the
specificational it-cleft. While all 7 of the OE specificational examples occur
in the first half of the eleventh century, 3 of the predicational tokens are
dated from around the middle of the tenth century. Balls (1991: 53) observation about predicational it-clefts therefore applies equally to this data; she
notes that they now appear as exceptions to generalizations about the cleft,
but they represent the earliest attested English cleft construction with neuter
pronoun subject. The question therefore presents itself: why is this variety
now exceptional in relation to the specificational it-cleft? In other words,
196
what accounts for the divergence in the development of these two related
constructions?
The predicational it-cleft seems to have changed little since Old English.
For instance, the OE predicational it-cleft in (21) has the same structure and
function as the nineteenth century token in (22).
(21)
Foram hit is swie ryht spell t Plato se uwita sde;

(coboeth,Bo:35.95.19.1836)
It is a very true saying that Philosopher Plato spake
(Sedgefield 1900: XXXV)
(22)
and the attention of the pupil must be engrossed in the first

instance with overcoming these difficultiesStill, it is but a divided
attention that we can give to the exercise. (BAIN-1878,383.350)
In contrast, the general increase in the specificational it-clefts frequency is

known to be accompanied by two main changes: it occurs with a greater
range of non-nominal foci over time and it acquires a new discourse
function, in which the cleft clause changes from expressing only given
information to also expressing new information presented as fact (see Ball
1994a; Patten 2010). I outline each of these developments in Sections 4 and
5, respectively, before discussing different ways of accounting for them in
Chapter 8.
4.
Changes to the clefted constituent
Table 2 shows the syntactic categories of specificational it-cleft foci in the

diachronic corpora. As anticipated, the data indicates that this construction
occurs with a wider and more diverse range of focal elements over time.
For instance, clear examples of it-clefts with clausal and adverb phrase foci
do not appear in the corpus until Early Modern English. Furthermore, while
it-clefts with prepositional foci are present in Middle English, these also
become much more frequent at the end of the Early Modern period, when
there is a substantial increase in the overall number of specificational itclefts, and AdvP-focus it-clefts start to appear. As shown in Table 2, itclefts with NP foci, which are (ordinarily) the most common type of it-cleft
at each stage of English, are also the oldest, dating back to Old English.
197
Table 2. The changing frequencies of it-clefts with a range of focal categories

OE
ME I
ME II
ME III
ME IV
EI
E II
E III
MBE I
MBE II
MBE III
NP
7 (2.4)
2 (3.9)
3 (10.2)
11 (11.3)
17 (32.0)
24 (20.8)
38 (29.1)
74 (65.5)
52 (87.0)
42 (56.9)
57 (101.3)
PP
AdvP
CL
1 (1.9)
1 (1.0)
1 (0.9)
2 (1.5)
19 (16.8)
26 (43.5)
54 (73.2)
28 (49.8)
1 (0.9)
9 (8.0)
2 (3.3)
7 (9.5)
6 (10.7)
1 (0.9)
2 (3.3)
2 (2.7)
4 (7.1)
Total
7 (2.4)
3 (5.8)
3 (10.2)
12 (12.3)
17 (32.0)
26 (22.6)
40 (30.6)
103 (91.1)
82 (137.2)
105 (142.4)
95 (168.8)
This change, involving the development of it-clefts with different categories of foci, is often taken to indicate the relaxation of a syntactic constraint: the focal position loses its NP specification and allows other kinds
of phrasal elements to enter into it (see Patten 2010). The data-set in Table
2 suggests that this expansion begins relatively early on, with the appearance of PP-focus it-clefts in Early Middle English. This it at odds with
Balls (1994a) suggestion that it-clefts with non-NP foci are the product of
a Late Middle English development, involving a partial merger between the
existing NP-focus it-cleft and the it-impersonals (see Chapter 8, Section
3.1). She cites LME examples of the PP-focus it-cleft dating from the early
fifteenth century.
On reexamining the LME data in Patten (2010), I came across a slightly
earlier token from the late fourteenth century (in period MEIII of the
PPCME2).
(23)
Me trowe at by e prayers of is holy mayde it is at at

Me believes that by the prayers of this holy maid it is that that
place was nevere it destroyed (a1387 CMPOLYCH,IV,125.869)
place was never yet destroyed
I think that it was by the prayers of this holy maiden that that place
was never destroyed
(Patten 2010: 230)
I argued that (23) requires an it-cleft analysis rather than an impersonal

interpretation because if we remove the prepositional phrase (in bold), the
meaning of the sentence is compromised. For instance, in (23), the writer
expresses his belief that Ethelbergas prayers prevented the destruction of
198
the abbey. However, on the impersonal interpretation, shown in (24), the

writer expresses a doubt about whether it is a fact that the abbey was never
destroyed. This is a far less intuitive gloss of [(23)], which seems instead to
presuppose the fact that the abbey was never destroyed (Patten 2010: 230).
This suggests that the postcopular PP is a focal element rather than a purely
optional adjunct; that is, it identifies the cause of the abbeys preservation,
rather than providing scene-setting temporal or spatial information.
(24)
I think it happened that that place was never destroyed.
A similar example occurs in the Early Middle English component of the

PPCME2. The following early thirteenth century token seems to be an itcleft rather than an it-impersonal because the information in the cleft clause
is presupposed. In the prior discourse, the devil provides a list of ways for
people to resist him and to protect themselves from the base desires which
lead them into filthy sin. One such weapon is given in (25): recognizing
the role of the devil in leading them to act on their shameful desires. This
interpretation is supported when we examine the following sentence, which
describes the devils evil nature.
(25)
enchen hit is urh me t hare lust leade ham to wurche to

wundre. enchen ef ha beie me? to hu bitter beast ha buhe.
(c1200-25 CMMARGA,78.379)
thinking that it is through me that their desire leads them to
shameful actions; thinking what a vicious creature they are paying
homage to if they submit to me
(Millet and Wogan-Browne 1990: 63)
If, on the other hand, we treat the postcopular prepositional phrase through
him as an optional adjunct, as in (26), the resulting reading is less plausible.
Here, as in (24), the relevant relation of agency is left unexpressed. This
fundamentally alters the meaning of the sentence.
(26)
thinking that it happens that their desire leads them to shameful

actions
The evidence from the early English corpora therefore suggests that itclefts could occur with prepositional foci as early as Early Middle English.
Indeed, it could be argued that the PP-focus it-cleft was actually available
in Old English. The YCOE data does not contain any clear-cut examples of
199
PP-focus it-clefts and consists of plenty of tokens which are unambiguously

it-impersonals, such as (27).
(27)
And hyt ws a on am ehtoan geare, e se mycla hungor heom

on becom, t hig for re hlafleaste a eoran ton.
(covinsal,VSal_1_[Cross]:14.2.105)
And it happened in the eighth year, when famine came upon them,
so that for want of bread they ate earth
(Cross 1996: 267)
Nevertheless, there are two genuinely borderline tokens in the YCOE, both
of which contain temporal adverbial phrases in postverbal position, expressing the duration of an event. On the one hand, these look like other
impersonals with measure of time adjuncts. On this basis, Ball (1991:
101) groups (28) with the OE impersonal from Bede in (29). On an impersonal reading then, (28) translates as Now, for 38 years, my daughter has
been lost to me. However, the example is also amenable to an it-cleft
analysis, on which the speaker Paphnutius specifies (and emphasizes) the
exact number of years that his daughter has been missing. This interpretation is not at odds with the discourse context and is the one opted for by
Filppula (2009: 273).
(28)
Nu hit is for eahta and ryttian gearan t min dohtor me losode

(coeuphr,LS_7_[Euphr]:270.283)
Now it is for eight and thirty years that my daughter hath been lost
to me
(Skeat 1881a: 351)
(29)
a ws twa gear t he t bsicopsetl swa st 7 heold.

(cobede,Bede_4:30.370.13.3702)
Then for two years he thus occupied and held the bishops seat
(Miller 1898: 371)
Here, I have chosen to exclude these tokens from the diachronic data-set
of it-clefts. As shown below, tokens with postcopular temporal expressions,
and which are clearly it-clefts rather than impersonals, do not appear in the
corpus until much later in the history of English. Therefore, a cleft analysis
for these borderline YCOE tokens is perhaps unmotivated (see also Ball
1991: 111). Nevertheless, the relatively early attestation of PP-focus itclefts in Early Middle English still raises the question of whether the it-cleft
construction ever actually had a syntactic constraint specifying for NP foci.
Although the diachronic data shows that the it-cleft occurs with a wider
200
range, and a greater number, of non-NP focal elements over time (see Table
2), the evidence suggests that this structural change is actually governed by
semantic conditions (see also Patten 2010).
Data from the YCOE and the PPCME2 suggests that, originally, the itcleft showed a preference for focusing NPs referring to the most discrete of
entities. In Old and Early Middle English, 10 out of 12 NP-focus it-clefts
(over 80%) contain proper name and pronominal foci which denote animate
individuals. Likewise, in Late Middle English, 22 out of 28 it-clefts with
NP foci refer to animate entities (almost 80%).3 This figure drops to under
50% in the PPCEME, with 64 animate NP foci in 136 relevant tokens (of
which 3 are indefinite NPs). In the PPCMBE, less than 40% of the 151 NPfocus it-clefts contain animate foci and only 52 of these 58 tokens have
definite, specific reference (under 35% of it-clefts with NP foci).
Abstract nouns, which do not denote discrete physical objects, occur as
it-cleft foci only from the Late Middle English period onwards. In LME,
they make up less than 15% of the NP-focus it-clefts (4 tokens out of 28).
This figure increases to over 20% in Early Modern English, with 29 out of
136 instances having abstract nouns in the focus position, and again to over
30% in Late Modern English, involving 48 out of the 151 it-clefts with NP
foci. As abstract nouns come to be used more frequently as it-cleft foci, we
find a strong tendency for them to occur in lists or accompanied by focusing adverbs. As we saw in Chapter 5, Section 3, such strategies are employed in present-day English as a way of individualizing characteristically
non-referring it-cleft foci, such as adjective phrases and manner adverbs, in
order to make them more acceptable (see also . Kiss 1998; Borkin 1984).
In Early Modern English, 27 out of the 29 examples containing postcopular abstract nouns (93%) occur with listed alternatives (19 tokens)
and/or with focusing adverbs (10 tokens). For example, in (30), the abstract
noun shame is presented as a discrete and distinct emotion, which stands in
contrast to pain as the most valuable basis for a childs punishment. Similarly, proud knowledge is defined and delimited through a contrast with
pure knowledge in (31). This example is especially interesting because here
contrastive listing also leads to a subtle reformulation of the cleft clause.
Although both clauses provide the same discourse-old information, the
former presents it in a much more neutral manner. I discuss this issue further in Section 5.
3. Ball (1991: 272) finds this same preference in her Late Middle English data.
She notes that 89 out of 111 NP-focus it-clefts (80%) contain animate foci.
201
(30)
for tis shame of the fault and the disgrace that attends it that they
should stand in feare of, rather then paine, if you would have them
a temper truely ingenuous.
(1685 LOCKE-E3-H,57.206)
(31)
that it was not the pure knowledg of nature and vniuersality

which gave the occasion to the fall; but it was the proude knowledge of good and euillwhich was the fourme of the temptation;
(1605 BACON-E2-P1,1,4R.26)
Contrastive focusing adverbs also serve to position the focus in relation to

excluded alternatives. In (32), the adverb alone emphasizes the uniqueness
of the abstract noun use. Likewise, in (33), the adverb only contrasts the
abstract opportunity (or alone time) with other, more concrete gifts.
(32)
Tis use alone hardens it and makes it more able to endure the cold
(1685 LOCKE-E3-P1,34.34)
(33)
Tis opportunity, tis a lone-hour only, that can make me happy.

(1688 BEHN-E3-P1,167.206)
However, after the Early Modern period, it-clefts containing postcopular

abstract nouns are less associated with listing and focusing adverbs. In the
PPCMBE, 27 out of 48 NP-focus it-clefts with abstract foci (56%) involve
listing (21 tokens) and/or focusing adverbs (7 tokens). What is more, this
association grows progressively weaker over time. At MBEI, 71% of itclefts with abstract nouns involve listing or focusing adverbs (10 out of 14
tokens); this drops to 54% at MBEII (7 out of 13 tokens) and to just under
48% at MBEIII (10 out of 21 tokens). This suggests that abstract nouns,
which do not denote discrete physical objects, have gradually become more
acceptable it-cleft foci and do not need to be quite so often coerced into the
referential focus position via individualizing techniques, as shown in (34).
(34)
but it was love for you that made me guilty
(COLLIER-1835,27.1002)
In addition to focusing more abstract entities over time, the it-cleft also
occurs with a greater number of relational foci. Table 3 shows an increase
in the number and variety of it-cleft foci expressing the relations of time,
place, means, reason and manner.
202
Table 3. The changing frequencies of it-clefts with a range of adverbial foci

Time
OE
ME I
ME II
ME III
ME IV
EI
E II
E III
MBE I
MBE II
MBE III
1 (0.8)
14 (12.4)
8 (13.4)
10 (13.6)
8 (14.2)
Place
12 (10.6)
3 (5.0)
12 (16.3)
10 (17.8)
Means
Reason
Manner
Total
1 (1.9)
1 (1.9)
1 (1.0)
1 (1.0)
4 (3.5)
5 (8.4)
17 (23.0)
7 (12.4)
2 (1.7)
1 (0.8)
6 (5.3)
6 (10.0)
10 (13.6)
5 (8.9)
4 (3.5)
4 (6.7)
13 (17.6)
5 (8.9)
2 (1.7)
2 (1.5)
40 (35.4)
26 (43.5)
62 (84.1)
35 (62.2)
As we can see, it-clefts with relational foci become much more frequent in
the second half of the Early Modern period. This corresponds to a general
increase in both the number of specificational it-clefts and the number of itclefts with non-NP foci at this time (see Table 2). Before this point, the few
it-clefts with adverbial foci are limited to expressing means or reason. For
example, the Middle English PP-focus it-cleft in (35), repeated from (23)
above, expresses a causal relationship. As noted in Patten (2010: 230), this
same relation can also be expressed by it-clefts with NP foci in Middle
English, such as (36). Here, the focal NP corresponds to the subject of the
cleft clause, rather than to an adjunct.
(35)
Me trowe at by e prayers of is holy mayde it is at at place

was nevere it destroyed
(a1387 CMPOLYCH,IV,125.869)
I think that it was by the prayers of this holy maiden that that place
was never destroyed
(Patten 2010: 230)
(36)
It es pride in aim at hyes aim

It is pride that elevates them
(a1425 CMBENRUL,11.370)
(Patten 2010: 230)
Clear examples of it-clefts containing temporal and spatial foci do not

occur until the latter half of the Early Modern period (see also Patten 2010).
This development sees the appearance of it-clefts with AdvP foci; at EIII,
all 9 AdvP-focus it-clefts involve reference to time and place (see Table 2).
In addition to examples with AdvP and PP foci, such as (37), we also find
NP-focus it-clefts expressing temporal and spatial relationships in this
period, as shown in (38). Indeed, at EIII, 11 out of the 26 tokens referring
to time and place involve NP foci.
203
(37)
It was in 78; but I am not certain of the Day of the Month: It was
on a Saturday he came
(1685 OATES-E3-P2,4.87.257)
(38)
It was, as near as I can remember, the 3d of August, that he went

out of Town
(1685 OATES-E3-H,4,73.C1.42)
Manner adverbials first appear as it-cleft foci during the late seventeenth
century. These tokens are especially interesting, since, unlike references to
time and place, manner expressions do not have conventional boundaries. It
is perhaps unsurprising then that over 60% of the 21 tokens from EIII to
MBEII occur either with focusing adverbs, shown in (39), or intensifying
modification used to turn a gradable expression into a well-defined
superlative, as in (40).
(39)
In a Word, twas in this flattering Light only, though not perhaps

so thoroughly considerd, I lookd upon the Life of an Actor when
but eighteen Years of Age
(CIBBER-1740,53.226)
(40)
They were so strict with us, as to landing any Goods, that it was
with extream Difficulty that I got on Shore three Bales of English
Goods
(DEFOE-1719,197.88)
Throughout this period, all manner adverbials occur in PP-focus it-clefts.

At MBEIII, however, there is a further development, with 3 out of the 5
tokens involving adverb phrase foci. As we saw in Chapter 5, Section 3,
manner adverbs often need to be placed in a contrast set with other members of this category before they are deemed acceptable it-cleft foci. This is
also true of the historical data. In the early twentieth century example in
(41), the focus is defined by what it isnt, rather than what it is, and the
description is shaped and reinforced by listing (wantonly nor altogether
willfully).
(41)
It is not wantonly, nor altogether willfully, that man has so often

lost his God.
(TALBOT-1901,95.93)
In addition to occuring increasingly with non-NP adverbial foci, it-clefts

also appear with PP foci which correspond to arguments of the verb embedded inside the cleft clause. For example, in (42) to this exercise is specified as matching the description given in the gapped relative clause the
supposed training more especially applies ___. Such tokens appear in the
diachronic corpus at the end of the Early Modern period (with 1 instance)
204
and become more frequent in Late Modern English (with 10 instances,

averaging just over 5 tokens every 500,000 words).4 In some of these
tokens, the focal PP expresses a meaning which is (at least) reminiscent of
the relational concept location. For example, (42) can be paraphrased as
where the supposed training more especially applies is to this exercise.
(42)
Accordingly, it is to this exercise that the supposed training more

especially applies.
(BAIN-1878,368.152)
However, in other cases, it is difficult to detect any kind of relational meaning whatsoever. For example, in (45), the postcopular PP is used to refer to
an animate individual. This token seems to function as an alternative to the
NP-focus it-cleft it is you to whom I am indebted, in which the focus corresponds to the object of the preposition in the embedded clause. Once
again, this suggests an increasing overlap in the functions of nominal and
prepositional it-cleft foci.
(43)
Lady C. So sir, it is to you that I am indebted for all the confusion

that has taken place in my house this afternoon. To Frank.
(COLLIER-1835,20.700)
5.
Changes to the cleft clause
In addition to the changes to the clefted constituent, the it-cleft construction

undergoes a parallel development which has consequences for the information status of the cleft clause. This change sees the rise of the informativepresupposition (IP) it-cleft, which provides brand-new information in a presupposed clause (see Chapter 5, Section 4).
Originally, the cleft clause was associated with expressing discourse-old
information. For instance, all 7 of the specificational it-clefts in the YCOE
contain information which is given in the immediate discourse context. In
(44), God has told Adam that he must prove his obedience by forgoing that
thing which God forbids. The fact that Adam will have to forgo something
is therefore old information; what is at issue is the identity of this thing (as
the fruit of one tree).
4. However, Ball (1991: 107) suggests that this may actually be a Late Middle
English development.
(44)
205
Swylce God cwde to him, Nast u na t ic eom in Hlaford

and t u eart min eowa, buton u do t ic e hate, and
forgang t ic e forebode. Hwt my hit onne beon t u
forgan sceole.
(cocathom1,CHom_I,_1:181.78.73)
As if God had said to him, Thou knowest not that I am thy Lord,
and that thou art my servant, unless thou dost that which I command, and forgoest that which I forbid thee. But what may it be
that thou shalt forgo?
(Thorpe 1844: 15)
In the Middle English corpus, however, we find two examples where the
information in the cleft clause is not given by the previous discourse, but is
nonetheless shared knowledge, including (45).
(45)
Abid
a while, I prey e, and taak good kep ho it is at
Abide.IMP a while, I pray thee, and take good keep who it is that
lene hym so boldely to Cristes
brest and slep so sauerly
leans him so boldly to Christ.GEN breast and sleeps so surely
in his lappe.
(c1400 CMAELR3,45.586)
in his lap.
Stay a while, I pray you, and take good note of who it is that leans
so boldly against Christs chest and sleeps so confidently in his
lap.
(Patten 2010: 236237)
In this example, the proposition that someone leant against Christs chest
has not been mentioned previously. Nevertheless, the immediate discourse
context is about the part of the Bible that this event is from. The purpose
of this extract is to remind us of this scene (and so assumes that we know
it), even though this particular event is not present in the previous discourse (Patten 2010: 236). In this MEII token, then, the information in the
cleft clause is shared, but non-salient.
It is not until Late Middle English that we begin to find examples which
can properly be called informative-presupposition it-clefts. There are three
such tokens in the PPCME2, one at MEIII and two at MEIV. For instance,
the relative clause in (46) contains discourse-new information which is not
necessarily shared knowledge. However, the token is still presuppositional,
and so gives the impression of expressing a fact known to a third party. It is
therefore unsurprising that this example, along with the two other instances
from LME, belongs to the genre history.
206
(46)
It was he at graunted Kyng Herri e Secunde to go into Yrlond

and turne hem to e feith,
(a1464 CMCAPCHR,108.2367)
It was he that granted King Henry the Second to go into Ireland
and turn them to the faith.
The IP it-cleft continues into the Early Modern period, with three instances
at both EI and EII. For example, (47) provides a discourse-new description
of Christ in the cleft clause. Here, the author presents this information as a
previously established fact by referring to the third-party source, St. Paul.
(47)
he is truthe and lyfe, he is alone our onlye medyator and advocate,

sytynge at the ryghte hande of hys Father. Yt ys he, as S. Powle
saythe, that ys our onlye redempsion, salvasyon, justyffycasyon,
and reconsylyation.
(c1555 MOWNTAYNE-E1-P1,190.273)
Towards the end of the Early Modern period, at EIII, the frequency of
IP it-clefts increases to 8 instances. In addition to the usual tokens, where
the cleft clause expresses information which we take to be certifiable fact,
as in (48), we also find example (49), which has a performative function.
(48)
I now see it was not without cause, that our good Queen Elizabeth
did so often wish her self a Milkmaid all the month of May, because
they are not troubled with fears and cares, but sing sweetly all the
day, and sleep securely all the night
(1676 WALTON-E3-P1, 234.278)
(49)
Aman. Pray be so just then to me, to believe, tis with a World of

