Appositive Relative Clauses in English

Studies in Discourse and Grammar (SiDaG)

Studies in Discourse and Grammar is a monograph series providing a forum
for research on grammar as it emerges from and is accounted for by discourse
contexts. The assumption underlying the series is that corpora reflecting
language as it is actually used are necessary, not only for the verification of
grammatical analyses, but also for understanding how the regularities we think
of as grammar emerge from communicative needs.

Research in discourse and grammar draws upon both spoken and written
corpora, and it is typically, though not necessarily, quantitative. Monographs in
the series propose explanations for grammatical regularities in terms of recurrent
discourse patterns, which reflect communicative needs, both informational and
socio-cultural.

Editor
Sandra A. Thompson
University of California at Santa Barbara
Department of Linguistics
Santa Barbara, CA 93106
USA

Volume 22
Appositive Relative Clauses in English. Discourse functions and competing
structures
by Rudy Loock
Appositive Relative Clauses
in English
Discourse functions and competing structures

Rudy Loock
University of Lille & STL laboratory, UMR 8163, CNRS

John Benjamins Publishing Company

Amsterdam/Philadelphia
 TM
The paper used in this publication meets the minimum requirements of
8

American National Standard for Information Sciences Permanence of

Paper for Printed Library Materials, ansi z39.48-1984.

Library of Congress Cataloging-in-Publication Data

Loock, Rudy.
Appositive relative clauses in English : discourse functions and competing structures /
by Rudy Loock.
p. cm. (Studies in Discourse and Grammar, issn 0928-8929 ; v. 22)
Includes bibliographical references and index.
1. Grammar, Comparative and general--Relative clauses. 2. Discourse analysis. 3. English
language--Semantics. I. Title.
P297.L66 2010
425--dc22 2010002102
isbn 978 90 272 2632 7 (Hb ; alk. paper)
isbn 978 90 272 8845 5 (Eb)

2010 John Benjamins B.V.

No part of this book may be reproduced in any form, by print, photoprint, microfilm, or any
other means, without written permission from the publisher.
John Benjamins Publishing Co. P.O. Box 36224 1020 me Amsterdam The Netherlands
John Benjamins North America P.O. Box 27519 Philadelphia pa 19118-0519 usa
This book is dedicated to the memory of my grandparents
Mireille and Camille Rusquart
 Table of contents

Acknowledgements xiii
Introduction 1

Part I. Appositive relative clauses (arcs): Definitions, previous studies,

typical and atypical syntactic realizations

chapter 1
Definitions and previous studies  7
1. The traditional appositive/determinative dichotomy 7
2. Distinctive criteria 8
2.1 A semantic/referential distinction 8
2.1.1 Explanation vs. determination 8
2.1.2 An independent speech act 10
2.1.3 Suppressibility vs. non suppressibility 11
2.1.4 Conclusion 12
2.2 The punctuation criterion 12
2.3 The morphosyntactic criteria 14
2.3.1 Relative pronoun 14
2.3.2 Antecedent 15
2.3.3 Stacking 17
2.3.4 Pied-piping 19
2.4 Other criteria 19
2.4.1 Idiomatic expressions 19
2.4.2 Quantifiers 20
2.4.3 Animals vs. humans? 20
2.5 The syntactic representation(s) of ARCs 21
2.5.1 The DRC: A unique constituent 21
2.5.2 The ARC: A unique constituent or orphanage? 22
2.5.3 The arguments in favor of the main clause/detachment
theory 24
2.5.4 Intermediate approaches 31
2.6 Conclusion on the DRC-ARC distinction 35
viii Appositive Relative Clauses in English

3. Problems 36
3.1 The limits of the distinction 36
3.2 The specific problem of DRCs with indefinite antecedents 37
3.3 An operational dichotomy? 40
3.3.1 Problems linked to the original binary opposition 40
3.3.2 A distinction irrelevant in other languages 40
3.4 Conclusion 47
4. Previous studies on the role of ARCs 47
4.1 ARCs as background information 48
4.1.1 Non-essential information? 48
4.1.2 The interaction between two propositional contents 48
4.1.3 An independent assertion 49
4.2 Previous attempts at defining taxonomies 49
5. Conclusion 52

chapter 2
Atypical appositive relative clauses 53
1. Starting point: Are you a good which or a bad which? 53
2. Definitions 53
2.1 Typical vs. atypical ARCs 53
2.2 Corpus 56
2.3 Description of the data 56
2.3.1 Gap-filled structures 56
2.3.2 Gapless structures 58
2.4 Dysfluencies vs. syntactic variation 59
2.5 Analysis 65
2.5.1 The role of the relative pronoun in gapless structures:
A plain connective? 65
2.5.2 Extension to gap-filled structures 67
2.5.3 The predominance of which 68
2.5.4 The Latin linking relative 68
2.5.5 Semantic and syntactic link with the main clause 69
2.5.6 A conversational strategy? 70
3. Conclusion 72
 Table of contents ix

Part ii. The discourse functions of ARCs: A suggested taxonomy

for written and spoken English

chapter 3
The corpus: ARCs in usage  77
1. The written corpus 77
1.1 Four sub-corpora 77
1.2 Morphosyntactic, semantic and pragmatic characteristics
of written ARCs 78
1.2.1 Punctuation 78
1.2.2 Morphosyntactic characteristics 79
1.2.3 Semantic characteristics 81
1.2.4 Pragmatic characteristics 84
2. The spoken corpus 87
2.1 Definitions 87
2.2 The corpus 88
2.3 Morphosyntactic, semantic and pragmatic characteristics
of spoken ARCs 90
2.4 Conclusion on the written/spoken distinction 92
3. Conclusion 93

chapter 4
The discourse functions of ARCs: A taxonomy 95
1. Introductory presentation 95
2. Continuative ARCs 95
2.1 Definition 95
2.2 Characteristics 101
2.2.1 Antecedent 102
2.2.2 Type and tense of verbs 102
2.2.3 Informational status of the elements 103
2.2.4 Suppressibility 103
3. Relevance ARCs 103
3.1 Relevance: A definition 104
3.2 Definition of relevance ARCs 105
3.3 The different discourse strategies 106
3.3.1 The leveling of the shared cognitive space 106
3.3.2 The legitimacy of the antecedent 109
3.3.3 Explanation, justification, concession: The implicit
inter-clausal link 111
3.3.4 Metatextual function: Relevance through transitions 115
 Appositive Relative Clauses in English

3.4 From relevance to politeness 117

3.5 Relevance vs. relevance 117
3.6 Characteristics 119
3.6.1 Antecedent and type of information in the ARC 119
3.6.2 Suppressibility 119
4. Subjectivity ARCs 120
4.1 Definitions 120
4.1.1 Explicit and implicit subjectivity 122
4.1.2 Polyphony and subjectivity ARCs 123
4.2 The different discourse strategies 125
4.2.1 Comment/judgment 125
4.2.2 Correction 126
4.2.3 Assessment and conclusion 127
4.3 Characteristics 127
4.3.1 Punctuation 127
4.3.2 Relative type 128
4.3.3 Antecedent 128
4.3.4 Suppressibility 128
5. Register variation: Spoken vs. written ARCs 129
5.1 Frequency 129
5.2 Discourse distribution 130
5.3 Specific (conversational) functions 130
5.3.1 Interpolation 131
5.3.2 Handing over to the addressee 133
6. The correlation between the discourse functions of ARCs and their
prosodic realizations 134
6.1 Introduction 134
6.2 The prosody of ARCs vs. the prosody of parentheticals
in general 135
6.3 The prosody of ARCs in relation to their functions in discourse 136
6.3.1 Results 136
6.3.2 Discussion: A (dis)continuity gradient? 137
7. Conclusion 138
 Table of contents xi

Part iii. ARCs and their competing allostructures

chapter 5
Definition of ARCs allostructures, with particular emphasis
on nominal appositives 143
1. Definitions 143
1.1 Information packaging 143
1.2 Allostructures: Definitions and methodology 146
1.3 Defining the allostructures of ARCs 147
2. Apposition: The natural allostructure of ARCs? 150
2.1 Apposition: An attempt at definition 151
2.2 Definitions 152
2.2.1 A secondary predication 152
2.2.2 Two types of semantic link 154
2.2.3 Co-reference: An illusion 155
2.2.4 Semantic vs. informational relation 155
2.2.5 Conclusion: The asymmetry of the two units 157
2.2.6 What kind of syntactic link? 157
2.3 Apposition and ARCs: Many similarities 161
2.4 A failed rendez-vous? 165
2.4.1 Apposition as abridged ARC 166
2.4.2 ARC as abridged apposition 168
3. Apposition and ARCs as allostructures 170
3.1 ARCs vs. simple nominal apposition 170
3.1.1 Identificational relation between the two units 171
3.1.2 Class membership relation between U1 and U2 176
3.2 ARCs vs. complex nominal apposition 177
3.2.1 Definitions 177
3.2.2 Presence/absence of the NP in U2: What constraints? 179
4. Conclusion 189

chapter 6
Appositive relative clauses and their other competing allostructures 191
1. What (other) allostructures for ARCs (alloARCs)? 191
2. Syntactic phenomena 194
3. Semantic and pragmatic phenomena 195
3.1 The hierarchization of information 196
3.1.1 Hot news vs. not so hot news information 196
3.1.2 Sentential parentheticals 198
3.1.3 The crucial role of intonation 199
xii Appositive Relative Clauses in English

3.1.4 The insertion of irrelevant information 201

3.1.5 Temporariness vs. elaboration 201
3.1.6 Continuative ARCs and ARCs with sentential
antecedents 202
3.2 Thematic progression: ARCs vs. preposing 206
3.3 The explicitness/implicitness of the link between
the information contents 208
4. Other influencing parameters 211
4.1 Register variation 211
4.2 Stylistic variation 212
5. Conclusion 214

Conclusion 217
References 219
Appendix 227
Index 231
 Acknowledgements

First of all, I would like to thank the CNRS laboratory Savoirs, Textes, Langage and
its current director, Professor Christian Berner, for providing me with excellent
working conditions during the writing of this book.
This work would not have been possible without the guidance of Philip H.
Miller, who led me down the path of research in linguistics. To him, I express my
most heartfelt gratitude.
Special thanks go also to Sandra A. Thompson, editor for the Studies in Dis-
course and Grammar series: her valuable comments on the first draft of the manu-
script led to significant improvements.
The research at the heart of this book has also been enriched by numerous ex-
changes and discussions with colleagues from various universities where my re-
search was presented, both at international conferences and laboratory seminars.
In particular, I would like to thank Doug Arnold, Anton Benz, Pierre Cotte, Ilse
Depraetere, Liliane Haegeman, Anne Jugnet, Mai Kuha, Peter Khnlein, Maarten
Lemmens, Emmanuelle Mathiot, Kathleen M. OConnor, Ellen F. Prince, Fayssal
Tayalati, Chrysanthie Therapontos and Danile Van de Velde. Special thanks go
to Cyril Auran of the joint research program on the discourse-prosody interface
that provided challenging material for my work.
I would also like to thank Chad Langford, who accepted the tedious task of
proofreading the first draft of the manuscript.
Finally, I would like to thank my family and friends for their unfailing sup-
port and encouragement.
 Introduction

Many books and journal articles have been published on relative clauses, treating
a wide range of languages and within the frameworks of many linguistic theories.
Most of them have focused on a now long and well-established dichotomy, between
restrictive/determinative relative clauses (henceforth in this book DRCs) and non-
restrictive/appositive relative clauses (henceforth ARCs). Linguists have provided
discriminating tests to distinguish between the two kinds of relatives, primarily
morphological (e.g. choice of relative pronoun) and syntactic (e.g. antecedent, link
with the main clause). Much of the description of ARCs has focused on their syn-
tactic status: they are considered either (i) syntactically integrated in the matrix
clause (integration theory, e.g. Jackendoff 1977) or (ii) integrated at discourse-level
only (e.g. Emonds 1979), hence not belonging to the syntax of the sentence. An-
other question raised is whether there is some kind of specific coordination rela-
tionship between the ARC and the main clause (DeVries 2002, 2006).
However, few studies have been specifically dedicated to the semantics and
especially the pragmatics of relative clauses, in other words to the functions of
relative clauses in discourse. Above all, the enormous amount of research on the
DRC/ARC dichotomy has led to the definition of an asymmetrical relationship
between the two kinds of relative clauses. While DRCs have been defined as re-
stricting their antecedent, that is to say playing a role in the construction of the
referent depicted by the antecedent, ARCs have been systematically defined as not
playing that role (hence the labels non-restrictive or non-defining for instance).
What this book aims to provide, therefore, is a positive definition of ARCs in dis-
course, hence the use of a positive term (appositive, instead of non-restrictive or
non-defining) to qualify these relatives. The choice of the label appositive is also
justified by the comparison we establish with appositives in general.

Organization of the book

The book is organized as follows. Part I is dedicated to definitions of the struc-

ture, from a morphological and syntactic point of view. Problems regarding
 Appositive Relative Clauses in English

the long-established DRC/ARC dichotomy are subsequently dealt with so as to

clearly define the category of relatives studied here. While the book deals with
ARCs in the English language, other languages, both Indo-European and non-
Indo-European, are considered in order to shed light on a dichotomy generally
presented as being systematic. Data from French, Spanish, German, Persian,
Arabic, Chinese and Middle Egyptian are provided. In Part I, special emphasis
is also paid to relative clauses whose syntactic configurations deviate from the
norm that is expected for them, i.e. gap-filled and gapless relative clauses, which
we patently refuse to treat as plain performance errors. Part II provides a study of
the discourse functions of ARCs, based on written and spoken data. A taxonomy
is presented, which distinguishes between 3 main categories: (i) relevance ARCs,
(ii) subjectivity ARCs and (iii) continuative ARCs. Each category is illustrated
with morphological, syntactic, semantic and pragmatic criteria. A study of the
prosodic realization of ARCs from each category is also provided for the ARCs
collected from spoken corpora. Part III is meant to complement Part II: after
defining what speakers use ARCs for in discourse, we also provide constraints
accounting for speakers choice for ARCs at the expense of other syntactic struc-
tures that could fulfill related discourse functions (relevance, subjectivity, conti-
nuity). Examples of such competing structures (which we label allostructures)
are sentential parentheticals, nominal appositives, independent juxtaposed
clauses or noun premodifiers among others.

Theoretical background

This study relies on attested data, systematically studied alongside the general
context from which it is extracted. The research at the origin of this book is there-
fore corpus-based. Very few examples in this book have been constructed for the
purpose of our demonstration, with the exception of syntactic discussions (Chap-
ter 1), which require no contextualization of the data. However, many original ex-
amples have been manipulated and modified to illustrate constraints that underlie
the pragmatic phenomena discussed within our analyses. Special emphasis has
been put on register variation: our examples come from written and spoken dis-
course, but finer-grained distinctions are made, such as the distinction between
different written registers (quality press, tabloid press, fiction scientific prose).
Our data comes from a 600-utterance, self-collected corpus of ARCs (to which
about 200 examples can be added for ARCs allostructures) but also from existing
electronic corpora such as The British National Corpus (BNC), The ICE-GB Cor-
pus, or The Aix-MARSEC corpus.
 Introduction 

As far as pragmatic concepts are concerned, our study falls under the scope
of discourse pragmatics. In particular, we rely on Princes (1981, 1992) taxonomy
of given/new information. Her taxonomy has been modified to fit our purpose,
which is to study the informational status of the semantic contents of the main
clause, the antecedent and the ARC. In particular we took into account the het-
erogeneity of intended addressees states of knowledge. Such a systematic study
has not been achieved before, even by the researchers whose aim was to provide a
pragmatics-based classification of ARCs (e.g. Cornilescu 1981). This is due to the
fact that researchers have worked primarily on constructed examples. Even recent
studies, such as Potts (2005), which defines ARCs (or supplementary relatives,
as he prefers to label them) as conventional implicatures, rely on constructed
data. This leads him to formulate erroneous generalities such as supplements [i.e.
ARCs but also appositive structures in general] are used to introduce new informa-
tion, often as a means of further clarifying and contextualizing the at-issue content
around them (Potts 2005:112). We will show that such statements are false: the
information conveyed by ARCs is not necessarily new nor is it necessarily old,
for that matter.
Also, as our aim is not only to analyze the discourse functions of ARCs, but
also to explain why speakers resort to this particular syntactic structure instead
of other competing structures (allostructures), our study naturally finds its place
within the information-packaging theory, as defined by Vallduv (1992). Along-
side studies within this theoretical framework, we consider that speakers have
different (syntactic) means to convey the same informational content and that
their choices are governed by specific, semantic-pragmatic constraints. Such an
approach is meant to account for why speakers use ARCs instead of other pos-
sible syntactic structures and is therefore meant to complement the taxonomy of
discourse functions that we suggest.
part i

Appositive relative clauses (arcs)

Definitions, previous studies, typical
and atypical syntactic realizations
 chapter 1

Definitions and previous studies

1. The traditional appositive/determinative dichotomy

Linguists traditionally distinguish two kinds of relative clauses:

(1) The Americans who are fascinated by space science are looking forward to
celebrating the 40th anniversary of the first manned mission to the surface of
the moon.
(2) The Americans, who are fascinated by space science, are looking forward to
celebrating the 40th anniversary of the first manned mission to the surface of
the moon.

In the first example, a subcategory of American people is created, i.e. only those
who are fascinated by space science. In the second example, the whole category of
Americans is concerned by the predicate be looking forward to celebrating the 40th
anniversary of the first manned mission to the surface of the moon. The ARC speci-
fies, incidentally, that the whole category is fascinated by space science.
This traditional opposition (1) vs. (2), which evaluates the narrowness of the
link between the relative clause and the antecedent (Cotte 2000:47, my transla-
tion), has been given many different labels, perhaps too many: determinative vs.
appositive (Malan 1999), determinative vs. descriptive (Souesme 1992), restric-
tive vs. appositive (Kleiber 1987; Khalifa 1999), defining vs. non-defining (Fowler
& Fowler 1949), attached vs. detached (Mlis 2000), restrictive vs. non-restric-
tive (Quirk et al. 1985; McCawley 1988; Fabb 1990; Biber et al. 1999; Huddleston
1984), integrated vs. supplementary (Huddleston & Pullum 2002; Potts 2005). In
this book, we use the terms determinative relative clauses (DRCs) and appositive
relative clauses (ARCs). The traditional opposition is sometimes controversial and
has been rejected by some linguists in favor of a gradient (Khalifa 1999) or contin-
uum (Mlis 1999, 2000), with DRCs and ARCs representing the two extremities.
We will not enter the debate here but rather will accept the traditional dichotomy
as it is, even though we are aware of the arguments against such a conception. This
problem is nevertheless addressed briefly at the end of Chapter 1 (Section 5), as
it has to be dealt with to provide an accurate definition of the object of this book.
 Appositive Relative Clauses in English

A number of differences, both semantic and morphosyntactic, legitimizes the

existence of the two sub-categories of relative clauses. Below, we list and discuss
them thoroughly, starting with the semantic criterion, given that the distinction
was originally defined by logicians in semantic terms. We then discuss the punc-
tuation criterion and the different morphosyntactic criteria used to distinguish
ARCs from DRCs. In spite of the abundant literature on the topic, the question
has not yet been given a definitive answer, as many occurrences, in particular rela-
tive clauses with indefinite antecedents (discussed in 3.2), pose genuine classifica-
tion problems. As our study deals exclusively with ARCs, we decided to put aside
occurrences which are problematic as far as their DRC/ARC status is concerned.
Above all, we aim to escape any asymmetrical definition of the role of ARCs in
discourse and thus voluntarily put problematic cases aside so as to escape the
endlessly thorny debate focusing on the DRC/ARC dichotomy and to provide
in the end a positive definition of the structure through the study of prototypical
yet attested examples. ARCs have mostly been studied in opposition to DRCs (cf.
labels like non restrictive or non defining), which may explain why this structure
has not sufficiently been studied in its own right. Finally, we discuss the idea of
the existence of two uses of relative clauses, instead of two different kinds, which
is suggested by the observation of other languages (Section 3.3).

2. Distinctive criteria

2.1 A semantic/referential distinction

2.1.1 Explanation vs. determination

The DRC-ARC distinction is actually much older than one might think, which ex-
plains the influence it still has today. It appeared as early as the 12th century with
the work of logicians, before being discussed in detail in the work of Port-Royal
(1662). It was in studying problems of co-referentiality that logicians brought
out the double role of relative clauses. In their Grammaire gnrale et raisonne
de Port Royal, Arnauld & Lancelot (1993:193197) drew a distinction between
French relative pronouns explaining qui and determinative qui. For them, a rela-
tive clause could be added for two purposes:

i. A plain explanation: the addition [of the RC] does not change anything to
the idea of the term, because what is added can be applied to it as a whole
(1993:193, my translation). The addition only develops what was enclosed in
the understanding of the idea of the first term, or at least what corresponds to
it as one of its accidents, provided it can be applied generally and completely.
(1993:227, my translation). An illustration of this subcategory is given in (3):
 Chapter 1. Definitions and previous studies 

(3) Les hommes, qui sont crs pour connatre et pour aimer Dieu
Men, who are created to know and to love God

ii. determinations: when what is added to a term does not apply to the term
as a whole but restricts and determines its meaning. In such conditions, it
[the term] is not to be taken as a general word in all its meaning, but only as a
part of this meaning (1993:228, my translation). They provide the following
example:
(4) Les hommes qui sont pieux
The men who are pious

It is therefore in terms of semantic contribution to a head noun that the distinction

was first defined, with explanations corresponding to ARCs and determinations
to DRCs. Arnault & Lancelots distinction concerns the semantic contribution of
the RC to what is traditionally called the antecedent. In the case of determina-
tions, the content of the RC takes part in the referential act that concerns the
NP in which the RC appears. It therefore restricts the head noun of the NP. On
the other hand, in the case of explanations, the content of the RC is not part of
the referential act concerning the NP, as the idea of the term, to use Arnauld &
Lancelots terminology, is not modified. An ARC therefore does not intervene in
the construction of the NP referent as a whole, but represents independent infor-
mational content and brings supplementary information about a referent that is
already determined otherwise.
It is crucial, then, to determine the kind of referential act an RC is involved
in. Arnauld & Lancelot mention the idea of the term to which the RC is added,
which is traditionally defined as the antecedent of the relative pronoun. Hence
the idea that a DRC restricts the antecedent while an ARC does not. But defining
the antecedent of the relative pronoun is actually more problematic than it seems
at first sight. In the case of a DRC, it is generally considered to be the NP that
precedes the relative pronoun with which it is co-referential. By taking a closer
look, though, one realizes that the notion of antecedent is more problematic when
considered in terms of referential construction. Consider the following example:

(5) The man that I saw at the party last night is John Sanders.

Traditional and pedagogical grammars consider that the NP the man is the ante-
cedent of the relative pronoun that and that the two form a more complex NP, as
the content of the RC (which is then determinative) is crucial for the construction
of the referent of the individual, corresponding to a determination, to use Arnauld
& Lancelots terminology. However, if we take a closer look at the referential con-
struction of the whole NP, it is clear that the noun man is modified first by the
10 Appositive Relative Clauses in English

RC, which restricts the referent. It is only then that determination through the
choice of the article/determiner intervenes and establishes the uniquely identifi-
able character of the individual (cf. Hawkins 1978). As a consequence, should we
say that the antecedent of the relative pronoun is man or the man? In the same
way, if we go back to our very first example in this chapter, it is only once the con-
tent of the RC has been applied to the head noun Americans that the determiner
the makes this subcategory of American people uniquely identifiable.
As a consequence, it seems safer to say that the antecedent of the relative pro-
noun is Americans in (1) and man in (5). Rather than saying that DRCs restrict
the referent of the antecedent, it is more appropriate to say that a DRC contributes
to, or plays a role in, the semantic construction of the referent depicted by the NP
in which it appears. On the contrary, an ARC does not take part in such referen-
tial construction but represents a secondary assertion about a referent that has
been semantically/referentially defined otherwise, independently of the embed-
ded RC. This is why proper names, which do not require the presence of any other
markers to refer to an entity that is immediately retrievable, can only be modified
by an ARC (with the exception of what Quirk et al. (1985:1257) call telescoped
relative clauses: The John I like is the John who brings me flowers.).

2.1.2 An independent speech act

Linguists have observed that certain sentential adverbs and connectives appear
only in independent clauses. ARCs seem to display the same behavior in this re-
spect, contrary to DRCs (Emonds 1979:239240):
(6) The boys, who have frankly lost their case, should give up. (= ARC)
(7) *The boys that have frankly lost their case should give up. (= DRC)

The behavior of parenthetical adverbials also highlights the distinction between

ARCs and DRCs:

(8) We persuaded Bill, who is a good person, most people agree, to leave immedi-
ately. (= ARC)
(9) *We persuaded anybody that has a good record, most people agree, to stay on
indefinitely. (= DRC)

We study this phenomenon in more detail below, but at this stage, the question is
to be considered within the general issue of recognizing an independent language
act in the ARC as opposed to the rest of the sentence (main clause, or MC for

. For a thorough analysis of the types of adverbs that can be found in ARCs (and nominal
appositives), see OConnor (2008).
 Chapter 1. Definitions and previous studies 11

short). As noted by Thorne (1972) and repeated in Potts (2005:127), some utter-
ance-modifying adverbs can appear in ARCs, all the while resisting true syntactic
embedding.
Also, as an independent predication, an ARC can contain a question-tag (10),
but a DRC cannot (10a). In the same way, an ARC is not under the scope of an
interrogation (11) or an injunction (12). The first pair of examples comes from
Khalifa (1999:12), the following two examples from McCawley (1988:445):
(10) Marcia, who you wanted to meet, didnt you?, has just arrived. (= ARC)
a. *The girl you wanted to meet, didnt you? has just arrived. (= DRC)
(11) Has John, who was talking to Mary a minute ago, gone home? (= ARC)
(12) Put the turkey, which is in the refrigerator, into the oven. (= ARC)

The speech act in the ARC is therefore independent of the speech act in the rest
of the sentence. As noted by McCawley (1988), these sentences contain two lan-
guage acts and can be rephrased with two sentences:

(10) b. Marcia has just arrived. You wanted to meet her, didnt you?
(11) a. Has John gone home? He was talking to Mary a minute ago.
(12) a. Put the turkey in the oven. Its in the refrigerator.

In the same way, an ARC can contain a performative verb, but a DRC generally
cannot:

(13) I bought a new DVD player, which I promise to lend you if you need it.
(= ARC)
(14) Welcome to the meeting, which I now declare open. (= ARC)
(15) Theatre of Blood, which I promise not to reveal the ending of, was slightly too
gory for me. (Davison 1975) (=ARC)
(13) a. ??The DVD player I promise to lend you next week is new. (= DRC)
(14) a. ??The meeting I now declare open will be a success. (= DRC)

ARCs, as opposed to regular syntactic embedding, do not cause performatives

to lose their performative status (Potts 2005:127). The independence of ARCs in
terms of speech act is therefore clear and is to be related to the explanation role
depicted by Arnauld & Lancelot, as opposed to the determination role of DRCs.

2.1.3 Suppressibility vs. non suppressibility

This traditional distinctive criterion is a direct consequence of the contribution
(or lack thereof) of the RC to the construction of the reference of the NP within
which an RC appears. If in the case of an ARC, the so-called antecedent can be
12 Appositive Relative Clauses in English

used alone in the sentence, then the RC is not necessary to establish the reference
of the complex NP that includes the RC. Hence the idea that an ARC is suppress-
ible, while a DRC is not. Souesme (1992:273, my translation) indeed claims that
an ARC provides just a precision that is absolutely unnecessary to the acceptability
and understanding of the utterance. Huddleston & Pullum (2002:1064) mention
an essential component of matrix message for DRCs and label ARCs supplemen-
tary, as if they were nothing but superfluous additions. This property has become
classic in the definition of the two different kinds of RCs and is found in many
pedagogical grammars, presented perhaps in too systematic a way. We will show
here that, although ARCs are not necessary for reference purposes, this does not
mean that they are not necessary to the understanding of the utterances in which
they appear. Their contribution to the definition of the antecedent is essential,
although not from a purely referential point of view.