Innocency I woud enquire, Whether you think those Women we
call Women of Reputation, do really scape all other Men, as they
do those Shadows of em, the Beaux. (1696 VANBR-E3-P1,43.108)
In this token, Amanda asks a question (and so performs a speech act) in the
cleft clause. However, by choosing a cleft formulation, she backgrounds the
question and its delicate subject matter while emphasizing her own position
of innocence. This concern to distance herself from the speech act is further
evidenced by the use of the modal woud. The example represents the first
performative it-cleft in the diachronic corpus, a subtype which will be discussed in more detail in Chapter 8. The development of the performative itcleft results from changes both to the cleft clause and the clefted constituent, since such sentences focus on the manner in which the discourse-
207
new/hearer-new act is performed. By the end of the Early Modern period,

the IP it-cleft is already an established construction and manner expressions
have begun to appear in the focal position (see Section 4).
In addition to the IP it-cleft tokens, the end of the Early Modern era also
sees some creative uses where a discourse-old presupposition is altered or
in some way manipulated. For example, in (50), the information in the cleft
clause is partly old and partly new. Here, it is given that Dunne is baulked
and confused, and therefore that something baulks thy Understanding, but
the rest of the proposition, that something baulks thy Honesty, is new to the
discourse (Patten 2010: 237).
(50)
Dunne. My Lord, I am so baulked, I do not know what I say myself; tell me what you would have me say, for I am cluttered out of
my Senses. L.C.J. Why, prithee Man, theres no body baulks thee
but thy own self; thou art asked Questions that are as plain as any
thing in World can be: it is only thy own depraved naughty Heart
that baulks both thy Honesty and Understanding, if thou hast any;
(1685 LISLE-E3-P2,4.118.362)
In this example, the Lord Chief Justice (L.C.J) manipulates a discourse-old

presupposition to include his individual opinion that Dunne is deliberately
being dishonest. The use of cleft structure here, therefore allows the speakers opinion to be presented as uncontroversial fact. A slightly earlier and
much more subtle example was given as (31) above. In this early seventeenth century token, the cleft clause information is made more subjective
as it is reformulated in a way that captures the authors negative attitude
towards the event discussed (see Section 4).
In the PPCMBE, the frequency of IP it-clefts doubles, accounting for
nearly 19% of all specificational it-clefts, compared with just over 8% in
the PPCEME. Counting only clear cases where the information in the cleft
clause is entirely discourse-new (and not merely non-salient) and where the
information is not inferable either from the situational or discourse context,
16 tokens occur at MBEI, 18 at MBEII and 19 at MBEIII.5 Some of the
5. There are a number of tokens in the PPCMBE which have the appearance of IP
it-clefts, but which actually contain highly non-salient, discourse-old information. For example, we can see that in (i), the narrators separation from his
father as he heads to university represents discourse-old information. However,
prior mention occurs on a previous page, before a long interval in which the
narrator recounts his early relationship with his father. By the time we get to the
208
more interesting cases include (51), which contains inaccurate information

in the presupposed clause. Here Maxwell makes the assumption (based
upon the experts accepted opinion) that the female worker bees must be, in
some way, impregnated. However, we now know that the queen bee is the
only fertile female in the hive. Indeed, the example contains a further incorrect presupposition by referring to an entity that does not exist (the King
bee). Nevertheless, (51) behaves in every other way like an ordinary IP itcleft; that is, the cleft clause information is presented as new, yet uncontroversial fact.
(51)
7. The greatest Part of the Writers on this Subject, having been of

the Opinion, That the Sovereign was a Male, and that it is by
Virtue of this King, that the Working-Bees, who are Females, are
impregnate;
(MAXWELL-1747,11.32)
The following example from Dickens is also an IP it-cleft. Here, the

cleft clause expresses information which is both new and unexpected, given
Mr. Pickwicks implicit belief. However, the it-cleft backgrounds the main
point before Dickens goes on to reinforce it (touched they were) and further
explain it. Dickens chooses to instead focus on Mr. Pickwicks foolish and
ungainly attempts to defend himself. This narrative technique allows the
reader to be touched by the same events that his adversaries are, so that, by
the time we read the information in the cleft clause, it no longer depicts a
controversial reaction but almost a shared experience.
(52)
It might have been Mr Pickwick's very unexpected gallantry, or it

might have been the complicated manner in which he had got
himself out of bed, and fallen all in a mass upon the hornpipe man,
that touched his adversaries. Touched they were; for, instead of
then and there making an attempt to commit manslaughter, as Mr
it-cleft, this episode in the narrators life is non-salient. However, the token
functions very much like an IP it-cleft, in that it (re)introduces a new chapter in
the narrators life and signals the start of a new segment in the narrative. Delin
(1992: 292) is therefore correct that, in many cases at least, the information
within an it-cleft presupposition appears to remind rather than inform.
(i)
In setting out for the university, I was to part with my father and
preceptorand it was not without many tears shed on both sides that we
parted, when I mounted the chaise in which I set out for Oxford.
(GODWIN-1805,66.208)
209
Pickwick implicitly believed they would have done, they paused,

stared at each other a short time, and finally laughed outright.
(DICKENS-1837,553.219)
The early twentieth century token in (53) is especially interesting
because it occurs right at the very end of the diachronic corpus. Out of all
of the other instances, this token best illustrates the distinct characteristics
we attribute to the present-day IP it-cleft. The example contains a postcopular temporal expression, it expresses new information in the cleft clause and
it occurs in discourse-initial position (introducing a section on Japanese
Gardening). While all of these properties are shared with the happen-class
impersonal construction (see Section 2.1), example (53) clearly has a different information structure. Although the fact that commercial cultivators
gave serious attention to the Japanese flora may be new to the reader, it is
presented as a commonplace occurrence. What is marked as surprising is
how long it took for this event to take place, not until some fifty years ago.
(53)
Although the introduction of the beautiful Japanese plants that now

contribute to the charm of British gardens belongs to the distant
past, it was not until some fifty years ago that commercial cultivators gave serious attention to the Japanese flora, with a view to
obtain some other of its members for the further enrichment of our
gardens.
(WEATHERS-1913,1,13.288)
There are also three further tokens of the early performative it-cleft in
the PPCMBE, including (54). In this example, Wetherell uses an it-cleft to
mitigate the face-threatening act of questioning the accuracy of the Lords
note. The construction highlights Wetherells deference to this high-status
individual and downplays the accusation that information is being withheld.
Performative it-clefts therefore function as a politeness strategy, enabling
the speaker to distance themselves from their act (marked as presupposed
or not at issue) by focusing on their emotional response, either to the act
itself or to the addressee.
(54)
Mr. Wetherell. It is with great deference I should question the

accuracy of your Lordships note, but I apprehend it will be found
to be admitted there, that you may enquire through what Officer of
Government the communication was made.
(WATSON-1817,1,160.2033)
210
In addition, the PPCMBE contains several more examples in which an

otherwise discourse-old presupposition contains elements of the speakers
opinion. Not all of these tokens are quite as playful or creative as the Early
Modern example in (50). Oftentimes, they involve the author placing their
own value judgements about the proposition inside the presupposed clause.
This encourages the reader to accept these opinions unchallenged. In (55),
for example, the author expresses his culturally-motivated opinion that for
women to work as manual labourers is a revolting proposition.
(55)
The fair sex are the farmers of the island; they condescend, or
submit, to dig the land, sow, harrow, and reap, and even patiently
thresh, and with handmills grind what their industry has obtained:
but it seems not to be brutality of disposition, but the imperious
necessities of situation, which impose such revolting toil upon the
weaker sex.
(TURNER1-1799,32.83)
However, towards the end of the nineteenth century, we come across yet
another new development. At this point, we find IP it-clefts which present a
statement of opinion in the cleft clause, rather than a statement of fact. For
example, in this it-cleft from Oscar Wildes An Ideal Husband, the information in the cleft clause is discourse-new. It is therefore not an example
involving the adaptation or manipulation of a discourse-old presupposition.
Nevertheless, this IP it-cleft is not based on factual information; the claim
that marriage is a hopeless, one-sided institution is the opinion of the
speaker.
(56)
Well, I will make her stand by her husband. That is the only thing
for her to do. That is the only thing for any woman to do. It is the
growth of the moral sense in women that makes marriage such a
hopeless, one-sided institution.
(WILDE-1895,75.992)
By placing this statement in a presupposed clause, it is expressed as a truism. This creates the sense that the speaker (and the writer, for that matter)
is an authority on the subject. The example is humorous, mainly because it
presents what is obviously a controversial claim as an incontestable fact.
The rhetorical device employed here is therefore assertion by presupposition.
Summary
6.
211
Summary
From this diachronic corpus investigation, we have seen that while the itclefts basic structure may have remained unaltered since Old English, the
construction has nevertheless undergone substantial changes in its function
and usage. In particular, the specificational it-cleft (unlike the predicational
it-cleft) undergoes a gradual and continual increase in frequency over time.
This is accompanied by changes to both the clefted constituent and the cleft
clause. As shown in Section 4, the it-cleft occurs with an increasingly wide
range of foci, which denote a more diverse array of semantic concepts. The
it-cleft originally showed a preference for focusing the most discrete of
entities, such as proper names and pronouns denoting animate individuals.
Over time, however, the construction begins to co-occur with less discrete
entities, such as abstract nouns (from Late Middle English onwards), before
taking a greater number and variety of relational and situational foci, which
express concepts such as time, place and manner (towards the end of Early
Modern English). This development has had consequences for the category
of it-cleft foci, initiating an increase in the number and variety of phrasal
categories that can enter into the focal slot.
Likewise, we saw in Section 5 that the it-cleft changes from expressing
only given information, to expressing shared but non-salient information,
before the inclusion of information which is factual or assumed to be factual in some sense. This development marks the introduction of the IP itcleft in Late Middle English. However, the data suggests that the change
continues further, as nonfactual opinion is slowly accommodated into the
cleft clause. For instance, towards the end of Early Modern English, creative uses appear in which elements of speaker opinion are integrated into an
otherwise discourse-old presupposition. This is followed, at the end of the
Modern English period, by examples where the entire cleft clause is comprised of opinion presented as fact.
This diachronic perspective therefore shows that the it-cleft has acquired
some of its more idiosyncratic properties over time, as a result of language
change. As was shown in Chapter 5, the present-day it-cleft can occur with
certain foci not found in related constructions and has a greater capacity for
expressing new information in the relative clause. These properties can now
be explained as the outcome of the it-clefts construction-specific historical
development. Nevertheless, two questions remain: how do we account for
these changes and why has only the it-cleft construction developed in this
particular way?
Chapter 8
The it-cleft and constructional change
1.
The two kinds of constructional change
In Chapter 6, we saw that the historical development of the English it-cleft

involves a process of fossilization; that is, the it-cleft schema is a relic from
an earlier time, retaining residual patterns which are no longer productive
outside of the cleft construction. On a usage-based account, fossilization is
the product of token frequency (that is, the number of times an individual
instance is activated). As the it-cleft becomes more frequent over time, the
schema is strengthened and reinforced. In other words, the cleft structure is
entrenched in the speakers grammar to such a degree that it is indifferent
to, and unaffected by, changes to the higher-order constructions from which
the it-cleft inherits its structural properties.1 Therefore, while the structured
inventory of the language has changed, such that the it-clefts superordinate
schemas have fallen out of use, their influence lives on in the lower-level
construction, which exhibits fossilized properties that no longer require the
support of an overarching taxonomy of constructions.
However, this is not the only way in which frequency affects the storage
and organization of grammatical knowledge. While token frequency results
in the entrenchment of the instance, type frequency (which involves the
activation of different types of instance) leads to the entrenchment of more
abstract schemas (see Chapter 2, Section 3). On this account, novel types of
instance are formed by extension from existing constructs. As these different types of instance coexist, speakers generalize over them to form a more
abstract (or schematic) and productive construction. This, in turn, may result in the reconfiguration of constructions higher up in the taxonomy, as the
system is reorganized in such a way that these novel instances are properly
integrated into the hierarchical network.
While fossilization results in the entrenchment of lower-level constructs,
this process of schematization therefore results in the entrenchment of (a
series of) higher-order constructions. As I explained in Chapter 2, this type
1. This is not to say that frequency is the only reason for the entrenchment of the
it-cleft construction. In Chapter 6, Section 7, I suggested that analogy and pattern-alignment may also have played an important part.
A grammatical constructionalization account
213
of change has been found to correspond to the traditionally item-based

change, grammaticalization. Assuming a constructional model of language,
in which atomic items and grammatical structures are treated as the same
fundamental object (as constructions), the cline from more lexical to more
grammatical is re-envisaged as a hierarchy from more substantive to more
schematic constructions. Consequently, the notion of becoming more
grammatical is understood as movement up the hierarchy, that is, as schematization, or grammatical constructionalization (see Trousdale 2008a).
In what follows, I show that the ongoing changes to the function and use
of the specificational it-cleft, outlined in Chapter 7, can be interpreted as an
example of grammatical constructionalization. The diachronic development
of the English it-cleft therefore involves both fossilization and schematization.
2.
On a grammatical constructionalization account, the changes to the range of

it-cleft foci and to the information status of the cleft clause originate in
language use, with speakers forming novel instances by extension from the
existing it-cleft schema. These changes proceed gradually, in incremental
stages, since they are governed by the degree to which the new instances
override the more general patterns of correspondence found in the original
construction; that is, by their degree of similarity to the it-cleft prototype
(see Chapter 2, Section 3). As I explained in Chapter 4, the it-cleft is a type
of specificational inversion construction. As such, it contains a focal referring expression and a discontinuous definite description (with an existential
presupposition). I suggest that the gradual trajectory of the changes to the
clefted constituent and the cleft clause is governed by the referential
semantics of the postcopular slot and the presuppositional nature of the
definite description.
For instance, we have seen that while the it-cleft originally focuses only
concrete objects (showing a preference for animate individuals), it occurs
with more abstract and relational foci over time, which denote less discrete
concepts. This is accompanied by an increase in the range of non-nominal
categories that can occur in postcopular position. As Croft (1991: 67) notes,
the semantic class of objects is the typological prototype of referring
constructions. Furthermore, since NPs typically denote objects, they are the
phrasal category most suited to performing a referring function. The change
to the range of it-cleft foci therefore involves extension from the prototype:
214
over time, semantic classes and syntactic categories which are less suited to
the referring function are accommodated into the referential slot.
This account is supported by the presence of coercion effects. Michaelis
(2003) observes that when there is a mismatch between the constructions
conventional meaning and that associated with any superimposed item, we
find an enriched interpretation, as the item conforms to the requirements of
the construction; she refers to this as coercion (see Chapter 2). For instance,
the it-cleft imposes referential characteristics onto non-nominal elements as
they are accommodated into the focal slot. As Bolinger (1972) observes,
prepositional phrases lose their relational properties when they occur as itcleft foci. Therefore, the meaning of (1) is not that the time is located in the
winter, but that the time is/equals in the winter. Such examples therefore
have little difference in meaning from those with NP-foci denoting discrete
periods of time, as in (2).
(1)
If it bee in the winter that your Hawke batheth, when no sunne

shineth, you may then drie her as well by the gentle aire of the fire
as otherwise.
(1615 MARKHAM-E2-P1,1,90.37)
If it is in the winter that your hawk bathes, when no sun shines,
you may then dry her as well by the gentle air of the fire as
otherwise
(2)
and if this be done in November, it will preserve the Trees for that
whole year, with that once doing, it being the winter time only that
they will feed upon the bark.
(1696 LANGF-E3-H,116.193)
In this way, non-nominal elements come to denote more discrete entities by

simple virtue of entering into the referential slot. As Davidse (2000: 1116)
notes, non-NP foci become like proper names; that is, they are cited in
their entirety and are given definite identification without being formally
marked as such.
Other times, less discrete concepts are individualized and made into
more acceptable it-cleft foci through listing or the use of focusing adverbs.
For example, we have seen that over 90% of abstract nouns functioning as
it-cleft foci in Early Modern English exhibit these overt coercion strategies
(see Chapter 7, Section 4). Towards the end of the Modern English period,
however, the figure drops to below 50%. This is indicative of a change to
the overarching it-cleft schema, as the occurrence of abstract nouns in focal
position becomes more conventionalized. These same coercion strategies
are required in present-day English as a way of enabling characteristically
215
predicative items, such as adjective phrases, to function as it-cleft foci, as in