2.1.4 Conclusion
The primary criterion for distinguishing ARCs and DRCs was therefore seman-
tico-referential. This opposition then had to be translated in morphosyntactic
terms, so as to create tools to support the fundamental semantic distinction. These
different criteria are listed and analyzed below, after a brief discussion on the role
of punctuation in the written register, as this criterion has, along with suppress-
ibility, become traditional to distinguish the two kinds of RCs.

2.2 The punctuation criterion

The fact that an RC may or may not contribute to a referential act leads to a dis-
tinctive criterion that has become (perhaps too) systematic to describe RCs in the
written register: the presence of punctuation traditionally a comma as being
the indication that an RC is appositive. On the contrary, a DRC is not marked
out with any punctuation. The presence between the antecedent and the relative
pronoun of punctuation such as a comma, a bracket, a parenthesis, a dash or even
a period in the case of sentential relatives, is treated as a characteristic of ARCs in
the written register, a transposition of the oral register, where ARCs are (suppos-
edly) marked off prosodically from the rest of the sentence.
However, as has been shown by many authors (Fowler & Fowler 1949;
Souesme 1992; Kleiber 1987; Malan 1999; Biber et al. 1999; Huddleston & Pullum
2002 inter alia), punctuation is not a test that can be systematically applied. While
its presence is a clue that is almost systematically reliable, its absence is absolutely
not significant. In other words, a relative clause without any punctuation between
the antecedent and the relative pronoun is not necessarily determinative, as is
 Chapter 1. Definitions and previous studies 13

shown in the following examples, where a determinative interpretation is clearly

impossible:

(16) Anthony Seaton () led a study of children in Saudi Arabia where rural com-
munities live in stark contrast to those in Westernised Jeddah.
(17) Currie who is understood to have secured a good six-figure sum for her diaries
from the publishers Little, Brown, revealed this weekend that she and Major had
had an affair between 1984 and 1988 when she was a backbench MP and he
was a government whip.

The relative clause follows the proper names of a country, Saudi Arabia, in (16),
and a person, Currie, in (17), proper names that are by definition referentially
autonomous and stabilized without the embedding of the RC. The RC is not nec-
essary to the identification of the referent and has to be interpreted as an ARC.
Yet, no punctuation mark is inserted between the antecedent and the pronoun.
Should we consider that it should have been there? Or should we conclude that it is
not a true appositive? And what about the following two sentences, found within
the same article, one with and the other without punctuation?

(18) There was no dispute, the prosecution said, that Abrams, who is from Liverpool,
carried out the attack on December 30 last year.
(19) The jury was told how Harrison and his wife Olivia fought with Michael Abrams
who was armed with a knife and part of a stone sword broken from a statue of
St George and the Dragon at the couples Oxfordshire mansion.

In particular, we noted the absence of punctuation with proper name anteced-

ents, for which no ambiguity is possible between a determinative and an apposi-
tive interpretation. This observation is similar to that made by McDavid (1977),
who observed the same phenomenon in his corpus study. It is also interesting to
note that in such examples, not only is the antecedent a proper name, but it is
also not its first mention in the discourse, therefore being more accessible to the
addressee.
In addition, ARCs whose antecedents are referentially autonomous and are
not separated from the relative pronoun by punctuation cover plural indefinites
with generic interpretation, as in (20) below. An obvious parallel is drawn between
two distinct groups of people, depicted by two NPs which, if we follow the punc-
tuation test, are modified by an ARC for one of them and a DRC for the other:

(20) The announcement was greeted with delight by biomedical scientists, who see
stem cells as a weapon against cancer, spinal injury, diabetes and Parkinsons
disease, and biotechnology companies who are anxious to begin the long process
of turning theoretical possibilities into practical treatments.
14 Appositive Relative Clauses in English

Therefore, the absence of a comma does not appear to be a reliable criterion. Its
presence, on the other hand, seems to be a satisfactory criterion to ensure the ap-
positive status of an RC. Only rarely is a DRC separated from the antecedent by
a comma. This is, however, possible with the insertion of parentheticals, for in-
stance; here, the presence of commas has no influence on the status of the RCs:

(21) Sure, there were right-on people, men and women, who would be repelled and
appalled by his logic, but that was fine by him.
(22) Karen tried to do what she was told, although sometimes she made mistakes;
and there were other things, like the sleepwalking, that she couldnt help.

In (21) the antecedent is the NP right-on people, which is separated from the DRC
who would be repelled and appalled by his logic by the NP men and women, which
corresponds to the second unit of an appositive construction. In (22) the anteced-
ent is not sleepwalking in which case the RC would be an ARC introduced by
that but other things, making the RC a DRC.
Fowler & Fowler (1949:86) provide a good summary of the punctuation cri-
terion in ARCs: Punctuation is a test that would not always be applicable even
if all writers could be assumed to punctuate correctly; but it is often a guide to the
writers intention. Consequently, this criterion has to be approached with some
caution. Nevertheless, as the presence of punctuation is an almost systematically
reliable test, we used this characteristic to determine the determinative/appositive
status of the RCs in our corpus, provided no ARC is separated from its antecedent
by punctuation provided one analyzes correctly examples like (21) or (22).

2.3 The morphosyntactic criteria

2.3.1 Relative pronoun

Prototypically, a DRC can be introduced by a wh- pronoun, that, or , depend-
ing on the syntactic function of the pronoun and the [+ HUMAN] / [ HUMAN]
status of the antecedent. On the other hand, the pronoun never introduces an
ARC. The pronoun that introduces an ARC only very rarely: our corpus con-
tains only a few examples, sometimes of debatable status. A corpus compiled by
Khalifa (1999) shows that 1.8% of ARCs are introduced by that, a possibility that
is generally considered as exceptional. Huddleston & Pullum (2002:1052) give a
few attested examples:

. The status of that as relative pronoun or complementizer is debatable. Some linguists in-
deed consider that that, which is not declinable depending on its syntactic function in the RC,
is in fact a complementizer, the main argument being the impossibility of having a preposition
 Chapter 1. Definitions and previous studies 15

(23) The patas monkey, that spends almost all of its time in open grassland, adopts
just such tactics.
(24) His heart, that had lifted at the sight of Joanna, had become suddenly heavy at
the sight of Ramadez thumping after her.
(25) February, that in other years held intimations of spring, this year prolonged the
bitter weather.
(26) She had long been accustomed to the solitary nature of her sons instincts, that I
had tried and failed to stifle.

Therefore, as is summed up by Huddleston (1984:401), non-restrictives are more

or less limited to the wh class. A corpus study led by McDavid (1977:80) has
shown that out of a 5,556-relative corpus, among which 1,242 are introduced by
that, none corresponds to an ARC introduced by this pronoun. For Cotte (1997),
who works within the French, utterer-centered, theory of enunciative operations,
this rarity can be accounted for by the inadequacy between, on the one hand, the
fact that that marks that the determination through the use of the RC is anterior
to another determination (i.e. that on the left side of the noun) and on the other
hand the fact that the ARC is the expression of a commentary upon a first real-
ity that is perfectly identified (Cotte 1997:284, my translation). Interestingly, this
distinction is reminiscent of that in the Port-Royal Grammar between determi-
native qui and explaining qui.

2.3.2 Antecedent
The syntactic categories to which what is traditionally called the antecedent of
the relative pronoun (cf. 2.3.1) belongs are more varied in the case of ARCs than
DRCs. Indeed, the pronoun of a DRC stands for an antecedent that is necessarily
nominal, while the pronoun of an ARC stands for an antecedent that can represent

before that (*of that, *in that), contrary to which (cf. e.g. Van der Auwera 1985). Others also treat
that as a non pronominal, invariable relativizer. We shall not discuss the issue further.
. While an ARC is traditionally introduced by a wh- pronoun, one must nevertheless note
the intriguing impossibility of using the relative pronoun why in ARCs, while this use is per-
fectly acceptable and common in DRCs (Huddleston 1984):
(i) The reason why he did it is not clear.
a. *This excellent reason, why he did it, was known.
On the other hand, which can also be used as what is called a relative determiner in an ARC,
but not in a DRC:
(ii) He bought me the new edition of Brave New World, which book I actually never read.
(= ARC)
a. *He gave me a novel which book he had never read. (= DRC)
16 Appositive Relative Clauses in English

an NP of course, but also a VP (27), a PP (28), an AdvP (29), AdjP (30), a whole
sentence (31), or even a whole paragraph (32), in which cases the antecedent is a
sentential antecedent.

(27) He went to Paris, which Mary never did. (VP)

(28) I put the book on the table, where Mary found it later that day. (PP)
(29) She reacted remarkably quickly, which is the way people do things when they like
their job. (AdvP)
(30) She is really helpful, which you should be when you are a social worker.
(AdjP)
(31) Tennis is possibly the least popular sport at the Olympics, which is quite some-
thing when it is up against Greco-Roman wrestling and walking. (S)
(32) The consequences of Americas obsession with domestic television ratings is
that the site for the Olympics remains at the mercy of NBC what Ebersol, to
his credit, calls the quintessential example of ugly Americanism. And if youre
wondering why they wanted to come to Sydney, they didnt. The only reason
the Games are here is that, in the aftermath of the 1989 Tiananmen Square
uprising, the IOC had to postpone their long-term plan to take the Olympics
to China. That would have suited NBC and the corporate sponsors who cant
wait to get their hands on the Chinese market. Sydney was considered a viable
option, facilities-wise, and won the next vote in 1993 but Australia, with fewer
than 20 million people, does not provide the sort of selling bonanza that China
would have done. As the AFR puts it, they would love to go to Beijing to foster
the spirit of Olympic marketing.
Which brings us to Athens. For several weeks now, rumours have grown that the
city will not be able to meet infrastructure and financial deadlines for the next
Games which could be switched to Seoul, where time is on the Americans side,
not to mention old business partners. (whole paragraph)

Also, while DRCs and ARCs allow the presence of an indefinite antecedent, in
ARCs the indefinite antecedent must have either a generic interpretation or a pre-
supposed interpretation, while in DRCs no such restriction exists. Example (33)
is therefore ungrammatical as the antecedent is neither generic nor presupposed:

(33) *I met a man, who was going to St. Ives. (OConnor 2008:260)

This specificity constraint also explains why the antecedent of an ARC cannot
contain a negative quantifier such as nobody or nowhere (De Vries 2002; Fabb
1990; OConnor 2008 inter alia), while a DRC can:
 Chapter 1. Definitions and previous studies 17

(34) *Nobody, who came to the party, had a good time. (= ARC)
(35) Nobody who came to the party had a good time. (= DRC)

2.3.3 Stacking
Linguists have called stacking the possibility of having two (or more) relative
clauses with the same antecedent in succession. It is traditionally considered that
it is impossible to stack two ARCs with the same antecedent, contrary to DRCs,
which can be stacked (McCawley 1988:419; Huddleston & Pullum 2002:1066 in-
ter alia). However, our corpus shows that stacked ARCs do exist with two or even
three successive ARCs for the same antecedent:

(36) Everyone was exceedingly sorry for Lady Marchmain, whose brothers names
stood in letters of gold on the war memorial, whose brothers memory was fresh
in many breasts.
(37) Others like David Shaw, who is fighting the Liberal Democrat seat of Kingston
(majority 56), who did not pose with Lady Thatcher, said: This view of William
Hague is what people said about Lady Thatcher in 1979.
(38) I had thought it all very suburban and up to date, now, under the stars, in the
walled city, whose streets were gentle, dusty stairways, and whose walls rose win-
dowless on either side, closed overhead, then opened again to the stars; where the
dust lay thick among the smooth paving stones and figures passed silently, robed
in white, on softer slippers or hard, bare soles; where the air was scented with
cloves and in sense and wood smoke now I know what had drawn Sebastian
here and held him so long.

Contrary to a widely-accepted claim (see De Vries 2002:197), ARCs can be stacked,

like DRCs. Novelist Graham Swift even used this as a central linguistic charac-
teristic of the narrator in his novel Waterland. The stacking of RCs is extremely
frequent in the very first chapter of the novel, whether the RC is introduced by a
coordinator or not. Most of the time, they represent independent sentences, the
antecedent a lock-keeper being in the end far from the relative pronouns:

(39) And by the Leem lived a lock-keeper. (1) Who was my father. (2) Who was a
phlegmatic yet sentimental man. (3) Who told me, when I was even younger
than you, that there was no one walking the world that hadnt once sucked
And that the stars (4) Who was wounded at the third battle of Ypres. And
had a brother killed in the same battle. (5) Who when asked about his memo-
ries of the War, would invariably reply that he remembered nothing. (5bis) Yet
who when he was not asked would sometimes recount bizarre anecdotes of
those immemorial trenches and mudscapes, as if speaking of things remote and
fantastical in which his involvement was purely speculative. How, for example,
18 Appositive Relative Clauses in English

the Flanders eels, countless numbers of which had forever made their abode
in those watery and low-lying regions, undeterred by the cataclysmic conflict
that was devastating their haunts, found their way into flooded saps and even
into shell craters, where there was no shortage of well-ripened food (6) Who
trapped eels himself in his native Fens. (7) Who showed me as a boy all the vari-
ous ways of cooking eels poached in vinegar and water; in a white sauce; in a
green sauce; in pies; in a stew with onion and celery; jellied, with horseradish;
chopped, skewered and roasted on an open fire and so I became just as partial
as he to their subtle and versatile flesh. And so did my brother. But my mother,
Fenwoman though she was and far from squeamish, could not abide them. She
would scream if she saw a not quite dead eel begin to slither on the kitchen table
(8) Who when he returned from the Great War in 1918, not only wounded in
the knee but profoundly dazed in the mind, was shunted for four years from this
hospital to that. But was dispatched in due course to Kessling Hall, until recently
country mansion of the Atkinson family, but now converted as a convalescent
home for war invalids. (9) Who spent many weeks in the spring and summer of
1922, sitting on the tree-girt and secluded lawns of that curative establishment
amongst several others be-scarred, be crutched and be-patched-up victims, all
of whom in that scene of apparent tranquillity (and four years after the guns
had stopped) were desperately attempting to find their peace-time bearings.
(10) Who fell in love with one of the nurses. (11) Who came home from the
war, a wounded soldier, and married the nurse who nursed him back to health.
A story-book romance. (12) Who, delivered from the holocaust, could scarcely
believe that this enchanted chapter of events was happening to him. (13) Whose
love was returned with surprising readiness. (14) Who married, in August
1922, this woman whom for several weeks his numbed brain had registered
only as nurse, brunette (14bis) and who even after his return to lucidity and
notwithstanding their growing mutual affection was reluctant to disclose her
name. (15) Who discovered only after a while that this white-aproned, war-
volunteer, now regular nurse, who was familiar in more ways than one with
Kessling Hall, was the daughter of a well known indeed notorious and come-
down-in-the-world brewer.
(16) Who through the mediation of this woman (her fathers residual influence
with the then still extant Leem Drainage and Navigation Board) acquired
the post of keeper of the New Atkinson Lock and Sluice. (17) Who learnt, so
it seemed to the boy who used to go with him to trap eels, to find both solace
and mysterious, never quite suppressed vexation in the situation:fixed home,
flowing river, flat land; beautiful wife. (18) Who became the father of two sons
(born 1923 and 1927), the first of whom turned out to be a semi-moron who
loved his motor-cycle.
 Chapter 1. Definitions and previous studies 19

Examples do exist, even though they are generally typical of the literary genre
the examples from Waterland clearly result from stylistic purposes. The stacking
of RCs is therefore a non-exclusive difference: ARCs seem to resist the phenome-
non, but the possibility exists. It is even sometimes acknowledged by linguists like
Emonds (1979), De Vries (2002, 2006), Potts (2005) (the latter qualifying stacked
ARCs as unexceptionable (Potts 2005:101)), or Cinque (2008), even though this
position remains in the minority.

2.3.4 Pied-piping
The term pied-piping describes a syntactic phenomenon in relative and interroga-
tive clauses concerning wh-movement. Consider the following sentences and
the two possibilities speakers have to form a relative clause or an interrogative
clause:

(40) John talked to a man who lives in New York.

a. The man who/that/ John talked to lives in New York.
a. The man to whom John talked lives in New York.
b. Who did John talk to?
b. To whom did John talk?

Sentences (a) and (b) illustrate preposition stranding (still condemned by some
prescriptivists never end a sentence with a preposition and yet very frequent
and natural) while sentences (a) and (b) illustrate pied-piping: the preposition
is not left stranded at the end of the clause but is brought along with the wh-
pronoun, hence the term pied-piping in reference to The Pied Piper of Hamelin.
Pied-piping is generally said to be more frequent in ARCs than DRCs. However,
this still needs to be confirmed, as the fact that DRCs can be introduced by a
non-wh- pronoun (that in particular), which does not authorize pied-piping (*to
that), necessarily entails more frequent preposition stranding in DRCs.

2.4 Other criteria

In this short section we would like to mention and discuss other distinctive crite-
ria mentioned in the literature to distinguish DRCs and ARCs.

2.4.1 Idiomatic expressions

Idiomatic expressions and collocations do not have the same distribution in ARCs
and DRCs. Indeed, while they are perfectly acceptable with a DRC, this is not the
case with an ARC (Emonds 1979:231):
20 Appositive Relative Clauses in English

(41) The headway that the students made was phenomenal. (= DRC)
(42) *That headway, which the students made last week, was phenomenal.
(= ARC)
(43) I will never forget the fun they made of me. (= DRC)
(44) *The fun, which they made of me, was delightful. (= ARC)

This can be accounted for easily by the fact that the NPs that headway in (42) and
the fun in (44), represent only fragments of idiomatic expressions. In such condi-
tions, the nouns headway and fun have no referent, and the NPs have no referen-
tial autonomy, and are thus incompatible with the embedding of an ARC.

2.4.2 Quantifiers
Many linguists have observed that quantifiers like any, no or every for instance do
not display the same behavior in the presence of a DRC as in the presence of an
ARC, as is illustrated by the following examples:

(45) Only the tourists who have any imagination go to visit Sicily. (= DRC)
(46) *Only the tourists, who have any imagination, go to visit Sicily. (= ARC)
(47) Every student who has a book should bring it to class. (= DRC)
(48) Every student, *who has a book/*who have their books, should bring it/them to
class. (= ARC)
(49) No candidate who scored 40% or more was ever failed. (= DRC)
(50) *No candidate, who scored 40% or more, was ever failed. (= ARC)

This traditional discriminating criterion requires however further discussion (see

Section 2.5.4.1).

2.4.3 Animals vs. humans?

According to Huddleston (1984:394), the choice of the relative pronoun with
nouns referring to animals is different depending on the status of the RC. In
DRCs, the relative pronoun who concerns humans only, while in ARCs it can be
used for both humans and animals. However, any quick corpus search using the
internet shows that who is frequently used with nouns referring to animals inde-
pendently of the status of the RC. This criterion is thus irrelevant for discriminat-
ing between DRCs and ARCs.

. The last pair of examples is taken from Huddleston & Pullum (2002:1060).
. We tested both pets (dog, cat, fish, sheep, bird) and other animals (deer, rhino, giraffe,
leopard).
 Chapter 1. Definitions and previous studies 21

2.5 The syntactic representation(s) of ARCs

Syntacticians, Chomsky being among the first of them, have of course been very
interested in the difference between the two kinds of RCs, wondering whether
the two sub-categories as defined by logicians are syntactically on the same level,
forming a whole constituent with the antecedent or not. If DRCs are less prob-
lematic for their representation there is a general consensus on the question
ARCs remain a thorny question. Quite a few theories are in conflict, ranging
from full integration to radical orphanage, as defined by Haegeman (1991) for
parenthetical adverbials. Below we expose and discuss briefly the different theo-
ries. This question is of particular interest here for two reasons: (i) the analysis of
what we label atypical RCs, found in our spoken corpus (see Chapter 2) and also
(ii) the syntax-discourse link we aim at establishing in this study, i.e. the correla-
tion between the discursive functions we define here and the syntactic status of
the clause, in particular for what we label continuative relative clauses.

2.5.1 The DRC: A unique constituent

On the level of syntactic representation, it seems to go without saying that a DRC
and its antecedent form one single constituent. The relative clause occurs post-
nominally in the referential definition of the NP in which it appears. As a conse-
quence, a DRC is a post-modifier of the head-noun, as an attributive adjective in
pre-nominal position would modify the head-noun, for instance:

(51) [NPThe man I had dinner with last night] is an architect.

(52) [NPThe tall man] is an architect.

Traditional syntactic constituency tests like pronominalization (51a), left dislo-

cation (51b), interrogative construction (51c), or use of a cleft structure (51d),
clearly show that the antecedent and the RC represent only one constituent:
(51) a. He is an architect.
b. The man I had dinner with last night, he is an architect.
c. Who is an architect? The man I had dinner with last night.
d. Its the man I had dinner with last night who is an architect.

In Jackendoff (1977)s theory, still a reference today, an ARC is represented under

the N, as are other modifiers and contrary to complements, and form a block
with the antecedent:
22 Appositive Relative Clauses in English

(53) The syntactic representation of DRCs [Source: Jackendoff 1977:169]

N
qp
Art N
qp
N S
6
The man who came to dinner
= DRC

2.5.2 The ARC: A unique constituent or orphanage?

Regarding ARCs, the question of their syntactic representation and degree of at-
tachment is much more complex. Two main theories have been in opposition in
the literature, to which intermediate theories have been added, and no consensus
has been reached. The two main theories are full integration (or the unique con-
stituent theory) on the one hand, as for DRCs, and the orphanage (sometimes
radical orphanage) theory on the other hand. Here we adopt Emondss terminol-
ogy: the Subordinate Clause Hypothesis (SCH) vs. the Main Clause Hypothesis
(MCH) to distinguish between the two approaches, which we define briefly be-
low, before turning to a more detailed analysis.

2.5.2.1 Full integration: The Subordinate Clause Hypothesis (SCH). Like a DRC,
an ARC and its antecedent represent only one constituent. This position is a minor-
ity position among specialists and is defended in particular by Jackendoff (1977).
(54) The syntactic representation of ARCs according to the SCH
[Source: Jackendoff 1977:169]
N

Art N S
ru 6
N S who Bill dislikes
6 = ARC
The man who came to dinner
= DRC

2.5.2.2 Detachment: The Main Clause Hypothesis (MCH). Like parenthetical

constructions, for instance, ARCs are considered detached elements and are seen
as a derivation from a main clause in deep structure. Such a position is defended
 Chapter 1. Definitions and previous studies 23

in particular by Ross (1967), Emonds (1979), McCawley (1988), Fabb (1990),

Bianchi (1999), Cinque (2008) inter alia.
(55) The syntactic representation of ARCs according to the MCH
[Source: Khalifa 1999:11]
NP ..

Det N C

We start here from this position, not because it represents the majority opinion,
but because the different arguments in favor of it seem to be more compelling.
We list and discuss them below, while at the same time insisting on their limits.
Within such a theory, an ARC is considered a juxtaposition to the whole sentence,
moved to the immediate right of the antecedent (McCawley 1988), as illustrated
by the shift from deep structure to surface structure, represented in (56):
(56) The syntactic representation of ARCs according to McCawley (1988)
[Source: Khalifa 1999:12]
So

S1 S2
qp qp
NPi VP NPi S
3 6 6
Det N have black hair who live in igloos
! ! J
The Eskimos
So

S1

NP S2 VP
6 6 6
The Eskimos who live in igloos have black hair

. Not all these authors agree on the way ARCs are integrated. In particular, they have differ-
ent views on the attachment level.
. One immediately notes that McCawleys (1988) suggested representation violates the tra-
ditional constraint on constituent structure by enabling branches to be crossed, which enables
him to represent discontinuous constituents.
24 Appositive Relative Clauses in English

According to Emonds (1979), ARCs are coordinated clauses interpolated within

a first independent clause (i.e. the MC) through a Parenthetical Formation Rule,
which enables the passage from (57) to (58):

(57) She works in that city for $300 a month, and her parents have never been
there.
(58) She works in that city, and her parents have never been there, for $300 a
month.

The parenthetical structure acquires a presupposed status, in opposition to the

MC which represents an assertion. The coordinator disappears in favor of a wh-
relative pronoun that occupies the complementizer position (S-Attachment Rule).
The comma intonation typical of ARCs is automatically predicted by Emonds
(1979:223224) following rule:
(59) A root S (immediately) dominated by another S is set off by commas.

The result is (60):

(60) She works in that city, where her parents have never been, for $300 a month.

An ARC, therefore, is external to the sentence it is detached from it as a par-

enthetical clause or a so-called peripheral adverbial clause would.

2.5.3 The arguments in favor of the main clause/detachment theory

Below we discuss the different arguments and counter-arguments to the detach-
ment/orphanage theory. The proponents of this approach suggest that an ARC
originates from an independent clause in deep structure and that it therefore
displays syntactic behavior similar to that of main clauses and not that of em-
bedded clauses. Most of the arguments and examples discussed here are taken
from Emonds (1979) and Mc Cawley (1988). As will be demonstrated, a certain
number of arguments are debatable, which explains why the question has yet to
receive a definitive answer. Other solutions to this problem, somehow interme-
diate between the SCH and the MCH, suggested in particular by Arnold (2004,
2007) and De Vries (2002, 2006), will be examined.

2.5.3.1 The separation antecedent RC. The fact that it is possible for an ARC
not to follow its antecedent immediately, and therefore be extraposed, tends to
prove that they do not form one constituent (examples from Huddleston & Pullum
2002:1066):
 Chapter 1. Definitions and previous studies 25

(61) Only the flower is used, which is not poisonous and is attached to the plant with
a very fine stem.
(62) She could hear her father in the next room, who was angrily complaining about
the horrific telephone bill.

In the same way, an ARC is separated from its antecedent when the latter is fol-
lowed by a DRC. The ARC and the DRC therefore have the same antecedent. The
acceptable order is DRC-ARC, the reverse order leading to an ungrammatical
sentence according to Huddleston & Pullum (2002:1066):
(63) The man that came to dinner, who was drunk, fainted.
(64) *The man, who was drunk(,) that came to dinner, fainted.

However, this discontinuity argument is not entirely reliable. Indeed, DRCs can
be extraposed as well, as in (65)(66), and such an argument cannot be used to
question the fact that a DRC and the antecedent of the relative pronoun form one
constituent:

(65) The man came in that I had met the day before.
(66) When the signal had actually been given, and the train was already in motion,
the door at the opposite side of the carriage opened, and the man came in who
had wrung the unconscious sleepers nose half off his face.

In the same way, an appositive structure can be inserted between the antecedent
and the DRC (examples already mentioned in Section 2.2):

(67) Sure, there were right-on people, men and women, who would be repelled and
appalled by his logic, but that was fine by him.
(68) Karen tried to do what she was told, although sometimes she made mistakes;
and there were other things, like the sleepwalking, that she couldnt help.

Also, a that-clause can be extraposed, without anyones questioning the fact that it
forms a single constituent with the head noun of the NP:

(69) The idea is surprising that you should go on holiday by yourself.

(70) The possibility cant be ruled out that she will call an early election. (Huddleston
& Pullum 2002:966)

Finally, and contrary to what Huddleston & Pullum (2002) claim, there are ex-
amples in which an ARC precedes a DRC, and these examples are perfectly ac-
ceptable. The DRC is separated from its antecedent by a digression marked by
26 Appositive Relative Clauses in English

punctuation. For that kind of sentence to be acceptable, the only condition is that
the DRC be the only element that follows the ARC (Emonds 1979:222):
(71) We found that movie, which cost plenty, that you so highly recommended.

Furthermore, if we consider again (63), in which the ARC follows the DRC, it
seems doubtful to say that the ARC and the antecedent are actually separated, as
the DRC plays a role in the referential construction of the NP the man that came
to dinner. This string represents the antecedent of who and is consequently not
separated from the ARC.