(3) (see also Chapter 5, Section 3). This suggests that the it-cleft undergoes
not only gradual, but also continual, grammaticalization (or grammatical
constructionalization).
(3)
Its not just intransitive that the verb has to be

(Ball 1991: 479, citing Horn 1978)
The change to the information status of the cleft clause, meanwhile, is

governed by the presuppositional nature of the constructions discontinuous
definite NP. Over time, speakers are able to further manipulate what sorts
of information can be marked as presupposed, overriding a general pattern
of correspondence between presupposed and old (or familiar) information
(see Chapter 5, Section 4). The change therefore proceeds in incremental
stages in a direction outwards from the prototype (with instances deviating
increasingly from it). Once again, this development involves coercion. As
Lambrecht (1994: 71) explains, by using an expression which requires a
presupposition, such as an it-cleft, to present information that is not shared
knowledge, the expression itself forces that presupposition and the hearer
accommodates it as such. In this way, new information and speaker opinion
are marked as known, despite the fact that they are not known to the hearer.
This creates a coerced interpretation, on which they acquire the status of
general knowledge or established fact (see Prince 1978).
The change leads to the creation of a new construction: the informativepresupposition (IP) it-cleft. The instances of this sub-construction override
both general patterns of correspondence (between presupposed and familiar
information) as well as inheritance from the existing it-cleft schema. Since
these instances have a distinct property a unique discourse function which
is in addition to the specifying function of other it-clefts speakers abstract
over the novel instances to form a new IP it-cleft sub-construction. At this
point, the accommodation of new information into the presuppositional
clause becomes conventionalized. As Lambrecht (1994: 71) notes, the IP itcleft therefore represents an extension of the existing it-cleft schema via
conventionalized pragmatic accommodation.
The discrepancy between the it-cleft schema and the less prototypical IP
it-cleft is subsequently resolved as the overarching category changes to suit
its new membership. In other words, the speaker inductively generalizes
over both the instances sanctioned by the existing specificational it-cleft
and the instances which make up the new IP it-cleft. This is illustrated in
Figure 11, where dashed arrows show the direction of the extension and the
216
subsequent abstraction. As a consequence of this, the specificational it-cleft

becomes a more schematic, productive, and higher-order construction. The
dashed lines surrounding the lower-level specificational it-cleft in Figure 11
show that it is supplanted by the more basic schema.
Informative-presupposition it-cleft
Figure 11. The emergence of the IP it-cleft and the subsequent schematization of
the specificational it-cleft construction
The more abstract it-cleft schema stipulates only those characteristics that
are shared by all of its members. The upshot is that particular requirements
regarding the discourse status, saliency, or familiarity of the information
that is accepted into the cleft clause are now almost non-existent. As Borkin
(1984: 125) notes for present-day it-clefts, Presupposed, when used with
respect to cleft sentences, then, means non-asserted or assumed to be
true, and no more than that.2
Nevertheless, the changes to the cleft clause are best characterized as a
process of pragmatic enrichment; over time, the it-cleft has developed new
uses, which are in addition to its original specifying function. For instance,
the it-cleft can be used to remind the hearer or reader of relevant but nonsalient information, to inform them of facts they may not already know and,
most recently, as an indirect way of communicating the speakers opinion.
In this way, the historical development of the it-cleft involves a change in
2. Lehmann (2008: 3.1.2) observes that, To the extent that pragmatic accommodation of the proposition presupposed by the extrafocal clause is conventionalized in the cleft sentence, the construction becomes more grammaticalized.
For Lehmann, the outcome of grammaticalization here is the levelling out of
the contrast between the presupposition (in the cleft clause) and the assertion
(in the postcopular, focal position). However, it should be noted that it-clefts
are inherently presuppositional, since they contain discontinuous definite NPs.
While this property may be manipulated, with the result that new information is
marked as presupposed, it is not in any way reduced or lessened over time.
217
perspective from what the hearer or reader knows to how the speaker feels.
As Traugott (1982, 1989) observes, grammaticalization often involves a
shift towards increasingly subjective meanings; that is, the grammaticalized
word or string of words comes to express the speakers beliefs or attitudes.
The change to the range of it-cleft foci might also be viewed as a move
towards subjectification, as it involves a gradual progression from focusing
discrete concrete entities to more abstract and interpretative qualities.
These parallel developments are instrumental in the emergence of the
performative it-cleft at the very end of the Early Modern period (shown in
Chapter 7, Section 5). In these tokens, the speaker performs a speech act in
the cleft clause and specifies the manner in which this act is performed in
the postcopular, focal position. In doing so, the speech act is backgrounded,
or taken for granted, while the speakers emotional response to it, or to the
hearer, is highlighted. In the early examples, the performance of the speech
act threatens to characterize the speaker in a negative way (as improper,
impolite or self-important). In contrast, the focused phrase shows what the
speaker wants to convey about themselves, their attitudes and their beliefs.
The performative it-cleft is therefore a highly subjective, speaker-oriented
construction. As shown in Figure 12, the performative tokens form a subtype of IP it-cleft.
It-cleft schema
IP it-cleft
Performative it-cleft
Proverbial it-cleft
Figure 12. Incorporating IP it-clefts and performative it-clefts into a taxonomy of

it-cleft constructions
In addition, speakers may recognize yet more substantive constructions. For

example, the multiword string it is with great pleasure that is common in
ceremonial contexts. As shown in (4), this now formulaic device acts as an
anticipatory signal to the audience that the ritual has begun.
218
(4)
It is with great pleasure that I present to you this award.
In sum, it seems that while the it-clefts basic structure has remained the
same since Old English, the specificational it-cleft has undergone a process
of gradual expansion. In accordance with Himmelmanns (2004) definition
of grammaticalization, the diachronic development of the it-cleft involves
both host-class expansion, whereby the construction allows a wider range
of components to enter into it, and semantic-pragmatic context expansion,
through which the construction develops new pragmatic functions (see also
Patten 2010).3 The history of the it-cleft construction is therefore in some
ways contradictory, in that it is both resistant to change and susceptible to
grammaticalization (i.e. grammatical constructionalization). Nevertheless,
on a usage-based model, both aspects of the it-clefts development are
understood in the same way, as involving a process of conventionalization;
that is, the entrenchment of schemas. Both fossilization and schematization
conspire together to make the it-cleft a more idiosyncratic construction: the
specificational it-cleft resists the wider changes affecting the grammatical
system while pursuing its own construction-specific development.
3.
Some alternative explanations
Nevertheless, a variety of alternative explanations have been put forward to

account for the it-clefts rising frequency, the development of it-clefts with
non-NP foci and the origin of the informative-presupposition it-cleft. In
what follows, I discuss three such accounts, which depend on mergers,
borrowings and word order changes, respectively. For each, I show that
they are not as successful at accounting for the diachronic facts (outlined in
Chapter 7) as an explanation involving grammatical constructionalization.
3. On Himmelmanns (2004) account, grammaticalization applies to constructions

rather than to isolated lexical items. However, here, construction is interpreted
as the surrounding context which influences an atomic, grammaticalizing element. In this work, and in Patten (2010), I adapt Himmelmanns definition to
examples of constructional change which do not result in the grammaticalization of lexical items.
3.1.
219
An impersonal account
In her (1994a) paper, Ball claims that the Old English impersonal construction, introduced in Chapter 7, Section 2.1, plays a central role in the development of it-clefts with non-NP foci and it-clefts with new information in
the cleft clause. On her account, the it-impersonals underwent a partial
merger with the specificational it-cleft in Late Middle English, resulting in
a new and separate construction: the AdvP/PP-focus IP it-cleft. According
to Ball (1994a), these tokens differ from the original NP-focus it-clefts in
that they contain sentential complements, rather than relative clauses,
which express propositions that are new to the discourse. The AdvP/PPfocus IP it-cleft construction then merges with the existing NP-focus it-cleft
to form another new sentence type: the NP-focus IP it-cleft.
Balls explanation of how the it-cleft acquired prepositional and adverb
phrase foci is not unreasonable, since we have seen that the impersonal
construction shows a strong superficial similarity to the it-cleft (see Chapter
7, Section 2.1). However, there are several aspects of the diachronic corpus
data that cannot be accounted for on this analysis. For one thing, there is no
evidence to suggest that it-clefts with PP and AdvP foci form a sentencetype distinct from the NP-focus it-cleft. As we saw in Chapter 7, Section 4,
the first PP-focus it-cleft in the PPCME2 appears some time before the Late
Middle English period (and even before what Ball (1991: 158) claims is the
first specificational NP-focus hit-cleft in her corpus). In contrast, it-clefts
with AdvP foci appear much later in the constructions development; clear
examples are only found towards the end of the Early Modern era (see also
Patten 2010: 231).
Instead, what seems to unite these two varieties for Ball (1994a: 605) is
that, as it-clefts with non-nominal focal elements, they are less amenable to
a relative clause analysis; in these tokens, the cleft clause typically expresses a complete sentence (without a perceptible gap) and is rarely introduced by elements other than that. Ball concludes that they therefore have a
different syntactic structure from NP-focus it-clefts, and contain a sentential complement rather than a restrictive relative clause. However, in
Chapter 5, Section 3, I showed that it-clefts with non-nominal foci do not
warrant a separate analysis from those containing postcopular NPs; once
we assume an extraposition-from-NP analysis of it-clefts, on which the
antecedent of the relative clause is the initial it rather than the clefted constituent, the syntactic category of the postcopular XP is immaterial. On this
account, we do not need to look to the impersonals to provide an independent source for a distinct AdvP/PP-focus it-cleft, which explains how it has
220
acquired a sentential complement. Once we take this step, there is no longer

any motivation to look for a historical derivation which goes beyond the
extension of a single it-cleft construction.
A further limitation of Balls account is that it predicts that any changes
to the clefted constituent and to the information status of the cleft clause are
originally limited to the new AdvP/PP-focus IP it-cleft. However, instead,
the diachronic data shows evidence of parallel developments affecting NPand non-NP-focus it-clefts simultaneously. For instance, we saw in Chapter
7, Section 4, that when time and place adverbials are attested as it-cleft foci
in the second half of the Early Modern period, some 40% of the relevant
tokens contain focal NPs. Similarly, there is little evidence of an early, and
obligatory, connection between IP it-clefts and it-clefts with AdvP/PP foci,
or of the proposed time-delay before IP it-clefts begin to occur with NP
foci. For instance, the earliest PP-focus it-cleft in the PPCME2 contains a
discourse-old cleft clause (see Chapter 7, Section 4). Likewise, while the
first IP it-cleft in the corpus contains a focal PP, it is soon followed by two
further NP-focus it-clefts, also in the LME period (see Chapter 7, Section
5). Furthermore, it is telling that the majority of IP it-clefts identified in the
PPCEME contain nominal foci.
Finally, there are several characteristics of the diachronic data that are
not anticipated on this account. For example, the appearance of a new kind
of it-cleft with AdvP/PP foci does not explain why NP-focus it-clefts occur
with more abstract foci over time or why such examples show evidence of
coercion; nor does it predict that the it-cleft will go on to permit a greater
range of non-nominal foci. Thus, in order to speculate about the it-clefts
development beyond the Late Middle English period, Ball (1994a: 614) has
to invoke a number of additional pressures; she notes, From the LME
period onwards, the it-cleft construction has taken in a greater variety of
non-NP foci, possibly in response to the decline of some alternatives and
functional change in others (e.g. preposing).
Likewise, the influence of the impersonals on the origin of the IP it-cleft
does not explain why aspects of speaker opinion are eventually accommodated into the cleft clause; nor can it account for the development of a
performative function for the it-cleft in Modern English. These later developments can only be explained once we recognize the fundamental role of
presupposition in governing the changes to the cleft clause. Tellingly, this
is one property that is not shared with the impersonal construction.
In conclusion, an account which relies solely on characteristics acquired
from the impersonal construction to license changes to the clefted element
221
and the cleft clause is unable to account for the full extent of the it-clefts
gradual and continual development.4
3.2.
A Celtic account
More recently, Filppula (2009) has appealed to Celtic influence in order to

explain the overall rise in English it-clefting and the expanding range of itcleft foci over time.5 He begins with an examination of the specificational
copular constructions of Celtic languages. These constructions differ from
the English it-cleft in that they do not occur with an introductory pronoun,
the copula be is optional, and it is unclear whether the subordinate clause is
a relative construction (see Filppula 2009: 271). Even so, Filppula (2009:
281) claims that they are both structurally and functionally close enough
to their English counterparts to be transferable in a contact situation.
For example, there is some evidence to suggest that the Celtic copular
constructions have impacted on it-clefting in Celtic-influenced varieties of
English. Filppula (2009: 270) notes that the Celtic constructions exhibit a
greater syntactic freedom than English it-clefts, since even the verb can be
put into focus position, as shown in the Irish example in (5). Similarly, it
is well known that the it-clefts of Irish English occur with a wider range of
property-denoting foci than those of other varieties of English, as shown in
the early twentieth century token in (6). In addition, the prominent role of
clefting in Irish is also found in Irish English (Filppula 2009: 282).
(5)
(Is) ag leigheamh at s
Reading he is (lit. (it is) reading that he is)
(Gregor 1980: 148, cited in Filppula 2009: 270)
(6)
Is it dead he is or living? Its little you care if its dead or living I

am.
(1903 Synge, Shadow of the Glen [Visser 1963: 50])
4. See Patten (forthcoming) for a discussion of how Balls (1994a) approach to

the diachronic data reflects, and is influenced by, a different set of theoretical
assumptions from those adopted in the current work.
5. Filppula does not discuss whether Celtic influence could also account for the
changes to the information status of the cleft clause. As I explained in Chapter
7, Section 2.1, Filppula (2009: 272) analyses tokens which I take to be OE impersonals as belonging to the class of informative-presupposition clefts. For
Filppula then, the IP it-cleft is present in Old English; it is not the outcome of a
subsequent development in the history of the it-cleft construction.
222
According to Filppula (2009: 288), the frequent use of cleft constructions in

Celtic-influenced varieties of English underlines the susceptibility of this
area of grammar to contact influences and offers an additional piece of
indirect evidence to suggest that influences from Celtic may have affected
the rise of it-clefting in English, more generally.
Filppulas theories are not incompatible with the constructionalization
account of it-clefts outlined in Section 2. For instance, Filppula (2009: 277)
considers the relaxation of syntactic constraints on the clefted constituent,
along with the overall rise in the frequency of it-clefts, to be a clear indication of the gradual grammaticalisation of clefting. Filppulas concern then
is not to explain how the changes to the it-cleft progress, but to ask why
grammaticalization has occurred; in particular, to ask whether influence
from Celtic languages could have possibly motivated this change. However, while it is plausible that Celtic influence could have had an impact on
English it-clefting more generally (i.e. beyond the Celtic-influenced varieties), Filppula (2009: 288) admits that this scenario is hardly conclusive
because of the indirect nature of much of the evidence. Furthermore, the
influence of Celtic languages is not really required to motivate the changes
to the it-cleft construction, since, as we have seen, they conform to general
principles and conventional pathways of language change (see Section 2).
3.3.
A word order account
Other authors have suggested that the it-cleft has developed in response to a
loss of alternative word order arrangements in English. As Jespersen (1937:
86) notes, this construction may be considered one of the means by which
the disadvantages of having a comparatively rigid grammatical word-order
(SVO) can be obviated. In support of this, the typological evidence shows
that cleft constructions seem to be favoured in languages with a rigid word
order, such as French, English and the Celtic languages (see Filppula 2009:
281282).
An example of this approach is the recent work of Los (2009) and Los
and Komen (forthcoming), which examines the hypothesis that the development of the English it-cleft was shaped by the loss of verb-second (V2).
The V2 construction allowed a variety of constituents (not only subjects) to
occur before a finite verb in second position. This meant the loss of a multifunctional first position which was associated with unmarked topics but
could also contain focused material. Los (2009) and Los and Komen (forthcoming) explore the rise of English it-clefting as a resolution strategy as
223
performing these same functions while conforming to the rigid SV word

order.
Los (2009) approach is to explore the association between first position
in V2 and unmarked topics. She suggests that the loss of V2 has meant that
the pre-subject position (involved in preposing) has become pragmatically
marked as expressing more prominent information and that the it-cleft provides us with a useful strategy to avoid positioning adverbials in a position
that is too prominent. She considers the following examples which differ in
acceptability.
(7)
#With great pleasure, we can inform you that your application was
successful
(8)
It is with great pleasure that we can inform you that your application was successful
(examples from Los 2009: 114)
Los (2009: 114) claims that the it-cleft in (8) functions to place with great
pleasure in end-focus position to make it less marked (italic original). Of
course, this argument is to some extent paradoxical, since the it-cleft is a
focusing construction and is not by any means a neutral way of presenting
information. While Los (2009: 114) recognizes this, she finds that nothing
else appears to explain the awkwardness of the more literal translation.
Los and Komen (forthcoming) take a different approach, which explores
the secondary role of the first position in V2 as a host to focused material.
They suggest that the loss of V2 has meant that focus-marking of first position adverbials is restricted in present-day English; the it-cleft develops as
a resolution strategy, placing focused material in a position that conforms
to the rigid SV word order while maintaining the relative ordering of constituents. They consider the following examples which differ in acceptability.
(9)
(*Precisely) to avoid such a conflict of interest I am resigning
(10)
Its (precisely) to avoid such a conflict of interest that I am resigning

(examples adapted from Los and Komen, forthcoming)
As shown in (9), the focus-marking of the first position clausal adverbial to

avoid such a conflict of interest is infelicitous. The it-cleft in (10) allows us
to place this focused adverbial in complement position while maintaining
the linear order of (9), in which the clausal adverbial comes before the verb
resigning. As a focusing construction, the it-cleft in (10) differs from (9) in
224
that it does not require the presence of the focus marker precisely to mark
the adverbial clause as focal.
In support of this hypothesis, Los and Komen (forthcoming) find that
the relative frequency of it-clefts with focusing adverbs increases between
Old English and Early Modern English, before undergoing a decrease in
Late Modern English. They account for this trend by noting that Old English has the alternative strategy of placing focus markers in first position;
the recent decline in it-clefts with focusing adverbs is attributed to the
survival of V2 reflexes in present-day English. However, the diachronic
development of it-clefts with focusing adverbs acquires a different explanation on my account. In Chapter 7, Section 4, I suggested that focusing
adverbs are used to coerce abstract nouns and relational items into the
referential position as these foci begin to occur more frequently in Early
Modern English. During the Modern English period, the relative frequency
of it-clefts containing abstract nouns and relative elements in the focal position increases, but these are less associated with focusing adverbs. This
suggests that their use has become conventionalized.
This exploratory research into the impact of the loss of V2 on English itclefting is very interesting and I am sympathetic to the view that word order
considerations are a significant factor influencing the early development of
the it-cleft construction. However, a potential difficulty with this approach
is that it prioritizes the role of it-clefts as an information-packaging device.
On the assumption that the it-cleft is a fully-fledged specificational copular
construction, containing a discontinuous definite noun phrase with existential presuppositions and a postcopular focal referring expression, there are
more factors than word order considerations involved in the differing
acceptability and use of (7) and (8), and (9) and (10). Certainly, word order
changes are unlikely to be the only factor influencing the it-clefts development. Ball (1991: 518) suggests that rigidity of word order cannot account for the continual increase in it-clefting after LME, since at this time,
alternatives involving movement have long since disappeared. Likewise,
Los and Komens (forthcoming) analysis does not explain why it-clefts
become more frequent in Late Modern English, despite the decrease in itclefts with focusing adverbs.
3.4.
Interim summary
In conclusion, while it is possible that word order changes, influences from