2.5.3.2 Anaphora with one. Another traditional argument to establish the syn-
tactic independence between the RC and the antecedent is the case of anaphors
that use the pronoun one. The antecedent and the RC cannot be referred to using
the pronoun one when the RC is an ARC, but this is possible when the RC is a
DRC. Below are Mc Cawleys (1988:420) famous examples:
(72) Tom has a violin that once belonged to Heifetz, and Jane has one too.
(= DRC)
(73) Tom has a violin, which once belonged to Heifetz, and Jane has one too.
(= ARC)

In (72), the pronoun one stands for the whole NP a violin that once belonged to
Heifetz: Jane owns, like Tom, a violin that used to belong to Heifetz. In (73), Jane
owns a violin but nothing is said about a former owner. The pronoun one refers
only to a violin:

(72) a. Tom has a violin that once belonged to Heifetz, and Jane has a violin that
once belonged to Heifetz too.
(73) a. Tom has a violin, which once belonged to Heifetz, and Jane has a violin
too.

This argument is very often given by linguists to illustrate the detached aspect of
an ARC, as opposed to a DRC. However, it has to be tempered. Indeed, Miller
(1992:7888) has convincingly shown that pronominalization with one is not a re-
liable test to determine subcategorization phenomena, in particular to distinguish
different kinds of complements. The distribution of one cannot be accounted for

. That is to say the difference between:

(iii) the king of England
(iv) the king from England
Only type (iv) complements can follow the pronoun one (Jackendoff 1977):
 Chapter 1. Definitions and previous studies 27

in subcategorization terms only, and as a consequence the possibility of using one

says nothing concerning the syntactic status of a string of words. Consider the
following example:
(74) Tom has a new violin. It is made of spruce wood. I want one too.

The anaphoric pronoun one stands for the NP a violin made of spruce wood, which
is not an NP in the preceding context. The pronoun does not, in such a case, stand
for a unique constituent.
Such an observation poses a real problem for McCawleys test. The fact that
only an NP containing a DRC (72) can be replaced with the pronoun one is not a
direct proof that the whole string is a constituent. Furthermore, it must be noted
that one stands for an antecedent that represents a referential entity: as ARCs do
not intervene in the referential construction of NPs, it is perfectly natural that the
pronoun one cannot stand for the content of the ARC.

2.5.3.3 The use of certain adverbs. We have already mentioned that one seman-
tic difference between ARCs and DRCs is the possibility for ARCs to represent
a speech act that is independent of the speech act in the MC. As a consequence,
adverbs behave differently depending on the status of the RC. We already men-
tioned the following pair, from Emonds (1979:239240):
(75) The boys, who have frankly lost their case, should give up.
(76) *The boys who have frankly lost their case should give up.

Adverbs of this type generally appear in main clauses only. The fact that they can
also appear in an ARC represents an argument in favor of the detachment theory.
In the same way, Emonds (1979) notes that an adverb like apparently does not
show the same semantic behavior in a DRC or an ARC:

(77) The woman, who John knows, apparently, cant get a job.
(78) The woman John knows, apparently, cant get a job.

If in the first example, apparently is ambiguous its scope is either on the main
predication or on the predication in the RC it can only have scope over the main
predication in the case of a DRC (78). For Emonds (1979), this is another argu-
ment in favor of a connection between ARCs and independent clauses.
The interpretation of adverbs thus tends to show that ARCs are interpreted
like main clauses, not embedded clauses.

(v) Jack met the king from England, and I met the one from France.
(vi) *Jack met the king of England, and I met the one of France.
28 Appositive Relative Clauses in English

2.5.3.4 ARCs as independent sentences. In the case of sentential relatives, the

sentential antecedent and the RC can be separated into two different sentences.
This is one of the strongest arguments in favor of the detachment/orphanage the-
ory. The ARC is juxtaposed to a first sentence while retaining all the characteris-
tics of a subordinate clause it is still semantically dependent on the MC:

(79) But if being Macaulay Culkin meant being good at drama, then forget it: he was
crap at drama, because he hated standing up in front of people. Which was why
he hated school. Which was why he wanted to be Macaulay Culkin. Which was
why he was never going to be Macaulay Culkin in a thousand years, let alone in
the next few days. He was going to have to go to school tomorrow.
(80) They are a great team and they are going to be around for a long time, said
Anderson. Who knows, they could go on to the best ever.
Which leaves Britain preparing for next years resumption of the Ashes series
looking for crumbs of comfort.
(81) Sixty years ago, all the things Will relied on to get him through the day simply
didnt exist: there was no daytime TV, there were no videos, there were no glossy
magazines and therefore no questionnaires and, though there were probably
record shops, the kind of music he listened to hadnt been invented yet. ()
Which would have left books. Books!

These examples can easily be rephrased as follows, with no significant semantic

or prosodic differences:

(79) a. But if being Macaulay Culkin meant being good at drama, then forget it: he
was crap at drama, because he hated standing up in front of people. This/
That was why he hated school. This/That was why he wanted to be
Macaulay Culkin. This/That was why he was never going to be Macaulay
Culkin in a thousand years, let alone in the next few days. He was going to
have to go to school tomorrow.
(80) a. They are a great team and they are going to be around for a long time,
said Anderson. Who knows, they could go on to the best ever.
This/That leaves Britain preparing for next years resumption of the Ashes
series looking for crumbs of comfort.
(81) a. Sixty years ago, all the things Will relied on to get him through the day
simply didnt exist: there was no daytime TV, there were no videos, there
were no glossy magazines and therefore no questionnaires and, though there

. The differences are situated on the discourse/pragmatics level. The third part of this book is
dedicated to this interesting question.
 Chapter 1. Definitions and previous studies 29

were probably record shops, the kind of music he listened to hadnt been
invented yet. () This/That would have left books. Books!

The ARCs in example (39), taken from Waterland, also illustrate this phenom-
enon: each ARC is an isolated sentence starting with a relative pronoun.
A very interesting example illustrating the possible syntactic independence of
ARCs is the following one:

(82) West was taking Modern History which wasnt modern history at all, it was
simply not Ancient History, which ended with the fall of Rome because he was
interested in folk songs and ballads.

The ARC, introduced by a dash, is immediately followed by an independent clause

used parenthetically. This possibility is yet another argument in favor of the inter-
pretation of ARCs as independent clauses.

2.5.3.5 ARCs and root transformations. According to the proponents of an or-

phanage analysis of ARCs, another series of syntactic phenomena tends to prove
that ARCs syntactically behave more like independent clauses than embedded
clauses. Emonds (1979) in particular mentions root transformations such as
Subject-Auxiliary Inversion (SAI), the fronting of negative elements or preposi-
tional phrases indicating direction. These transformations are predicted to con-
cern main clauses only, whether juxtaposed or coordinated. They are predicted to
generate ungrammatical sentences if applied to embedded clauses:

SAI:
(83) Will James ever finish reading that book?
(84) *I wonder whether will James ever finish reading that book?
(85) Fred didnt go to the rock concert, and neither did Will.

Negative element fronting:

(86) Never in my life have I seen such a crowd.
(87) *The claim that never in my life have I seen such a crowd is not a lie.
(88) Ive been out of work before, but never have I had to borrow money.

Directional pp fronting:
(89) Up the street trotted the dog.
(90) *The fact that up the street trotted the dog surprised me.
(91) Up the street trotted the dog and down the street trotted the cat.
30 Appositive Relative Clauses in English

This series of predictions by Emonds means that ARCs should not undergo such
transformations. However, it is possible to find root transformations in ARCs, as
is attested by the possibility of finding only followed by an SAI in the following
examples:

(92) Under the previous government there was a loss of direction, they did not insist
that the Palestinian Authority fight terrorism, until we reached the large wave
of attacks, which only in its wake, did the Palestinian Authority begin to operate
in this area.
(93) Perhaps a more risky involvement is the one that Ohio has chosen in India,
where only in the 90s did the government become interested in foreign direct
investment.

The following constructed examples with ARCs (94)(96) are also accepted by
native speakers of English, examples with DRCs being systematically rejected (97)
except when the antecedent is indefinite (98):

(94) Last week, I went to London, where only at night can you enjoy good restau-
rants.
(95) She was airlifted to hospital, where only hours later did she die.
(96) We went to Turkey, where only in the 90s did people realize that joining the EU
might be beneficial for them.
(97) *I went to the pub where only on Friday evenings can you stay all night long.
(98) ?This country is a country where only in the 90s did people realize that joining
the EU might be beneficial.

These results corroborate Hooper & Thompson (1973)s analysis, according to

which the discrimination for the existence of root transformations does not have
to do with embedding vs. non-embedding, but rather with presupposition vs. as-
sertion. Hooper & Thompson claim that ARCs and DRCs with indefinite ante-
cedents are not presupposed but asserted, which explains why root transforma-
tions are felicitous (Hooper & Thompson 1973:489491).
It is thus difficult to draw any real conclusions concerning the significance
of root transformations in ARCs. The fact that syntactic phenomena tradition-
ally typical of independent clauses are possible in ARCs is still interesting in that
it demonstrates the semantic/pragmatic independence of the informational con-
tents in ARCs vis--vis that in the main clause.

2.5.3.6 Conclusion? Although many arguments seem to favor a detachment/or-

phanage approach to ARCs, some of the arguments used by the supporters of
 Chapter 1. Definitions and previous studies 31

such a syntactic analysis are debatable, which explains why some researchers re-
ject such an approach and adopt the single-constituent hypothesis. Other coun-
ter-arguments to the detachment theory are even stronger. In particular, a very
strong counter-argument is word order in V2-languages such as Dutch. In such
languages, the verb in embedded clauses is in final position, including ARCs:

(99) Ik heb het boek aan Jan gegeven, die het dan heeft doorgegeven aan Marie.
I have the book to Jan given, who it then has given to Marie.
I gave the book to Jan, who then gave it to Marie.

2.5.4 Intermediate approaches

Recent research (Arnold 2004, 2007 and De Vries 2002, 2006 in particular) has
argued in favor of intermediate approaches between detachment/orphanage and
full integration. De Vries in particular argues in favor of the existence of a specify-
ing coordination link between an ARC and the main clause while Arnold argues
in favor of syntactic integration associated with semantic independence. Below
we explain the origin for a coordination analysis before turning to Arnolds and
De Vriess analyses.
Many pedagogical grammars mention the apparent affinity between ARCs
and coordinated sentences, suggesting glosses using and-clauses:

(100) I gave the book to John, who gave it to Mary.

a. I gave the book to John, and he gave it to Mary.
(101) Luckily she landed on the bag itself, which burst.
a. Luckily she landed on the bag itself, and it burst.
(102) Macaulay Culkin probably got three hundred and fifty pounds a week, maybe
even more, which meant that if he were Macaulay Culkin he could pay his mum
to teach him.
a. Macaulay Culkin probably got three hundred and fifty pounds a week,
maybe even more, and it meant that if he were Macaulay Culkin he could
pay his mum to teach him.

Register variation (spontaneous speech vs. planned written discourse) also high-
lights the affinity between ARCs and coordinated clauses. In writing, the follow-
ing two examples would probably include an ARC instead of an and-clause:

(103) So we borrowed a car, and it was an electric 225 I dont know if such a car
exists today.
(104) Things are coming that are going to be able to keep us alert and awake as we
drive, and thats gonna save a lot of lives.
32 Appositive Relative Clauses in English

Interestingly, Gresset (1984), with an enunciativist approach, claims that which, as

opposed to that, establishes a relationship with the main clause that is reminiscent
of coordination. Starting from the fact that which can often be replaced with and
this, he suggests that which stands for an element that has already been established
as a discourse object. Quirk et al. (1985:1258) also claim that the link between an
ARC and its main clause is more similar to coordination than subordination.
Interestingly, some ARCs can be introduced by a coordinator (and, especially)
starting a new sentence. Examples (105) and (106) below, taken from the written
part of our corpus, show that such examples do exist in spite of prescriptive gram-
mars claiming the opposite.

(105) Labour present is Neil Kinnock, who speaks for 63 minutes and quotes Robert
Frost about miles to go and promises to keep, and about a brighter dawn when
poverty, privilege, fear, disease, woe, and war shall live only in the memory of
time. And who then talks about the security and liberty of the planet.
(106) And by the Leem lived a lock-keeper. Who was my father. [] Yet who when he
was not asked would sometimes recount bizarre anecdotes of those immemorial
trenches and mudscapes, as if speaking of things remote and fantastical in which
his involvement was purely speculative. ()10

In these two examples and/yet can be analyzed as coordinating two RCs. But oth-
er examples (here, taken from the British National Corpus, spoken section) show
that and can introduce an ARC that is not coordinated to a first RC:

(107) Any commercial vans or lorries seemed to be made out of what was originally
private cars. And which is natural because you know, the very early motor cars
were only a sort of toy for the rich as it were and er when it came to you see er
grocers vans or, or er laundry vans made out of old pr private cars.
(108) But when the I L P wanted to lay down more militant lines the Labour Party
wanted to throw them out anyway, and so the I L P-ers all decided that er the
best thing to do was to leave rather than get thrown out. And which is different
for today, but that that was what really happened.
(109) But being me I couldnt stand that very long, maybe two or three months, and I
asked to be moved. And which was termed, Going down the roads. Er because
the pit bottom was lit up and it meant going down into the dark, an exciting
thought for a young fella.

10. There is no consensus on the status of yet as a coordinator. Many researchers however ac-
knowledge that it is similar to a coordinator (Biber et al. 1999:80), if not a coordinator per se
(Quirk et al. 1985:263).
 Chapter 1. Definitions and previous studies 33

There thus seems to be some affinity between ARCs on the one hand and coordi-
nation on the other.
However, it must be pointed out that the supporters of the single-constituent
analysis question the possibility to rephrase ARCs with coordinate clauses and the
introduction of ARCs with a coordinator. They underline the non-systematicity
of the replacement of the wh- pronoun with and + pronoun. Jackendoff (1977)
provides the following examples:

(110) *Go to Cincinnati, and it is on the Ohio River.

a. Go to Cincinnati, which is on the Ohio River.
(111) *Are we landing in Washington, and it is on the Potomac?
a. Are we landing in Washington, which is on the Potomac?

This counter-argument to the coordinative analysis of ARCs does not hold. As

is shown by Emonds (1979), Jackendoff seems to be confusing coordination and
coordinator. The main problem actually lies in the impossibility to coordinate
sentences with different speech acts. The following sentences, semantically equiv-
alent to (110a)(111a) are perfectly acceptable:

(110) b. Go to Cincinnati. It is on the Ohio River.

(111) b. Are we landing in Washington? it is on the Potomac.

2.5.4.1 Arnold (2004, 2007). Within the theoretical framework of HPSG,

Arnold (2004, 2007) argues in favor of an analysis of ARCs as independent clauses
in spite of their retaining the syntactic characteristics of embedded clauses. His
analysis aims to reconcile the semantic intuition that ARCs are semantically like
independent clauses and syntactically similar to embedded clauses. ARCs have
both wide and narrow scope: wide because the illocutionary force of ARCs is
independent of that in MCs, and narrow if we consider antecedents with quanti-
fiers. Arnold shows that, contrary to what is generally assumed, antecedents of
ARCs can include quantifiers like every and no, and other quantifiers as well:

(112) Many/most/few/no/at least 10 candidates, all/some/three of whom have sent in

their CVs, have agreed to a face to face interview.
(113) Every/no modern plane, which may or may not have an engine in its tail, is
prone to this sort of problem.

ARCs, then, display paradoxical characteristics when both wide scope and nar-
row scope seem to apply, as in the following example:

(114) Sam believes every chess set comes with a spare pawn, which Kim thinks is usu-
ally taped to the top of the box with its base uppermost.
34 Appositive Relative Clauses in English

The ARC is simultaneously within the scope of every (narrow scope) but outside
the scope of the verb believe (narrow scope). Commenting on these observations,
Arnold (2004:18) suggests that ARCs are interpreted semantically like indepen-
dent clauses while retaining all the syntactic characteristics typical of embedded
clauses. Such an analysis provides an explanation for the Dutch data we men-
tioned earlier: ARCs are similar to embedded clauses concerning the (final) posi-
tion of the verb in particular.11

2.5.4.2 De Vries (2002, 2006). Several researchers have explicitly argued in fa-
vor of the existence of a coordination link between an ARC and its host. Among
them is De Vries, who suggests the existence of a specifying coordination link
(symbolized &:P in his work) between the ARC and its antecedent. Specifying
coordination means that the second element (the ARC) is used to specify the first
element (the antecedent), providing extra information on it (specification of A by
B means that B is a logical subset of A (De Vries 2002:214)). The fact that apposi-
tive structures (ARCs but also other types of appositives) can be introduced by a
coordinator (or in particular) suggests, according to this analysis, that appositives
are a kind of coordination:

(115) The American president, Barack Obama

&:P
qp
DP &:
ru ru
D NP &: DP
the American Barack Obama
president

There is not enough space here to detail De Vriess analysis, all the more so as such
an analysis is not without problems (see OConnor 2008:288295), but this is one
of the most recent developments in syntax concerning ARCs.

11. In Arnold & Borsley (2008), which investigates the combination of ellipsis/anaphora and
the use of ARCs, the authors are even stricter in their rejection of an orphanage approach and
show that the data provided below clearly argue in favor of a syntactic integration for ARCs:
(vii) A: Who owns a dog? B: Kim, which is regrettable./*Kim, which is a daschund.
(viii) A: Does Kim own a dog? B: Yes, which is regrettable./*Yes, which is a daschund.
 Chapter 1. Definitions and previous studies 35

Table 1. Summary on the DRC-ARC opposition

DRC ARC
1. s emantic/referential The RC is required to identify The antecedent is identifiable
independence of the the referent of the antecedent independently of the
antecedent embedding of the RC
2. p unctuation No punctuation between Traditional presence of
antecedent and relative punctuation between
pronoun antecedent and pronoun, but
not systematic
3. s uppressibility Non suppressible without Suppressible without any
referential modification of referential modification to the
the antecedent antecedent
4. relative pronoun wh-, that or (almost) exclusively wh-
5. a
 ntecedent NP only NP, VP, PP, AdjP, AdvP,
sentential antecedent
6. s yntactic Antecedent and RC are one Different theories from
representation constituent full integration to (radical)
orphanage

2.6 Conclusion on the DRC-ARC distinction

Table 1 provides a summary of the main differences between DRCs and ARCs.
What this table shows is that, even though the DRC-ARC distinction is not
completely clear-cut, a certain number of criteria aid in maintaining a distinction
between the two types of relative clauses in quite a relevant way. As our aim is not
to enter into this debate and write yet another book on the DRC-ARC dichotomy,
we shall accept the distinction and study only the relative clauses that can un-
ambiguously be labeled appositive, the primary discriminating criterion being
the referential independence of their antecedents. Ambiguous or problematic ex-
amples have been carefully removed from our corpus.
One must not forget, however, that such a traditional distinction remains
problematic. Below we investigate the major problems that such a dichotomous
vision entails.
36 Appositive Relative Clauses in English

3. Problems

3.1 The limits of the distinction

Though well established, the dichotomous vision of relative clauses is not as clear-
cut as grammarians would like it to be. When confronted with real data, there
are numerous cases for which the status of the RC cannot be determined with
certainty. A study by Fuchs on French relative clauses (Fuchs & Rivire 1987, up-
dated in Rivire & Rivire 2000) has shown that even linguists could not agree on
many examples.
Also, some linguists consider that, because of the existence of problematic
cases, a clear-cut border between DRCs and ARCs cannot be established. As a
consequence they prefer describing a gradient or continuum between two ex-
treme poles represented by DRCs on the one hand and ARCs on the other. This
continuum would be a semantic continuum that would tolerate intermediate in-
terpretations in the case of RCs where the two values are mixed and the DRC-
ARC distinction irrelevant (Khalifa 1999:187). We strongly reject this vision of
RCs: what kind of relative clause could be found half way through on the DRC-
ARC continuum (which can be defined as a coherent whole characterized as a
collection, sequence, or progression of values or elements varying by minute degrees
[Source: Merriam Websters Dictionary])? The existence of a continuum/gradient
does not seem to be operating, but simply reveals the inadequacy of the tradi-
tional distinction and especially the discriminating criteria.
In order to solve the classifying problems, some researchers have suggested
the introduction of new categories or even a redefinition of the binary vision.
The first option is illustrated by Lambrecht (2000), who worked on French data, a
language for which the traditional DRC-ARC distinction is comparable to that of
English. Lambrecht suggests new distinctions to solve the problem posed by the
following examples:

(116) Il tait une fois une belle princesse qui vivait dans un vieux chteau.
It was once a pretty princess who lived in an old castle.
Once upon a time there was a pretty princess who lived in an old castle.
(117) Jai eu mon beau-frre qui a fait Paris-Nice.
I have had my brother-in-law who did Paris-Nice.
My brother-in-law traveled from Paris to Nice.
(118) Ya le tlphone qui sonne.
Theres the phone thats ringing.
Phones ringing.
 Chapter 1. Definitions and previous studies 37

(119) Je vois le facteur qui arrive.

I see the postman that is coming.
I can see the postman coming.
(120) Voil le facteur qui arrive.
Here is the postman who comes.
Heres the postman coming.

To classify these examples which are problematic for the traditional DRC-ARC
dichotomy, Lambrecht suggests the creation of a new category: presentative RCs,
which simultaneously introduce a new discourse entity and predicate something
about it (Lambrecht 2000:50).
The second option (redefining the binary vision) is illustrated by Cotte (2000)
inter alia, who suggests, following an approach based on the French theory of
enunciative operations, that RCs should be distinguished according to the mo-
ment of their cognitive construction, before or after that of the main clause. A rel-
ative clause is therefore (i) constitutive (see (121)) when cognitively constructed
before the main clause (in other words preconstructed) or (ii) qualifying (qualifi-
cative in French, see (122)) when its construction is posterior to that of the main
clause, constituting some kind of addition to the main clause. This distinction
does not match the DRC-ARC distinction, even though qualifying RCs and ARCs
have properties in common. Consider the following two examples:

(121) I wont read any book that has been badly reviewed.
(122) She said she was forty, which I dont believe.

In (121), the event review pre-exists to the contents of the main clause and the
RC is constitutive. The RC in (122) conveys a comment that is posterior to the
predicative relation [she/say that she was forty] and is therefore a qualifying RC
(Cotte 2000:4851).
We shall not go into more detail here; what these studies do show, however,
is that the traditional dichotomy does not cover all occurrences of relative clauses
and that other kinds of classification are possible. Lambrechts examples, in partic-
ular, show that RCs can function in a way that is distinct from ARCs and DRCs.

3.2 The specific problem of DRCs with indefinite antecedents

This configuration is very common in English and yet, grammars do not men-
tion this problematic case for the traditional DRC-ARC dichotomy. An indefinite
antecedent can have a specific or non-specific value, and this has consequence on
the semantic interpretation of RCs. Consider the following examples:
38 Appositive Relative Clauses in English

(123) Id like a book that deals with the Civil War.

(124) He bought a house which hell move into in June.

These two RCs are traditionally considered as DRCs. And yet, by looking closer,
one realizes that while the RC in (123) plays a role in the identification of the
antecedents referent (I would like a specific book, not just any book), the RC in
(124) plays a different role. The referent of the antecedent can be identified inde-
pendently of the embedding of the RC: he bought an entity that corresponds to a
house. And then we learn that he will move into this house in June:

(124) a. He bought a house. After wondering when to move in for about a month,
he decided to move into it in June.

The two RCs react differently to the following tests:

i. Punctuation insertion between the antecedent and the RC:

(123) a. ?/#Id like a book, which deals with the Civil War.
(124) b. He bought a house, which hell move into in June.

ii. Replacement of the RC with a coordinated clause:

(123) b. ?/#Id like a book and it deals with the Civil War.
(124) c. He bought a house and hell move into it in June.

iii. Replacement of the block antecedent+RC with a proper name:

(123) c. Id like The American Civil War. (where The American Civil War is the
name of the book)
(124) d. #He bought Lovely View. (where Lovely View is the name of the house)

iv. Traditional constituency tests like the question test:

(123) d. A: What would you like?

B: Id like a book that deals with the Civil War.
(124) e. A: What did he buy?
B: #He bought a house which hell move into in June.

v. Topicalization of antecedent+RC:

(123) e. A book that deals with the Civil War Id like.

(124) f. ??A house which hell move into in June he bought.
 Chapter 1. Definitions and previous studies 39

This contrast in syntactic possibilities reveals a difference depending on the kind

of indefinite antecedent. But does this mean that the RC in (123) is determina-
tive and the RC in (124) appositive? The solution some researchers (e.g. Fuchs &
Rivire 1987) have found is to distinguish contrastive DRCs (125) and non-con-
strastive DRCs (126):

(125) People who smoke low-tar cigarettes tend to inhale more deeply, bringing
particles to the smallest and most vulnerable tissues in the lungs where these
cancers start.
(126) Snows that set records for this early in the season move out of New England
today after shutting down roads, airports and shopping trips on a two-day
march up the East Coast.

In (125), there is a contrast between two sub-categories of smokers, those smok-

ing low-tar cigarettes and the other smokers. In (126), no such contrast exists
between snows that set records and other kinds of snows. The speaker mentions
the snows that have recently fallen and which have set records. The RC can be
removed and a comma could be inserted between the antecedent and the relative
pronoun in (126), but not in (125):
(125) a. *People, who smoke low-tar cigarettes, tend to inhale more deeply, bringing
particles to the smallest and most vulnerable tissues in the lungs where these
cancers start.
b. #People tend to inhale more deeply, bringing particles to the smallest and
most vulnerable tissues in the lungs where these cancers start.
(126) a. Snows, which set records for this early in the season, move out of New
England today after shutting down roads, airports and shopping trips on a
two-day march up the East Coast.
b. Snows move out of New England today after shutting down roads, airports
and shopping trips on a two-day march up the East Coast.

Such a distinction between contrastive and non-contrastive DRCs provides a rel-

evant analysis for examples like (125)(126). But it does not really solve the status
of the RC: in (126) for instance, is the RC still determinative or appositive if we
consider that the antecedent has found its reference independently of the embed-
ding of the RC? The question remains open to debate, which is why this study on
ARCs does not include such examples and relies only on traditional ARCs.
40 Appositive Relative Clauses in English

3.3 An operational dichotomy?

3.3.1 Problems linked to the original binary opposition

From the very beginning, the dichotomy was based on considerations formu-
lated to define the properties of restrictive/determinative RCs as opposed to
other kinds of RCs. This is where the origin of the problem lies: ARCs (or non-
restrictives) have been defined asymmetrically as not being and not having the
properties of restrictive/DRCs. Gathering all the RCs that do not share the same
semantic properties as DRCs in no way guarantees that this newly-created cat-
egory is homogeneous. However, the distinction has remained intact, corrobo-
rated by the works of Chomsky (1966, in particular), who considered the distinc-
tion established by logicians as the expression of a difference between the deep
structure and the surface structure of a sentence. Although the distinction has
been questioned in recent works (on English and French), it is mentioned in all
pedagogical grammars and taken for granted, illustrated with constructed, arti-
ficial examples such as The Eskimos, who have black hair, live in igloos (ARC) vs.
The Eskimos who have black hair live in igloos (DRC).12

3.3.2 A distinction irrelevant in other languages

The distinction between different kinds of relative clauses according to semantic/
morpho-syntactic criteria seems to be primarily an Indo-European debate. While
most Indo-European languages do in fact maintain a difference between DRCs
and ARCs, there are other languages for which such a distinction is completely
irrelevant (Asian languages in particular, see below).
In English, French or Spanish, DRCs and ARCs display different morpho-
logical and syntactic characteristics. In English, in addition to the punctuation
criterion which is not totally relevant the distribution of relative pronouns
depends on the status of the RC (see Section 2.3.1). In French, complex relative
pronouns like lequel, laquelle, lesquels (combination of definite article followed
by interrogative pronoun) appear in ARCs but not in DRCs (127). French ARCs
with sentential antecedents are not introduced by que/qui but ce que/ce qui or
NP+ que/qui (128):
(127) Jai trouv un livre par terre, lequel avait t perdu par un tudiant. (= ARC)
I have found a book on the floor, the which had been lost by a student.
I found a book on the floor, which had been lost by a student.
a. *Jai trouv un livre lequel avait t perdu par un tudiant. (= DRC)
I found a book the which had been lost by a student.

12. We are not beyond reproach in this matter, especially in Chapter 1.

 Chapter 1. Definitions and previous studies 41

(128) Il est parti trois jours Londres, ce qui est une opportunit formidable pour
lui.
He is gone three days to London, this/that which is an opportunity great for
him.
He is gone to London for three days, which is a great opportunity for him.
a. *Il est parti trois jours Londres, qui est une opportunit formidable pour
lui.
He is gone three days to London, which is an opportunity great for him.