Celtic languages and the loss of the it-impersonals could have contributed
A construction-specific development
225
to an increase in the frequency of English it-clefts at certain points in time,

none of these factors can satisfactorily account for the it-clefts gradual and
continual development, explaining both the changes to the range of it-cleft
foci and the information status of the cleft clause. As we saw in Section 2,
this is only possible with a grammatical constructionalization account, on
which the it-clefts construction-specific development is shown to follow a
conventional pathway of change.
4.
Why do it-clefts undergo a construction-specific development?
In Section 2, we saw that the historical development of the it-cleft involves

extension from the prototype. Over time, less discrete entities and relational
or situational information (often expressed by non-nominal items) are
accommodated into the referential slot. Likewise, nonfamiliar information
gains a coerced interpretation in the cleft clause, as a result of being marked
as presupposed. However, as I explained in Chapter 3, most specificational
copular constructions contain a referring expression and a definite description (with an existential presupposition). From this, we might expect that
the different kinds of specificational sentence would likely undergo this
same diachronic development. Nevertheless, in Chapter 5, we saw that the
exact range of it-cleft foci is construction-specific and other specificational
copular constructions are unable to occur with brand-new information in
the relative clause. In this section, I identify some subtle semantic and
pragmatic properties which may have enabled and shaped the it-clefts construction-specific historical development. I frame this discussion in a comparison of it-clefts and wh-clefts (a topic which has received considerable
interest in the cleft literature).
4.1.
Why do it-clefts develop a construction-specific range of foci?
Despite their close familial relationship, it-clefts and wh-clefts occur with a
different range of foci. For example, we saw in Chapter 5, Section 3, that
certain prepositional phrases can occur as the focus of an it-cleft which are
not permitted in the wh-cleft configuration, see examples (11) and (12).
(11)
It was as a Guardsman that he came to the Second Battalion

(S2A-011 124, BBC Radio 4 Trooping the Colour)
226
(12)
*How he came to the Second Battalion was as a Guardsman
However, in many other respects the wh-cleft construction is more relaxed

than the it-cleft about what can enter into the referential slot. For instance,
while wh-clefts commonly occur with VP foci and permit a full range of
clausal foci, shown in (13) and (15), it-clefts do not occur with verb phrases
and non-factive clauses in the focal position, shown in (14) and (16).
(13)
What hes done is spoil the whole thing
(14)
*Its spoil the whole thing that hes done

(examples from Leech and Svartvik 2002: 218)
(15)
What Im saying is that hes stupid
(16)
*Its that hes stupid that Im saying
Furthermore, adjective phrases also make for more acceptable wh-cleft foci,
as demonstrated by the differing acceptability of (17) and (18). As we have
seen, APs can only be accommodated into the focal position of an it-cleft
under specific circumstances, such as listing, shown in (19).
(17)
What he is is obsessive
(18)
*Its obsessive that he is
(19)
It isnt obsessive that Bill is, just manic-depressive

(Delahunty 1984: 76)
In what follows, I suggest that these distributional differences ultimately

result from a subtle semantic difference in the constructions introductory
elements; that is, the initial it of it-clefts and the wh-words of wh-clefts. As
we saw in Chapter 6, Section 5, the cleft it is a semantically underspecified
pronoun. This means that the determinative head of the it-clefts discontinuous NP can be given all manner of different interpretations. In contrast,
while the what of wh-clefts (or what-clefts) has a very general meaning, it
is nevertheless specified as near synonymous with the thing.
This distinction is made particularly apparent in predicational clefts. As
Ball (1991: 60) notes, predicational it-clefts often have a more restrictive
interpretation than predicational wh-clefts. For example, while the it-cleft
in (20) is understood to mean the dress that she wore was wonderful, the
what-cleft in (21) tells us that the thing she wore was a wonderful dress. In
227
this example then, the cleft it is interpreted as having a much more specific
meaning than the thing, gaining its semantic interpretation from the postcopular noun (see Chapter 5, Section 2). This shows that the cleft pronoun
is therefore underspecified, rather than truly general, preferring a specific
(or restrictive) interpretation.
(20)
It was a wonderful dress that she wore
(21)
What she wore was a wonderful dress
The semantic flexibility of the determinative it explains why the it-cleft

is such a useful (and relatively productive) construction. Not only can the
discontinuous NP be predicated of human and plural referents (see Chapter
6, Section 5), but it can also characterize entities which cannot be classified
by other nouns. For example, the it-cleft in (22) is more acceptable than the
corresponding wh-cleft and th-cleft given in (23) and (24). Here, the prepositional phrase with great determination does not quite fall within the
semantic scope of how or the adverbial noun way, which typically denote
instrument or means. As a result, in these examples, the referring expression does not match up with the head noun of the definite description.
(22)
It was with great determination that he climbed to the top
(23)
*How he climbed to the top was with great determination
(24)
*The way that he climbed to the top was with great determination
As Declerck (1994: 215) notes, the range of wh-words or noun phrases

that can function as pronouns (the place/thing/one/way) is too narrow for
every cleft to be matched by a pseudo-cleft. For Declerck, the it-cleft is
compatible with a greater variety of foci because the that-clause is headless
(see Declerck 1984a: 281). Here, I suggest that it is the semantically underspecified meaning of the cleft it which means that it can cover the same
ground as the adverbial nouns and much more (Bolinger 1972: 122).
A similar explanation could perhaps help to account for the acceptability
of it-clefts like (25), which contain PP foci referring to animate individuals.
Although these foci are in some sense directional, they cannot be captured
by the nominal relative where, as shown by the unacceptable wh-cleft in
(26) (see also Chapter 7, Section 4). Thus, the semantically underspecified
nature of it could provide an answer as to how the it-cleft has developed
this alternative to NP-focus it-clefts where the NP functions as the object of
228
a preposition, as in (27), and why these examples do not have corresponding th-clefts, shown in (28). On this account, it in (25) is not really comparable to the head noun person. Instead, such examples involve the extension
of a more relational concept. This would explain why (25) does not require
a preposition in the precopular (discontinuous) NP, much like (29).
(25)
It was to Ed that she was referring
(Huddleston 1984: 460)
(26)
*Where she was referring was to Ed
(27)
It was Ed that she was referring to
(28)
*The person that she was referring was to Ed
(29)
Where she left it was with Ed
(Huddleston 1984: 460)
However, we saw above that while the cleft pronoun it is semantically

underspecified, the interpretation it is given is usually specific. This means
that the discontinuous NP in it-clefts characterizes the postcopular XP uniquely. In other words, the description is not general enough to potentially
classify any other referent. This accounts for some of the restrictions on the
range of it-cleft foci and explains why listing is an effective coercion strategy for accommodating items which denote properties into the referential
slot. As Declerck (1984b: 144) notes, properties are not mutually exclusive: if X has property A, there is no reason why it should not have other
properties as well. For example, in (30) and (32) the descriptions the thing
that he was and the thing that he is do not exclusively apply to the adjective
phrase sick and the predicative noun secretary. However, by contrasting
sick with tired in (31), the property sick is established as a member of a
limited set of potential ailments. Likewise, in (33), a secretary is contrasted
with the profession that uniquely matches the more restrictive (or specific)
description the thing that Id wanted to be.
(30)
*Its sick that he was
(. Kiss 1998: 262)
(31)
Its not sick that he was but tired
(. Kiss 1998: 262)
(32)
*Its a secretary that I am
(33)
They made me a secretary, but it wasnt a secretary Id wanted to

be
(Ward, Birner, and Huddleston 2002: 1418)
229
Nevertheless, not all characteristically predicative elements can be individualized in this way. For example, providing a more specific description
and establishing a limited set of alternatives in (35) does not make the verb
phrase eat much more acceptable than in (34). This suggests that in order
for the semantic class of actions to fill the referential slot of the it-cleft,
they must become more noun-like, as with the gerund eating in (36). This
example is closer to the prototypical it-cleft instances, which contain
nominal foci denoting discrete entities (see Section 2).
(34)
*Its eat that she does
(35)
??Its eat that she likes to do best, not shop
(36)
Its eating that she likes best, not shopping
In addition to the restrictions on verb phrase foci, it-clefts cannot occur

with non-factive clauses in the postcopular position (see Delahunty 1984;
Delin 1989). For example, (37) contains the non-factive verb say. Here, the
proposition that he never replied to her letters is not assumed to be true
because it is attributed to the thoughts and beliefs of the person who said it
and could therefore involve conjecture or falsehood. The example in (38),
on the other hand, contains the factive verb regret. Here, the clausal complement is understood to express a true proposition. In other words, he
regretted the fact that he never replied to her letters.
(37)
*It was that he had never replied to her letters that he said
(38)
It was that he had never replied to her letters that he (most)

regretted
Again, the reason for this difference in acceptability seems to lie in the
fact that sentences such as (38) are closer to the it-cleft prototype. Here, the
factive clause refers to an actually occurring event. In such sentence, the
that-clause can be replaced by other noun-like elements, such as gerunds
(his insistence that he was in the right) and adjectival nominalizations (his
carelessness) (see Kiparsky and Kiparsky 1970). However, these nominal
constructions cannot replace that-clause complements of non-factive verbs.
It makes sense then, that when presented with that-clause foci, the hearer
interprets them as referring to existentially presupposed events, rather than
to some abstract concept such as the words somebody said. In example
230
(37) then, the non-factive verb say conflicts with our prior assumption that
the proposition expressed in the postcopular clause is true.
However, while it-clefts were originally restricted to NP foci denoting
the most discrete of entities, the early wh-clefts do not share this same characteristic. According to Traugott (2008), the specificational wh-cleft did not
emerge until the late seventeenth century, at a time when other specificational copular constructions were already well established. From her data, it
appears that the early what-cleft never had a particular association with NP
foci, occurring frequently with a range of clausal foci (including both factive and non-factive clauses) and verb phrase foci (such as to-infinitives).6
From the very beginning then, the wh-cleft could occur with categories
of foci which are not acceptable in the it-cleft. For example, the wh-cleft in
(39), containing a focal non-factive clause, is perfectly grammatical.
(39)
What he said was that he had never replied to her letters
There are two possible reasons for this. First, since wh-clefts were never
really associated with nominal foci, the that-clause will not necessarily be
interpreted as a noun-like concept, such as an actually occurring event. Secondly, wh-clefts have a different linear order to it-clefts, which contain extraposed relative clauses. In example (39) then, the non-factive verb say is
given prior to its clausal complement, marking it as non-presupposed. As a
result, the hearer cannot misinterpret the focal clause as factive. However,
the linear order of the description and the referent which matches this description cannot be the only reason for the grammaticality of wh-clefts with
non-factive foci. Otherwise, we could not explain why reverse wh-clefts,
6. Traugott (2008) suggests that wh-clefts with infinitival foci may have developed from non-specificational tokens. Originally, examples such as (i) had only
a purposive meaning (the thing I do is in order to please you). With the emergence of the specificational wh-cleft, such examples could be reanalyzed. On
this new reading, the to-infinitive is identified as the thing that I do rather than
my purpose in doing it. The independence of this new meaning from the original purposive construction is accomplished by the fact that to finally becomes
an optional element, allowing bare infinitive foci, shown in (ii). See Mair and
Winkle (2012) on the change from to-infinitive to bare infinitive in specificational pseudoclefts.
(i)
What I do is to please you
(ii) What I do is please you
231
such as (40), are deemed more acceptable than their corresponding it-clefts
(see also Delin 1989: 97).
(40)
?That he had never replied to her letters was what he said
The reason why wh-clefts are open to a greater variety of non-nominal

categories than it-clefts seems to lie in the constructions introductory
element. As noted above, while the cleft it is semantically underspecified,
the what of what-clefts is specified as near synonymous with the thing. In
the wh-cleft construction, this very general concept covers not just objects,
but also actions and properties. As shown above, these sorts of entities are
not mutually exclusive and, as a result, NPs headed by the semantically
general what do not always provide a description which is truly unique to
the postcopular referent. For example, in (41), (42) and (43) below, there is
not only one possible thing that John is or Sarah was or that she can do.
Nevertheless, these action/property-denoting foci can enter into the wh-cleft
construction without requiring the use of coercion strategies.
(41)
What John is is stupid
(42)
What Sarah was was an idiot
(43)
What she can do is eat
We have seen then that differences in the range of foci found in it-clefts
and wh-clefts can be traced to a subtle difference in the semantics of their
introductory elements. That the cleft it is semantically underspecified helps
to explain why it-clefts can occur with prepositional phrases that cannot be
classified by other nouns; that the cleft it prefers a specific interpretation
means that prototypically predicative elements (which are not mutually
exclusive) will not make for acceptable it-cleft foci without the use of coercion strategies. In contrast, the what of wh-clefts is semantically general but
is nevertheless specified as near synonymous with the thing. This provides
a reason as to why the wh-cleft is open to a wider range of syntactic categories, which denote all manner of things, including objects, properties and
actions. In turn, this shows why wh-clefts are historically less associated
with NP foci and why only it-clefts exhibit a clear preference for focusing
noun-like elements.
232
4.2.
Why do it-clefts develop construction-specific discourse functions?
It-clefts and wh-clefts also display different discourse functions. As we saw

in Chapter 7, the it-cleft has developed an informative-presupposition subtype, in which brand-new information is expressed in the presuppositional
cleft clause. This discourse-function is specific to the it-cleft construction;
it is not shared by any other kind of specificational copular sentence (see
Chapter 5, Section 4). For instance, corpus studies of wh-clefts have not
found any evidence to suggest that the construction can occur with brandnew information in the relative clause (see for example Prince 1978; Delin
1989; Collins 1991a). Based upon such evidence, Delin (1989: 180) concludes that wh-clefts cannot carry presuppositions that appear at the NEW
end of the range (emphasis original).7 Ward and Birner (2004: 157)
demonstrate this with the following examples. In (44), the description what
he does is salient because of the earlier mention of Triggs profession.
However, when the wh-clause is non-salient, this same wh-cleft becomes
infelicitous. For instance, in (45), the mere sighting of a friend does not
give rise to the issue of what he does.
(44)
Hey look! Thats my friend Jeremy Triggs over there. Hes a

lexicographer. What he does is find alert readers who recognize
new words or new usages for ordinary ones.
(45)
Hey look! Thats my friend Jeremy Triggs over there. #What he

does is find alert readers who recognize new words or new usages
for ordinary ones.
(examples from Ward and Birner 2004: 157)
Thus, the IP it-cleft is the result of a construction-specific development;

there is no corresponding variety of wh-cleft. Furthermore, while the it-cleft
has subsequently developed a specialized performative function, the whcleft has acquired a construction-specific use as a presentational device (see
7. Reverse wh-clefts exhibit a similar restriction. For instance, Oberlander and
Delin (1996: 200) find that reverse wh-clefts are frequently used for re-evoking information, rather than communicating entire new propositions (see also
M. Johansson 2001). Likewise, Hedberg and Fadden (2007) find that reverse
wh-clefts are unlike IP it-clefts in that the focal slot and the relative clause
cannot both contain hearer-new information. It is only when the subject is
discourse-old that new information can occur in a postcopular fused relative.
As shown in Chapter 3, such sentences are analysed here as predicational rather
than specificational constructs.
233
below). In what follows, I suggest that these discourse-functional differences stem from a subtle distinction in the constructions prototypical
information structures. In particular, while the relative clause of the it-cleft
was originally associated with expressing discourse-old information (see
Chapter 7, Section 5), wh-clefts have always been associated with inferable
information (see Traugott 2008).
The corpus studies of Prince (1978) and Collins (1991a) suggest that
wh-clefts are commonly found with inferable information in the relative
clause. As Prince (1978) notes, information is inferable if it is appropriate
to the speech situation and can therefore be assumed to be already in the
hearers consciousness (see also Prince 1981). Wh-clefts therefore differ
from it-clefts in that they express information that is not necessarily already
known, but which is understood to be relevant to the discourse context and
which is therefore assumed to be salient for the hearer. Prince (1978: 894)
concludes that while it-clefts are associated with known, old, or factual
information (and so are less relevant to speaker-hearer interaction), It is
Chafes [1974, 1976] notion of givenness THAT WHICH THE SPEAKER
ASSUMES TO BE IN THE HEARERS CONSCIOUSNESS that is pertinent to whclefts (emphasis original).8 Prince (1978: 896) demonstrates this distinction by considering the following examples, which differ in acceptability;
she notes that in the attested token given in (46), it is known or knowable
8. In contrast, Gundel (1985: 98) suggests that it is the it-cleft and not the whcleft whose relative clause must contain material that is already in the addressees consciousness at the time of utterance. Gundel claims that the it-clefts
sentence-final topic clause refers an entity which is not only familiar but
activated. In other words, it is the focus of attention. On this account, we would
expect that the wh-cleft can occur at the very beginning of a discourse when
the addressees attention can generally not be expected to be focused on the
topic, but that the corresponding it-cleft will not be appropriate in such a
context (Gundel 1985: 97). However, as Collins (1991a: 105) notes, it is only
IP it-clefts that can occur in initial position in such contexts; wh-clefts are only
found discourse-initially when the information in the relative clause is inferable
from the speech situation. For instance, in Gundels (1985: 98) example below,
the fact that the lecture is on some topic is inferable from our existing knowledge of how lectures work and is therefore present in the hearers consciousness.
(iii) (At the beginning of a lecture)
What I would like to talk about today is conversational implicature.
(Gundel 1985: 98)
234
from the fact that the speaker sews books, that something is rotten or
amiss with them. However, the theme of this discourse, i.e. what it is about,
is books, not what is rotten; thus the it-cleft is ideally suited to the task.
(46)
So I learned to sew books. Theyre really good books Its just

the covers that are rotten.
(Bookbinder in Terkel 1974 [Prince 1978: 896])
(47)
So I learned to sew books. Theyre really good books. Whats

rotten is just the covers.
(Prince 1978: 896)
The early association between it-clefts and discourse-old (rather than

inferable) information provides an explanation as to why this construction
has developed an informative-presupposition subtype. As we have seen, the
it-cleft differs from the wh-cleft in that the relative clause expresses old or
known information, which is not so dependent upon speaker-hearer interaction. Prince (1978: 898) notes, The fact that it-clefts may present information as known without making any claims that the hearer is thinking about
it (or, in fact, even knows it) presents the speaker with a strong rhetorical
temptation: what is to prevent him/her from putting new information into
the that-clause?. In IP it-clefts then, hearer-new information is marked as
known albeit known to a third party. Such sentences therefore have the
function of expressing this information as uncontroversial fact, signalling
for the hearer to accept it unchallenged (see Delin 1992). Thus, the development of the IP it-cleft depends upon, and ultimately reinforces, the association between it-clefts and known/factual information.
Since IP it-clefts are associated with expressing previously established
facts, they enable the speaker to distance themselves from the information
expressed in the cleft clause. As Prince (1978: 900) comments, IP it-clefts
are used when the speaker does not wish to take personal responsibility for
the truth or originality of the statement being made. This explains how the
it-cleft configuration came to be used to express elements of speaker
opinion in the cleft clause and why the it-cleft developed a performative
function, which is unique among the family of specificational copular constructions (shown in Chapter 7, Section 5). As we have seen, performative
it-clefts mitigate the force of a speech act by marking it as presupposed.
Instead, the speakers emotional response to the act becomes the focus of
the sentence. Thus, performative it-clefts function much like announcements embedded under factive verbs of emotion, such as regret in (48). As
Abbott (2000: 1432) notes, here the speaker removes herself from respon-
235
sibility for the content of the announcement, and goes on record instead as
expressing an appropriate emotion.
(48)
We regret to inform you that your insurance policy is hereby

cancelled.
(Abbot 2000: 1430)
The it-clefts construction-specific development is therefore shaped by

its early association with known, or factual, information. Interestingly, this
association seems to permeate the entire it-cleft configuration not just the
information status of the cleft clause. For instance, Bolinger (1970: 69,
1977: 71) finds that, in it-clefts, the specificational relationship between the
clefted constituent and the cleft clause must have a prior basis. He discusses
the following discourses, which differ in acceptability.
(49)
(50)
A: When will we know?