In Spanish, relative pronouns are also distributed according to the determinative/

appositive status of the RC. The relative pronoun quien appears only in ARCs:

(129) El hombre, quien estaba de mal humor, se fue a dormir. (= ARC)

The man, who was in a bad mood, went to sleep.
a. *El hombre quien estaba de mal humor se fue a dormir. (= DRC)

Likewise, the pronoun cuando (when) can only appear in DRCs. Finally, as in
French, ARCs with sentential antecedents are introduced by a combination of a
determiner and a relative pronoun, viz lo que or lo cual:

(130) Tuve que abandonar mi hogar, lo que (lo cual) me dej tristsimo.
I had to abandon my house, REL (= which) made me very sad.
a. *Tuve que abandonar mi hogar, que/cual me dej tristsimo.

If we consider a more distant European language (i.e. non Germanic or Ro-

mance) such as Persian, the DRC-ARC distinction also appears through different
morphosyntactic realizations. Persian RCs are introduced by ke,13 whether the
RC is determinative or appositive. The difference lies in the suffixation of the head
noun with the suffix -i in the case of DRCs only (Taghvaipour 2003:13).14 The fol-
lowing pair illustrates the difference between DRCs and ARCs in Persian:

(131) daneju-i ke be iraz rfteh+bud braym nameh-i nevet.

student-i REL to Shiraz gone for me letter-a wrote
The student who went to Shiraz wrote me a letter. (= DRC)

13. We shall not label ke a relative pronoun, as some researchers claim it is a complementizer
instead (see Taghvaipour 2003:511, especially). This is reminiscent of the debate concern-
ing that.
14. This suffix signals the noun as being close to definiteness, although not completely: [T]he
-i in restrictive RC constructions makes the head noun of the RC move towards definiteness,
though it does not make it completely definite (Taghvaipour 2003:15).
42 Appositive Relative Clauses in English

a. Ali ke be iraz rfteh+bud braym nameh-i nevet.15

Ali REL to Shiraz gone for me letter-a wrote
Ali, who went Shiraz, wrote a letter to me. (= ARC)

But the distinction applies far beyond the confines of Indo-European languages.
Interestingly, Tagalog, an Austronesian language spoken mainly in the Philip-
pines by about 20 million people, displays the difference between DRCs and
ARCs through the relative position of the relative clause and the antecedent.
DRCs can precede or follow the antecedent but ARCs can be in post-nominal po-
sition only (Del Gobbo 2001:24). Lancri (1999) has shown that a geographically
and chronologically distant language like Middle Egyptian (2000 BC1300 BC)
also displayed an opposition that can be compared, at least partially, with the
DRC-ARC dichotomy. Although Lancri prefers to refer to a distinction linked to
the degree of cohesion between the antecedent and the RC (minimal degree of
cohesion for RCs introduced with nty m; maximal degree of cohesion for RCs in-
troduced with imy),16 her analysis of Middle Egyptian data17 brings out semantic
differences that are clearly reminiscent of the DRC-ARC debate, including the
limits of such a binary vision.
However, such a seemingly universal, perfect picture should not conceal the
fact that not all languages maintain a distinction between the two types of relative
clauses. Below we discuss the case of three languages for which the traditional
distinction is not operational with such surface differences: a Germanic language
(German), an Asian language (Chinese) and an Afro-Asiatic language (Arabic).

3.3.2.1 The case of German. Even within the Germanic branch of Indo-Euro-
pean families, the DRC-ARC distinction is problematic from a formal point of
view. In writing, there is no formal difference between the two types of RCs as
both types are signaled with a punctuation mark and introduced with similar
relative pronouns (inflected forms of der, die, das). In both cases, the verb is in
final position. The following two examples are therefore ambiguous:

15. The absence of -i in this example has nothing to do with the fact that the proper name Ali
already ends with an i.
16. A third category in the micro-system exists: RCs introduced with wnn.ty.fy m. These refer
to a future-oriented event.
17. She compares Middle Egyptian and Old English relative clauses. This might seem specula-
tive but her analysis of fundamental similarities in the role of relative clauses and the distribu-
tion of relative pronouns is very convincing.
 Chapter 1. Definitions and previous studies 43

(132) Die Amerikaner, die Bush nicht mgen, essen keine Bretzeln mehr.
(= DRC/ARC)
The Americans PUNC who Bush not like PUNC eat no pretzels any more.
The Americans (,) who do not like Bush (,) do not eat pretzels any more.
(133) Die Kinder, mit denen wir spielten, waren sehr begeistert. (= DRC/ARC)
The children PUNC with whom we played PUNC were very delighted.
The children (,) with whom we played (,) were very delighted.

Speakers of German consequently resort to other strategies to specify the (non)-

restrictive role of the RC: intonation, disambiguation through context, or also the
introduction of a demonstrative before DRCs antecedents (134) vs. the introduc-
tion of an adverb in ARCs (135):

(134) Diejenigen Amerikaner, die Bush nicht mgen, essen keine Bretzeln mehr.
(= DRC)
Those Americans PUNC who Bush not like PUNC eat no pretzels any more.
The Americans who do not like Bush do not eat pretzels any more.
(135) Die Amerikaner, die brigens Bush nicht mgen, essen keine Bretzeln mehr.
(= ARC)
The Americans PUNC who by the way Bush not like PUNC eat no pretzels any
more.
The Americans, who by the way do not like Bush, do not eat pretzels any
more.

3.3.2.2 Chinese. In some Asian languages, Chinese in particular, the traditional

distinction is also problematic. This is how Del Gobbo (2001:1) summarizes the
situation:
We know what an appositive relative clause is supposed to look like in English
or Italian. When we turn to languages such as Chinese or Japanese however, the
demarcation line becomes blurry and facts are not so clear; at best we find that
in those languages no syntactic distinction can be made between a restrictive
and an appositive relative clause. Since all the languages for which such a dis-
tinction is not clear-cut have pre-nominal relatives (in most cases, exclusively),
it is easy to suspect that there could be a correlation between the linear order of
the relative clause with respect to its head and the availability of an appositive
interpretation.

Because of the absence of relative pronouns in Chinese relative clauses being

introduced by the marker de the situation of RCs is very different from that
in Indo-European languages. Recent research (Del Gobbo 2001; Zhang 2001)
has shown that only one type of RC exists in Chinese. All Chinese RCs display
44 Appositive Relative Clauses in English

the properties that DRCs have in languages for which the distinction is relevant:
type of antecedent (e.g. impossibility for RCs to modify a proper name (136) or
a sentential antecedent (137)), impossibility to introduce certain adverbs such
as tanbaideshuo (frankly) which appear in English ARCs only (138), influence of
quantifiers such as meiyige (every) on the interpretation of the RC, necessarily
restrictive (139). Del Gobbo concludes on the non-existence of ARCs in Chinese,
more precisely on the impossibility to interpret a Chinese RC as an ARC.

(136) ??Ni zuotian kanjian de Zhangsan shi wo pengyou.

You yesterday saw DE Zhangsan is my friend.
Zhangsan, who(m) you saw yesterday, is my friend.
(137) *Zhangsan hen Lisi conglai jiu bu de congming.
Zhangsan very Lisi never just not DE clever.
Zhangsan is clever, which Lisi has never been.
(138) *Wo tanbaideshuo kai de zhe yiliang qiche zhuangzai bu wanhao.
I franlkly am driving DE this a car condition not excellent.
This car, which I am frankly driving, is not in excellent condition.
(139) Meiyige Op chuan wazi de xuesheng dou shi tiaowude.
Every wear socks DE student all be dancers
Any student who is wearing socks is a dancer.

Zhang (2001) reaches the same conclusion using different argumentation. Her
analysis starts from the observation that null-head appositives (i.e. ARCs with-
out a head noun) do not exist in Chinese. Units in apposition have to be full
NPs and include the antecedent of the RC, which cannot be ellipted. Examples
(140)(141) are grammatical: the first one shows a simple noun phrase in apposi-
tion, the second one a complex NP containing a noun that is post-modified by a
relative clause:

(140) Bolin, zhe ge Ouzhou chengshi you bashi duo jia bowuguan.
Berlin this Europe city has 80 more museums.
There are more than 80 museums in Berlin, the European city.
(141) Bolin, zhe ge renren dou zhidao de chengshi, you bashi duo jia
Berlin this every one all know DE city has 80 more
bowuguan.
museums.
There are more than 80 museums in Berlin, the city that every one knows.

However, (142) is ungrammatical in Chinese:

 Chapter 1. Definitions and previous studies 45

(142) *Bolin, renren dou zhidao de, you bashi duo jia bowuguan.
Berlin every one all know DE have 80 more museums
There are more than 80 museums in Berlin, which every one knows.

In the light of such data, Zhang (2001) concludes that ARCs do not exist in Chi-
nese, although they can appear in units in apposition provided their antecedent
is present, which means the RCs are then DRCs. Attempting to tease apart DRCs
and ARCs therefore makes no linguistic sense in Chinese. The same phenomena
apply to Japanese (Kuno 1973)18 and Korean (Del Gobbo 2001:24), but we shall
not enter into a detailed analysis here.

3.3.2.3 Arabic. The Asian languages just mentioned are not the only languages
where relative clauses can trigger a determinative interpretation only. Arabic is
another example of a language for which ARCs per se do not exist. Speakers of
Arabic to whom we gave French sentences (such as (143)(144)) including ARCs
to translate into Arabic have found it impossible to provide a literal translation.
The ARC had to be replaced with a full NP in apposition or completely rewritten.

(143) Les Franais, qui aiment le fromage, aiment le vin.

The French, who like cheese, like wine.
(144) Marie, qui est alle au march, est revenue avec des oranges.
Marie, who went to the market, came back with oranges.

When translating (143), the informants replaced the ARC with an NP corre-
sponding to amateurs of cheese.
(143) a.
Al faranciouna mouhibou aljoubnati youhibouna al khamra
The French, amateurs of cheese, like wine.

When translating (144), Arabic speakers transformed the relative clause into a
main clause, whether juxtaposed (144a) or coordinated (144b):
(144) a.
Maria Dahabat ila souqi wa ahdarat minhou al bourtoqala
Marie went to the market. She came back with oranges.
b.
Maria dahabat ila souqi wa ahdarati lbourtoqala
Marie went to the market and came back with oranges.

18. Recent research, however, has proposed that in spite of the widely-accepted claim, Japa-
nese syntax does discriminate between the two types of relatives via word order (see Ishizuka
2008).
46 Appositive Relative Clauses in English

The exclusive DRC status of Arabic RCs is also apparent in RCs with sentential
antecedents. A sentence like He went to London, which is an opportunity for him
cannot be translated using a relative clause as (145) and (146) illustrate:

(145) ,
Laqad dahaba ila London, forsatoun binisbati lahou
He went to London, an opportunity for him.
(146) ,
Laqad dahaba ila London, achayou ladi youtabarou forsatan lahou
He went to London, thing that is an opportunity for him. (sic)

As in French or Spanish, the relative pronoun has to be preceded by an element,

in this instance a noun (thing). The RC is then determinative, not appositive, al-
though the whole NP is in apposition to the main clause.
What this brief survey shows is that the DRC-ARC is not universal. In lan-
guages like Chinese or Arabic, ARCs simply do not seem to exist. This is not a real
problem for our study, dedicated to English, a language for which the dichotomy is
quite operational (if we exclude the problematic cases of indefinite antecedents).
Interestingly, this survey of RCs in languages other than Indo-European lan-
guages could lead us to form the following hypothesis: RCs are determinative by
default, but in some languages, according to constraints that would require defin-
ing, RCs can appear (alone) in apposition to a first unit, nominal or not, while in
other languages a head is required (either a noun or a demonstrative as in Chinese
or Arabic or even French and Spanish for sentential relatives). If such a hypothesis
were to be confirmed, there would not be two types of relative clauses, but two
uses of relative clauses, systematically determinative. This could explain why all
the morphosyntactic tests used to discriminate between DRCs and ARCs men-
tioned in Section 2 are not one hundred percent reliable (e.g. that can be found in
ARCs,19 root transformations apply in ARCs, etc.)

19. At this stage, it is important to remember that the specialization of wh-relative pronouns
in ARCs is the result of prescriptivism. In Merriam-Websters Concise Dictionary of English Us-
age (2002:727728), the authors claim that, although the use of that in ARCs had decreased
between the 17th and 19th centuries, it is the Fowler brothers rule, according to which an ARC
should not be introduced by that in spite of writers usage at the time that restricted the choice
of relative pronouns to the wh- pronouns. This rule is still taught in schools today, which might
explain the (artificial?) rarity of ARCs introduced by that.
 Chapter 1. Definitions and previous studies 47

3.4 Conclusion

When we consider all the problems the traditional DRC-ARC dichotomy engen-
ders, one easily understands why it is repeatedly and systematically questioned.
This is not easy, as is underlined by Fuchs (1987:122), as the binary opposition is
so convenient to explain two types of referential constructions.20

4. Previous studies on the role of ARCs

What is striking when one considers the research that has been carried out on
the use of ARCs is its asymmetrical definition with DRCs. Most studies rely on
constructed examples to illustrate the differences, mostly morphosyntactic, with
DRCs. As DRCs play a role in the definition of the referent depicted by the an-
tecedent, ARCs are often considered as not fulfilling this role, and therefore as
being somehow optional and therefore suppressible. ARCs are seen as bringing
extra information, some sort of secondary information, precision or elaboration,
hence the choice of supplementary by Huddlestion & Pullum (2002) as a label to
name this subcategory of RCs. To determine the status of a relative clause, Fowler
& Fowler (1949:85) even suggest:
To find whether a clause defines or does not define, remove it, and see whether
the statement of which it formed a part is unaltered: if not, the clause defines.21

In comparison with the endless debate on the DRC-ARC dichotomy, there are few
studies on ARCs themselves, apart from their syntactic characteristics; they are,
furthermore, always presented in opposition with DRCs. Discourse studies have
been quite rare: only a few researchers have provided work on the uses of ARCs
in discourse. This is what this book aims to provide, but we did not start from
scratch the research of linguists (Jespersen and Cornilescu, especially) have pro-
vided useful starting points. Below we provide and discuss the different analyses
linguists have provided concerning ARCs.

20. However, it is important to note that when we asked consultants in particular German
speakers to translate sentences into their own language, the fact that they resorted to syntactic
structures different from RCs seem to reinforce the idea that RCs have two uses in discourse.
21. A still more extreme position on the supposedly removable character of ARCs is Potts
(2005)s: [J]ust as we can slide a newspaper advertising supplement into the trash without loss of
journalistic content, so too can we trim a sentences semantic parsetree () without changing the
at-issue content (Potts 2005:92).
48 Appositive Relative Clauses in English

4.1 ARCs as background information

4.1.1 Non-essential information?

As mentioned earlier, ARCs have often been considered supplementary infor-
mation, therefore non-essential/optional. This has become a traditional defining
criterion of ARCs. We of course strongly disagree with the optional character
or suppressibility of ARCs. What is certain, however, is that the embedding of
the information gives the RC another status: the informational content is taken
into the background, as opposed to the informational content of the main clause,
in the foreground. Loetscher (1973) and Cornilescu (1981) distinguish between
hot news and not so hot news, which is an interesting way to consider the
information conveyed by the MC and the ARC. Cotte (1997:134135, my trans-
lation) mentions information that has some value while retaining some subordi-
nate character, which in the end makes the ARC partially integrated. A relative
clause that is appositive is then considered as a syntactic means of backgrounding
information.
The informational content conveyed by the ARC is not on the same level as
that conveyed by the main clause. Its role in discourse is not that of a DRC. But
what, then, is its role in discourse? Why do speakers use such a syntactic structure?
What do they intend to convey when they make this particular linguistic choice?

4.1.2 The interaction between two propositional contents

The existence of a complex inter-clausal link between the ARC and its main clause
is a property that has been traditionally mentioned in the literature. ARCs enable
the connection of two clauses with an implicit semantic link to be reconstructed
inferentially. The informational contents of the main clause (P) and the ARC (Q)
do not simply add to each other (P & Q): the meaning of the whole sentence goes
beyond such a plain addition. An inter-clausal relationship (cause, consequence,
concession among others) is often present, and this is precisely why speakers re-
sort to the use of an ARC (instead of a coordinated clause for instance).
This implicit link between MC and ARC needs to be defined precisely, some-
thing which previous studies have not really done. The interaction with the co-
text and context is crucial, but so is the addressees supposed encyclopedic knowl-
edge. Consider the following example:

(147) Salim, who passed all his exams, is happy.

The meaning of the whole sentence goes beyond the plain addition of the infor-
mational contents of the MC (Salim is happy) and the ARC (Salim passed all his
exams). Most addressees will make the inference that a clausal link exists between
the MC and the ARC:
 Chapter 1. Definitions and previous studies 49

(147) a. Salim is happy as/because/since he passed all his exams.

The inference is made possible because of the hierarchization of the two informa-
tional contents: the ARC is in the background, which triggers the interpretation
of [Salim/pass all his exams] as background information shedding particular
light on the main assertion. The ARC then functions semantically as an adver-
bial. Because in the world as we know it, success is a source of happiness and
failure a source of unhappiness (encyclopedic knowledge), the link is inferred
as being causal. We could however imagine a co-text/context where passing his
exams would be for Salim a source of unhappiness. In such a context, the link to
be inferred would be concessive, not causal:

(147) b. Salim is happy although/despite the fact that he passed all his exams.

4.1.3 An independent assertion

Some researchers have reached the conclusion that ARCs represent a semi-in-
dependent clause, sometimes a completely independent clause. Fowler & Fowler
(1949:85), for instance, consider ARCs as the insertion of an independent com-
ment, description [or] explanation. Such an analysis is reminiscent of the syntac-
tic debate concerning ARCs and the orphanage approach: Emonds (1979), for
instance, considers that ARCs are interpolated at discourse-level. The link be-
tween syntax and discourse here is obvious and merits investigation.

4.2 Previous attempts at defining taxonomies

One kind of distinction is very often mentioned to discriminate between different

types of ARCs: sentential ARCs vs. ARCs with nominal antecedents. This crite-
rion is however purely morphosyntactic and does not represent a relevant subcat-
egory in a discourse-based taxonomy.
In quite the same way, some attempts at defining discourse-based taxono-
mies mix semantico-pragmatic and syntactic criteria. This is the case of Jespersen
(1970), who nonetheless defined a subcategory of ARCs (which he labels loose
relative clauses) that has become quite traditional and which will be useful for
our study. Jespersen distinguishes between (i) Continuative Relative Clauses
(CRCs for short in his work), (ii) mental parentheses, and (iii) other, nameless
ARCs (which in itself is problematic). An example of continuative RC provided
by Jespersen is reproduced in (148):

(148) () he had seen my aunt give this person money outside the garden rails in the
moonlight, who then slunk away and was seen no more.
50 Appositive Relative Clauses in English

Among the properties of CRCs that he lists are: (i) they are added to what could
have been the end of the sentence; (ii) they can be replaced with an independent
sentence introduced by and and a personal pronoun; (iii) they belong to a formal,
literary register; (iv) time indicators such as immediately, presently, then can
be found; (v) they are separated from the main clause with a punctuation mark
(comma, semi-colon or period); (vi) they may contain a wish or a question.
As far as mental parentheses are concerned, they correspond to an interrup-
tion of the natural flow of discourse, enabling the speaker to express an idea or a
commentary. We are close to insertions signaled by expressions such as in other
words, as it were, truth to say or curiously enough (Jespersen 1970:113). Example
(149) is a prototypical example of what Jespersen considers a mental parenthesis:
(149) I am interested in wine. And which is the point I know that the wine is always
there.

What Jespersen suggests for continuative relative clauses is very interesting and
has been very often referred to in the subsequent literature on relative clauses.
However, this subcategory has become somehow blurry, as linguists use the term
continuatives to qualify all sorts of ARCs as long as they appear as some kind of
continuation of the sentence. This is particularly the case for ARCs with senten-
tial antecedents such as found in (150)(151):

(150) Norway won the 2009 Eurovision contest, which was quite unexpected.
(151) On the same show Stephanie revealed she and Triple H were trying for a baby,
which hints at another awful gimmick in the future.

The subcategory of continuative ARCs mentioned today in quite a number of

pedagogical grammars confuses discourse and syntactic criteria and has become
not homogeneous enough to be relevant. This is probably due to the fact that
Jespersen himself does not provide a precise definition he provides only two ex-
amples alongside a series of properties. This subcategory therefore needs a thor-
ough redefinition to be relevant. As for the subcategory defined as mental paren-
theses, the main problem is that this does not define the kind of information that
is conveyed by the ARC. The concepts of idea and commentary are extremely
vague. This subcategory does not seem relevant. Finally, the definition of these
two subcategories suggests that the leftovers could constitute a third category,
by default. Nothing guarantees the homogenous character of the ARCs defined in
such an asymmetrical way.
Cornilescu (1981), who based her research on Jespersen (1970), Ziv & Cole
(1974) and Carabulea (1959), suggests a three-category classification of ARCs
(which she prefers to label nonrestrictives):
 Chapter 1. Definitions and previous studies 51

i. Appositive Relative Clauses (ApRCs)

ii. Continuative Relative Clauses (CRCs)
iii. Adverbial Relative Clauses (AvRC)

Cornilescu specifies from the very start that her subcategories are defined in
terms that are exclusively pragmatic. Appositive relative clauses are a syntactic
means for the backgrounding of information; their use in discourse consists in
a conversational strategy that distinguishes between hot news and not so hot
news. Their semantic contents are presupposed. Continuative Relative Clauses
are in a logical relation (causal or temporal) with the matrix and represents a con-
tinuation of the sentence, in final position. CRCs are derived from a coordinated
sentence and belong to the category of asserted relative clauses, being on the same
level as the main clause.

(152) Eric gave the book to Mary, who then returned it to the library.

They can be extraposed, whereas ApRCs cannot:

(153) A boy, whom I had never seen before, was trying to break in. (= ApRC)
a. ?A boy was trying to break in, whom I had never seen before.
(154) John, who, by the way, Ill never invite to a party again, really bothered me at
the party last night. (= CRC)
a. John really bothered me at the party last night, who by the way, Ill never
invite to a party again.

This argument seems highly debatable as (i) not all extraposed ARCs are continu-
ative RCs, even if we consider Cornilescus definition; (ii) the example provided
in (154) does not represent a continuative relative clause. This property does not
seem relevant.
The third category, labeled Adverbial Relative clauses (AvRC), finds its origin
in the grammar of Romanian. It gathers the ARCs which are similar with adverbi-
als: they are involved in some asymmetrical semantic relationship with the main
clause. AvRCs are either close to ApRCs (their contents are presupposed) or to
CRCs (their contents are asserted):

(155) He asked to be allowed to leave the answering of that question, which was very
complicated, to the witness Hall.
a. He asked to be allowed to leave the answering of that question to the witness
Hall, because the question was very complicated.

Cornilescus (1981) suggestion of a taxonomy for ARCs has a major problem: it

does not create three distinct subcategories but only isolates two particular cases:
the ApRC category finally corresponds to the RCs which are neither continuative
52 Appositive Relative Clauses in English

nor adverbial. Also, Cornilescu claims that her taxonomy is based on pragmatic
criteria and yet, no mention is made of the informational status of the semantic
contents conveyed by the ARCs. This failing in her analysis stems from the fact
that she uses constructed, decontextualized examples. This is exactly the kind of
failings that a relevant discourse-based analysis of ARCs aims to prevent.

5. Conclusion

In spite of the many objections and problems that can be formulated against the
traditional DRC-ARC dichotomy, such a distinction seems operational for a lan-
guage like English, provided one pays particular attention to problematic cases
such as RCs with indefinite antecedents. Once the category of ARCs has been cau-
tiously defined and delimited, it still remains that this subcategory of RCs has not
received an adequate, thorough study: most of the definitions are asymmetrical,
taking place within the context of the DRC-ARC dichotomy and the search for
morphological and syntactic criteria to discriminate between the two subcatego-
ries. In particular, the role played by ARCs in discourse is generally overshad-
owed. This, specifically, is what this book aims to provide its readers with: if ARCs
play no role in the referential definition of their antecedents, then what is their
specific role in discourse? To answer such a question, a systematic corpus study
needs to be completed, focused on criteria linked with the informational status of
the contents of ARCs, their antecedents and their main clauses.
Before turning to the corpus study, however, we need, in order to complete
the definition of our object, to mention a particular syntactic configuration for
ARCs. Some ARCs display a syntactic pattern that deviates from the norm as
defined by grammars for this structure. We name them atypical ARCs. The next
chapter is dedicated to this phenomenon.
 chapter 2

Atypical appositive relative clauses

1. Starting point: Are you a good which or a bad which?

Working on attested data, in particular on spontaneous speech, naturally brings

up examples that are problematic for traditional grammatical classifications. There
are, in addition to dysfluencies, cases that have not been taken into consideration
when constructing examples for a syntactic analysis. There are also examples that
simply do not fit traditional linguistic descriptions, as unexpected data can and
do show up. This is exactly what happened when we started analyzing the spoken
part of our different corpora (both self-collected and electronic). While compil-
ing examples of ARCs, we came across RCs whose syntactic structures apparently
deviated from the norm established by pedagogical (and reference) grammars.
We posit that their frequency of occurrence is in fact too high for such apparently
deviant structures to be simply a random phenomenon. In particular, the analysis
of this data sheds interesting light on the discussion concerning the role of the
relative pronoun and the syntactic status of the ARC.

2. Definitions

2.1 Typical vs. atypical ARCs

According to traditional analyses of ARCs (and also DRCs, of course), a gap is left
by what syntacticians call wh- movement. This gap is called a trace by genera-
tivists (symbolized t) or copy in more recent analyses such as Chomsky (1995),
and is co-referential with the wh- element that has been fronted to the beginning
of the relative clause:

(1) This law, whichi some lawyers would regard ti as tendentious, prohibits employ-
ers from HIV testing.
(2) Dennis, whoi we were talking about ti last night, says youre absolutely mental.
54 Appositive Relative Clauses in English

Traces are copies of the moved elements that have been deleted at PF (Phonologi-
cal Form), but are available for interpretation at LF (Logical Form). The empty
position is filled by a trace that retains the syntactic and semantic properties of
the moved element, to which it is linked (Cook & Newson 2007:134). This analy-
sis predicts that the following sentences cannot be generated, as a trace cannot
be filled:

(1) a. *This law, whichi some lawyers would regard iti as tendentious, prohibits
employers from HIV testing.
(2) a. *Dennis, whoi we were talking about himi last night, says youre absolutely
mental.

This is what we call a typical analysis of RCs in general, of ARCs in particular.

The problem for such an analysis (found in all pedagogical grammars without
exception) is that in spontaneous speech, relative clauses in English are not al-
ways that neat. Some might say that spontaneous speech is full of dysfluencies
and/or that we should not care about the jumble in peoples heads as a colleague
working in syntax once told us, but the frequency of occurrence of such exam-
ples suggests there is something more than jumble. Alongside typical ARCs, we
found a significant number of what we label atypical ARCs. We reject the terms
ungrammatical or even non-conforming, as we would like to treat them as
part of the system.
Atypical ARCs in our data fall into two categories: (i) those in which the gap
is filled with what traditional grammar calls a resumptive (or shadow) pronoun/
phrase (gap-filled examples) and those in which no gap can be found (gap-
less examples). Below we provide a few attested examples to illustrate these
two configurations. We will carry out a more detailed discussion of the data in
Section 2.5:

Gap-filled examples:
(3) My nickname is Pan whichi I dont like iti so much (BNC)
(4) My foot is narrow in the arch area whichi I wouldve expected iti to widen
(MK)

. The references indicated in brackets at the end of each example concern the sub-corpus
from which the example has been extracted: BNC stands for British National Corpus; ICE-
GB stands for ICE-GB corpus; MK stands for Mai Kuha, who analyzed such examples from a
sociolinguistic point of view and who kindly transmitted her corpus to me. When no reference
is mentioned, this means the example has been taken from our own corpus (see below for
description). We removed all punctuation marks from the examples, even when these were
included by the transcribers of the BNC or ICE-GB.
 Chapter 2. Atypical appositive relative clauses 55

(5) If you do wash your hair then youre washing away some of the goodness out
of you | and whichi you dont want to do thati because youre carrying a child
inside you
(6) Thats a suggestion of yours which I followed whichi I didnt even want to do
thati
(7) Id also recommend er er our website er the Environmental Resources Trust
wherei theres information about my solar home and how to build one [on the
website]i its www.ert.net

Gapless examples:
(8) She gained a half pound which they were predicting shed gain five pounds
(MK)
(9) She wont do it on the carpet which I dont blame her (MK)
(10) Those are warm jammies which its been real warm lately anyway (MK)
(11) And she decided to move out which I think shes crazy
(12) And, you got a side dish with it which I had a gratin dauphinois
(13) Im gonna have someone there, just so books arent stolen which I think one was
taken last year. (MK)
(14) And we asked for time off for a seminar, for a one day seminar and which we
werent granted the time off we were granted time off without pay but not paid
leave.