B: Its tomorrow that well know
(Bolinger 1970: 69)
A: When will you tell me?

B: #Its tomorrow that Ill tell you
(Bolinger 1970: 69)
The discourse in (49) would be entirely appropriate if speakers A and B are

waiting for exam results which are to be made available on tomorrows
date. As a result, in this example, the time of knowing has been previously
established. In (50), on the other hand, the time of telling has no prior basis
and is decided by the speaker there and then. Consequently, an it-cleft is
not the best choice for expressing this information. Additional evidence for
Bolingers claim can be found in the corpus data from the ICE-GB, which
suggests that it-clefts collocate with words and phrases which demonstrate
that the specificational relationship is grounded in fact. More than 5% of all
it-clefts (24/404) occur with the words fact, therefore and thus in the same
text unit, as in (51). Many of these examples play a summative role, occurring at the end of a discourse segment, including (52) and (53).
(51)
It was in fact the physiotherapist who confirmed the suspicion of

the doctor that it was a cartilage problem.
(W1B-020 050, business letter to insurance company)
(52)
It is thus the slacks that have soils most resembling those of true
terrestrial situations, and the richest plant and animal communities. 5.2.3 The relationships of dune sands to terrestrial soils.
(W2A-022 079, academic writing)
236
(53)
I began this thesis with images of the positive, generous sexuality

of the ancient Turkish deities and I conclude, with the inescapable
fact that, bar the Tantric tradition, it is the sexuality of the
Goddess, and consequently real women, that has suffered most in
the process of transition that the Aryan heroes brought into the
worlds symbolic art forms so variously enshrined in each religious
tradition.
(W1A-008 011, student essay, MA dissertation)
In contrast, wh-clefts are not so associated with the expression of factual

information. As we saw in Section 4.1, wh-clefts commonly occur with foci
denoting actions and properties entities which are not mutually exclusive.
Therefore, rather than providing a factually correct referent which uniquely
matches a previously established description, the wh-cleft exhibits a certain
flexibility, allowing the speaker to choose between different, but equally
correct alternatives. For Herriman (2003: 3), The wh-cleft takes up a position which acknowledges the existence of alternatives. At the same time,
however, it closes down the discussion by expressing the authorial position
in preference to all other alternative positions (emphasis added). Consequently, the wh-cleft construction actually serves to emphasize the speakers role in the specificational process. For example in (41) and (42) above,
repeated here as (54) and (55), the speaker selects the characteristic which
they believe appropriately sums up the individual in question. In doing this,
the speaker expresses their own opinions about that individual. Likewise, in
(43), repeated here as (56), the speaker selects an activity which she does
well. By highlighting one particular activity over many possible others, the
speaker is able to make a comment about the individual in question, for
example that she is greedy.
(54)
What John is is stupid
(55)
What Sarah was was an idiot
(56)
What she can do is eat
The often-noted association between wh-clefts and inferable information

also provides opportunities for the speaker to actively construct, rather than
simply report, a specificational relationship. For instance, we have seen that
while some wh-clefts have antecedents in the prior discourse, oftentimes
the information in the relative clause is inferable from the metalinguistic
context. This relies upon the hearers understanding of the norms of the
speech situation, including the role of the speaker. For instance, we can
237
cooperatively assume that the fact that the speaker intends to express their
meaning and that they have relevant thoughts and opinions will be in the
hearers consciousness (see Prince 1978: 890891). Thus we find wh-clefts
introduced by expressions such as what I mean is, what I mean to say is,
what worries me is and what Ive noticed is, to name but a few. Sentences
like these, including (57), provide the speaker with a forum to express their
views through the construction of a specificational relationship which has
neither a prior basis nor a grounding in fact. As Prince (1978: 891) comments, such WH-clefts are often used not simply for clarifying previous
assertions, but also for remaking them.
(57)
What I mean is we need something to answer somebody.

(The Presidential Transcripts 1974 [Prince 1978: 890])
Thus, in many ways, the wh-cleft is a more subjective, speaker-oriented

construction than the it-cleft. While wh-clefts highlight the speakers role in
the specificational process, it-clefts reduce the speakers responsibility for
the information expressed by treating it as a pre-established fact (although
this is also what allows the speaker to manipulate this construction as a way
of expressing their opinions). This explains why the wh-cleft construction
has never developed informative-presupposition or performative subtypes.
Instead, the wh-cleft has undergone a construction-specific development,
acquiring a presentational function.
As shown in (57) above, wh-clefts with antecedents in the metalinguistic
context often express very general information in the precopular clause. In
presentational wh-clefts, the initial clause is so uninformative that it simply
functions as a presentational device which introduces a postcopular proposition. For example, in (58), the postcopular constituent expresses the main
informational content of the sentence; the fused relative and the copular
verb function as a unit, separated from the rest of the sentence by a comma.
(58)
What Im saying is, you shouldnt let her boss you about
The clausal focus is presented here as a complete sentence; in other words,

it does not have to be introduced by that, unlike the clausal foci of it-clefts.9
9. Such examples provide support for the claim made by den Dikken, Meinunger,
and Wilder (2000) and Schlenker (2003) that NPI connectivity in wh-clefts can
be accounted for by invoking ellipsis. The wh-cleft differs from other specificational constructions in that it licenses negative polarity items in the focal posi-
238
Bolinger (1977: 11) suggests that the word that is inherently anaphoric. The
lack of that in (58) therefore indicates that the postcopular clause expresses
a brand-new assertion.
In these presentational wh-clefts, the proposition expressed in the focal
clause is thus more informative than both the information in the precopular
clause and also the relationship of specification that exists between them.
The development of this discourse function seems to be related to a gradual
change in the information status of the precopular fused relative. According
to Koops and Hilperts (2009) diachronic study, the wh-cleft has come to
be introduced by increasingly general descriptions over time. For example,
while the highly general verb do was often found in the precopular phrase
of early wh-clefts (see also Traugott 2008), the construction now commonly
occurs with the even less specific verbs happen and be, see (59). The verb
happen is first attested in the wh-cleft in the late nineteenth century, and is
used frequently by the middle of the twentieth century. Most recently, the
verb be has become especially common in wh-clefts in present-day spoken
English.
(59)
I think what happened was, I just launched in.

(Nick Hornby, How to be Good, p.5)
Koops and Hilpert (2009: 235236) note that while the verb do requires
that the agent of the dynamic event is specified (what he did was), the
verb happen does not (what happened was); in turn, the verb be is maximally general since it can encompass both dynamic and stative events
(what it was was). As Prince (1978: 893) notes, such expressions are also
highly inferable, since there seems to be a pragmatic principle that says
that events keep occurring and that in our culture, at least, they are our
proper and constant concern.
tion, shown in (iv). If we assume that these sentences are in some way related
to wh-clefts with sentential foci, such as (v), then NPI connectivity in wh-clefts
can be explained. However, since tokens like (v) form a separate presentational
subtype of wh-cleft, an ellipsis approach is unlikely to be valid analysis for all
instances of the construction, contra Schlenker (2003). See Chapter 3, Section
4.3.2, for other accounts.
(iv) What he didnt buy was any wine
(v)
What he didnt buy was, he didnt buy any wine
239
This usage is said to be most especially common in spoken language.

Weinert and Millers (1996) study examines the use cleft constructions in
spontaneous, informal discourse, using dialogue data from the Map Task
experiments (see Anderson et al. 1991). They find that all 20 of the wh-cleft
tokens contain the low content verbs do or happen.10 Weinert and Miller
pay close attention to the suprasegmental features of these tokens, which
lead them to question whether any of their examples are classic clefts;
that is, clefts which do not have a pause separating the cleft clause + copula
from the clefted constituent. Instead, they suggest that the clefted constituent has a potentially independent status in all of the wh-cleft tokens, including (60).11 They note, Even without the cleft clause, the clause or verb
phrase carries all the necessary information and can be interpretedThis,
we suggest, is what allows the fast tempo and phonetic reductions on the
cleft clause (Weinert and Miller 1996: 197).
(60)
Cause what youre doing is youre going up the side of the

allotments
(Weinert and Miller 1996: 194)
Weinert and Miller (1996: 197) suggest that these wh-clefts have a macrodiscourse function: they all serve as the climax to a period of negotiation
between the instruction-giver and the instruction-follower, through which
the instruction-giver expresses their conclusions. However, wh-clefts are
also forwards pointing in that they function as a bridge to the next section
of discourse, in this case, leading into the next subset of instructions.
Hopper (2001) also finds that presentational wh-clefts play a significant
role in conversational turn-taking. He suggests that the precopular clause
delays the delivery of the main assertion, by adumbrating (foreshadowing)
the continuation in general terms without giving away the main point
(Hopper 2001: 114). This has the dual purpose of informing the reader that
10. Similarly, Hopper (2004: 156) finds that the pattern what (NP) {do, happen,
say} {is was} accounts for almost 90% of all occurrences in his corpus; he
concludes that the pseudocleft could almost be called formulaic.
11. This conclusion is supported by the use of punctuation in this striking token
taken from a college students written work. Here, a full stop separates the cleft
clause + copula from the clefted constituent; the latter is presented as a completely independent sentence.
(vi) The navy box like object he had placed in front of me caught my eye. I
was baffled by it. What I didnt realise was. I would later learn this would
be the most important item to save someones life.
240
what follows is worthy of attention as well as buying time for the speaker
to formulate the postcopular assertion, therefore enabling them to hold the
floor. In wh-clefts then, the specificational relationship between the precopular description and the postcopular referent is often completed online, decided by the speaker during the course of the utterance. This often
gives the wh-cleft a fragmentary appearance (Hopper 2004: 156), shown
in (61).12
(61)
Okay, Right, Oh. Right what we need to do right is like eh...

(Weinert and Miller 1996: 194)
Indeed, Hopper and Thompson (2008) suggest that many apparent cases of
incomplete wh-clefts are instead instances of a distinct construction, in
which a wh-clause is used to project the upcoming stretch of discourse. In
other words, it tells us that there is more to come. They discuss example
(62), in which the wh-phrase is not followed by the copular verb, but by an
independent sentence functioning as a directive, cook all the fish. Although
there is still a specificational meaning here, in that the initial phrase classifies the next sentence as the thing she should do, Hopper and Thomson
(2008: 108) suggest that the introductory expression serves mainly to
frame the following talk.
(62)
what you oughta do though Mar,cook all the fish. cause we

wont use it,if you dont cook it.
(Hopper and Thompson 2008: 108)
Over time then, it seems that the wh-cleft has developed a new discourse
function in which the act of specification is so general that it functions only
as a simple presentational device. Indeed, for some speakers it may be that
this subtype of wh-cleft no longer has a specificational meaning at all. This
would explain the existence of the double is construction, exemplified by
(63), which contains two instances of the matrix copula.
(63)
What it is is, is that I just cant see the point of doing it
12. See also Schmid (2001: 1536) on the use of definite NPs containing abstract
head nouns, such as the thing is, as a useful hesitation device. As Tuggy
(1996: 725) comments, The thing is is such a strongly entrenched automatic
unit, that it is easy for speakers to bring it out from their minds as a unit, even
before they have figured out what they want to say next.
241
In such sentences, the initial description (in this case, what it is) functions
as a unit with the first matrix copula be (see Brenier and Michaelis 2005).
Since this unit of information functions only as a formulaic presentational
device, a further copula is provided after the intonation break to reinforce
the specifying relationship between the postcopular clause and the initial
description. This construction has a further turn-taking function, providing
even more down time for the speaker to formulate their main assertion.13
We have seen then that the development of different discourse functions
for the it-cleft and wh-cleft constructions is related to their association with
known (or factual) information and with Chafe-given (or hearer-activated)
information, respectively. This, in turn, can be traced back to an early link
between it-clefts and discourse-old information and wh-clefts and inferable
information. These subtle differences in information-structure explain why
only the it-cleft has developed an informative-presupposition subtype and
why the wh-cleft has instead acquired a use as a presentational device. Both
of these developments involve subjectification. We have seen that while the
it-cleft can be employed as an indirect way of communicating the speakers
opinion in the cleft clause (under the guise of uncontroversial fact), the whcleft configuration can be used as a way of introducing the speakers views
in the focal phrase (which is also dependent on the fact that wh-clefts allow
a full complement of clausal and verb phrase foci).
4.3.
Summary
The construction-specific development of the it-cleft therefore seems to be

influenced and shaped by two, relatively minor, semantic and pragmatic
properties: the underspecified semantics of the cleft pronoun it and an early
13. Tuggy (1996) provides a cognitive account of what he calls the 2-B construction, outlining a family of constructs which are more or less entrenched, including the thing is is that CLAUSE and the problem is is that CLAUSE. However,
Tuggy does not discuss wh-clefts with double be. Pavey (2009) attempts to fill
this gap and notes a relationship with the amalgam construction shown in (vii).
Here, the focal NP hand sanitizer functions as the referent for both a specificational inversion sentence (its hand sanitizer) and a reverse wh-cleft (hand
sanitizer is what it is). Such constructs could also be said to perform a reinforcing role. See Massam (1999) for a generative account of the double is
construction.
(vii) Its hand sanitizer is what it is
(interview transcript [Pavey 2009: 27])
242
association with discourse-old information. For instance, we have seen how

the cleft it could have influenced the changes to the range of it-cleft foci.
Because this pronoun is maximally underspecified, it can describe elements
which cannot be classified by other nouns, and which are therefore unacceptable in other copular constructions. However, while it is underspecified, the interpretation it is given is usually specific. This gives a reason as
to why it-clefts show a preference for focusing discrete, mutually-exclusive
entities. The early association with discourse-old information, on the other
hand, has helped shape the development of new discourse functions for the
it-cleft construction. In particular, it results in an association between itclefts and the presentation of pre-established facts. This explains why the
it-cleft alone has developed informative-presupposition and performative
subtypes constructions which function to downplay the speakers responsibility for the information expressed.
As a point of comparison, wh-clefts differ from it-clefts in that the
pronoun what is both specified (as near synonymous with the thing) and
general. This explains why wh-clefts are able to focus syntactic categories
denoting properties, actions and events (entities which are not mutually
exclusive). Furthermore, wh-clefts show an early association with inferable,
as opposed to discourse-old, information, including that which can be
inferred from the metalinguistic context. As shown above, wh-clefts with
metalinguistic antecedents often express highly general information in the
precopular clause. This explains why the wh-cleft has developed a presentational function, in which the act of specification is merely used as a
device for introducing a focal proposition.
This, rather cursory, comparison of it-clefts and wh-clefts outlines some
possible reasons why, despite there being a familial relationship among all
specificational copular constructions, the it-cleft has undergone a separate
development, acquiring a greater range of idiosyncratic properties. We have
seen that small semantic and pragmatic differences in the make-up of these
constructions mean that the it-cleft and the wh-cleft have subtly different
prototypes, which govern and shape their construction-specific pathways of
change (see Chapter 2, Section 3).
Chapter 9
Conclusions
This book has provided a synchronic and diachronic constructional account

of the English it-cleft construction. In Chapter 1, we saw that the it-cleft is
an unusual configuration, with a non-standard structure that does not conform to highly general grammatical rules. In addition, it-clefts exhibit various pragmatic and discourse-functional characteristics which, at first sight,
are difficult to make sense of given their structural properties. In Chapter 2,
we saw that the framework of construction grammar is ideally suited to the
treatment of this specialized linguistic pattern, since it not only offers a way
of capturing and representing idiosyncratic information (through the makeup of the symbolic construction), but it also encourages us to explain how
less regular properties are motivated by identifying generalizations, albeit at
a local level, in the constructional taxonomy of grammatical knowledge.
As a result, I claimed that a constructional perspective is most consistent
with an approach which examines it-clefts in relation to a family of copular
constructions (rather than as information structure variants of syntactically
more basic sentences). In Chapter 3, I built up an analysis of specificational
copular sentences which draws heavily from the inverse tradition. On this
account, specificational meaning is quite literally the inverse of predication;
that is, it derives from the same nominal predication relation of class inclusion but involves interpreting this relation from the opposite perspective, as
listing the membership of a set rather than attributing a property to a referent. I showed further that the inherent meaning of definite NP predicates
makes them especially well-suited to enabling a specificational interpretation, since they describe the totality of a restricted set. This interpretation
also relies on a particular information structure, in which the referring
expression (the membership of the set) is in focus.
The account of specificational copular sentences sketched here captures
the same intuitions as other inverse analyses, but focuses on the question of
what constitutes specificational meaning, rather than on the conditions
affecting syntactic movement operations. It has the advantage of treating
specification as a phenomenon separate from (but associated with) inversion. From this, it follows that so-called reverse specificational copular
sentences are accounted for straightforwardly and the relation between
specification and inversion is explained, rather than assumed. Furthermore,
244
Conclusions
in recognizing the central role of definiteness in the creation of specificational meaning, the current account is more successful at explaining the
restrictions on indefinite specificational subjects. In particular, it shows that
while indefinite NPs provide a unique and useful function, their felicitousness is nevertheless governed by their similarity to definite NPs; that is,
they must exhibit, or at least not be at odds with, some characteristics associated more with definiteness than indefiniteness.
On the assumption that it-clefts are instances of the specificational
inversion construction, it follows that they too will involve nominal predication, containing a definite NP predicate and a postcopular referring
expression. This led me to argue for a discontinuous constituent analysis of
it-clefts in Chapter 4. On this account, the cleft clause is an extraposed
restrictive relative, which modifies the cleft pronoun it to form a discontinuous definite description. While this meant that we had to accept and
(momentarily) overlook certain idiosyncrasies in cleft structure, including
the obligatory extraposition of the relative clause and the restrictive modification of the pronoun it, it also allowed us to make several pertinent generalizations. For instance, on this account, the behaviour of the cleft clause is
consistent with a restrictive relative clause analysis. Likewise, the properties of focus, presupposition, exhaustiveness and contrast fall out from this
analysis, and are explained either as characteristics of definiteness more
generally or as products of specificational meaning.
While other authors have recognized such correspondences between itclefts and definite NPs, my account advances the current literature by offering support for the hypothesis that definiteness is fundamental to the it-cleft
and to specificational copular sentences in general. In this way, the analysis
of specificational meaning outlined in Chapter 3 provides further justification for a discontinuous constituent account of it-clefts. I concluded Chapter 4 by situating my account in relation to the cleft literature. I claimed that
the analysis argued for here is able to capture more generalizations than
both expletive analyses (which view the pronoun it as a dummy element)
and other constructional accounts (which make little use of inheritance
the explanatory mechanism of construction grammar). It also differs from
other recent discontinuous constituent accounts in that it defends a more
straightforward extraposition-from-NP analysis of it-clefts. While the more
complicated analyses claim to be able to account for some of the it-clefts
contradictory properties, I have explored the extent to which we can
account for them in other ways, without adding to the structural complexity
of the it-cleft.
Conclusions
245
In Chapter 5, I showed that this analysis extends naturally to incorporate