To sum up, Figure 1 distinguishes between the different kinds of atypical ARCs:

Structures with a
resumptive pronoun
Gap-filled structures

Atypical ARCs Presence of the NP/PP

itself (resumptive phrase)
Gapless structures

Figure 1. Atypical ARCs

56 Appositive Relative Clauses in English

2.2 Corpus

Our corpus consists of examples personally collected from radio and television
programs or from private conversations. We completed this self-collected cor-
pus with examples taken from electronic corpora namely the British National
Corpus and the ICE-GB Corpus. Finally, Mai Kuha, from Ball State University
in the United States, who worked on speakers attitudes towards such examples
with a sociolinguistic approach, generously offered her self-collected corpus.
This method of data collection has one major flaw: we cannot provide statistics
that would reflect the frequency of such examples in comparison with typical
relative clauses.
What all these examples have in common is that they are extracted from spon-
taneous, unplanned speech. For instance, in the radio and television programs we
investigated, we only retained the parts where speakers are spontaneous, such as
interviews or debates.

2.3 Description of the data

2.3.1 Gap-filled structures

Gap-filled examples consist in examples where the gap is filled with a pronoun,
either a personal pronoun like it, as in example (3) or a deictic pronoun like that
as in (5)(6) or there (see (15) below):

(15) They were just towed across the Midway onto the bridle path, wherei they were
just sitting therei peacefully. (Prince 1990)

Gap-filled examples can also consist in ARCs where the gap is filled with a re-
sumptive phrase, as in example (7), where the PP on the website has been pro-
nominalized (where) and fronted to the beginning of the clause (wh- movement),
which in a typical ARC would have left a trace in the place of the PP. Such cases
are never mentioned and, as far as we know, no study addressing them exists.
Examples with resumptive pronouns are quite well-known in English, as is
shown by the abundant literature on the topic, in particular within the genera-
tivist framework (Kroch 1981; Doron 1982; Sells 1984; McCloskey 1990; Prince
1990, 1995; Erteschik-Shir 1992; Taghvaipour 2003, to name a few). What the
literature says to explain why resumptive/shadow pronouns appear is generally
linked to a last resort solution due to syntactic constraints: examples (16) and (17)
below are traditionally considered ungrammatical, (16a) and (17a) making the
sentences at least interpretable (Kroch 1981):
 Chapter 2. Atypical appositive relative clauses 57

(16) *Thats the man who I wonder whether Mary saw t last week.
a. ?Thats the man who I wonder whether Mary saw him last week.
(17) *People are coming out with symptoms that the doctors dont know what t are.
a. ?People are coming out with symptoms that the doctors dont know what
they are.

However, we claim, alongside Prince (1990, 1995) for instance, that not all exam-
ples can be explained by the existence of syntactic constraints. Prince (1990), who
used the attested data collected by Kroch (1981), studied such examples where no
syntactic constraints are involved, and found out that in English and in Yid-
dish resumptive pronouns are more frequent with ARCs and DRCs which play
no role in the construction of the reference of the NP-antecedent, like DRCs with
indefinite antecedents which have a specific interpretation:

(18) He bought a house which hell move into it in June. (Prince 1990:14)

What we can conclude from Princes study is that there is a correlation between
the referential independence of the antecedent on the one hand and the potential
for a resumptive pronoun on the other hand. Prince concludes that in examples
where no syntactic constraints are involved, the pronoun acts like an ordinary
pronoun, as it would in an independent clause. However, she does not address
the question of the role of the relative pronoun, which we shall do below. An-
other study, by Cann et al. (2004), in the theoretical framework of dynamic syntax
and based on constructed data, suggests that resumptive pronouns belong to the
grammar of English and agrees with Prince when she claims that not all examples
can be explained by the presence of syntactic constraints. They explain that the
presence of the resumptive pronoun is due to an extra pragmatic effect (Cann
etal. 2004:26ff.) such as a contrastive interpretation or a special emphasis on the
syntactic subjects agentivity.
These studies are interesting insofar as they demonstrate that resumptive pro-
nouns are not restricted to syntactic constraints phenomena. However, they are
limited to the analysis of the resumptive pronoun and do not offer analyses for
the relative pronoun.
The examples from our corpus belong to this category: no syntactic con-
straints are involved; furthermore, it is noteworthy that removing the resumptive
pronouns does not worsen the acceptability and interpretability of the sentences.
From a purely prescriptive point of view, they are actually better than the ver-
sions with the pronoun:
58 Appositive Relative Clauses in English

(3) a. My nickname is Pan which I dont like t so much.

(4) a. My foot is narrow in the arch area which I wouldve expected t to widen.
(5) a. If you do wash your hair then youre washing away some of the goodness out
of you. And which you dont want to do t because youre carrying a child
inside you
(6) a. Thats a suggestion of yours which I followed which I didnt even want to
do t.
(15) a. They were just towed across the Midway onto the bridle path, where they
were just sitting t peacefully.

As far as examples with resumptive phrases are concerned, removing the resump-
tive phrase also improves the acceptability of the sentence from a purely prescrip-
tive point of view:

(7) a. Id also recommend er er our website er the Environmental Resources Trust

where theres information about my solar home and how to build one.

Finally, one last counter-argument to the syntactic constraint explanation is the

existence of resumptive pronouns that occupy a subject position in the relative
clause. Such examples supposedly do not exist. However, our corpus study shows
that such examples can be found:

(19) He did a couple of things on the roof thats something but this thing reaches pub-
licity stage whichi iti could in a couple of weeks time appear in a paper. (BNC)
(20) But it was no different from me starting work up here that time. After so long
after so long whichi iti was a long time. (BNC)

In these examples the use of the resumptive pronoun it cannot be explained by the
presence of syntactic constraints. Once again, removing the pronoun improves
the acceptability of the examples from a purely prescriptive point of view:

(19) a. He did a couple of things on the roof thats something but this thing reaches
publicity stage which t could in a couple of weeks time appear in a paper.
(20) a. But it was no different from me starting work up here that time. After so
long after so long which t was a long time.

2.3.2 Gapless structures

Such examples, where no gap is present, are even more problematic for gram-
matical analyses: not only is it impossible to locate a trace/copy, but it also seems
that no antecedent can be systematically retrievable for the relative pronoun. If
we consider the examples (8)(14) and attempt to reformulate them with two
 Chapter 2. Atypical appositive relative clauses 59

separate sentences (a manipulation that can easily be done with typical and gap-
filled examples: see (1b)(6b) below), such a manipulation is impossible, as no
antecedent can be retrieved for the relative pronoun:

(1) b. This law prohibits employers from HIV testing. Some lawyers would regard
this law as tendentious.
(2) b. Dennis says youre absolutely mental. We were talking about Dennis last
night.
(3) b. My nickname is Pan. I dont like Pan so much.
(4) b. My foot is narrow in the arch area. I wouldve expected (the arch area of)
my foot to widen.
(5) b. If you do wash your hair then youre washing away some of the goodness out
of you. And you dont want to wash your hair because youre carrying a
child inside you.
(6) b. Thats a suggestion of yours which I followed. I didnt even want to follow
your suggestion.

What we have is two syntactically independent clauses which are linked, in a way
that we will discuss below, by a relative pronoun which seems to be playing the
role of a plain connective. No syntactic constraints can explain the use of such
structures; the explanation then has to lie elsewhere. We will argue in favor of a
conversational strategy.
Before turning to the analysis of gap-filled and gapless occurrences, we deal
with the debate on their legitimacy in the grammatical system of the English
language.

2.4 Dysfluencies vs. syntactic variation

When facing such data, some researchers are sometimes (mis)led to wonder about
the linguistic legitimacy, the official grammatical character of atypical examples,
gapless examples in particular. In a general way, such a debate is tantamount to

. One might legitimately wonder whether the relative pronoun still has a role as an anaphoric
marker in examples such as (9) or (11): wont do it on the carpet and decided to move out might
seem appropriate antecedents. Nevertheless this is not the case for all our examples; we argue
that the possibility to retrieve an antecedent is guided by speakers conviction that a relative
pronoun has to have an antecedent. In the analysis we suggest here, we consider that when the
anaphoric role is fulfilled by another element in the clause in the case of gap-filled structures or
simply not necessary in the case of gapless structures. According to such analyses, the relative
pronoun does not fulfill that role.
60 Appositive Relative Clauses in English

deciding where the border lies between dysfluencies/performance errors and a

significant phenomenon of English; this can only be done if one takes the position
of an arbiter. Our corpus aims to cover the uses that speakers make of ARCs and
naturally includes gap-filled and gapless examples. This goes against all grammars
of English, including the most recent academic grammars like Biber et al. (1999)
or Huddleston & Pullum (2002), who consider such examples to be ungrammati-
cal. Only the examples with resumptive pronouns are mentioned, generally in
footnotes, and are systematically identified as performance errors, as is exempli-
fied by the following two quotations:
These characteristics are associated with the difficulties of online production,
sometimes resulting in constructions that might be considered non-standard or
even a disfluency. (Biber et al. 1999:622)
Pronouns used in place of a gap in relative clauses are known as resumptive pro-
nouns. In some languages they represent a regular feature of relative clause for-
mation, but in English they are ungrammatical
 (Huddleston & Pullum 2002:1091n)

As far as gapless examples are concerned, they simply are never mentioned. It
should however be noted that the Oxford English Dictionary (1989, Vol XX:226)
somehow mentions gapless relative clauses in its entry dedicated to which: in
vulgar use, without any antecedent, as a mere connective or introductory particle.
Surprisingly, they provide examples taken from Swift and Thackeray, not particu-
larly known for their vulgarity.
It is true that English speakers have very negative reactions when confronted
with such examples. Kuha (1994), who deals with native speakers reactions to
gapless examples, has shown that consultants systematically reject this kind of
structure although they do produce them spontaneously. As a consequence, ex-
periments with native speakers are hard to conduct and bound to fail. It is also
true that some examples seem to correspond to dysfluencies: hesitation markers
such as er, erm, repetitions, rephrasing etc.:

. Kuha found such atypical structures in speakers of all ages and geographical/social origins:
It seems that users include men and women, high school graduates and university professors, na-
tives of California, Indiana, Michigan, and Ohio; their ages range from 20 to 50.; [t]he use of this
construction is not necessarily limited to informal situations () [but] is found more frequently in
speech than writing (Kuha 1994:1).
 Chapter 2. Atypical appositive relative clauses 61

(21) Well Id like Dave Girt Leeds City Id like s some recognition of West Yorkshires
problems to be evident in er the deliberations which er at the moment its its
absent it may it may have been taken into account but its absent in the exposi-
tion (BNC)
(22) Erm but er I learnt to write a fair hand, erm without too many flourishes which
er because I was always told that somebody might be looking at this in a hundred
years time or more. (BNC)
(23) Especially in the case of this fil, where the underlying reality truth the underlying
truth of the film is revealed to be something very different from what the narra-
tor himself is presenting it as I mean its a bizarre comparison in a way. (BNC)

However, their recurrence suggests that gap-filled and gapless examples are not a
random phenomenon of the English language. Gap-filled examples in particular
represent a well-known phenomenon in many languages of the world (one fifth
of the worlds languages are resumptive languages according to Hagge (1982))
including Yiddish, Arabic, Hebrew, Irish and Greek.
These structures are also found in dialectal English. As far as gapless examples
are concerned, they are said to be frequent in Scottish English (Miller 1988):

(24) you can leave at Christmas if your birthdays in December to February which
I think is wrong like my birthdays March and I have to stay on to May which
when Im 16 in March I could be looking for a job.

A very interesting study by Van den Eydn (1993) has shown that both gapless
and gap-filled exist in Dorset English, alongside other non-standard features
(e.g. the use of the relative pronoun what in ARCs and DRCs and that in ARCs,
a rare use of who...). She treats such differences with standard relativization as
syntactic variation. She shows that non-standard relativization, which includes
the existence of gap-filled (under the label relpro-cum-perspro) and gapless
constructions (under the label connectors), is not a random phenomenon but is
governed by an intricate network of language-internal, as well as language-exter-
nal factors (Vanden Eydn 1993:84). By language-internal she means semantic
(e.g. restrictive or non-restrictive status of RCs; the (in)definiteness of the ante-
cedent) as well as syntactic factors (e.g. syntactic function of the antecedent in
the MC/of the relative pronoun in the RC; RC length). As for language external

. Unfortunately, because of the difficulty to collect such examples mentioned above, we can-
not provide statistics that would detail the frequency of their occurrence.
. These also include left dislocation in her corpus in the case of resumptive pronouns outside
the relative clause (Van den Eydn 1993:71, 156): Well, the hairdresser that comes here, shes been
coming now to me five weeks and shes getting my hair to grow.
62 Appositive Relative Clauses in English

factors, these refer to the social background and gender of the speakers. Below
are two examples taken from her corpus, first of a gapless example, second of a
gap-filled example:

(25) you had the Norman invasion, which a lot of our words now have got a lot of
French origin, havent they? (Van den Eydn 1993:87)
(26) An you had to buy products that feed the pigs on it. (Van den Eydn 1993:156)

Finally, another argument, which is for us the strongest argument showing that
gap-filled and gapless examples are not simply dysfluencies and must be integrat-
ed into our corpus, is their existence in writing, in dialogues or internal mono-
logues found in fiction. In such cases, the writer wants to give a character some
specific traits and resorts to linguistic devices in order to do so. Written examples
of atypical RCs exist and their aim is to convey a certain orality, a certain reality,
to the written text. Below we provide a few examples of both gap-filled and gapless
ARCs taken from our written corpus:

(27) Your Show was on a poster, which Dora got hold of one of them, and pinned it
to the laundry wall, to liven it up; and I knew at once it was you, even though
you have another name and have grown your beard very wild. (Alias Grace, M.
Atwood. 1996. Virago: 305)
(28) At the funeral we barely talked. So you avoided me a little, which I understand
it, believe me, and can sympathize. (The Body Artist, Don DeLillo. 2001.
Picador: 5859)
(29) Oh, I dreamed of a true class reunion of our family. Just imagine their faces, if
they saw this place. Which, I might add, none of them came. (The Poisonwood
Bible, B. Kingslover. 1999. Harper Perennial: 463)
(30) She and her friends raised the money and bought a good Land Rover with a
rebuilt engine in Atlanta. Which, by the way, Mothers group has never raised
one red cent for me, to help put in upstairs plumbing at the Equatorial, for
example. (The Poisonwwod Bible: 475)
(31) Come on, N, she said, lets what the fuck! and we gone to this massive like
stately home except it werent it was a hotel, but thats what youd think, youd
think, Brideshead Refuckingvisited, which my mum loved that programme,
give her ideas, and she gets us this room like the size of a church, starts ordering
salmon and champagne and shit and dancing around in her underwear, which I
dont know why she was down to that but she was, I remember it certain. (Poppy
Shakespeare, C. Allan. 2007. Bloomsbury Publishing: 2)
 Chapter 2. Atypical appositive relative clauses 63

(32) Is this a bad time? () Because if it is, Carol said, we can talk whenever.
(). Which, by the way, Carol said, did you get the card I sent? (Falling Man,
Don DeLillo. 2007. Scribner: 24)
(33) But here, with these people, you cant even think it. You dont know what to do.
Because theyre a million miles outside your life. Which, besides, theyre dead.
(Falling Man: 81)

In a very interesting way, as noted by Kjellmer (1988), Victorian writer Charles

Dickens resorted to the use of gapless structures when writing dialogues that
staged one of the key characters in his famous novel Great Expectations. To check
on the systematicity of this usage of gapless structures, we used the electronic ver-
sion of the text to list all the occurrences of atypical instances of the relative pro-
noun which, as it is this pronoun that is part of the phenomenon. We listed about
thirty occurrences, all of which were attributed to the character of Joe Gargery.
The variety of English that Dickens put into Joe Gargerys fictitious mouth is
sprinkled with lexical, grammatical and pronunciation features (the latter being
rendered by faulty spelling). This is thus a conscious strategy for Dickens. Below
we provide a few examples:

(34) Yes, Joe; but what I wanted to say, was, that as we are rather slack just now, if
you would give me a half-holiday to-morrow, I think I would go up-town and
make a call on Miss Est Havisham.
Which her name, said Joe, gravely, aint Estavisham, Pip, unless she have been
rechrisened.
(35) For, Joe had actually laid his head down on the pillow at my side and put his
arm round my neck, in his joy that I knew him.
Which dear old Pip, old chap, said Joe, you and me was ever friends. And when
youre well enough to go out for a ride what larks!
(36) For the tenderness of Joe was so beautifully proportioned to my need, that I
was like a child in his hands. He would sit and talk to me in the old confidence,

. The use of the relative pronoun which actually signals here a shift in the topic. This is close
to the expression speaking of which, where which has often no antecedent but allows a speaker
to shift the topic without doing so too abruptly.
. Many examples exist with DRCs as well:
(i) I mean I left schools and places I didnt even know I was leaving them. (The Catcher in
the Rye, J. D. Salinger. Penguin: 8)
(ii) He stuck around till around dinner-time, talking about all the guys at Pencey that he
hated their guts. (The Catcher in the Rye: 38)
. Available on http://etext.lib.virginia.edu/toc/modeng/public/DicGrea.html.
64 Appositive Relative Clauses in English

and with the old simplicity, and in the old unassertive protecting way, so that I
would half believe that all my life since the days of the old kitchen was one of the
mental troubles of the fever that was gone. He did everything for me except the
household work, for which he had engaged a very decent woman, after paying
off the laundress on his first arrival. Which I do assure you, Pip, he would often
say, in explanation of that liberty; I found her a tapping the spare bed, like a
cask of beer, and drawing off the feathers in a bucket, for sale. Which she would
have tapped yourn next, and drawd it off with you a laying on it, and was then
a carrying away the coals gradiwally in the soup-tureen and wegetable-dishes,
and the wine and spirits in your Wellington boots.

This use of which is the exact linguistic feature thanks to which Pip, the central
character and stepson of Joe, will be able to recognize Joe at a time when, lying on
a bed, he hears him before being able to see him properly:

(37) At last, one day, I took courage, and said, Is it Joe?

And the dear old home-voice answered, Which it air, old chap.

We maintain that the use of atypical examples used by some writers to define some
of their characters is significant, attesting so to speak their actual presence in the
language. The use of gapless structures in particular was often used as a symbol of
non-standard language in the satirical magazine Punch in the nineteenth century
(Kjellmer 1988:158). We may legitimately consider that if writers had a certain
awareness of the linguistic phenomenon, both its use in spontaneous speech but
also its register distribution, then the phenomenon was common enough to be
noticed. Such written ARCs then give a specific flavor to the discourse and are
interpreted with no difficulty.
Of course atypical examples can also be found on internet blogs, where the
register is generally closer to speech than writing:

(38) I found this picture, which I thought I had lost. If you know me, which I dont
know why someone would be reading this if they didnt, Im a big Ohio State Fan.
(http://mforeman.blogspot.com/2007_03_01_archive.html)
(39) Things are changing around here. We are getting new staff members, which
they are awesome, and we are losing staff members, which is not so awesome.
(http://mforeman.blogspot.com/2007/01/things-are-changing-because-they-
are.html)
(40) We are supposed to have it tomorrow, but we did good at performance yesterday,
which introduced our high schools football team and the two junior high teams,
which I dont know why they added them, but whatever. (http://dreamerforever.
buzznet.com/user/)
 Chapter 2. Atypical appositive relative clauses 65

(41) Ah I love these types of writes. Ive written a few myself, which one of your lines
automatically made me think of one of my own. (http://www.eliteskills.com/
z/126975)

2.5 Analysis

After presenting the data in 2.3 and arguing for their legitimacy as ARCs in 2.4,
we now provide an analysis of these examples, starting with the role of the rela-
tive pronoun in gapless structures. The provided analysis will then be extended to
gap-filled structures to shed new light on resumptive examples. Finally, we explain
that the use of such structures in discourse represents a conversational strategy.

2.5.1 The role of the relative pronoun in gapless structures:

A plain connective?
The role of a relative pronoun is twofold (see Jespersen 1970 or Port Royal gram-
mar for instance):

First, as a pronoun, it stands for the referent that has been previously men-
tioned in the main clause. It clearly has a role as an anaphoric marker.
Second, it relates, or links, two clauses, one of which is typically subordinate
to the other. It also has a connective role.

This double role anaphoric and connective renders the pronoun complex,
since the two functions are synthetized in one marker. Evidence for the two func-
tions of the wh- relative pronoun can be brought to the surface if we consider
examples from diachronic (42)(43) and dialectal English (44)(45), where the
two roles are dissociated. In Old English in particular, the invariant relativizer e
could be followed by a pronoun, as is shown by the following examples (CHEL,
Vol. 1:229):
(42) & ic gehwam wille rto tcan e hiene (ACC)
and I whomever shall thereto direct Rel him
his lyst ma to witanne
of-it would-please more to know
and I shall direct anyone to it who would like to know more about it.

. This type of syntactic configuration still existed in Middle English but disappeared with
the introduction of wh- pronouns (fourteenth century). The frequency of relative pronoun +
pronoun in the embedded clause then decreased rapidly (CHEL, Vol. 2:308ff.), which suggests
that their presence was made necessary by the absence of case, gender, and number markers for
e and that.
66 Appositive Relative Clauses in English

(43) Swa bi eac am treowum e him (DAT)

So is also to-those trees Rel to-them
gecynde bi up heah to standanne
natural is up high to stand
so it is also with trees to which it is natural to stand up straight.

In some dialects of English, the dissociation of the two functions of the relative
pronoun can also be found, such as in Scottish English, in particular for the geni-
tive case (Chevillet 1991:135; Milroy & Milroy 1993:111113):
(44) the girl that her eighteenth birthday was on that day
the girl whose eighteenth birthday was on that day
(45) thats the man that his house was burnt
thats the man whose house was burnt

Most of the examples of ARCs taken from our spoken corpus display a traditional
synthetization of the double role of relative pronouns in one marker, even when
the spontaneous character of the discourse leads to hesitations, repetitions or dys-
fluencies:

(46) he was convinced the battle for the hearts and minds of the people was being
won especially among the Ovambo who form the majority of SWAPOs support
(AM)
(47) Well I got involved through Celeste who I used to dance with with London
Contemporary going back twenty years or so (ICE-GB)
(48) (RAD06_ARC1/2) But unfortunately when I go out to other countries which I
do quite often and I do enjoy it er you know, I I feel like I mean there was a quote
in the paper the other day that I said I was a British ambassador for food which
wasnt actually quite right I said I feel like the British ambassador for food
(49) No <,> there s one there s one girl friend friend of Jane who we thought prob-
ably would would be good for that post (ICE-GB)

But in our atypical examples, in particular gapless examples, the role of the rela-
tive pronoun is reduced to one function only: connective, as the anaphoric role is
simply irrelevant. This predominance of the connective function appears through
the reformulation of our examples with the replacement of who/which with a con-
nective word in the broad sense of the word, therefore introducing explicitly a
link that was implicit between the two clauses (see Chapter 1, Section 4.1.2 for a
brief discussion):
 Chapter 2. Atypical appositive relative clauses 67

(8) c. She gained a half pound although/while they were predicting shed gain
five pounds
(9) c. She wont do it on the carpet, so/and I dont blame her
(10) c. Those are warm jammies and its been real warm lately anyway
(11) c. And she decided to move out and/but I think shes crazy
(12) c. And you got a side dish with it, and/while I had a gratin dauphinois
(13) c. Im gonna have someone there, just so books arent stolen, as/because I
think one was taken last year

The sentences thus obtained are perfectly acceptable. The so-called relative pro-
noun introduces a new, independent clause, in which no gap is present. The sen-
tences shown in (8c) to (13c) have almost the same interpretation as the original
sentences. In order to define this almost same interpretation one needs to won-
der about the semantic link between the two clauses. We turn to this point in
Section 2.6, but before doing so, we would like to extend this analysis to gap-filled
examples in the following subsection.

2.5.2 Extension to gap-filled structures

If we manipulate gap-filled examples as we did with gapless examples in (8c)
(13c), we see that this is perfectly possible:

(3) c. My nickname is Pan but/although I dont like it so much

(4) c. My foot is narrow in the arch area but/yet/while I wouldve expected it to
widen
(5) c. If you do wash your hair then youre washing away some of the goodness
out of you | but you dont want to do that because youre carrying a child
inside you
(6) c. Thats a suggestion of yours which I followed but I didnt even want to do
that
(7) c. Id also recommend er er our website er the Environmental Resources Trust
as/since theres information about my solar home and how to build one on
the website its www.ert.net
(14) c. They were just towed across the Midway onto the bridle path, and they were
just sitting there peacefully.

What we claim for the role of the relative pronoun in these examples is this: the
anaphoric function is fulfilled by the resumptive pronoun/phrase while the con-
nective function is fulfilled by the wh- relative pronoun. This second role is its
sole role in such examples.
68 Appositive Relative Clauses in English

2.5.3 The predominance of which

A word needs to be said about the predominance of which, as nearly all our atypi-
cal examples involve this relative pronoun. This is no coincidence. Contrary to
other relative pronouns such as who(m), when, or where, the pronoun which can
have, and only in an ARC, a greater variety of antecedents, including sentential
antecedents (see previous chapter, Section 2.3.2). The specificity of which is that
its antecedent can be vaguer, looser in a way, as is the case for this and that (recall
that ARCs with sentential antecedents can easily be rephrased with coordinated
clauses starting with and this/that). This possibility makes which a marker that
is linked to its antecedent in a less constrained way. This may explain the at-
tenuation (even loss) of its anaphoric role to the benefit of its connective role,
especially in an ARC that is linked to the main clause in a looser way too (as
opposed to a DRC).

2.5.4 The Latin linking relative

The phenomena described above are reminiscent of the Latin linking relative that
can be found at the beginning of Latin relative clauses. This particular type of
relative clause corresponds to an independent clause and starts with the pronoun
which has first and foremost a connective role although an antecedent can be re-
trieved. This establishes a semantic link between the two clauses such as causality
or concession, but plain coordination is also involved. The pronoun then plays
the role of a coordinator and cannot be preceded by any conjunction (Lavency
1998:10). The following two examples illustrate this phenomenon (Ernout &
Thomas 2002:438):
(50) Centuriones nutu vocibusque hostes
centurions-NOM.PL voice-ABL.SG gestures-ABL.PL-and enemies-ACC.PL
vocare coeperunt: quorum progredi
call/invite-INF.PRS start-3PL.IND.PRF: REL-GEN.PL move forward-INF.PRS
ausus est nemo
dare3SG.IND.PRF no one-NOM.SG
By gestures and by voice they started to invite the enemies; (but) none of
them dared to move forward.
(51) perutiles Xenophontis libri sunt;
very useful-NOM.PL Xenophon-GEN.SG books-NOM.PL be-3PL.IND.PRS;
quos legite, quaeso, studiose.
REL-ACC.PL read-2PL.IMP.PRS, ask-1SG.IND.PRS, carefully.
The books by Xenophon are very useful; (so) read them carefully, I pray
you.
 Chapter 2. Atypical appositive relative clauses 69

2.5.5 Semantic and syntactic link with the main clause

2.5.5.1 Inferences. Suggesting that which has a peculiar connective function
means wondering about the kinds of complex inter-clausal semantic relations
that exist in such atypical examples. The reformulations with connectives show
that these include concession and causality for instance. This is not limited to
atypical ARCs. We mentioned, in Chapter 1, Section 4.1.2, that an implicit, inter-
clausal link could exist between the ARC, then interpreted as an adverbial clause
(Quirk et al. 1985:1240; Cornilescu 1981) and the MC. Below are the examples
we provided:

(52) Salim, who passed all his exams, is happy. (implicit causality)
(53) Salim, who failed all his exams, is happy. (implicit concession)

As in these typical examples, in atypical examples, one can reconstruct inferentially

a semantic link between the two clauses: causality in (7), (9) and (13) for instances;
concession in (4), (6) and (8). We must however specify that this is not always the
case: in some examples, the informational contents of the main clause and of the
atypical ARC merely add to each other without any extra semantic link.
Interestingly, this brings out another debate, on the syntactic relationship be-
tween the two clauses this time.