predicational (including proverbial) it-clefts, owing to inheritance from the
predicate nominal construction. I claimed that such sentences present a particular problem for expletive analyses of it-clefts and demonstrate further
support for a discontinuous constituent account. Furthermore, I showed that
this data can be accounted for more successfully if we assume an inverse
analysis of specificational copular sentences. From this perspective, specificational and predicational it-clefts have similar structures and the same
predicate nominal semantics. They differ primarily in how their elements of
form and components of meaning correspond to one another. While in the
specificational examples, the discontinuous definite NP functions as the
semantic predicate, in predicational it-clefts, this is treated as an act of
reference.
However, most discontinuous constituent accounts assume that it-clefts
(and other specificational copular sentences) are semantically equative.
From this, it follows that specificational and predicational it-clefts have
very different semantic structures, often involving different kinds of copula.
Furthermore, since on an equative approach, the discontinuous definite NP
functions as a referring expression in both varieties of it-cleft, it is unable to
account for the different realization of the cleft pronoun in specificational
and predicational it-clefts with plural clefted constituents. Finally, in
Chapter 5, we saw that there are examples of the specificational it-cleft
which exhibit additional, construction-specific properties that cannot be
accounted for by inheritance. In particular, the it-cleft allows some constituents to be clefted that are not found in other specificational copular
constructions and is able to occur with brand-new information in the cleft
clause, a characteristic which seems to be unique to this particular type of
specificational construction. I showed that while these tokens can be
accommodated into the it-cleft analysis argued for here, and do not require
a separate analysis, they nevertheless require explanations from outside the
present-day language system.
The historical component of the book began in Chapter 6. Here, we saw
that that the cleft structure is actually an Old English construction, which
has remained largely unchanged over time. When examined in relation to
the language system of earlier periods of English, the it-clefts idiosyncratic
structural properties, including the restrictive modification of the pronoun
it, the obligatory extraposition of the cleft clause, and the unusual pattern of
agreement, quickly lose their mystique. These properties were shown to be
originally inherited from once productive patterns, including the determinative pronoun construction, the paratactic relative clause construction and
246
Conclusions
the (semantically underspecified) OE/EME lexeme it. While these patterns

have since changed or fallen out of use, their influence remains entrenched
(or fossilized) within the it-cleft schema.
The early historical evidence therefore provides an explanation for these
idiosyncratic structural properties as well as demonstrable support for a discontinuous constituent account of it-clefts. The OE hit-clefts were shown to
be especially instructive in this regard, since the gender agreement system
provides an additional means with which to identify the structure of these
early tokens. With only one possible exception, all of the relevant instances
in the YCOE demonstrate agreement between the neuter cleft pronoun and
the relative pronoun (t), exactly as we would expect on an extrapositionfrom-NP account. This data has, to my knowledge, never been used in support of an extraposed/discontinuous structure for all types of it-cleft, despite
its clear preference for this type of analysis. Indeed, an expletive analysis is
unable to account for these early hit-clefts. We saw in Chapter 6 that this
has often resulted in the discounting of the OE specificational tokens from
the recorded history of the English it-cleft.
In Chapter 7, I conducted a diachronic investigation, which follows the
it-cleft from Old to Modern English. Making use of data extracted from the
YCOE, the PPCME2, the PPCEME and the PPCMBE, I showed that the
frequency of the specificational it-cleft has increased, and its function has
expanded, over time. In particular, we saw that the it-cleft occurs with an
increasingly wide range of foci. Originally focusing only concrete objects
(showing a preference for animate individuals), the it-cleft gradually occurs
with more abstract and relational foci, which denote less discrete concepts.
This is accompanied by an increase in the range of non-nominal categories
that can occur in postcopular position. In addition, the it-cleft acquires new
discourse functions as it changes from expressing old information in the
cleft clause, to also expressing non-salient information, before the inclusion
of information which is new to the hearer but assumed to be factual in some
sense, before allowing even speaker opinion to occur in this position. As a
result, the it-clefts idiosyncratic range of foci and its ability to occur with
brand-new information in the cleft clause were shown to be the outcome of
the it-clefts construction-specific historical development.
In Chapter 8, I argued that these changes to the function and use of the
specificational it-cleft are best interpreted as an example of schematization.
On this account, new instances are formed by extension from the prototype,
overriding the more general patterns of correspondence found in the original construction. In particular, the changes to the clefted constituent and the
cleft clause are governed by the referential semantics of the postcopular slot
Conclusions
247
and the presuppositional nature of the definite description. Over time,

semantic classes and syntactic categories which are less suited to the referring function are accommodated into the referential slot, and speakers are
able to further manipulate what sorts of information can be marked as
presupposed. The upshot is that the it-cleft gradually becomes a more schematic (or open) and productive construction, sanctioning a wider range of
instances. I concluded Chapter 8 by identifying two subtle semantic and
pragmatic properties which may have enabled and shaped the it-clefts
construction-specific historical development: the underspecified semantics
of the cleft pronoun it and an early association with old (or known) information.
In sum, this book has argued for a straightforward extraposition-fromNP analysis of the English it-cleft, drawing on evidence from three main
areas: (a) the central role of definiteness in the creation of specificational
meaning, (b) the existence and make-up of predicational (and proverbial) itclefts, and (c) the early, historical it-cleft data. The approach is constructional in that it aims to maximize motivation for the it-cleft construction, by identifying generalizations at a local level in the constructional
taxonomy. We have seen that the it-cleft inherits properties from specificational copular and predicate nominal constructions, from definite NPs and
restrictive relative clauses. On this account, many of the it-clefts properties
are shown to conform to more general patterns of correspondence. However, idiosyncrasies are also tolerated on this constructional account. Rather
than incorporating them into a complex (and counterintuitive) synchronic
analysis, this book has made use of historical evidence to explain how and
why the it-cleft has developed construction-specific properties over time. In
this way, the diachronic investigation informs and supports an otherwise
simple and intuitive analysis of the English it-cleft.
There are several areas of research which could follow from this study.
For instance, the proposed relationship between specification and the characteristics of definiteness should really be tested on a larger set, and a wider
range, of naturally occurring indefinite specificational subjects. Furthermore, if, as has been suggested here, it-clefts are instances of NP inversion,
it would be useful to situate the it-cleft within a history of specificational
inversion sentences, more generally. On a much smaller scale, while this
book has highlighted the common structure and shared history between itclefts and demonstrative clefts, our focus here has been on the it-cleft construction. A data-led study comparing the frequency, use and behaviour of
these cleft types over time would be most revealing.
Corpora and data sources
Hornby, Nick
2001
How to be Good. London: Penguin.
Kroch, Antony, and Ann Taylor
2000
The Penn-Helsinki Parsed Corpus of Middle English (PPCME2).
Department of Linguistics, University of Pennsylvania. CD-ROM,
second edition, http://www.ling.upenn.edu/hist-corpora/.
Kroch, Antony, Beatrice Santorini, and Lauren Delfs
2004
The Penn-Helsinki Parsed Corpus of Early Modern English
(PPCEME). Department of Linguistics, University of Pennsylvania.
CD-ROM, first edition, http://www.ling.upenn.edu/hist-corpora/.
Kroch, Antony, Beatrice Santorini, and Ariel Diertani
2010
The Penn-Helsinki Parsed Corpus of Modern British English
(PPCMBE). Department of Linguistics, University of Pennsylvania.
CD-ROM, first edition, http://www.ling.upenn.edu/hist-corpora/.
Nelson, Gerald, Sean Wallis, and Bas Aarts
1998
International Corpus of English Great Britain (ICE-GB) Release 2.
http://www.ucl.ac.uk/english-usage/projects/ice-gb/.
Taylor, Ann, Anthony Warner, Susan Pintzuk, and Frank Beths
2003
The York-Toronto-Helsinki Parsed Corpus of Old English Prose
(YCOE). Department of Linguistics, University of York. Oxford
Text Archive, first edition,
http://www-users.york.ac.uk/~lang22/YcoeHome1.htm.
References
Abbott, Barbara
2000
Presuppositions and nonassertions. Journal of Pragmatics 32: 1419
1437.
2004
Definiteness and indefiniteness. In The Handbook of Pragmatics,
Laurence R. Horn, and Gregory Ward (eds.), 122150. Oxford:
Wiley-Blackwell.
2006
Where have some of the presuppositions gone? In Drawing the
Boundaries of Meaning: Neo-Gricean Studies in Pragmatics and
Semantics in Honor of Laurence R. Horn, Betty Birner, and Gregory
Ward (eds.), 120. Amsterdam/Philadelphia: John Benjamins.
2010
Definite and indefinite. In Concise Encyclopedia of Philosophy of
Language and Linguistics, Alex Barber, and Robert J. Stainton
(eds.), 131138. Oxford: Elsevier.
Akmajian, Adrian
1970
On deriving cleft sentences from pseudo-cleft sentences. Linguistic
Inquiry 1 (2): 149168.
Allen, Cynthia
1980
Topics in Diachronic English Syntax. New York: Garland Publishing.
Anderson, A. H., M. Bader, E. G. Bard, E. Boyle, G. Doherty, S. Garron, S. Isard,
J. Kowtko, J. McAllister, J. Miller, C. Sotillo, H. Thompson, and R. Weinert
1991
The map task dialogues: a corpus of spoken English. Language and
Speech 34 (4): 351366.
Atlas, Jay David, and Stephen C. Levinson
1981
It-clefts, informativeness, and logical form: Radical Pragmatics
(Revised Standard Version). In Radical Pragmatics, Peter Cole (ed.),
162. New York: Academic Press.
Ball, Catherine N.
1977
Th-clefts. Pennsylvania Review of Linguistics 2: 5769.
1991
The historical development of the it-cleft. Ph. D. diss., University of
Pennsylvania.
1994a
The origins of the informative-presupposition it-cleft. Journal of
Pragmatics 22: 603628.
1994b
Relative pronouns in it-clefts: The last seven centuries. Language
Variation and Change 6: 179200.
1996
A diachronic study of relative markers in spoken and written English. Language Variation and Change 8: 227258.
250
References
1999
A short history of agreement in it-clefts. Paper presented at the

Modern Language Association annual convention (session 434: New
Work in Language Theory III), Chicago IL, 29 December.
Barsalou, Lawrence W.
1992
Cognitive Psychology: An Overview for Cognitive Scientists. Hillsdale NJ: Lawrence Erlbaum Associates.
Bardal, Jhanna
2008
Productivity: Evidence from Case and Argument Structure in Icelandic. Amsterdam: John Benjamins.
Birner, Betty J.
1994
Information status and word order: An analysis of English inversion.
Language 70 (2): 233259.
1996
The Discourse Function of Inversion in English. New York/London:
Garland Publishing.
Birner, Betty J., Jeffrey P. Kaplan, and Gregory Ward
2007
Functional compositionality and the interaction of discourse constraints. Language 83 (2): 317343.
Blom, Alied, and Saskia Daalder
1977
Syntaktische Theorie en Taalbeschrijving. Muiderberg: Coutinho.
Bolinger, Dwight
1970
The lexical value of it. Working Papers in Linguistics (University of
Hawaii) 2 (8): 5776.
1972
A look at equations and cleft sentences. In Studies for Einar Haugen;
Presented by Friends and Colleagues, Evelyn S. Firchow, Kaaren
Grimstad, Nils Hasselmo, and Wayne A. ONeil (eds.), 96114. The
Hague/Paris: Mouton.
1977
Meaning and form. London/New York: Longman.
Borkin, Ann
1984
Problems in Form and Function. Norwood NJ: Ablex Publishing
Corporation.
Bourcier, Georges
1977
Les Propositions Relatives en Vieil-Anglais. Paris: Editions Honor
Champion.
Brenier, Jason M., and Laura A. Michaelis
2005
Optimization via syntactic amalgam: Syntax-prosody mismatch and
copula doubling. Corpus Linguistics and Linguistic Theory 1 (1):
4588
Bresnan, Joan
1994
Locative inversion and the architecture of Universal Grammar.
Language 70 (1): 72131.
Bring, Daniel
1998
Identity, modality, and the candidate behind the wall. In Proceedings
of SALT 8, Devon Strolovich, and Aaron Lawson (eds.), 3654.
Ithaca NY: Cornell University Linguistic Publications.
References
251
Burton-Roberts, Noel
1981
Review of Hawkins (1978). Language 57: 191196.
Bybee, Joan
1985
Morphology: A Study of the Relation between Meaning and Form.
Amsterdam: John Benjamins.
1995
Regular morphology and the lexicon. Language and Cognitive
Processes 10 (5): 425455.
2003
Mechanisms of change in grammaticalization: The role of frequency.
In The Handbook of Historical Linguistics, Brian D. Joseph, and
Richard D. Janda (eds.), 602623. Oxford: Blackwell.
2006
From usage to grammar: The minds response to repetition. Language 82 (4): 711733.
Calude, Andreea S.
2007
Demonstrative clefts in spoken English. Ph. D. diss., University of
Aukland.
2008a
Clefting and extraposition in English. ICAME Journal 32: 734.
2008b
Demonstrative clefts and double cleft constructions in spontaneous
spoken English. Studia Linguistica 62 (1): 78118.
Cann, Ronnie
1993
Formal Semantics: An Introduction. Cambridge: Cambridge University Press.
Chafe, Wallace L.
1974
Language and consciousness. Language 50: 111133.
1976
Givenness, contrastiveness, definiteness, subjects, topics, and point
of view. In Subject and Topic, Charles N. Li (ed.), 2555. New
York: Academic Press.
Chomsky, Noam
1977
On wh-movement. In Formal Syntax, Peter W. Culicover, Thomas
Wasow, and Adrian Akmajian (eds.), 71132. New York: Academic
Press.
2000
New Horizons in the Study of Language and Mind. Cambridge:
Cambridge University Press.
Cockayne, Thomas O. (ed.)
1866
Hali Meidenhad. Early English Text Society. London: Trbner and
Co. http://archive.org/details/halimeidenhadfr00cockgoog (accessed
02 February 2012).
Collins, Peter
1991a
Cleft and Pseudo-cleft Constructions in English. London/New York:
Routledge.
1991b
Pseudocleft and cleft constructions: A thematic and informational
interpretation. Linguistics 29: 481519.
Croft, William
1990
A conceptual framework for grammatical categories (or: A taxonomy of propositional acts). Journal of Semantics 7: 245279.
252
References
1991
Syntactic Categories and Grammatical Relations. Chicago IL: The

University of Chicago Press.
2001
Radical Construction Grammar: Syntactic Theory in Typological
Perspective. Oxford/New York: Oxford University Press.
2007
Construction grammar. In The Oxford Handbook of Cognitive
Linguistics, Dirk Geeraerts, and Hubert Cuyckens (eds.), 463508.
Oxford/New York: Oxford University Press.
Croft, William, and D. Alan Cruse
2004
Cognitive Linguistics. Cambridge: Cambridge University Press.
Cross, J. E. (ed.)
1996
Two Old English Apocrypha and Their Manuscript Source: The
Gospel of Nichodemus and The Avenging of the Saviour.
Cambridge: Cambridge University Press.
Culicover, Peter W., and Ray Jackendoff
2005
Simpler Syntax. Oxford: Oxford University Press.
Culicover, Peter W., and Michael S. Rochemont
1990
Extraposition and the complement principle. Linguistic Inquiry 21
(1): 2347.
Curme, George O.
1931
Syntax. New York: Heath.
Davidse, Kristin
2000
A constructional approach to clefts. Linguistics 38 (6): 11011131.
Declerck, Renaat
1983a
It is Mr. Y or He is Mr. Y? Lingua 59: 209246.
1983b
Predicational clefts. Lingua 61: 945.
1984a
The pragmatics of it-clefts and wh-clefts. Lingua 64: 251289.
1984b
Some restrictions on clefts that highlight predicate nominals. Journal
of Linguistics 20: 131154.
1986
Two notes on the theory of definiteness. Journal of Linguistics 22:
2539.
1987
Definiteness and inclusive reference. Journal of Literary Semantics
16: 1229.
1988
Studies on Copular Sentences, Clefts and Pseudo-clefts. Leuven:
Leuven University Press; Dordrecht: Foris.
1992
The inferential it is that-construction and its congeners. Lingua 87:
203230.
1994
The taxonomy and interpretation of clefts and pseudo-clefts. Lingua
93: 183220.
Declerck, Renaat, and Shigeki Seki
1990
Premodified reduced it-clefts. Lingua 82: 1551.
Delahunty, Gerald P.
1982
Topics in the Syntax and Semantics of English Cleft Sentences.
Bloomington IN: Indiana University Linguistics Club.
References
1984
253
The analysis of English cleft sentences. Linguistic Analysis 13 (2):

63113.
Delin, Judy L.
1989
Cleft constructions in discourse. Ph. D. diss., University of Edinburgh.
1992
Properties of it-cleft presupposition. Journal of Semantics 9: 289
306.
Diewald, Gabriele
2006
Context types in grammaticalization as constructions. In Constructions special volume 1.
http://elanguage.net/journals/constructions/article/view/24/29
(accessed 02 February 2012)
Dikken, Marcel den
2005
Specificational copular sentences and pseudoclefts. The Blackwell
Companion to Syntax. Vol. 4, Martin Everaert, and Henk van
Riemsdijk (eds.). Oxford: Blackwell.
2006
Relators and Linkers: The Syntax of Predication, Predicate Inversion, and Copulas. Cambridge MA: MIT Press.
2009
Predication and specification in the syntax of cleft sentences. Ms.,
CUNY Graduate Center.
http://web.gc.cuny.edu/dept/lingu/dendikken/docs/predication_specif
ication_clefts_short_paper.pdf (accessed 05 May 2010)
Dikken, Marcel den, Andre Meinunger, and Chris Wilder
2000
Pseudoclefts and ellipsis. Studia Linguistica 54 (1): 4189.
Donnellan, Keith, S.
1966
Reference and definite descriptions. The Philosophical Review 75:
281304.
Dorgeloh, Heidrun
1997
Inversion in Modern English: Form and Function. Amsterdam/Philadelphia: John Benjamins.
. Kiss, Katalin
1998
Identification focus versus information focus. Language 74 (2): 245
273.
Emonds, Joseph E.
1976
A Transformational Approach to English Syntax: Root, StructurePreserving, and Local Transformations. New York: Academic Press.
Erdmann, Peter
1987
It-Stze im Englischen. Heidelberg: Winter.
Erteschik-Shir, Nomi
2007
Information Structure: The Syntax-Discourse Interface. Oxford: Oxford University Press.
Evans, Vyvyan, and Melanie Green
2006
Cognitive Linguistics: An Introduction. Edinburgh: Edinburgh University Press.
254
References
Fillmore, Charles J., Paul Kay, and Mary Kay OConnor