2.5.5.2 Coordination or subordination? When we manipulated our original ex-

amples to replace the relative pronouns (which most of the time) with what we
called connectives in the broader sense of the term, we replaced the relative pro-
nouns with subordinators (because, although, since, while) but also sometimes
with coordinators (and, but, yet10). This specific use of the relative pronoun as a
connector in gap-filled and gapless examples therefore brings up the question of
the nature of the syntactic link between the main clause and the relative clause.
This question finds a natural connection with the syntactic debate we men-
tioned in Chapter 1, Section 2.5 concerning the syntactic status of ARCs. Among
the different approaches (ranging from radical orphanage to full integration) is
the derivation of ARCs from coordinated independent sentences (e.g. Emonds
1979). De Vries (2002, 2006) recent proposal on specifying coordination is also
relevant here.

10. The status of yet (coordinator/subordinator) is problematic (in particular, the fact that it
can be preceded by and questions its status as coordinator). However, yet is very similar to a
coordinator (see Biber et al. 1999:80, for instance, or Quirk et al. 1985:263, who claim uncon-
ditionally that yet is a coordinator).
70 Appositive Relative Clauses in English

The fact that in our examples, the connective role of the relative pronoun
establishes a link that is closer to coordination than subordination is another
argument to fuel the debate syntacticians have had for several decades now. In
addition, such an analysis also questions the very legitimacy of this traditional
coordination/subordination dichotomy, considering clause-combining as a more
complex phenomenon than a binary vision.

2.5.5.3 Conclusion. Gapless and gap-filled ARCs do not correspond to tradi-

tional depictions of ARCs such as the ones found in reference and pedagogical
grammars, which therefore do not provide a complete picture of the use of ARCs.
The analysis we provide here confers a connective role to the relative pronoun
while the anaphoric role traditionally fulfilled by the same marker in standard
structures is fulfilled either by the resumptive/shadow pronoun (gap-filled struc-
tures) or simply not relevant because non-existent (gapless structures). The na-
ture of the link between the MC and the atypical (A)RC shows that inferential
links can exist and that the kind of syntactic link that exists between the main
clause and the ARC is problematic for a restricted, binary vision that considers
only coordination and subordination.
However, one question remains to be answered: since we know that syntactic
considerations are not the sole explanation for the use of such ARCs, what is then
the linguistic motivation behind speakers use of such structures? We argue below
in favor of a conversational strategy.

2.5.6 A conversational strategy?

One may wonder about speakers choice for gapless or gap-filled structures when
an independent clause, coordinated or juxtaposed, could suffice. It seems legiti-
mate to think that a very specific role can be attributed to such a use of the relative
pronoun, a use that corresponds to a conversational strategy. And this use seems
to be supported by prosodic evidence as well.
According to Kjellmer (1988), who discussed gapless examples, the use of
the relative pronoun, which he calls adverbial or conjunctional, in such atypical
examples corresponds to the necessity of oral English clauses to be introduced
rather than juxtaposed without any link between them. His corpus study shows
that most clauses in spoken English are introduced by some element (coordinator,
connective). The relative pronoun which (Kjellmer discusses only examples with
which) appears to take on that role in what he calls substandard oral English:
 Chapter 2. Atypical appositive relative clauses 71

The use of clause-opening which will be thought of as a mark of correctness gen-

erally. It is reasonable to suppose that clause-opening relatives thus come to be
regarded as practically redundant elements, to be used at the speakers discretion
for stylistic purposes. (Kjellmer 1988:161)

This idea is interesting, but we would like to go further by speculating that such a
use of the relative pronoun in discourse corresponds to a floor-keeping maneuver.
The idea is that when starting a clause with a relative pronoun, a speaker has few-
er chances of being interrupted than when uttering an independent clause. The
speaker then starts with the relative pronoun to show his/her intention to pro-
ceed, but actually utters an independent clause, in which no gap appears, either
because it is filled with a resumptive pronoun or because it simply does not exist.
In both cases the part following the relative pronoun is an independent clause
which is linked to the main clause thanks to the so-called relative pronoun.
This strategy is supported by prosodic evidence. It is generally assumed, al-
though systematized in an erroneous way (see for instance Bolinger 1989), that
ARCs are marked off prosodically from the rest of the sentence, in particular
through the use of a pause or a break in the melody, specifically between the
antecedent and the relative pronoun.11 However the study of our corpus of ARCs
as a whole shows that pauses sometimes occur after the relative pronoun, as is il-
lustrated by (54) and (55) below:12

(54) I think wed be looking at a different situation where {1030ms} companies would
be dropping from record profits to er to breaking even in a in a month

11. Quirks (1957) study on spoken relative clauses established that ARCs (or non-restrictives
to use his terminology) are characterized by open juncture (), a fresh intonation contour, and
a change () in the degree of loudness (Quirk 1957:101). Many grammars of contemporary
English signal the presence of pauses as a systematic criterion: A supplementary relative is
marked off prosodically from the rest of the sentence by having a separate intonation contour;
there is typically a slight pause separating it from what precedes and, if it is non final in the sen-
tence from what follows (Huddleston & Pullum 2002:1058). ARCs have been classified as par-
entheticals (e.g. Bolinger 1989) traditionally characterized by the presence of pauses, a higher
speech rate and a lower pitch range among others, although all criteria are not systematically all
present simultaneously (see Auran & Loock 2010 for more details). In particular the presence
of pauses is not compulsory, as is shown by quite a few examples in our corpus and by studies
such as Tao & McCarthy (2001). Nevertheless, the presence of pauses is one of the most signifi-
cant criteria: Garro & Parker (1982:157) have shown in a study on planned English that in
the case of ARCs, pauses were ten times as long as pauses in the case of DRCs.
12. These pauses are longer than biologically-constrained pauses, estimated at 200 ms (Morel
& Danon Boileau 1998:1415).
72 Appositive Relative Clauses in English

(55) You know everything its an entire roller coaster for the audience and the
audience just dont sit there and watch there but become actually engulfed in
what theyre seeing which {480ms} is why its won you know Pulitzer Prize for
drama

This is something we also noticed with atypical examples. We unfortunately have

no recorded examples to offer. This does not mean that the intonation unit is
closed after the pronoun instead of before it; it simply means that the pause occurs
after the relative pronoun and not before it, as is assumed by most descriptions
of relatives in spoken English. The speaker then interrupts his/her speech after
the relative pronoun, thus leaving a clause unfinished with the relative pronoun
in mid-air, clearly indicating s/he has not finished speaking. The addressees
are therefore invited not to interrupt. This kind of strategy is called floor-keeping
maneuver and is recognized as a conversational strategy by Biber (1988) and
Wennerstrom (2001) in particular. As is emphasized by the latter, it is safer to
pause in midphrase than at a phrase boundary if one wishes to avoid interruption
(Wennerstrom 2001:173). Wichmann (2000) shows that such configuration is very
frequent in spontaneous speech, the boundary of a clause occurring not before
but after a conjunction. Interestingly, Schleppergrell (1991) found a similar use of
the subordinator because, which in some instances can go beyond the expression
of causality, playing a role in the organization of discourse, as in (56) below:

(56) The fifth position break is in a lot of dances. Especially in a lot of Latin dances.
Because this is the fifth position break. [Demonstrates]

The role of because is then to express the speakers intention to proceed, which
Schleppergrell calls further elaboration to hold the floor.
The clear advantage of leaving the relative pronoun in mid-air, used accord-
ing to the analysis suggested here as a connector between two clauses, instead
of a coordinator like and, for instance, is that the addressee(s) are less likely to
interrupt the speaker after what sounds like a relative pronoun, even if the rest
of the sentence shows no gap as a traditional RC would but instead represents an
independent clause.

3. Conclusion

The aim of this chapter was to complete the definition of Appositive Relative
Clauses provided in Chapter 1 by examining ARCs whose syntactic configura-
tions do not correspond to traditional descriptions found in reference and peda-
gogical grammars. Gap-filled and gapless examples are considered here a part of
 Chapter 2. Atypical appositive relative clauses 73

the system of relativization in present-day English. Atypical ARCs thus belong to

our corpus.
Although many questions remain unanswered,13 the definition of our object
of study is now considered complete and we can turn to what forms the crest of
this study: the discourse functions of ARCs, both written and spoken.

13. In particular, one could wonder about a possible process of grammaticalization. The relative
pronoun which is used regularly with animate antecedents in Tyneside English for instance (see
Milroy & Milroy 1993) but in configurations where who would be expected in standard English.
One could argue that which might then be moving on one extra step towards the grammatical
pole, taking on new grammatical functions, such as the role of a plain connective. Strangely
enough, such a situation would be reminiscent of Old English, where the invariable particle
e, which had no syntactic link with an antecedent provided one could retrieve one , could
introduce a relative clause with a resumptive pronoun (see for instance Lancri 1999:2627).
Such a tentative point of view naturally requires further investigation.
 part ii

The discourse functions of ARCs

A suggested taxonomy for written
and spoken English

In this part, we will address the question of what speakers use ARCs for in dis-
course. Given that ARCs play no role in the referential definition of their ante-
cedents, what then is/are their role(s) in discourse? We saw in Chapter 1 that this
question has not really been addressed by researchers, who have, with a few ex-
ceptions only, systematically defined the functions of ARCs negatively, essentially
saying that ARCs are not DRCs. In an attempt to provide a positive definition of
ARCs, we conducted a corpus study of written and spoken contemporary English
and investigated different registers such as fiction, press (quality and tabloids),
scientific texts, and spoken English, planned and unplanned.
Part II is organized as follows: Chapter 3 is dedicated to the presentation of the
corpus, which shows the characteristics of ARCs in usage, sometimes quite differ-
ent from the artificially constructed examples found in pedagogical grammars,
both in the written and spoken registers. We justify in Section 2.1 the existence of
two separate studies. For both registers we also investigate the morphological, syn-
tactic and semantic-pragmatic characteristics of ARCs so as to compare with gen-
eralizations given by grammars and also to compare spoken and written English.
These criteria are then used in Chapter 4 to illustrate the proposed taxonomy.
Chapter 4 is dedicated to the presentation of our taxonomy of the discourse
functions of ARCs. The general aim of this chapter is to establish a link between
syntax and discourse, between syntactic form and contextual meaning. Each cat-
egory is defined and illustrated with examples that correspond to different dis-
course strategies. The existence of each category is justified by a series of tests.
Each category is also correlated with morphosyntactic, semantic and pragmatic
characteristics defined in Chapter 3, as well as with prosodic characteristics for
spoken ARCs through a study to check whether differences in discourse functions
for one specific syntactic structure might correspond to differences in prosodic
realizations.
 chapter 3

The corpus
ARCs in usage

1. The written corpus

1.1 Four sub-corpora

Our written corpus consists of 450 sentences each containing an ARC as defined
in part one: so as to avoid the DRC-ARC distinction, the ARCs that we collected
are unambiguously appositive: their antecedents are determined independently
of the embedded RC, and punctuation separates the antecedent from the relative
pronoun.
As we wanted to avoid having a restricted view of the discourse functions
fulfilled by ARCs, we collected examples from different registers, with a variety
of addressees and writers. Our 450-utterance written corpus thus consists of con-
textualized examples taken from four sub-corpora, QUAL, TABL, FIC and SPEC
defined below:

QUAL[ITY NEWS]: about 200 examples of ARCs taken from what is tradi-
tionally called the quality press, British and American (e.g. The Times (GB),
The Guardian, The Observer, The International Herald Tribune).
TABL[OIDS]: about 100 examples of ARCs taken from tabloid newspapers,
mostly British (e.g. The Sun, News of the World).
FIC[TION]: about 100 examples of ARCs taken from contemporary fiction
(e.g. Margaret Atwoods The Robber Bride, Nick Hornbys About a Boy, Gra-
ham Swifts The Sweet Shop Owner).
SPEC[IALISED TEXTS]: about 50 examples of ARCs extracted from journal
articles from the fields of psychology and medicine.

The corpus of examples was then systematically studied in order to provide

quantitative data giving a precise picture of the characteristics of ARCs found
in real, usage-based occurrences as opposed to the constructed examples of
pedagogical grammars, and also to illustrate the taxonomy we will propose in
78 Appositive Relative Clauses in English

Chapter 4. The relevance of morphosyntactic criteria in particular will become

clearer when we proceed to the study of the characteristics of each category of
ARC in the taxonomy.
The examples cited to illustrate our study are systematically specified with
the sub-corpus they were extracted from. Our labeling system also distinguishes
sentential relatives (SR), i.e. relatives with a non-nominal, sentential antecedent;
other examples are simply labeled ARC. Additional, off-corpus examples have
been added to the original corpus when interesting and strategy-revealing occur-
rences of ARCs were discovered. These have no specific references.

1.2 Morphosyntactic, semantic and pragmatic characteristics

of written ARCs

Table 1 sums ups the different characteristics that were examined.

Table 1. Criteria for the quantitative study

Punctuation type: comma, dash, (square) brackets, period

Morphosyntactic nominal/sentential ARC
criteria middle/final position in the sentence
choice of relative
type and syntactic function of antecedent
syntactic function of relative pronoun
tense of the verb in the ARC, aspect
Semantic criteria type of verbs (state/action)
subjectivity markers
Pragmatic criteria informational status of antecedents referent
informational status of the information conveyed in the ARC

In the following subsections we provide quantitative generalities on written ARCs,

based on the study of our corpus. These will enable a comparison with the charac-
teristics of ARCs according to their function in discourse.

1.2.1 Punctuation
A written ARC is traditionally separated from its antecedent with a comma (more
than 9 times out of 10). Dashes are also a frequent option (6%), which suggests
some affinity with parenthetical structures (see Part III). Separation with a period,
 Chapter 3. The corpus 79

that is, separating antecedent and ARC into two independent sentences, is quite
marginal but possible, although generally severely condemned by usage guides:

(1) (FIC_SR203) Sixty years ago, all the things Will relied on to get him through
the day simply didnt exist: there was no daytime TV, there were no videos,
there were no glossy magazines and therefore no questionnaires and, though
there were probably record shops, the kind of music he listened to hadnt been
invented yet. (Right now he was listening to Nirvana and Snoop Doggy Dogg,
and you couldnt have found too much that sounded like them in 1933.) Which
would have left books. Books!
(2) (FIC_SR206-7-8) But if being Macaulay Culkin meant being good at drama,
then forget it: he was crap at drama, because he hated standing up in front
of people. Which was why he hated school. Which was why he wanted to be
Macaulay Culkin. Which was why he was never going to be Macaulay Culkin
in a thousand years, let alone in the next few days. He was going to have to go
to school tomorrow.

1.2.2 Morphosyntactic characteristics

The written ARCs in our corpus are mostly juxtaposed to a nominal antecedent;
sentential relatives (SRs) are marginal, although the difference is less significant
in fiction texts, in which 25% of ARCs are SRs. Concerning the position of ARCs,
most of them are in final position (65% for the whole corpus, up to 90% for ARCs
extracted from the fiction corpus) but one register-related exception needs to be
noted: ARCs extracted from the tabloid corpus (TABL) are mostly in middle po-
sition (66% of them), which shows that a link exists between the position of the
ARC and the type of register. As we will see, this can be linked with discourse
functions as well.
ARCs are mostly introduced by the pronoun which, except for the TABL cor-
pus, where who is the most frequent pronoun. (This is to be explained by the
informative goal of this kind of press, which provides information on celebrities).
The pronoun that is extremely marginal (an example is provided in (3)), as is
whom, on its way to fossilization according to Khalifa (1999) and generally re-
placed by who. However, whom strangely appears in cases where it is supposedly
ungrammatical (see example (4)). The pronoun where is quite frequent (7% of
ARCs), but its meaning is not always purely locative (see (5) and (6)):

(3) (QUAL_ ARC52) Nomuras withdrawal immediately cast into doubt the last
tranche of 30m promised by the Millenium Commission, that was to be repaid
out of the purchase money from Nomura.
80 Appositive Relative Clauses in English

(4) (QUAL_ARC116) Now the attention turns to Merkov, whom Bulgarian press
agencies claim returned positive tests for a banned diuretic two weeks before
the games.
(5) (QUAL_ARC150) Sometimes new names come to you easily, Mr Powell said.
When they dont, my favorite trick is looking at the Monday sports results,
where you get all kinds of names and nationalities that you can corrupt into
something pronounceable.
(6) (QUAL_ARC151) Occasionally we run a competition among the staff, where
everyone writes down 100 names.

Antecedents are referentially stable independently of the embedded ARC. Quite

a significant number of them are proper names (60% for TABL for instance, 40%
for QUAL). Definite NP antecedents represent 30% of ARCs in the whole corpus
(example (7) illustrates this configuration) while indefinite NP antecedents rep-
resent 17% (see example (8)). The type of antecedent will also be related to the
discourse functions of ARCs.

(7) (FIC_ARC229) What reasons could he give for not inviting Suzie round to his
flat, where there were no toys or cots or nappies or bowls, where there was no
second bedroom even?
(8) (QUAL_ARC4) The government of President Thabo Mbeki has been sitting on
a report sponsored by the EU, which suggests that giving retro-viral drugs to
pregnant women could save the lives of nearly 14,000 babies.

Example (8) shows that indefinite antecedents are possible with ARCs. The NP a
report sponsored by the EU has a specific interpretation. But generic indefinites are
also possible, as illustrated by (9), where the antecedent is a plural indefinite NP:

(9) (QUAL_ARC88) Gay people are likely to acquire rights over the long term
which would put them on a par with heterosexual couples who can get mar-
ried.

. Although the relative pronoun is the syntactic subject of the verb returned, it bears the
accusative case. The presence of Bulgarian press agencies claim triggers this use of the accusa-
tive form instead of who. Although condemned by purists and newspapers style guides (see
for instance The Times recommendations on http://www.timesonline.co.uk/tol/tools_and_
services/specials/style_guide/article986741.ece), this linguistic phenomenon is not so mar-
ginal and has been studied by Haegeman (2000) in particular, who analyzes such examples
as case reassignment: wh- movement brings the pronoun under the scope of the verb (here,
claim), which triggers the use of accusative.
 Chapter 3. The corpus 81

Antecedents are mostly syntactic subjects in the main clause (35% for the gen-
eral corpus, a percentage that reaches 65% for TABL). A frequently occurring
combination is a relative pronoun that is also subject in the ARC (80% of rela-
tive pronouns occupy a subject position in our written corpus). These combined
results are highly significant in terms of discourse functions. In such conditions,
information is provided about the discourse topic twice: through the predicate
associated with the subject in the MC and through the informational content con-
veyed by the ARC. We show below that the discrepancy between this position as
a discourse topic/syntactic subject, on the one hand, and the informational status
of the referent (new information), on the other hand, is a trigger for the use of an
ARC. A significant number of ARCs have antecedents that are adjuncts (time in
(10), place in (11)), a configuration that is paralleled in the ARC. This is the sec-
ond most frequent configuration in our corpus.

(10) (QUAL_ARC48) C. Dyas () confirmed that the blood donated by the seven
vCJD victims was taken before 1999, when the authority introduced new
safety procedures designed to remove prions, the infectious proteins thought
to transmit vCJD.
(11) (QUAL_ARC137) After Mohammeds death in 632, the faith of Islam spread
under the rule of successive caliphs () across Central Asia and into the
Western Mediterranean, where a corner of southern Spain remained Islamic
until the end of the 15th century.

The results concerning the tense and aspect of the verbs in the ARCs are quite
uniform across the sub-corpora: simple tenses (present and past) are by far the
most frequent. Aspects be + -ing and have + -en are marginal.

1.2.3 Semantic characteristics

The first semantic criterion that we controlled for concerns the type of event de-
scribed by the verb in the ARC, which determines the semantic role of the ARC
as a whole. In particular we wanted to determine whether this second predication
established a permanent property (12) or a specific event in relation with the an-
tecedent (13).

(12) (FIC_ARC272) Those of you brave enough will, in time, be allowed to play
rugby (which is similar to American football, but does not involve stopping for
a rest every twenty seconds or wearing full kevlar body armour like nancies).
(13) (QUAL_ARC149) Loving Graham Greene, it must be said, has its derivative
side, though Emerson owes a stylistic and thematic debt less to Greene than to
Joan Didion, who long ago patented the terse, cool, elliptical, ironic style that
Emerson adopts.
82 Appositive Relative Clauses in English

We first used the traditional state verbs vs. activity verbs distinction. However,
this proved unsatisfactory as the combination of an activity verb and a simple
present tense results in the attribution of a permanent property to the subject. We
thus classified the events into general properties ((14)(15)) and specific events
((16)(17)). This distinction proved useful to define the discourse-based catego-
ries we identified (see below).

(14) (QUAL_ARC57) Welcome to the Costa del Cardigan, where the sun never sets
and you never want to go home.
(15) (QUAL_ARC130) Japan is situated in one of the most seismically active areas
in the world, where several tectonic plates are coming together, causing frequent
tremors.
(16) (QUAL_ARC56) So we asked a man, who shrugged his shoulders and disap-
peared into a nearby shop.
(17) (SPEC_ARC435) But it is also likely to want reassurance from Mr Peyrelevade,
who took over at Crdit Lyonnais in 1993, that he did not know about the alleg-
edly illegal operation, run via Altus Finance, a former subsidiary.

Another semantic criterion that proved useful for our study was the presence or
absence of subjectivity markers in the ARC. This criterion is difficult to evaluate.
What we mean by subjectivity is the presence, through the use of certain lin-
guistic markers, of the speakers/writers opinion, judgment or assessment on the
antecedent or the subject/predicate relation in the main clause. The interpretation
or evaluation of facts and events was also considered as subjective. While some
examples are obviously evaluative ((18)(21)), others convey more subtle subjec-
tivity ((22)(24)):
(18) (FIC_ARC217) And theyd been to have a look at his school, which was big and
horrible, and wandered around their new neighbourhood, which was called
Holloway, and had nice bits and ugly bits, and theyd had lots of talks
(19) (FIC_ARC216) He had been to see two films with his mum, Home Alone 2,
which wasnt as good as Home Alone 1, and Honey, I Blew Up the Kid, which
wasnt as good as Honey, I Shrunk the Kids, and his mum had said that modern
films were too commercial, and that when she was his age
(20) (FIC_ARC220) He never managed to strike up much of a rapport with Maisy,
Angies mysteriously sombre five-year-old, who seemed to regard him as frivo-
lous to his core.
(21) (QUAL_ARC113) After Atlanta, where southern hospitality was killed off by
incompetence and corporate greed, the movement desperately needed a feel-
good Olympics.
 Chapter 3. The corpus 83

In these examples, the markers of subjectivity are obvious: in (18) and (19) the
use of the adjectives horrible and (not as) good (as) clearly shows the presence of
an evaluation by the narrator of the quality of the school in (18) or of the movies
in (19). In (20) the use of the verb seem provides the reader with the narrators
assessment of the situation. Finally in (21) the use of nouns incompetence and
corporate greed shows the journalists evaluation of what took place during the
Atlanta Olympic Games. In all these examples, there is a rupture or disjunction
between the main clause (factual) and the ARC (evaluation). The discourse is
interrupted by a judgement or an evaluation of the narrator in fiction texts or the
journalist in press articles. As a consequence, the use of an adjective like big in
(18) also conveys the evaluation/perception of the narrator, more than a factual,
objective description of the school.
In (22)(24), the subjectivity is not so explicitly marked:

(22) (TABL_ARC343) The singer, who believes she is a rock diva trapped in the body
of a pop star, launched into her new single
(23) (FIC_ARC248) [I]f they can avoid calling her little lady, or saying they werent
expecting her to be so feminine, by which they mean short.
(24) (FIC_SR222) West smiles, which means he knows she cant hear what he hears
but likes her for not saying so.

In (22), it is the use of the verb believe, which presupposes she might not be such a
rock diva in the body of a pop star, that triggers the sarcastic and therefore sub-
jective interpretation that the ARC conveys. In (23), the relationship feminine=
short is the narrators interpretation of what they really mean when using the
word feminine. This is not an objective association. In (24), the event West smiles
gives rise to an interpretation of the facial gesture by the narrator: the fact that he
smiles is interpreted (with, by definition, a certain degree of subjectivity). Once
again, in such examples, a rupture or disjunction is present between the content
of the main clause and that of the ARC. This will give rise to effects of polyphony
and will be studied in detail in the next chapter.
Even more subtle subjectivity is found in the following example:

(25) (QUAL_ARC169) Mr Blair, whose son attends a selective school, complained

that too many comprehensives had a one-size-fits-all mentality

This time the subjectivity is totally implicit and has to be inferred by the addressee,
who needs the whole context to understand the subjective aspect of the necessar-
ily factual information content of the ARC. This sentence, in its context, is to be
understood as a criticism of Tony Blairs own criticisms on comprehensive schools
in Britain. It is to be understood as a de-legitimization of Blairs comments since
84 Appositive Relative Clauses in English

his son, at the time, attended a selective school. The journalists implicit comment
is: what does he know about the situation in comprehensive schools? However,
this subjective comment is not linguistically encoded: it has to be inferred by the
addressee(s). This example will be commented on again later in the next chapter.
To classify the ARCs in our corpus according to the degree of subjectivity ex-
pressed in the clause, we resorted to three categories: (i) S+ for the cases where the
informational content is overtly subjective (use of evaluative adjectives, adverbs,
verbs; nouns with positive/negative denotations and connotations; modal aux-
iliaries or verbs that express the speakers/writers evaluation or interpretation);
(ii) S for the subjectivity that is conveyed less explicitly or in an ambiguous way;
(iii) S for ARCs where the conveyed information is purely factual and objective,
without any shift of point of view as compared with that in the main clause.
The corpus study revealed that the majority of written ARCs could be clas-
sified as S+ and S, i.e. conveying at least a certain degree of subjectivity. How-
ever, these global results conceal the heterogeneity of the results for the different
sub-corpora: there is a clear-cut opposition between QUAL and SPEC on the one
hand, for which most ARCs convey non-subjective information, and FIC and
TABL on the other hand, for which ARCs convey subjective information, either
through the use of explicit commentaries, judgments and opinions or through the
use of polyphony effects due to shifts in focalization for fiction texts. This opposi-
tion seems perfectly natural if we consider the informative goal of the different
types of discourse: objective reports for QUAL and SPEC; scandals, debates and
harsh criticisms for TABL and natural intermingling of external and internal fo-
calization (and the use of direct and indirect speech) in literary texts.
The corpus study also reveals a strong correlation between subjectivity and
sentential ARCs (as opposed to ARCs with nominal antecedents): more than 9
out of 10 sentential ARCs are classified as S+ or S:
(26) (QUAL_SR7) Tennis is possibly the least popular sport at the Olympics, which
is quite something when it is up against Greco-Roman wrestling and walking.
(27) (TABL_SR302) On the same show Stephanie revealed she and Triple H were
trying for a baby, which hints at another awful gimmick in the future.
(28) (TABL_SR301) We are all working too long hours, which is making us ill, less
productive and is not supporting family life.
(29) (FIC_SR228) Karen is far away, which is just as well.