1988
Regularity and idiomaticity in grammatical constructions: The case
of let alone. Language 64 (3): 501538.
Filppula, Markku
2009
The rise of it-clefting in English: Areal-typological and contactlinguistic considerations. English Language and Linguistics 13 (2):
267293.
Fischer, Olga, Ans van Kemenade, Willem Koopman, and Wim van der Wurff
2000
The Syntax of Early English. Cambridge: Cambridge University
Press.
Fodor, Janet D., and Ivan Sag
1982
Referential and quantificational indefinites. Linguistics and Philosophy 5 (3): 355398.
Fowler, Henry W., and Francis G. Fowler
1908
The Kings English, 2nd edn. Oxford: Clarendon Press Bartleby.com.
1999. www.bartleby.com/116/ (accessed 02 February 2011)
Francis, Elaine, J.
2010
Grammatical weight and relative clause extraposition in English.
Cognitive Linguistics 21 (1): 3574.
Francis, Elaine J., and Laura A. Michaelis
2003
Mismatch: A crucible for linguistic theory. In Mismatch: FormFunction Incongruity and the Architecture of Grammar, Elaine J.
Francis, and Laura A. Michaelis (eds.), 127. Stanford CA: CSLI
publications.
Fried, Mirjam
2008
Constructions and constructs: Mapping a shift between predication
and attribution In Constructions and Language Change, Alexander
Bergs, and Gabriele Diewald (eds.), 4779. Berlin/New York: Mouton de Gruyter.
Fox, Barbara A., and Sandra A. Thompson
1990
A discourse explanation of the grammar of relative clauses in
English conversation. Language 66 (2): 297316.
Goldberg, Adele E.
1995
Constructions: A Construction Grammar Approach to Argument
Structure. Chicago IL: University of Chicago Press.
2003
Words by default: The Persian complex predicate construction. In
Mismatch: Form-Function Incongruity and the Architecture of
Grammar, Elaine J. Francis, and Laura A. Michaelis (eds.), 117146.
Stanford CA: CSLI publications.
2006
Constructions at Work: The Nature of Generalization in Language.
Oxford/New York: Oxford University Press.
Gmez-Gonzlez, Mara de los ngeles
2004
A three-dimensional account of it-clefts in discourse: a corpus-based
study. Southwest Journal of Linguistics 23 (2): 81120.
References
255
It was you that told me that, wasnt it? It-clefts revisited in

discourse. In Structural-Functional Studies in English Grammar,
Mike Hannay, and Gerard J. Steen (eds.), 103139. Amsterdam:
John Benjamins.
Gmez-Gonzlez, Mara de los ngeles, and Gonzlvez-Garca, Francisco
2005
On clefting in English and Spanish. In The Dynamics of Language
Use: Functional and Contrastive Perspectives, Christopher S. Butler,
Mara de los ngeles Gmez-Gonzlez, and Susana M. DovalSurez (eds.), 155196. Amsterdam/Philadelphia: John Benjamins.
Graff, Delia
2001
Descriptions as predicates. Philosophical Studies 102 (2): 142.
Gundel, Jeannette K.
1977
Where do cleft sentences come from? Language 53 (3): 543559.
1985
Shared knowledge and topicality. Journal of Pragmatics 9: 83107.
Gundel, Jeannette K., Nancy Hedberg, and Ron Zacharski
1993
Cognitive status and the form of referring expressions in discourse.
Language 69 (2): 274307.
2001
Definite descriptions and cognitive status in English: Why accommodation is unnecessary. English Language and Linguistics 5 (2):
273295.
Halliday, M.A.K.
1967
Notes on transitivity and theme in English, part II. Journal of
Linguistics 3: 199244.
Halvorsen, Per-Kristian
1978
The syntax and semantics of cleft constructions. Ph. D. diss.,
University of Texas, Austin.
Han, Chung-hye, and Nancy Hedberg
2008
Syntax and semantics of it-clefts: A Tree Adjoining Grammar
analysis. Journal of Semantics 25: 345380.
Hankamer, Jorge
1974
On the non-cyclic nature of WH-clefting. Chicago Linguistic Society
(CLS) 10: 221233.
Hasselgrd, Hilde
2004
Adverbials in IT-cleft constructions. In Advances in Corpus Linguistics. Papers from the 23rd International Conference of English
Language Research on Computerized Corpora (ICAME 23) Gteborg 2226 May 2002, Karin Aijmer, and Bengt Altenberg (eds.),
195211. Amsterdam/New York: Rodopi.
Haugland, Kari E.
1993
A note on cleft and existential sentences in Old English. English
Studies 5: 407413.
Hawkins, John A.
1978
Definiteness and Indefiniteness: A Study in Reference and Grammaticality Predication. Atlantic Highlands NJ: Humanities Press.
2007
256
References
1991
On (in)definite articles: Implicatures and (un)grammaticality prediction. Journal of Linguistics 27: 405442.
1994
A Performance Theory of Order and Constituency. Cambridge:
Cambridge University Press.
2004
Efficiency and Complexity in Grammars. Oxford: Oxford University
Press.
Hedberg, Nancy
1990
Discourse pragmatics and cleft sentences in English. Ph. D. diss.,
University of Minnesota.
2000
The referential status of clefts. Language 76 (4): 891920.
Hedberg, Nancy, and Lorna Fadden
2007
The information structure of it-clefts, wh-clefts and reverse wh-clefts
in English. In The Grammar-Pragmatics Interface: Essays in Honor
of Jeannette K. Gundel, Nancy Hedberg, and Ron Zacharski (eds.),
4976. Amsterdam/Philadelphia: John Benjamins.
Heggie, Lorie A.
1988
The syntax of copular structures. Ph. D. diss., University of Southern
California.
Heim, Irene R.
1982
The semantics of definite and indefinite noun phrases. Ph. D. diss.,
University of Massachusetts.
Heller, Daphna
2002
On the relation of connectivity and specificational pseudoclefts.
Natural Language Semantics 10 (4): 243284.
2005
Identity and information: Semantic and pragmatic aspects of specificational sentences. Ph. D. diss., The State University of New Jersey.
Herriman, Jennifer
2003
Negotiating identity: The interpersonal function of wh-clefts in English. Functions of Language 10 (1): 130.
Herzfeld, George (ed.)
1900
An Old English Martyrology; Re-edited from Manuscripts in the
Libraries of the British Museum and of Corpus Christi College
Cambridge. Early English Text Society. London: Trbner and Co.
Heycock, Caroline, and Anthony Kroch
1999
Pseudocleft connectedness: Implications for the LF interface level.
Linguistic Inquiry 30 (3): 365397.
2002
Topic, focus, and syntactic representations. In West Coast Conference on Formal Linguistics (WCCFL) 21, Line Mikkelsen, and Chris
Potts (eds.), 101125. Somerville MA: Cascadilla Press.
Higgins, F. Roger
1979
The Pseudo-cleft Construction in English. New York: Garland Publishing.
References
257
Himmelmann, Nikolaus P.
2004
Lexicalization and grammaticization: Opposite or orthogonal? In
What Makes Grammaticalization? A Look from its Fringes and its
Components, Walter Bisang, Nikolaus Himmelmann, and Bjorn
Wiemer (eds.), 2142. (Trends in Linguistics, Studies and Monographs 158.) Berlin/New York: Mouton de Gruyter.
Hopper, Paul J.
2001
Grammatical constructions and their discourse origins: Prototype or
family resemblance? In Applied Cognitive Linguistics 1: Theory and
Language Acquisition, Martin Ptz, Suzanne Neimeier, and Ren
Dirven (eds.), 101129. (Cognitive Linguistics Research 19). Berlin/
New York: Mouton de Gruyter.
2004
The openness of grammatical constructions. Chicago Linguistic
Society (CLS) 40: 153175.
Hopper, Paul J., and Sandra A. Thompson
2008
Projectability and clause combining in interaction. In Crosslinguistic
Studies of Clause Combining, Ritva Laury (ed.), 99123. Amsterdam/Philadelphia: John Benjamins.
Horn, Laurence R.
1981
Exhaustiveness and the semantics of clefts. In Papers from the 11th
Annual Meeting of NELS, V. Burke, and J. Pustejovsky (eds.), 124
142. Amherst: GLSA publications.
1986
Presupposition, theme and variations. In Papers from the Parasession on Pragmatics and Grammatical Theory, 22nd Regional Meeting
of the Chicago Linguistics Society, Anne M. Farley, Peter T. Farley,
and Karl-Erik MacCullough (eds.), 168192. Chicago: Chicago Linguistics Society
Huddleston, Rodney
1984
Introduction to the Grammar of English. Cambridge: Cambridge
University Press.
Hudson, Richard A.
1990
English Word Grammar. Oxford: Blackwell.
2007
Language Networks: The New Word Grammar. Oxford: Oxford University Press.
Jacobson, Pauline
1994
Binding connectivity in copular sentences. In Semantics and Linguistic Theory (SALT) IV, Mandy Harvey, and Lynn Santelmann (eds.),
161178. Ithaca NY: Cornell University.
Jackendoff, Ray
1977
X-bar Syntax: A Study of Phrase Structure. Cambridge MA: MIT
Press.
Jespersen, Otto
1927
A Modern English Grammar on Historical Principles, Part 3.
Heidelberg: Winter.
258
References
1937
1949
Analytic Syntax. London: Allen and Unwin.

A Modern English Grammar on Historical Principles, Part 7. Copenhagen: Munksgaard.
Johansson, Christine
2008
The use of the it-cleft construction in 19th century English. In Corpora and Discourse: The Challenges of Different Settings, Annelie
del, and Randi Reppen (eds.), 243266. Amsterdam/Philadelphia:
John Benjamins.
Johansson, Mats
2001
Clefts in contrast: A contrastive study of it clefts and wh clefts in
English and Swedish texts and translations. Linguistics 39 (3): 547
582.
Kay, Paul, and Charles J. Fillmore
1999
Grammatical constructions and linguistic generalizations: The whats
X doing Y? construction. Language 75 (1): 133.
Kaltenbck, Gunther
2004
It-Extraposition and Non-Extraposition in English: A Study of Syntax
in Spoken and Written Texts. Wien: Braumller.
Kiparsky, Paul, and Carol Kiparsky
1970
Fact. In Progress in Linguistics, Manfred Bierwisch, and Karl E.
Heidolph (eds.), 143173. The Hague: Mouton.
Koops, Christian, and Martin Hilpert
2009
The co-evolution of syntactic and pragmatic complexity: Diachronic
and cross-linguistic aspects of pseudoclefts. In Syntactic Complexity:
Diachrony, Acquisition, Neuro-cognition, Evolution, Talmy Givn,
and Masayoshi Shibatani (eds.), 215238. Amsterdam: John Benjamins.
Lakoff, George
1987
Women, Fire and Dangerous Things: What Categories Reveal About
the Mind. Chicago IL: University of Chicago Press.
Lambrecht, Knud
1994
Information Structure and Sentence Form. Cambridge: Cambridge
University Press.
2001
A framework for the analysis of cleft constructions. Linguistics 39
(3): 463516.
Langacker, Ronald
1987
Foundations of Cognitive Grammar, Vol I: Theoretical Prerequisites. Stanford CA: Stanford University Press.
1990
Subjectification. Cognitive Linguistics 1: 538.
1991
Foundations of Cognitive Grammar, Vol II: Descriptive Application.
Stanford CA: Stanford University Press.
Leech, Geoffrey, and Jan Svartvik
2002
A Communicative Grammar of English. 3d ed. London: Pearson
Education.
References
259
Lehmann, Christian
2008
Information structure and grammaticalization. In Theoretical and
Empirical Issues in Grammaticalization, Elena Seoane, and Mara
Jos Lpez-Couso (eds.), 207229. Amsterdam/Philadelphia: John
Benjamins.
Los, Bettelou
2009
The consequences of the loss of verb-second in English: Information
structure and syntax in interaction. English Language and Linguistics 13 (1): 97125.
Los, Bettelou, and Erwin Komen
Forthc. Clefts as resolution strategies after the loss of a multifunctional first
position. In The Oxford Handbook of the History of English, Terttu
Nevalainen, and Elizabeth Closs Traugott (eds.) New York/Oxford:
Oxford University Press.
Mair, Christian, and Claudia Winkle
2012
Change from to-infinitive to bare infinitive in specificational cleft
sentences: Data from world Englishes. In Mapping Unity and Diversity World-Wide, Marianne Hundt, and Ulrike Gut (eds.), 243262.
Amsterdam/Philadelphia: John Benjamins.
Massam, Diane
1999
Thing is constructions: The thing is, is whats the right analysis?
English Language and Linguistics 3 (2): 335352.
Matsunami, Tamotsu
1961
A historical consideration of the disjunctive formula: It is I that am
to blame. Einungaku Kenkyu (Studies in English Literature) 38: 1
15. Tokyo: English Literary Society of Japan.
Michaelis, Laura
2003
Headless constructions and coercion by construction. In Mismatch:
Form-Function Incongruity and the Architecture of Grammar,
Elaine J. Francis, and Laura A. Michaelis (eds.), 259310. Stanford
CA: CSLI publications.
Mikkelsen, Line
2002
Two types of definite description subjects. In Proceedings of the 7th
ESSLLI student session (at ESSSLI 14), Malvina Nissim (ed.), 141
153. Trento, Italy.
http://linguistics.berkeley.edu/~mikkelsen/papers/ESSLLI02_pro.pdf
(accessed 05 July 2011)
2004
Reexamining Higgins taxonomy: A split in the identificational
class. Paper presented at Linguistic Society of America (LSA)
Meeting, Boston MA, 10 January.
2005
Copular Clauses: Specification, Predication and Equation. Amsterdam: John Benjamins.
260
References
2007
On so-called truncated clefts. In Kopulaverben under Kopulastze:

Intersprachliche und Intrasprachliche Aspekte, Ljudmila Geist, and
Bjrn Rothstein (eds.), 4768. Tbingen: Niemeyer Verlag.
Miller, Thomas (ed.)
1898
The Old English Version of Bedes Ecclesiastical History of the
English People. Early English Text Society. London: Trbner and
Co.
Millet, Bella, and Jocelyn Wogan-Browne (eds.)
1990
Medieval English Prose for Women: Selections from the Katherine
Group and Ancrene Wisse. Oxford: Clarendon Press.
Mitchell, Bruce
1985a
Old English Syntax, Vol. 1: Concord, the Parts of Speech and the
Sentence. Oxford: Clarendon Press.
1985b
Old English Syntax, Vol. 2: Subordination, Independent Elements,
and Element Order. Oxford: Clarendon Press.
Mitchell, Bruce, and Fred C. Robinson
2007
A Guide to Old English. 7th ed. Oxford/Malden: Wiley-Blackwell.
Morris, Richard (ed.)
1880
The Blickling Homilies of the Tenth Century: From the Marquis of
Lothians Unique Ms. A.D. 971. Early English Text Society. London:
Trbner and Co. http://archive.org/details/oldenglishhomil00morr
Moro, Andrea
1997
The Raising of Predicates: Predicative Noun Phrases and the Theory
of Clause Structure. Cambridge: Cambridge University Press.
Neale, Stephen
1990
Descriptions. Cambridge MA: MIT Press.
Nelson, Gerald
1997
Cleft constructions in spoken and written English. Journal of English
Linguistics 25 (4): 340348.
Nunberg, Geoffrey, Thomas Wasow, and Ivan Sag
1994
Idioms. Language 70 (3): 491538.
Oberlander, Jon, and Judy Delin
1996
The function and interpretation of reverse wh-clefts in spoken discourse. Language and Speech 39 (2/3): 185227.
ONeil, Wayne
1977
Clause adjunction in Old English. General Linguistics 17 (4): 199
211.
Partee, Barbara H.
1998
Copula inversion puzzles in English and Russian. In Formal
Approaches to Slavic Linguistics: The Seattle Meeting 1988, Katarzyna Dziwirek, Herbert Coats, and Cynthia Vakareliyska (eds.),
361395. Ann Arbor: Michigan Slavic Publications.
References
2004a
261
Reprint. Ambiguous pseudoclefts with unambiguous be. In Compositionality in Formal Semantics: Selected Papers by Barbara H. Partee, Barbara H. Partee (ed.), 190202. Malden MA/Oxford: Blackwell Publishing. Original edition, in Proceedings of NELS XVI, Steve
Berman, Jae-Woong Choe, and Joyce McDonough (eds.), 354366.
Amherst MA: University of Massachusetts at Amherst, 1986.
2004b
Reprint. Noun phrase interpretation and type-shifting principles. In
Compositionality in Formal Semantics: Selected Papers by Barbara
H. Partee, Barbara H. Partee (ed.), 203230. Malden MA/Oxford:
Blackwell Publishing. Original edition, in Studies in Discourse
Representation Theory and the Theory of Generalized Quantifiers,
Jeroen Groenedijk, Dick de Jong, and Martin Stokhof (eds.), 115
143. Dordrecht: Foris, 1987.
Patten, Amanda L.
2008
Accommodating predicational and proverbial clefts. Paper presented
at the Cleft Workshop (Cleft08), Centre of General Linguistics, Berlin, 29 November.
2010
Grammaticalization and the it-cleft construction. In Gradience,
Gradualness and Grammaticalization, Elizabeth C. Traugott, and
Graeme Trousdale (eds.), 221243. Amsterdam: John Benjamins.
Forthc. The historical development of the it-cleft: A comparison of two different approaches. Studies in Language 36 (3).
Pavey, Emma
2003
An analysis of it-clefts within a role and reference grammar framework. In University of North Dakota-SIL Working Papers 47.
http://www.und.edu/dept/linguistics/wp/2003Pavey.PDF
(accessed 02 February 2012)
2004
The English it-cleft construction: A Role and Reference Grammar
analysis. Ph. D. diss., University of Sussex.
2009
What it is is pseudoclefts changed. Southern Journal of Linguistics
33 (1): 2438.
Percus, Orin
1997
Prying open the cleft. In The 27th Annual Meeting of the North East
Linguistics Society, Kiyomi Kusumoto (ed.), 337351. Amherst MA:
GLSA publications.
Prez-Guerra, Javier
1998
Integrating right-dislocated constituents: A study on cleaving and
extraposition in the recent history of the English language. Folia
Linguistica Historica 19 (12): 725.
1999
Historical English Syntax: A Statistical Corpus-Based Study on the
Organisation of Early Modern English Sentences. Mnchen/Newcastle: Lincom Europa.
262
References
Forthc.
A corpus-based analysis of it-clefts in the recent history of English:

What ist you lack? In Creation and Use of Historical English Corpora in Spain, Nila Vzquez (ed.). Newcastle upon Tyne: Cambridge
Scholars Publishing.
Pinkham, Jessie, and Jorge Hankamer
1975
Deep and shallow clefts. Chicago Linguistic Society (CLS) 11: 429
450.
Prince, Ellen F.
1978
A comparison of wh-clefts and it-clefts in discourse. Language 54
(4): 883906.
1981
Toward a taxonomy of given-new information. Radical Pragmatics,
Peter Cole (ed.), 223255. New York: Academic Press.
1992
The ZPG letter: Subjects, definiteness, and information-status. In
Discourse Description: Diverse Linguistic Analyses of a Fundraising
Text, William C. Mann, and Sandra A. Thompson (eds.), 295325.
Philadelphia: John Benjamins.
Postal, Paul M.
1970
On so-called pronouns in English. In Readings in English Transformational Grammar, Roderick A. Jacobs and Peter S. Rosenbaum
(eds.), 5682. London: Ginn.
Reeve, Matthew
2010
Clefts. Ph. D. diss., University College London.
2011
The syntactic structure of English clefts. Lingua 121: 142171.
2012
Clefts and their Relatives. (Linguistics Aktuell/Linguistics Today
185). Amsterdam/Philadelphia: John Benjamins.
Rochemont, Michael
1986
Focus in Generative Grammar. Philadelphia: John Benjamins.
Rothstein, Susan
2001
Predicates and their Subjects. Dordrecht: Kluwer.
Sag, Ivan A., Hans C. Boas, and Paul Kay
2012
Introducing sign-based construction grammar. To appear in SignBased Construction Grammar, Hans C. Boas, and Ivan A. Sag (eds.).
Stanford: CSLI Publications.
http://lingo.stanford.edu/sag/papers/sag-boas-intro.pdf.3
(accessed 04 April 2012)
Santorini, Beatrice
2010
Annotation Manual for the Penn Historical Corpora and the
PCEEC: Release 2.
http://www.ling.upenn.edu/hist-corpora/annotation/index.html
Schlenker, Phillipe
2003
Clausal equations: A note on the connectivity problem. Natural
Language & Linguistic Theory 21: 157214.
References
263
Schmid, Hans-Jrg
2001
Presupposition can be a bluff: How abstract nouns can be used as
presupposition triggers. Journal of Pragmatics 33: 15291552.
Sedgefield, Walter John (ed.)
1900
King Alfreds Version of the Consolations of Boethius, Done into
Modern English. Oxford: Clarendon Press.
http://oll.libertyfund.org/title/1178/78834 (accessed 07 July 2011).
Sharvit, Yael
1999
Connectivity in specificational sentences. Natural Language Semantics 7 (3): 299339.
2003
Tense and identity in copular constructions. Natural Language
Semantics 11 (4): 363393.
Skeat, Walter W. (ed.)
1881a
lfrics Lives of Saints. Vol. 2. Early English Text Society. London:
Trbner and Co. http://archive.org/details/aelfricslivesof02aelf (accessed 07 July 2011)
1881b
lfrics Lives of Saints. Vol. 1. Early English Text Society. London:
Trbner and Co. http://archive.org/details/aelfricslivesof01aelf (accessed 07 July 2011)
Smith, Alphonso
1906
Studies in English Syntax. Boston/ New York: Ginn.
http://archive.org/details/studiesinenglis01smitgoog (accessed 8 September 2011)
Sornicola, Rosanna
1988
It-clefts and wh-clefts: Two awkward sentence types. Journal of
Linguistics 24: 343379.
Sperber, Dan, and Deirdre Wilson
1986
Relevance: Communication and Cognition. Oxford: Blackwell Publishing.
Stanley, Jason, and Zoltn Gendler Szab
2000
On quantifier domain restriction. Mind & Language 15 (2/3): 219
261.
Surez-Gmez, Cristina
2006
Relativization in Early English (9501250): The Position of Relative
Clauses. Bern: Peter Lang.
Szabolcsi, Anna
1983
Focusing properties, or the trap of first order. Theoretical Linguistics
10: 125145.
Taylor, Ann
2003
The YCOE annotation manual. http://wwwusers.york.ac.uk/~lang22/YCOE/doc/annotation/YcoePosToc.htm
Taylor, John R.
2002
Cognitive Grammar. Cambridge: Cambridge University Press.
264
References
Thorpe, Benjamin (ed.)

1844
The Homilies of the Anglo-Saxon Church: The First Part, containing
the Sermones Catholici, or Homilies of lfric. Vol. 1. London:
lfric Society. http://archive.org/details/homiliesofanglos01aelfuoft
(accessed 10 October 2011)
Tognini-Bonelli, Elena
1992
All Im saying is: The correlation of form and function in
pseudo-cleft sentences. Literary and Linguistic Computing 7 (1): 30
42.
2001
Corpus Linguistics at Work. Amsterdam/Philadelphia: John Benjamins.
Traugott, Elizabeth Closs
1982
From propositional to textual and expressive meanings: Some
semantic-pragmatic aspects of grammaticalization. In Perspectives
on Historical Linguistics, Winifred P. Lehmann, and Yakov Malkiel
(eds.), 245271. Amsterdam: John Benjamins.
1989
On the rise of epistemic meanings in English: An example of
subjectification in semantic change. Language 65 (1): 3155.
2007
The concepts of constructions mismatch and type-shifting from the
perspective of grammaticalization. Cognitive Linguistics 18 (4):
523557.
2008
All that he endeavoured to prove was: On the emergence of
grammatical constructions in dialogic context. In Language in Flux:
Dialogue Coordination, Language Variation, Change and Evolution,
Robin Cooper, and Ruth Kempson (eds.), 143177. London: Kings
College Publications.
Trousdale, Graeme
2008a
Constructions in grammaticalization and lexicalization: Evidence
from the history of a composite predication construction in English.
In Constructional Approaches to English Grammar, Graeme Trousdale and Nikolas Gisborne (eds.), 3367. Berlin: Mouton de Gruyter.
2008b
A constructional approach to lexicalization processes in the history
of English: Evidence from possessive constructions. Word structure
1: 156177.
Tuggy, David
1996
The thing is is that people talk that way. The question is is why? In
Cognitive Linguistics In the Redwoods: The Expansion of a New
Paradigm in Linguistics, Eugene Casad (ed.), 713752. Berlin:
Mouton de Gruyter.
Vallduv, Enric, and Elisabet Engdahl
1996
The linguistic realization of information packaging. Linguistics 34:
459519.
References
265
Visser, F. Th.
1963
An Historical Syntax of the English Language. Part 1: Syntactical
Units with One Verb. Leiden: Brill.
Ward Gregory, and Betty Birner
2004
Information structure and non-canonical syntax. In The Handbook of
Pragmatics, Laurence R. Horn, and Gregory Ward (eds.), 153174.
Oxford: Wiley-Blackwell.
Ward Gregory, Betty Birner, and Rodney Huddleston
2002
Information packaging. In The Cambridge Grammar of the English
Language, Rodney Huddleston, and Geoffrey Pullum (eds.), 1363
1447. Cambridge: Cambridge University Press.
Wasow, Thomas
1997
Remarks on grammatical weight. Language Variation and Change 9:
81105.
2002
Postverbal behaviour. Stanford: CSLI Publications.
Watson, Nicolas, and Jacqueline Jenkins (eds.)
2006
The Writings of Julian of Norwich: A Vision Showed to a Devout
Woman and a Revelation of Love. Pennsylvania: The Pennsylvania
State University Press.
Weinert, Regina, and Jim Miller
1996
Cleft constructions in spoken language. Journal of Pragmatics 25:
173206.
Wierzbicka, Anna
1980
Lingua Mentalis: The Semantics of Natural Language. New York:
Academic Press.
Williams, Edwin
1980
Predication. Linguistic Inquiry 11 (1): 203238.
1983
Semantic vs. syntactic categories. Linguistics and Philosophy 6:
423446.
Wirth, Jessica R.
1978
The derivation of cleft sentences in English. Glossa 12 (1): 5881.
Index
agreement
case, 9799, 168170, 172173
gender, 30, 133, 149, 170176, 181,
183, 246
number, 30 39n, 9697, 99102,
112113, 132133, 148149,
162168
person, 9699, 105, 168170, 191
all-clefts, 10, 62, 6869
analogy and pattern alignment, 104,
158, 160161, 165, 178179, 212n
Celtic influence, 221222, 224
cleft clause
changes to information status of, 13,
204211, 213, 215218, 225,
232235, 242
obligatory extraposition of, 12, 73,
109, 148149, 157162, 183,
245246
structure of, 1, 3, 8893, 104105,
116118, 123124, 136, 161
162
cleft it
expletive, see expletive accounts
definite, see discontinuous
constituent accounts
semantically underspecified, see
underspecified it
cleft schema, 7576, 78, 152, 161
clefted constituent
categories of foci, 3, 121, 134140,
221, 225229, 231
changes to, 1213, 196204, 211,
213215, 217221 , 224225,
227228, 242, 246247
coercion, 2426, 4143, 139140, 201,
214215, 220, 224, 228, 231
see also individualizing techniques
Cognitive Grammar, 16, 4347

compositionality, degree of, 9, 1718,
23, 114, 116117, 119
connectivity effects, 6567, 237n
constituency, VP, 8788, 9396, 102
103, 111112
construction grammar, 45, 912, 14,
1626, 4042, 47, 74, 115119,
146147, 212213, 243244
constructionalization, 24, 213218,
222, 225
contrast, 6, 63, 8587, 103, 137, 139,
181182, 200201, 203, 228
conventionality, degree of, 13, 1718,
2122, 26, 119, 153, 185, 214215,
216n, 218, 224
copula
be of identity, 2, 29, 43, 66n, 110
112, 117, 132
classifying be, 43, 46, 57
non-relational be, 4344, 46
corpora, search and selection process,
1415, 184194
see also ICE-GB; PPCEME;
PPCMBE; PPCME2; YCOE
definiteness and inclusiveness, 11, 36
37, 41, 47, 5455, 5758, 6364,
69, 72, 8287, 244
demonstrative (this-, that-) clefts, 74
76, 104
history of, 151153, 156, 160, 173,
179, 185, 247
determinative pronouns, 153160, 169,
177178, 180, 186
discontinuous constituent accounts, 11,
7278, 109115
double is construction, 240241
Index
Early Modern English, 159n, 161162,
165166, 189190, 194, 196197,
200203, 206207, 211, 214, 217,
219220, 224, 230
end weight, see information structure
generalizations
entrenchment, 12, 2122, 131, 157,
178179, 212, 218, 246
equative approach, 2831, 33, 61n,
110115, 117, 131134, 167, 245
equative sentences, 28, 3031, 42, 59n
exclusiveness, see indefiniteness
exhaustiveness, 2, 11, 3537, 5758,
75, 8286, 103104, 110, 117, 137
existential there be/it be sentences, 52n,
158n, 190191, 193
expansion, see schematization
expletive accounts, 57, 11, 71, 102
107, 112, 114, 116, 123127, 136,
244245
influence on historical accounts,
169170, 172174, 176, 181,
185, 246
extraposed relative clauses, 73, 93, 95
96, 102, 113n, 148, 154, 156159,
161, 173
see also cleft clause, obligatory
extraposition of
extraposed subject clauses, 4, 15, 19
20, 160161, 193
extraposition accounts, 79, 11, 7173,
107115, 136 137
factive clauses, 226, 229230, 234235
familiarity conditions, 3, 36, 39, 5055,
8081, 8687, 121122, 140146,
215216, 232237, 241 242
see also information structure
generalizations; presupposition
focus-first it-clefts, 92, 149, 191
focusing
in it-clefts,12, 57, 73, 7980, 116,
140, 223224
267
in specificational sentences, 35, 37

38, 5759, 7980, 102
in V2 construction, 222224
see also clefted constituent
fossilization, 13, 22, 25, 177179, 212
213, 218, 246
garden-path readings, 91, 113n
givenness hierarchy, 141
grammaticalization, 16, 2325, 213,
215, 216n, 217218, 222
see also constructionalization
ICE-GB, 14, 9192, 9798, 100, 120
121, 124, 126, 128, 130, 134, 138
139, 225, 235236
identificational copular sentences,
174176, 180182
identity statements, see equative
sentences
impersonal/presentational sentences,
186190, 193, 197199, 209, 219
220, 224
inclusiveness, see definiteness
indefiniteness and exclusiveness, 37,
4758, 84, 128, 143144
individualizing techniques
focusing adverbs, 66n, 139, 200
201, 203, 214, 223224
listing, 139, 200201, 203, 214,
226, 228229
nominalization, 138, 229
information structure generalizations,
40, 73, 129, 158159, 161, 179
see also familiarity conditions
informative-presupposition it-clefts, 3,
121122, 140146, 186, 232n,
233n, 234
development of, 188, 204211,
215220, 221n, 232, 234, 237,
241242
inheritance, 1013, 16, 1821, 25, 78,
102, 115119, 177178, 244245,
247
268
Index
inverse approach, 3032, 34, 38, 50

51, 7778, 106, 112, 115, 127128,
132133, 167, 169, 243
inversion sentences, 3840, 57, 59
Irish and Irish English, 221
it-cleft schema, 128129, 131, 135,
149, 178, 212216, 246
Late Modern English, 161162, 189
190, 193197, 200204, 207211,
214, 220221, 224, 238
Middle English, 149150, 154159,
163166, 168170, 176177, 179
181, 182n, 190191, 194200, 202,
204n, 205206, 211, 219220
mismatch, 1920, 24, 26, 40, 59, 72,
114, 116, 213214
motivation, 1013, 2021, 25, 40, 59,
73, 119120, 156, 158, 161, 177
179, 183184, 243, 247
Old English, 22, 150160, 163164,
170178, 180183, 186189, 192,
194200, 204205, 218219, 221n,
224, 245
pattern alignment, see analogy
performative it-clefts, 206207, 209,
217218, 220, 232, 234235, 237,
242
PPCEME, 15, 159n, 166, 169, 185,
189190, 194, 197, 200203, 206
207, 214, 220, 246
PPCMBE, 15, 185, 190, 193194, 196
197, 200204, 207210, 246
PPCME2, 15, 154, 159n, 165166,
185, 190191, 194195, 197198,
200, 202, 205206, 219220, 246
predication, nominal, 1920, 2830,
3437, 4049, 57
predicational clefts, 4, 92n, 120135,
226227, 245, 247
history of, 151152, 155 156, 160,

166168, 171, 173174, 176,
180181, 194196
presupposition, 2, 11, 8082, 86, 103,
110, 140, 186, 244
governing change, 210, 213, 215
217, 220, 225, 247
see also informative-presupposition
it-clefts
proverbial it-clefts, 4, 120, 122124,
127, 129131, 245
history of, 149, 156, 179
pseudoclefts, 78, 6263, 6970, 98
99, 145
as a source for it-clefts, 8, 107109,
136137
see also all-clefts; th-clefts; whclefts
quantification, universal, 41, 45, 6869,
112, 117
relationality, see valence
relative clauses
establishing relatives, 5355, 143
145
history of relative markers, 158n,
161162, 170171
nonrestrictive relatives, 8890,
111112, 116118, 123, 192
reduced relatives, 63, 9192, 113n,
152
restrictive relatives, see cleft clause,
structure of; extraposed relative
clauses
zero realization of relative marker,
8992, 193, 158n
reverse specificational sentences, 32,
3538, 5859, 6669, 73, 74n, 80,
144145, 160n, 169, 181, 231232,
233n, 241n, 243
right-dislocation, 8, 107n, 108, 124,
161n, 173
Index
schematization and expansion, 13, 23
26, 197, 212213, 215216, 218,
246
specificational meaning, 2, 3238, 40,
4749, 5358, 6062, 80, 182, 243
244
value-variable relationship, 2, 32
33, 61, 80, 83, 117118, 125,
181
see also copula, be of identity
subjectification, 207, 217, 237, 241
th-clefts, 7, 10, 6264, 66, 6869, 72,
76, 9899, 107109, 135136, 145,
227228, 240n
there-clefts, 7576, 84, 104
token and type frequency, see usagebased approach
truncated clefts, 14, 7678, 94
underspecified it, 87, 129130, 226
228, 231, 242, 247
human and non-human, 155156,
164165, 167168, 177178,
plural and singular, 99101, 132
133, 163168, 177178, 246
uniqueness, see definiteness
usage-based approach, 10, 13, 16, 18,
2125, 178179, 212218
valency and relationality, 4041, 44,
46, 49, 57, 82
value-variable relationship, see
specificational meaning
verb-second, 222224
wh-clefts, 7, 10, 62, 6469, 101n, 107
108
categories of foci, 136, 138, 140,
225228, 230231
development of, 230, 232, 237238,
240242
discourse functions, 232233, 236
242
269
predicational, 127128, 226227

presentational, 232, 237242
Word Grammar, 19n
YCOE, 15, 150152, 154, 164, 170
172, 175, 176n, 180, 184189, 192,
194200, 204205, 246

It Cleft PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

It Cleft PDF

Uploaded by

Copyright:

Available Formats

The English It-Cleft

Topics in English Linguistics

Elizabeth Closs Traugott

The English It-Cleft

Chapter 1 Introduction and background. . . . . . . . . . . . . . . . . . . . . .

Chapter 2 A model of language structure and language change. . . .

Chapter 3 Specificational copular constructions. . . . . . . . . . . . . . . .

4.3. Other specificational copular constructions . . . . . . . . . . .

Chapter 4 It-clefts as specificational copular sentences. . . . . . . . . .

Chapter 5 Other varieties of it-cleft . . . . . . . . . . . . . . . . . . . . . . . . .

Chapter 6 The it-cleft and earlier periods of English . . . . . . . . . . . .

4. An obligatorily extraposed relative clause?. . . . . . . . . . . . . . . . . .

Chapter 7 The it-clefts development over time. . . . . . . . . . . . . . . .

Chapter 8 The it-cleft and constructional change. . . . . . . . . . . . . . .

Corpora and data sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

An outline of the project

[It] [was] [Frank] [that complained]

Introduction and background

It was FRANK that complained

A further property of the it-cleft is that it exhibits exhaustiveness effects.

It wasnt Frank that complained

Cleft sentences also have a specificational (or identifying) meaning. For

The one that complained was Frank

An outline of the project

Its in October that hes leaving

Another domain of variation relates to the information status of the

A: Did Max write the letter?

Introduction and background

It is a long road that has no turning

The road that has no turning is a long one

It was the Colonel [that __ survived]

It is a miracle [that he survived]

In this book, I provide answers to these questions by examining it-clefts

An overview of the literature on cleft sentences

how larger (multi-word) constructions undergo change. I discuss this aspect

An overview of the literature on cleft sentences

As I noted in Section 1, authors tend to view it-clefts either from the

The expletive approach

Introduction and background

while the remaining semantic content is backgrounded into a sentence-final

It was [[Frank]i [that ____i complained]]

[It was] Frank [that] complained

According to Jespersen, this analysis explains why it-clefts are used as a

An overview of the literature on cleft sentences

The extraposition approach

For others, it-clefts are considered foremost as specificational copular

Its grape soda that I like best

What I like best is grape soda

The thing that I like best is grape soda

My favourite drink is grape soda

Introduction and background

Other accounts that view it-clefts in relation to specificational copular

A constructional approach to it-clefts

That which he stole was money

That was money which (that) he stole

It was money which (that) he stole

Consequently, for these authors, there is little consistency as to the exact

A constructional approach to it-clefts

So which of these approaches is the better one? From the perspective of

Introduction and background

A constructional approach to it-clefts

Introduction and background

A diachronic approach to it-clefts