1.2.4 Pragmatic characteristics

Pragmatic-related characteristics of ARCs are at the core of our study. In particu-
lar, the informational status of ARCs, their antecedents and the main clauses they
are attached to, are all crucial in determining the discourse functions of ARCs. By
 Chapter 3. The corpus 85

ASSUMED FAMILIARITY

NEW INFERRABLE EVOKED

qp qp qp
Brand new Unused Noncontaining Containing Textually Situationally
inferrable inferrable evoked evoked

Figure 1. Princes (1981) taxonomy

informational status we refer to the old/given or new character of the information

that is conveyed. Such concepts have often been used in different ways by re-
searchers: as shown by Prince (1981:225232), at least three different definitions
of givenness can be found in the literature:
Givennessp (predictability/recoverability): the speaker assumes that the hear-
er can predict or could have predicted that a particular linguistic item will or
would occur in a particular position within a sentence.
Givennesss (saliency): the speaker assumes that the hearer has or could ap-
propriately have some particular thing/entity/ in his/her consciousness at the
time of hearing the utterance.
Givennessk (shared knowledge): the speaker assumes that the hearer knows,
assumes, or can infer a particular thing (but is not necessarily thinking
about it).
Prince defined a useful taxonomy of given and new information, a taxonomy that
we use here to determine the information status of the different elements in our
corpus examples (ARC and antecedent/main clause). We further complement her
taxonomy, suggesting extra categories. Below we define her taxonomy and ex-
plain how it has been used and extended for our own corpus study.
Prince (1981) starts by defining assumed familiarity, which she divides into
three categories (see Figure 1).
In Prince (1992), she refines her analysis on the informational status of con-
stituents (mostly NPs) and distinguishes between three types of new/old infor-
mation:
discourse new vs. discourse old information: has the entity been men-
tioned or not in the prior discourse?
hearer new vs. hearer old information is the information given/new de-
pending on the speakers assumption as to the state of knowledge of his/her
addressee(s)?
inferrables: information whose existence can be inferred by the ad
dressee(s)
86 Appositive Relative Clauses in English

Our corpus study relies on this taxonomy of given/new information, as discrep-

ancies between the discourse new/old status and the hearer new/old status of the
information is frequent. An essential discourse function of ARCs is based on
such a discrepancy, thus representing a compromise we explain in detail in the
next chapter.
For the purpose of our study, we extended Princes (1981, 1992) taxonomy in
two different directions. First, while Prince applies it to entities that are designat-
ed exclusively by NPs, we apply it to the informational content of clauses as well
as to sentential antecedents. Such an extension is not innovative: in particular,
Birner & Ward (1998) (for topicalization and dislocation among others) or Miller
(2001) (for the extraposition of subject that-clauses) also apply the taxonomy to
the informational contents of clauses. Second, as we needed to take into account
the diversity and heterogeneity of a journalists addressees, we added extra catego-
ries to the taxonomy. Since the hearer new/old status of information depending
on the assumed shared knowledge between the speaker and his/her addressees is
impossible to determine precisely, we added an intermediate category, viz the in-
determinables. We divided the indeterminables into the following categories:

hearer new+ (HN+): the information is likely to be unknown by the ad

dressee(s)
hearer old+ (HO+): the information is likely to be known by the addressee(s)
hearer new/old (HN/HO): cases for which it is impossible to decide whether
the addressee(s) know(s) the information or not

We now provide the results obtained for antecedents and then for ARCs. A large
majority of antecedents (with the exception of the TABL corpus) are discourse
new (78% for QUAL, 73% for FIC, 64% for SPEC and 47% for TABL). What we
can conclude from this is that ARCs are added to an element that is mentioned for
the first time in discourse, requiring further elaboration in addition to its refer-
ential stability. On the other hand, most antecedents are hearer old or inferrables
(70% in our global corpus). The use of an ARC, then, is not to provide addressees
with information for the identification of referents but rather to provide extra
information, the type of which remains to be determined.
Almost exclusively, ARCs convey discourse new information. At the same
time, however, the hearer new/old distinction provided results that are not ho-
mogeneous: apart from the FIC corpus, where ARCs naturally bring hearer new
information to the readers, only about half of the other ARCs bring hearer new
information (60% for QUAL, 55% for TABL and 64% for SPEC). Only very rarely
is the information hearer old, although it can be:
 Chapter 3. The corpus 87

(30) (QUAL_ARC88) Gay people are likely to acquire rights over the long term
which would put them on a par with heterosexual couples who can get mar-
ried.

The subject-predicate relation [heterosexual couples/can get married] conveys defi-

nitely hearer old information for all addressees. Such a configuration corresponds
to a specific discourse strategy: opposing two categories of people, heterosexual
people vs. homosexual people. The discourse old information is relevant in that
it defines in an asymmetrical way the properties of the other category of people
(homosexuals), who at the time the article was written could not get married.
The most striking result of the study on the informational status of ARCs is
the number of cases that belong to the indeterminables category (HN+, HO+,
HN/HO, as defined previously): these ARCs represent respectively 37%, 44% and
34% in the QUAL, TABL and SPEC sub-corpora. The significance of such inde-
terminable informational status will be revealed by the main discourse function
of ARCs to be discussed in the next chapter.
Finally, one of the pragmatic properties that is systematically mentioned in
pedagogical and reference grammars, as well as in the linguistic literature on rela-
tive clauses in general, is the systematic suppressibility of ARCs (see Chapter 1,
Section 4). Suppressibility is probably one of the greatest myths concerning ARCs.
Our study shows that it is not systematic, even if we consider a pure referential
analysis of the examples: removing an ARC can lead to semantically unacceptable
sentences within the context from which they have been extracted. Our definition
of the different discourse functions of ARCs below reveals that, first of all, such
systematicity is false, and, second, that suppressibility depends on the discourse
function of the ARC.

2. The spoken corpus

2.1 Definitions

Modern linguists long gave priority to the written register, considering speech
as secondary because it was seen as filled with performance errors, dysfluencies,
and disintegrated discourse; the written language was seen as more amenable to
systematic analysis. In the mid-20th century, linguists started studying speech,
both planned and spontaneous. The development of new technologies, which
first allowed the existence of recordings that provided data to study and, later
in the 20th century, the existence of electronic corpora has given linguists the
tools enabling spoken language to be studied more easily than before, when they
88 Appositive Relative Clauses in English

had to resort to attested examples they had heard and written down without any
possibility for verifications. Although some linguists have adopted the extreme
attitude of considering speech only (see Bloomfield 1933:21, and his writing is
not language credo), what we witness in the early 21st century is research ad-
dressing both written and spoken language. Speech and writing are considered
as two systems of communication, different but with the same prestige (Biber
1988:7; Crystal 1997:180). Linguists have studied the differences between the two
registers and adopted different theoretical stances, some considering that the two
registers are governed by two different grammars and others considering that the
same grammar is applied in different ways depending on the register. Biber (1988,
1995), Biber et al. (1999) and Leech & Svartvik (1994) adopt the latter point of
view. Biber, in particular adopts a multi-dimensional approach, and shows that
the same rules are applied differently depending on the register.
This is the view we adopt here as well; the study of our spoken corpus aims at
complementing the corpus of written English analyzed in the previous sections.
We consider speech to be a register, and our aim is to compare the distribution
and characteristics of ARCs in the spoken language, as well as their discourse
functions, with those of written ARCs. Consequently, both contrasts and similari-
ties will be determined in this section, between speech and writing for the same
syntactic construction.

2.2 The corpus

The corpus of spoken English that we used for our study required, as for the cor-
pus of written English, a precise, practical definition of the object of study. In
writing, the most-often quoted discriminating feature distinguishing (construct-
ed examples of) DRCs and ARCs is punctuation. Of course, this criterion is irrel-
evant in speech. We thus used the same defining criterion as for written examples
of ARCs, for which we showed that the punctuation criterion was not one hun-
dred percent reliable. We considered the referential stability of the antecedent: if
the RC takes part in the referential definition of the antecedent, then it is a DRC;
if not, then it is an ARC. We are aware that such a definition is problematic for
ARCs with indefinite antecedent, both in writing and speech (see Chapter1, Sec-
tion3.2), but it seems to us that the referential independence of antecedent vis--
vis the embedding of the RC is the most reliable criterion.
In speech, the discrimination between the determinative/appositive status of
the RC can be established thanks to prosody, in particular the presence of pauses,
although as we mentioned in Chapter 2, pauses are not systematic. Example (31)
illustrates the crucial role of prosody:
 Chapter 3. The corpus 89

(31) (RAD16_ ARC1) And somebody would always volunteer to take me up the
hillside | where the massacred happened.

The existence of a pause after hillside indicates that the speaker considers that
the hillside has been referentially determined independently of the embedding
of the RC (in the left co-text). The antecedent and the RC belong to two different
intonation units (see Du Bois 1991; Du Bois et al. 1993), hence the use of the sign
|. The content of the ARC aims to specify for the addressee what happened to the
hillside in question, both the massacre and the hillside being discourse old.
The problems one encounters when working on a spoken corpus do not end
with defining the object of study. Teasing apart DRCs and ARCs sometimes relies
on prosodic phenomena (see example (31)), and some researchers have therefore
found it technically difficult to label the examples in their corpora. For instance,
Biber et al. (1999), who rely on statistics to provide readers with a precise infor-
mation on usage, were unable to distinguish between DRCs and ARCs, and more
globally between restrictive and non-restrictive postmodifiers:
Because our corpus is not annotated for prosodic features, we do not attempt to
distinguish between restrictive and non-restrictive postmodifiers in conversa-
tion. (Biber et al. 1999:603)

To build our corpus, we decided to collect our own examples. We recorded radio
and television programs, British, American and Irish (summer 2001). From this
corpus, which focused on unplanned, spontaneous English, were extracted about
60 ARCs, labeled below either RAD or TEL, depending on the source they came
from (radio or television). Two motivations guided for our choice: (i) we wanted
to be certain of the status of the RCs we collected and (ii) within a discourse-
based study, as for the written corpus, we wanted to know the context and co-text
of the extracted utterances.
This self-collected corpus was complemented by approximately 40 examples
taken from the spoken-language portions of the British National Corpus, provid-
ed that these examples were unambiguously ARCs, independently of any pro-
sodic contour or presence of pauses. These examples are labeled BNC below.
Additional examples were taken from other electronic corpora to illustrate our
analysis of the discourse functions of ARCs, and in particular to conduct a dis-
course-prosody interface study (see next chapter). These additional corpora are
the International Corpus of English, British Component (ICE-GB) and the Aix-
Marsec corpus; the examples extracted from these corpora are labeled ICE-GB
and AM respectively. Below we provide a brief definition of the three electronic
corpora we used to conduct our different studies:
90 Appositive Relative Clauses in English

i. The British National Corpus is a 100-million-word collection of samples of

written and spoken language from a wide range of sources (national and re-
gional newspapers, academic writing, popular fiction for the written section;
radio shows, speeches and informal conversations for the spoken section)
ii. The Aix-MARSEC Corpus (cf. Auran, Bouzon & Hirst 2004) is constituted of
BBC recordings from the 1980s and provides natural sounding but mostly
scripted English. It constitutes a second evolution from the original SEC (Spo-
ken English corpus, cf. Knowles, Wichmann & Alderson 1996), the MARSEC
(Machine Readable SEC) constituting the first one (Roach et al. 1993).
iii. The ICE-GB corpus (cf. Greenbaum 1996) contains 1,000,000 words of written
and spoken English from the 1990s, the latter being both scripted and un-
scripted. The ICE-GB corpus now provides audio files alongside the written
scripts. Unfortunately, for accessibility reasons, we were able to use this cor-
pus only for the discourse-prosody study (Chapter 4, Section 6), and Chap-
ter3 relies on the BNC and a self-collected corpus.

This corpus of spoken ARCs complements the written corpus. A comparative

study with written ARCs is provided below in order to establish the differences
between the two different modes and the specificity of spoken ARCs. The criteria
that were used are therefore similar: morphosyntactic (choice of pronoun, type of
antecedent), semantic (type of verb, subjectivity) and pragmatic (informational
status of antecedents and ARCs).

2.3 Morphosyntactic, semantic and pragmatic characteristics

of spoken ARCs

The first thing to be noted is that the relative pronoun which is the most common
introducer of spoken ARCs (two thirds of them are introduced by which). This is
to be linked with the predominance of sentential relatives, necessarily introduced
by which, which are much more frequent in speech than in writing (three out of
ten as opposed to one out of ten for the written corpus). The relative pronoun
where is also more frequent in speech than in writing, to refer to antecedents that
go beyond pure location:

(32) (RAD03_ARC6/ARC7) Its named after a position that we take | a physical

position in our bodies at the impulse the moment of the impulse to engage
in violence | where we open our palms forward and step and lean in to an
engaged er violent stand | and we now as artists face this incredible task and
challenge | which was difficult anyway | of looking at and reflecting at the
impulse of violence that each of us has inside of us and where does that come
from, what is its origin and what are the alternatives to it
 Chapter 3. The corpus 91

(33) (RAD11_ARC1) Especially in the case of this film where | the underlying real-
ity truth the underlying truth of the film is revealed to be something very
different from what the narrator himself is presenting it as

As a consequence, the syntactic function of the pronoun is more frequently that

of an adverbial (2 ARCs out of 10), while other pronouns fulfill the subject posi-
tion of the ARC.
Most spoken ARCs are final (9 out of 10), a proportion that differs from
written ARCs. As an interruption of an on-going clause, middle ARCs constrain
speakers and addressees memories and are therefore avoided, although they do
exist:

(34) (RAD14_ARC2) You you slip too easily between | an Athenian world | where
after all women were only one step above slaves | and the way in which Aristotle
uh discussed friends of virtue and communicative friends
(35) (RAD09_ARC1) Unfortunately | the Gilbert Library | which is a great favourite
of mine in Pearse Street | is closed for the moment | its refurb being refur-
bished
(36) (BNC_ARC29) Hes pretty good on snooker as well, the, the Rothmans Grand
Prix, which er is held at the Hexagon in Reading, starts tonight, goes on the
October the twenty first

It should however be noted that the memory constraints imposed on speakers

by the middle position of ARCs may engender dysfluencies and repetition, as in
(37)(38):

(37) (BNC_ARC33) His brother Melvin, er who has another removal firm in York,
erm he had a funny experience too that Ive not hac had.
(38) (BNC_ARC2) But she made certain that my sister who was six years younger,
she made certain that she went to the county school

One of the strategies developed by speakers is then the extraposition of ARCs so

that they appear in final position. This is illustrated by the following examples:

(39) (RAD01_ARC1) I would have | I would recommend | er Odno Powers | phone

number | er whos er actively involved in promoting solar homes | and that is a
toll-free number: 8772229288
(40) (TEL01_ARC1) Youve met | youve met uh Louis Theroux as well | (?he) whos
made your documentary about you for for er |
(41) (RAD16_ARC4) And then came Nicholson came down | whos one of the
most talented of the younger people at that time.
92 Appositive Relative Clauses in English

(42) (BNC_ARC3) Chris mum may have remarried, who Andy calls Mum, erm
remarried when Andy was, you know, fifteen, and Chris was twenty-five

As far as the antecedent is concerned, results clearly diverge from those for the
written corpus. Fewer antecedents are proper names, just as fewer antecedents are
the syntactic subject of the verb in the main clause, whereas this was a frequent
combination of the written corpus, particularly in journalistic prose. The fact that
antecedents are only rarely syntactic subjects goes along with the small number
of middle ARCs, and corresponds to the dynamic character of the spoken regis-
ter. Antecedents occupy syntactic positions that canonically appear at the end of
sentences: direct/indirect objects, subject complements and adjuncts represent
more than 8 occurrences out of 10 (this proportion is only 1 out of 2 for the writ-
ten corpus).
A majority of the verbs in the ARC are in the simple present or simple past
(85% of verbs) and depict a state rather than a process, which is in sharp contrast
with the results for the written corpus (6 verbs out of 10 are state verbs in spoken
ARCs, as opposed to only 2.5 for the written ARCs).
Subjectivity is more characteristic of spoken ARCs than written ARCs with
a S+/S proportion amounting to more than 80%. As a consequence, we already
know that this will lead to a different distribution in terms of discourse functions
depending on the mode of communication. A study by Tao & McCarthy (2001)
shows that there is a privileged configuration for spoken ARCs: which + modal
expression + be, modal expression standing for modal auxiliaries but also ad-
verbs like really, hopefully, actually, probably, evidently and expressions like I think,
I suppose. The verb be is generally in the present tense.
Finally, following Princes (1981, 1992) taxonomy of given/new information,
which we extended for our study, in particular with the creation of a category
for indeterminables, we also studied the informational status of antecedents and
ARCs. Antecedents are more frequently new for the addressees in speech than in
writing (40% vs. 22%). Similarly, ARCs are more frequently hearer new in speech
than in writing.

2.4 Conclusion on the written/spoken distinction

The major characteristics of spoken ARCs are the following: they are generally
in final position and introduced by which, whose function is to be the subject
of the verb in the ARC, while the antecedent tends to be anything but a subject.
Sentential antecedents are frequent and the information conveyed by the ARCs
is newer and more subjective than for written ARCs.
 Chapter 3. The corpus 93

Adopting a multi-dimensional approach like Biber (1988)s, one might won-

der about the affinity between the spoken corpus and the different sub-corpora
(QUAL, TABL, FIC and SPEC) that constitute our corpus of written English. In-
terestingly, the results of our quantitative study reveal that the spoken corpus has
a certain affinity with the FIC written subcorpus. In both corpora, the pronoun
relative which is the most represented, fulfilling a function that is primarily sub-
ject; the ARC is mostly final (9 ARCs out of ten for both corpora) and sentential
ARCs are represented with nearly identical proportions (27% for spoken ARCs
and 28% for FIC ARCs). The syntactic function of the antecedent is mostly the
subject position in all sub-corpora with the exception of spoken and FIC ARCs.
The statistical results provided in Appendix 1 provide all the figures and allow a
complete comparison between the different sub-corpora.
Such statistical similarities show that the two modes of communication share
a certain affinity. This can very likely be explained by the communicative goals of
these two modes of communication: in both cases, the speaker has a good rep-
resentation of his/her addressee(s) as s/he knows the state of the knowledge that
they have. In the case of spontaneous spoken English, which includes private con-
versations, the speaker knows the knowledge of his/her addressees and if not, the
fact that this mode of communication is in praesentia communication gives clues
as to the way the information is received, while repairing informational misjudg-
ments is always a possibility. As for novelists, they have total control over what
the readers know or do not know about the story unfolding as they are the ones
providing the information.

3. Conclusion

This chapter has provided a contemporary usage-based description of ARCs in

discourse, both written and spoken, giving a representative picture of ARCs char-
acteristics in attested examples from various registers, sometimes quite remote
from both artificial and constructed examples found in some pedagogical gram-
mars and linguistic articles and also from the recommendations of prescriptive
style guides.
The characteristics that have been observed and quantified will be useful to il-
lustrate the different discourse functions we define in the next chapter, thus estab-
lishing a link between morphosyntax and discourse for one specific syntactic con-
struction, more generally between form and meaning. We now turn to this central
chapter of the book, on the discourse functions of appositive relative clauses.
 chapter 4

The discourse functions of ARCs

A taxonomy

1. Introductory presentation

The taxonomy we suggest here to define the discourse functions of ARCs com-
prises three main categories:

Continuative Appositive Relative Clauses

Relevance Appositive Relative Clauses
Subjectivity Appositive Relative Clauses

In Sections 2, 3 and 4 below, we provide, for each category, a general definition

that is then illustrated by the specific discourse strategies involved for both writ-
ten and spoken data; we then list each categorys formal characteristics, based on
the quantitative study summarized in the previous section.
Additional specific roles will be added in Section 5.3 for spoken ARCs, which
have more to do with turn taking between speakers than separate discourse func-
tions per se. In Section 6, finally, we investigate the existence of a possible link be-
tween the discourse functions that have been defined and the prosodic realization
of the ARCs in each category.

2. Continuative ARCs

2.1 Definition

The existence of this sub-category of ARCs has previously been established

by other researchers (e.g. Jespersen 1970; Cornilescu 1981; Daalder 1989 and
Huddleston & Pullum 2002). However, their definitions are not interchangeable
in any systematic way. In particular, definitions have mixed discourse and syn-
tactic criteria and the label continuative is applied in contemporary grammars
to examples (1) and (2) without any distinction:
96 Appositive Relative Clauses in English

(1) (QUAL_ARC56) So we asked a man, who shrugged his shoulders and disap-
peared into a nearby shop.
(2) (FIC_SR205) Macaulay Culkin probably got three hundred and fifty pounds
a week, maybe even more, which meant that if he were Macaulay Culkin he
could pay his mum to teach him.

Our definition is discourse-based and takes into consideration semantic and

pragmatic criteria only. Continuative ARCs (or continuatives for short) represent
a specific kind of ARC which enables, by the depiction of an extra-linguistic event
subsequent to that in the main clause, a forward movement within narrative time.
As such, continuatives are exceptional, as narrative dynamism is traditionally
seen as belonging to the foreground as opposed to the background. These notions
of foreground vs. background have been used frequently, with definitions that are
not systematically interchangeable, making the concepts sometimes quite con-
fusing. In this book, foreground and background are to be understood as per-
taining to informational considerations independently of any syntactic consider-
ations. Both terms refer to the notion of grounding, as developed in particular by
Hopper & Thompson (1980), who define foregrounded clauses (or foreground)
as compris[ing] the backbone or skeleton of the text, forming its basic structure
while backgrounded clauses (or background) put flesh on the skeleton, but are
extraneous to its structural coherence (Hopper & Thompson 1980:281). Clauses
that belong to the foreground are generally ordered in a temporal sequence and
a change in their order will result in a change in the temporal sequence of the
original semantic interpretation (Labov 1972:360). On the other hand, clauses
that belong to the background are not ordered with respect to each other, and
can even be removed (Hopper & Thompson 1980:282). It is this definition we
adopt here when we oppose foreground and background information. This dis-
tinction cannot be correlated with the syntactic main clauses/embedded clauses
distinction, as is shown by Reinhart (1984) or Thompson (1987) for instance, who
shows that embedded (or subordinate, to use her term) clauses can be considered
as foregrounded clauses and be part of the main narrative line: about 10% of the
predicates in her corpus occur in temporally sequenced clauses and correspond

. To be complete, we must specify that importance of the event (backbone) and temporal
ordering do not necessarily go hand in hand. (Thompson 1987:436). To make the picture even
more complicated, researchers have shown that foreground and background may not represent
a strict binary opposition: instead there seems to be a gradient along which different degrees of
backgroundedness can be found (e.g. Givon 1984).
 Chapter 4. The discourse functions of ARCs 97

to narrative subordination. As a consequence, although independent clauses in-

tuitively seem to be the natural candidates for foregrounding, general claims that
foreground information is conveyed by independent clauses while background
information is conveyed by embedded clauses are too general to be completely
accurate. The most we can say, following Reinhart (1984:796), is that the use of
syntactic embedding represents a powerful means to mark backgrounded infor-
mation. Relative Clauses in English can convey foregrounded information (see
for instance Depraetere 1996). In particular, continuative ARCs as we define them
here belong to this category of embedded clauses representing foreground infor-
mation and naturally correspond to what Thompson labels narrative subordina-
tion. This is to be opposed to relevance and subjectivity ARCs, which represent
background information.
Given the definition we have provided above, the following examples are thus
considered continuative ARCs (while (2) is not):

(1) (QUAL_ARC56) So we asked a man, who shrugged his shoulders and disap-
peared into a nearby shop.
(3) (TABL_ARC301) She was found face down in the water and airlifted to hospi-
tal, where she died hours later.
(4) (QUAL_ARC51) Robinson College asks candidates to complete a questionnaire
on their education background, which is then used to help decide whether they
are offered a place.
(5) Northern Scotland will have occasional light rain which will be followed dur-
ing the day by colder but still mainly cloudy weather with a few sleet and snow
showers (AM)
(6) You can understand such a lot of poetry at a childs level which later on becomes
much more complicated (ICE-GB)

The informational content of each ARC depicts an event that is successive to a

first event depicted in the main clause (MC) and concerns an element quoted in
the main clause, traditionally an element that is clause-final (see Figure 1).

. Thompsons definition of subordinate clauses includes four types of clauses: (i) the comple-
ments to another predicate, (ii) non-finite clauses, (iii) clauses that begin with a conjunction
that marks it as an adverbial clause, (iv) relative clauses. She shows that narrative subordinate
clauses accomplish, in addition to temporal sequencing, a a text-creation goal (Thompson
1987:451), i.e. specific discourse functions that independent clauses do not fulfill, such as an
orienting function for adverbial clauses, which orient readers back to the time line after a
digression.
98 Appositive Relative Clauses in English

event 1 event 2
(1) we/ask a man the man/shrug his shoulders
(3) RC/ask to fill in a questionnaire the questionnaire/be used
(4) she/be airlifted she/die
(5) northern S/have rain it/be followed by
(6) you/understand poetry it/become more complicated

Figure 1. MC and continuative ARC as the depiction of two successive events

Even when the event depicted is generic, as in example (3), the advancement
of narrative time holds for each specific instance of the generic event, and we
consider the ARC continuative.
The use of ARCs provides certain narrative dynamism, the two sequential
events being presented as being linked to one another; this link can trigger a cau-
sality interpretation (see below). This is to be opposed to the juxtaposition of two
independent clauses, which somehow presents the two events separately, as the
following reformulations illustrate:

(1) a. So we asked a man. He shrugged his shoulders and disappeared into a

nearby shop.
(3) a. She was found face down in the water and airlifted to hospital. She died
there hours later.
(4) a. Robinson College asks candidates to complete a questionnaire on their
education background. It is then used to help decide whether they are
offered a place.
(5) a. Northern Scotland will have occasional light rain. It will be followed
during the day by colder but still mainly cloudy weather with a few sleet
and snow showers
(6) a. You can understand such a lot of poetry at a child s level. It later on
becomes much more complicated

It is within this type of ARC that adverbials such as then, now, later, in turn or
time adjuncts are to be found. They can be inserted without any difficulty. Given
that other types of ARCs (relevance and subjectivity) cannot be rewritten using
such adverbials, the possibility of inserting them can be used as a test to deter-
mine whether an ARC is to be viewed as continuative or not. For example, (1) is
continuative but (2) and (7) are not, although the syntactic configuration is iden-
tical (syntactic subject followed by an action verb in the preterit form):
 Chapter 4. The discourse functions of ARCs 99

(1) b. So we asked a man, who then/after that shrugged his shoulders and
disappeared into a nearby shop.
(2) a. #Macaulay Culkin probably got three hundred and fifty pounds a week,
maybe even more, which then/after that meant that if he were Macaulay
Culkin he could pay his mum to teach him.
(7) (TABL_ARC319) Businessman Stuart Crane was impaled by the 5ft stake,
which hurtled through his car windscreen when he crashed into a field beside a
dual carriageway.
a. #Businessman Stuart Crane was impaled by the 5ft stake, which then/
later/after that hurtled through his car windscreen when he crashed into
a field beside a dual carriageway.

It is not solely the sequence [predicate 1 relative pronoun predicate 2] with

process verbs that triggers the continuative interpretation of an ARC: the seman-
tics of the verb and the informational content of the two clauses play a crucial
role. Example (8), taken from Depraeteres (1995) study on the interpretation of
tenses in English relative clauses, was interpreted in different ways by her con-
sultants, either as (8a) or (8b), which shows that the interpretation of an ARC as
continuative is not exclusively dependent on syntactic and semantic criteria:

(8) Tom met his girlfriend in London, where he lived for three years.
a. Tom met his girlfriend in London while he was already living in London.
b. Tom met his girlfriend in London and subsequently moved to London.

Because of the existence of a chronological sequence, a semantic dependence

between the informational content of the two clauses exists: the content of the
ARC is dependent on the information conveyed by the main clause. Negating the
event in the MC then prevents the event in the continuative ARC from happen-
ing, which will not be the case for other discourse categories:

. Depraeteres consultants were asked to specify the order in which the two events (1 = Tom/
meet his girlfriend in London; 2 = Tom/live in London for three years) happened: either Tom
met his girlfriend while he was already living in London (order 21, see (8a)) or he first met his
girlfriend in London and subsequently moved to the British capital (order 12, see (8b)), with a
potentially causal relationship between the two events, to be reconstructed inferentially on the
encyclopaedic assumption that when you are in relationship with someone, you naturally want
to live near him/her. Only the second option corresponds to a continuative use of the ARC.
Depraeteres results were as follows: 19% chose the order 12 (viz continuative interpretation)
as the only possible interpretation while 21 was chosen by 30%. The majority chose one option
but admitted that the other interpretation was also possible, although not the most obvious one
(31% for 12 primarily/21 possibly and 20% for 21 primarily/12 possibly).
100 Appositive Relative Clauses in English

(1) c. #So we did not ask a man, who shrugged his shoulders and disappeared
into a nearby shop.
(3) b. #She was not found face down in the water and airlifted to hospital, where
she died hours later.

Such semantic dependence can lead to the existence of a causality link that has
to be inferentially reconstructed. This implicit link can be made explicit with the
insertion of thus or as a consequence for instance:

(9) (FIC_ARC239) Luckily she landed on the bag itself, which burst.
a. Luckily she landed on the bag itself, which thus/as a consequence burst.

The process land triggers the second process burst, just as the process ask trig-
gers the process shrug in (1): in these examples causality is inferentially added to
temporal continuity. However, such semantic dependence is not compulsory for
all continuative ARCs, as examples (3) and (5) illustrate:

(3) c. #She was found face down in the water and airlifted to hospital, where she
thus died hours later.
(5) b. #Northern Scotland will have occasional light rain which will thus be
followed during the day by colder but still mainly cloudy weather with a few
sleet and snow showers

A finally defining characteristic of this subcategory of ARC as opposed to the

other subtypes defined below is that the informational contents of the MC and the
ARC seem to be on the same semantic, or even syntactic, levels: each can be in-
terpreted as belonging to the foreground as defined above in spite of the semantic
dependence between the two clauses. This has been noted by other researchers,
although, once again, one has to be careful, as the definitions provided sometimes
combine semantic and syntactic criteria:
Whereas elsewhere the information conveyed in a supplementary relative is
somewhat backgrounded relative to that conveyed in the clauses containing the
anchor, the continuative relative has equality of informational status, presenting
a further event in a narrative chain. (Huddleston & Pullum 2002:1064)

Such an absence of hierarchy between the two clauses paves the way for the sug-
gestion sometimes found in the literature that continuative ARCs share syntactic
characteristics with independent clauses (Emonds 1979 among others). The rela-
tive pronoun seems to take on the role of a coordinator, and no longer that of a
subordinator:
 Chapter 4. The discourse functions of ARCs 101

In what O. Jespersen called a continuative relative clause and O. Behaghel

weiterfhrender Relativstaz, the relative pronoun has indeed a tendency to
lose its fundamental role as a subordinator to acquire that, paradoxically, of a
coordinator.  (Kuntsman 1994:1)
Interestingly however, in contrast to the other cases, there does seem to be a
reason here to deny a subordinate status to the sentence part following who. The
event character of the clause can be accounted for by having the means of the
relevant elements then, copied and it perform their specifying roles in accumu-
lative fashion. We do not have here some person involved in an independently
understood activity then copying it which could characterise in essentially the
same way other persons than such as get a letter placed on their desk. The point
of the clause is the gradual leading up to the achievement, which then as such
completes the first half of the sentence.  (Daalder 1989:204)

Interestingly, this refers back to the debate on the syntactic status as a whole (see
Chapter 1, Section 2.5) but also to the problematic status of the relative pronoun
in atypical ARCs (see Chapter 2). Arnolds (2004, 2007) suggestion that ARCs do
behave like independent clauses in certain respects, while retaining subordinate
syntax, is particularly relevant for continuative ARCs.
We now turn to the formal characteristics (morphological, syntactic, infor-
mational status) of continuative ARCs.

2.2 Characteristics

Paradoxically, although this subcategory is the only one in our taxonomy that
has been previously defined by other researchers (bearing in mind the restric-
tions mentioned earlier), examples of continuative ARCs remain quite rare. Only
20 examples from our written corpus correspond to the definition we provided.
As far as our spoken corpus is concerned, the limited number of tokens (100)
brought a result that corpus-based studies naturally have to face: no examples
from our self-collected spoken corpus can be labeled continuative. However, this
does not mean that spoken continuative ARCs do not exist (corpora, no matter
how large, are by definition finite selections of speech). Continuatives seem to be
particularly prolific in live reports of soccer matches (examples taken from the
ICE-GB corpus):

(10) and again Michalichenko left-footed switches the play to the far side to
Kanchelskis who comes in field

. Daalder bases his analysis on the following example: He gave the letter to the clerk, who then
copied it.
102 Appositive Relative Clauses in English

(11) He picked out Allan Smith <,> who had to stretch for it <,> deep inside the
area on the far side about ten yards inside the area about five yards from the
by-line
(12) Leaves it for Geoff Thomas who plays it square to Derigo on the far left
(13) Plays it square infield to Parker who s pushed forward
(14) Finds Smith who lays it off neatly to Barnes

The characteristics specific to continuative ARCs were established on the basis of

the statistics compiled for the written corpus (as no example were found in our
corpus for spoken English). The results reveal a high degree of homogeneity with-
in the class, as well as a high degree of divergence with respect to other ARCs.

2.2.1 Antecedent
Antecedents modified by a continuative ARC are systematically NPs, never
predicates (which are possible antecedents for sentential relatives). However, it
should be noted that such a configuration is not impossible and intuitively
seems quite natural:

(15) He finally decided to go and work in London, which later enabled him to apply
for a permanent job in Britain.

Antecedents are generally the object, direct or indirect, of the verb in the MC
(55% of continuative antecedents), a result which is at odds with the results for the
whole corpus, where the subject function is the most common. Adjuncts of time
and place are also represented (35%). These results are to be expected, since the
position of the ARC is necessarily final, whereas a syntactic subject is canonically
at the beginning of the sentence.

2.2.2 Type and tense of verbs

Verbs in continuative ARCs are typically process verbs, which seems perfectly
natural, given the definition we provided for continuative ARCs. However, it
should be noted that temporary states can represent the second extra-linguistic
event required for an ARC to be continuative. Consider the following example:

(16) I went straight to Bucks office and checked with the secretary, who was at first
reluctant to give me the details, but when I threatened to take the matter up with
Buck himself, she told me that Rusty had been arrested (Mlis 2000)

The ARC verb is a state verb that describes a temporary state of the secretary.
Here Carlsons (1977) distinction between individual-level predicates and stage-
level predicates is relevant. Individual-level predicates describe the permanent
 Chapter 4. The discourse functions of ARCs 103

property of an individual, while stage-level predicates describe a temporary char-

acteristic that corresponds to a stage of the individuals existence. In (16), the
predicate be reluctant is a stage-level predicate, which enables narrative time to
move forward. The passage from state A to state B can be considered as an event,
with the use of at first explicitly mentioning a change in state.
The tense generally associated with the verb is the preterit in its simple form
(more than the three quarters of our examples), which illustrates the fact that
continuatives are predominantly found in narratives, in fiction, newspapers or
specialized texts. The simple present can also be found when used within narra-
tions in the present tense:

(17) (QUAL_ARC183) In the book the consul is portrayed as a rather sad man who
drinks too much and marries an Indian girl out of loneliness, who eventually
leaves him.

Whether writers use the preterit or the present tense, we notice that the tense is
similar for the verb in the main clause and the verb in the continuative ARC. This
result is also at odds with the results for the other sub-categories in our taxonomy,
where different tenses are used in the MC and the ARC, reflecting a shift in the
discourse, even a digression.

2.2.3 Informational status of the elements

The informational status of the referent of the antecedent, as well as that of the
content of the ARC, is clearly homogeneous. The antecedent is mostly discourse
new and systematically hearer old or inferrable, and is assumed to be retrievable
by the addressees. As for the ARC itself, its content is systematically discourse
new/hearer new.

2.2.4 Suppressibility
The consequence of what has just been said about the informational status of the
antecedent and the ARC makes the suppressibility of the continuative ARC im-
possible, as it would entail a gap in the sequence of events and in understanding.
Once again, these results are at odds with the results for the other ARC types that
will be defined below.

3. Relevance ARCs

This subcategory of ARC is the most frequent in our corpus, both in writing and
speech. The starting-point of the definition of relevance ARCs is the relationship
between the speaker and his/her addressee(s). More specifically, the key element
104 Appositive Relative Clauses in English

is represented by the formers assumptions about the knowledge that the latter
has/have. The term relevance is borrowed from the concept of relevance as
first defined by Grice (1975, 1989) in his Cooperative Principle (Maxim of Rela-
tion Be relevant) and developed by Sperber & Wilson (1986 principally). We
first discuss the concept of relevance as developed by these researchers and then
provide our definition of relevance ARCs before detailing the different conversa-
tional strategies relevance ARCs illustrate.

3.1 Relevance: A definition

One of the Maxims developed by Grice is the Maxim of Relation or Relevance:

any contribution should contribute to the relevance of discourse as a whole. This
Maxim was later developed by Sperber & Wilson within their theory of relevance
(Sperber & Wilson 1986, 1993, 2002). Each communicative act necessarily im-
plies a presumption of relevance, which Sperber & Wilson (1986:158ff.) define
as follows:

a. The set of assumptions I which the communicator intends to make manifest

to the addressee is relevant enough to make it worth the addressees while to
process the ostensive stimulus.
b. The ostensive stimulus is the most relevant one the communicator could have
used to communicate I.

Relevance is then defined as a comparative concept (Sperber & Wilson 1986:125):

Extent condition 1: an assumption is relevant in a context to the extent that
its contextual effects in this context are large.
Extent condition 2: an assumption is relevant in a context to the extent that
the effort required to process it in this context is small.

To sum up, the concept of relevance corresponds to the search for adequate con-
textual effects for no gratuitous processing effort. A contextual effect consists in
a modification of what they call the background, i.e. the context of hypotheses
known by the addressee, while the mental effort is the amount of effort necessary
for the addressee to process the contextual effects. In effect, the greater the effort,
the weaker the relevance. It is this concept of relevance that is at the core of our
definition of relevance ARCs.
 Chapter 4. The discourse functions of ARCs 105

3.2 Definition of relevance ARCs

The use of an ARC in discourse can be triggered by the speakers will to opti-
mize the relevance of his/her utterances by reducing the mental effort of the
addressee(s), while at the same time increasing the contextual effects. By insert-
ing supplementary information, the aim of which is not to determine the refer-
ence of the antecedent, the speaker makes the antecedent or the subject-predi-
cate relation relevant within the discourse, thus optimizing the relevance of the
whole discourse. More precisely, the speaker takes precautions with respect to the
knowledge s/he assumes is shared with the addressees. This is particularly crucial
in journalistic prose, where the journalist is trying to reach as wide a readership
as possible, a readership that is necessarily heterogeneous and unknown by the
journalist (in absentia communication). The use of a relevance ARC is then trig-
gered by the speakers intention to avoid any violation of the relevance principle
as defined by Sperber & Wilson.
Relevance ARCs thus represent a compromise between too much and too
little information. The results of the study on the informational status of the refer-
ent of the antecedent and/or the ARC itself (discourse new/old; hearer new/old)
reveal the ambiguous character of the information. Using an ARC takes the in-
formation to the background, away from the hot news, as opposed to the not
so hot news, if we refer to Cornilescu (1981)s terms, used to comment on ARCs
(which she labels Ap[positive]RCs) as a whole:
The speaker lets the hearer distinguish between hot news and not so hot news
One can avoid treating the hearer as uninformed by embedding the respective
proposition as an ApRC, thereby indicating that the proposition is supplemen-
tary information, recalling an old fact, which does not belong to the intended
message. (Cornilescu 1981:58)

To sum up, here is our definition of relevance ARCs:

A relevance ARC is used to make relevant the antecedent or the predicate in

which it appears. Though the antecedent has been identified from a purely
referential point of view, it is judged to be not sufficiently determined to be
used alone in discourse, at least for some of the targeted addressees. The
relevance ARC is therefore a prevention strategy such that the ARC is pre-
vented from being non-relevant for the uninformed reader, while at the same
time it provides background information in a form which avoids its being
non-relevant for the informed reader. In this, the relevance ARC represents a
compromise.
106 Appositive Relative Clauses in English

3.3 The different discourse strategies

Having provided our definition of relevance ARCs, we turn below to the differ-
ent conversational strategies our definition encompasses, namely (i) the leveling
of the shared cognitive space; (ii) the legitimization of the antecedent; (iii) the
explanation, justification or concession strategy.

3.3.1 The leveling of the shared cognitive space

The insertion of an ARC serves to insert supplementary information that aims
to compensate for any discrepancies in the amount of knowledge shared by the
participants. The addressee(s) or some of the addressees may not be familiar
with the referent of the antecedent and/or the informational content of the ARC
(hearer new information), and the speaker simply aims at leveling the shared
knowledge to reduce any gap between informed addressees (for whom the infor-
mation is hearer old) and the uninformed addressees (for whom the information
is hearer new). The following two examples illustrate this specific use of relevance
ARCs, extremely frequent in journalistic prose:

(18) (QUAL_ARC128) At the Dayton conference in 1995, which ended the war in
Bosnia, he was still central to American diplomatic efforts, winning friends with
his focus on the big picture and his evocative language.
(19) (QUAL_ARC135) It is hard to square his action with the Energy Policy and
Conservation Act, which authorizes the president to tap the reserve.

In (18), the ARC informs the reader for whom this information is hearer new (or
forgotten) about the aim and result of the 1995 Dayton conference. In (19), the
content of the act is supplied in the ARC. In both cases, the information conveyed
by the ARC aims to maximize the relevance of the whole utterance by inserting
information that might not be known by at least some of the addressees but which
is however crucial to obtain the appropriate contextual effects: importance of the
1995 Dayton conference for (18) and discrepancy between the current action and
the Energy Policy and Conservation Act for (19).
Such examples are also found in spontaneous private conversations, when
the speaker is not sure whether his/her addressee(s) know(s) information that is
judged necessary by the speaker for a good understanding of the utterance and a
maximum relevance of it but which might be unknown by the addressee(s). Ex-
ample (20) illustrates the phenomenon:

(20) (BNC_ARC29) Hes pretty good on snooker as well, the, the Rothmans Grand
Prix, which er is held at the Hexagon in Reading, starts tonight
 Chapter 4. The discourse functions of ARCs 107

It is in this kind of ARC that expressions like you know will be found, thanks to
which the speaker indicates that he is not considering that the addressee does not
know the information but is only reminding him/her of it (or pretending to):

(21) (RAD11_ARC4) a couple of months went by | and about a month before we

started shooting | I got this call from Dick Heimann | whos Woodys long-time
music supervisor and arranger and kind of a genius of of orchestration and
arrangement | n you know and I he said uh Im gonna have you get together
with a a a you know voice person in Los Angeles
(22) A: I mean I m going to see a play next week with an actress friend of mine in
it who is now who s doing very well now thank you very much whether
B: oh whos that
A: Jo Jo Lumley who s you know doing the Revenger s Tragedy (ICE-GB)

A very interesting example is provided in (23), taken from Eurostar Magazine,

which is distributed on Eurostar trains between continental Europe and London
and which sometimes provides the same articles in different languages English,
French and Dutch. The sentence was originally written in English (23), and then
translated into French (23a):

(23) When Steven Soderbergh [who has already directed Clooney in Out of Sight]
phoned to tell me he was planning to do a remake of the film, I jumped for joy.
a. Lorsque Steven Soderbergh (qui a dj dirig lacteur dans Hors datteinte,
ndlr) ma appel pour me dire quil comptait refaire le film, jai saut de joie.
(Eurostar Magazine 49:10)

When translating (23) into French, the translator added ndlr (note de la rdac-
tion, i.e. editors note in French), which brings the relevance strategy to the sur-
face: the information conveyed by the ARC was added by somebody exterior to
the discourse (the speaker is actor George Clooney) to provide information that
might be unknown for some of the addressees but considered important by the
journalist/editor to optimize the relevance of the whole discourse: the fact that
Soderbergh and Clooney have already worked together implies a certain familiar-
ity between them and a certain professional collaboration.
The leveling of the shared cognitive space defined here as a discourse strategy
aiming to optimize the relevance of a speakers discourse implies that the strategy
is addressee-oriented. As a consequence, we might expect differences depending
on the identity of the intended addressees. And indeed we do. The following ex-
ample is taken from The Sunday Times, a British newspaper whose readers do not
necessarily have any specific medical knowledge (or at least are not expected to):
108 Appositive Relative Clauses in English

(24) (QUAL_ARC47) Blood from donors is usually split into its components, such as
red cells, which carry oxygen; plasma, with white cells that help fight infection;
and platelets, which aid clotting.

The informational contents of the two ARCs which carry oxygen and which aid
clotting aim to provide readers with the role of red cells and platelets. In articles
taken from medical journals such as The British Medical Journal, the compounds
red cells and white cells are never followed by ARCs whose aim is to level the
shared cognitive space, as the information that red cells carry oxygen and plate-
lets aid clotting is supposedly known by all the readers of the medical journal. A
search on the BMJ website (www.bmj.com) concerning the top 50 results for red
cells shows that none of the occurrences of red cells are followed by an ARC that
would specify their role.
In the same way, example (18) also reveals this leveling strategy. This exam-
ple was found in an article from The International Herald Tribune (Issue 36,576
October 7th8th, 2000) which had first been published in The New York Times
(October 6th, 2000). Interestingly, the original example was slightly different:

(18) a. At the Dayton conference in 1995 that ended the war in Bosnia, he was still
central to American diplomatic efforts, winning friends with his focus on
the big picture and his evocative language.

In the original example, the relative clause is introduced by the relative pronoun
that, not which, and no commas or other punctuation mark indicate the appositive
status of the relative clause. One could argue that the Dayton conference in 1995 is
referentially determined independently of the embedding of the RC and that the
RC is an ARC. However, the antecedent and relative clause sequence in (18a) can
be understood as the conference that took place in Dayton in 1995 and ended the
war in Bosnia (similarly to the guy I met last night mentioned to addressees that
have no idea that I met anyone last night). In (18), the fact that the Dayton confer-
ence ended the war in Bosnia is presented as separate information, as something
people might know but are nevertheless reminded of in case they have forgotten
or simply ignore the information. Such a difference could have to do with the
intended readerships of the two newspapers (Americans (almost) exclusively for
The New York Times; international readership two thirds of the readers are not
American for The International Herald Tribune, published in Paris), and to the
assumed shared knowledge between the journalist and the readers. The use of the
which-clause seems to assume that readers might be more familiar (the informa-
tion conveyed by the ARC is intended to be a plain reminder) with the Dayton
Conference than the use of the that-clause, which constructs the reference of the
 Chapter 4. The discourse functions of ARCs 109

conference through the use of several modifiers and forces the reader to accom-
modate to this new information.

3.3.2 The legitimacy of the antecedent

Sometimes, it is the presence of the antecedent itself as the syntactic subject or
object within a subject-predicate relationship that might seem irrelevant for at
least some of the addressees. The role of the juxtaposed ARC is then to ensure the
relevance of the antecedent. Again, in journalistic prose, it is quite frequent for a
journalist to mention someone through the use of a proper name that is discourse
new and hearer new, i.e. completely new for the addressee(s). This is the case for
witnesses of an event, for instance, or experts unknown to the general public. The
journalist, by mentioning some properties of the referent of the antecedent, then
stresses the referents privileged position as a witness within the subject-predi-
cate relation. The aim of the relevance ARC is then to justify and legitimize the
referents presence. Being legitimized, the referent is presented as entitled to give
an opinion, a judgment, an analysis or a comment. Examples (25)(28) illustrate
this phenomenon:

(25) (QUAL_ARC9) Tony Sewell, who has just finished an inquiry into soaring levels
of exclusions among black pupils from a London school, claimed that too much
concern with money and consumer goods was almost as damaging to black
pupils chances as racism.
(26) (QUAL_ARC106) The story of genetic research on humans, which took 10 years
to uncover, is likely to shake the world of anthropology to its core, according to
professor Terry Turner of Cornell University, who has read the proofs.
(27) (QUAL_ARC89) Law firm Belmont Hodgson, which specialises in gay issues, is
planning to challenge the Inheritance Tax situation under which gay couples
miss out on the more favourable tax treatment for husbands and wives.
(28) (RAD16_ARC2) And somebody would always volunteer to take me up the
hillside | where the massacred happened.

In (25), it is because the content of ARC [T. Sewell/just finished an inquiry into
soaring levels of exclusions among black pupils from a London school] is true that
the content of the MC [T. Sewell/claim that too much concern with money and
consumer goods is almost as damaging to black pupils chances as racism] is relevant
within the context of the press article. Tony Sewell is unknown to the general pub-
lic and his name is discourse new/hearer new for the addressees. The information
provided by the ARC validates the suitability of the referent as a reliable source
who acts as a witness: s/he provides evidence to illustrate the journalists thesis.
The utterance as a whole (MC+ARC) then serves as evidence to the journalists
110 Appositive Relative Clauses in English

argument. Such a strategy is very frequent in journalistic prose, when journalists

report the comments of a person whose function, role or experience is in fact
more important than his/her name. In (26), the phenomenon is similar: Terry
Turner is both discourse new and hearer new for the addressees, and the content
of the ARC legitimizes his privileged position to provide a relevant opinion: it
is because he has read the proofs that his opinion is relevant in the discussion.
In (27), it is because the law firm B. Hodgson is a specialist in gay issues that its
position is privileged to plan a challenge of the Inheritance Tax and its mention
relevant in an article dealing with tax-related gay issues. Finally, in example (28),
taken from our spoken corpus, the mention of the hillside whose reference has
been established prior in the discourse becomes relevant when the addressees
learn that this is where the massacre [of Serbs] happened. Without the contents
of the ARCs, addressees might wonder why Terry Turner, the law firm B. Hodgson
or the hillside are mentioned. In other words, the ARCs ensure their relevance in
the discourse.
This strategy is also used with antecedents that represent a date or a period of
time. Consider (29)(30):

(29) (QUAL_ARC48) C. Dyas () confirmed that the blood donated by the seven
vCJD victims was taken before 1999, when the authority introduced new safety
procedures designed to remove prions, the infectious proteins thought to trans-
mit vCJD.
(30) (SPEC_ARC438) That switch from new to old sectors was also a feature of
trading on Wall Street on Thursday, when the Dow Jones Industrial Average
pushed up 96 points and the Nasdaq Composite could only manage a 10 point
gain.

In these examples, it is not the dates per se that matter, but rather what took place
on those particular dates: it is because 1999 was the year when prions were re-
moved that this date is relevant as a boundary before which contaminated blood
was taken and after which this was no longer supposed to be the case; it is because
the Dow Jones Industrial Average pushed up 96 points and the Nasdaq Compos-
ite could only manage a 10 point gain on Thursday that this date is relevant in
the main clause. Without the ARCs, the addressees might wonder why 1999 or
Thursday has been chosen.
Such a discourse strategy is not restricted to journalistic prose. We also
find such examples in fiction texts and conversations. Consider the following
examples:
 Chapter 4. The discourse functions of ARCs 111

(31) (FIC_ARC273) The piece shes playing is called Gavotte. Ettovag. Its a good
word; she will think of a use for it, later. The piano reeks of lemon oil. Ethel,
who comes in to clean, has been told not to polish the keys with it shes only
supposed to use a damp cloth but she pays no attention, and Tonys fingers will
smell of lemon oil for hours.
(32) I wrote it down sort of uh after I d had the conversation obviously but it was
a few days after when I decided that this was really what I would talk about
<,> and it s a conversation with uhm Paul who is the coordinator of the O D P
school where I teach (ICE-GB)

At the point of the novel from which example (31) is extracted, the readers have
never encountered the proper name Ethel, whose referent is discourse new and
hearer new. The character has never been mentioned before (and will never be
mentioned again after this sentence). Without the ARC, readers might wonder
about the relevance of mentioning the character, whose raison dtre as fictitious
character would not be established. It is the content of the ARC that legitimizes
the presence of Ethel in the discourse, as the subject of has been told not to polish
the keys with it. Example (32), extracted from a face-to-face private conversation,
shows that discourse new/hearer new referents can be introduced in the discourse
but legitimized by the presence of the ARC, without which the addressee might
wonder about the mention of this new element in the discourse.

3.3.3 Explanation, justification, concession: The implicit inter-clausal link

In Chapter 1, we mentioned that the existence of a complex inter-clausal link
between the ARC and its main clause is a property traditionally mentioned in
the literature. Using an ARC can then enable speakers to provide an explanation
for, a justification of, or an opposition to the informational content in the MC.
This implicit link, to be reconstructed implicitly by the addressee(s), can shed
new light on the content of the MC or the antecedent itself, therefore optimizing
the relevance of the discourse as a whole. The implicit link the relative pronoun
itself does not establish any semantic link between the MC and the ARC enables
the speaker to say more than the plain addition of the informational contents of
the two clauses. Following Sperber & Wilsons terminology, ARCs then can rep-
resent an explicature, which enables the construction of an appropriate hypoth-
esis through decoding, disambiguation, reference resolution and other pragmatic

. Paul is clearly a discourse new element, as is shown by the left co-text provided by the
ICE-GB corpus. Although its hearer new status cannot be asserted with absolute certainty,
the right co-text, in which Paul is mentioned only twice, does not provide any specific refer-
ence to the participants acquaintance with him before the discourse. In any case, the infor-
mation Paul is the coordinator of the ODP school where I teach is necessarily hearer new.
112 Appositive Relative Clauses in English

processes. The meaning of an utterance with an ARC is to be understood as going

beyond the plain addition of the content of the two clauses (Khalifa 1999:2728;
Cornilescu 1981).
Interestingly, although this type of relevance ARC does not display specific
formal characteristics in English, this is not the case for Latin. In Latin, the exis-
tence of an implicit inter-clausal link is openly marked by the subjunctive mood.
A relative clause whether determinative or appositive in the subjunctive can
be interpreted as cause (33), concession (34), consequence (35), condition (36) or
goal (37) (Saussy 1977:394305), as our glosses and translations suggest:
(33) O fortunate adulescens, qui tuae
O happy-VOC.SG young man-VOC.SG, REL-NOM.SG your-GEN.SG
virtutis Homerum praeconem inveneris
virtue-GEN.SG Homer-ACC.SG herald-ACC.SG find-2SG.SUBJ.PRF
Happy young man, because/since you have found a Homer to be a herald of
your worth.
(34) Exercitui Caesaris luxuriem objiciebant,
army-DAT.SG Caesar-GEN.SG luxury-ACC.SG criticize-3PL.IND.IPFV,
cui semper omnia defuissent.
REL-DAT.SG always everything-ACC.PL lack in-3PL.SUBJ.PST.PRF
They criticized Caesars army for its luxury, although it had always lacked in
everything.
(35) Is est quem omnes admirentur.
he-NOM.SG be-3SG.IND.PRS REL-ACC.SG all-NOM.PL admire-3PL.SUBJ.PRS
He is such that everybody admires him.
(36) Haec qui videat, nonne cogatur
this-ACC.PL REL-NOM.SG see-3SG.SUBJ.PRS, not? force-3SG.SUBJ.PRS.PASS
confiteri deos esse?
concede-INF.PRS god-ACC.PL be-INF.PRS
When one sees this, does one not have to recognize the existence of gods?
(37) Misit legatos, qui pacem
send-3SG.IND.PRF envoy-ACC.PL, REL-NOM.PL peace-ACC.SG
peterent.
ask for-3PL.SUBJ.IPFV
He sent envoys to ask for peace.

According to Ernout & Thomas (2002:336), this use of the subjunctive in relative
clauses aims to emphasize the intervention of the speaker, who is then providing
more than a plain addition of the semantic contents conveyed by the MC and the
RC, by establishing an inferrable semantic relationship between MC and RC.
 Chapter 4. The discourse functions of ARCs 113

Below we provide examples (38)(43) that can easily be paraphrased by mak-

ing the semantic link explicit through the use of a connective (38a)(43a):

(38) (QUAL_ARC177) The late Democratic governor of Missouri, Mel Carnahan,

who died in a plane crash three weeks ago, also made history of a kind as the
first dead man to win election to the US Senate.
(39) (FIC_ARC235) Its the mix of domestic image and mass bloodshed that does it
to them; a mix that would have been appreciated by Zenia, who enjoyed such
turbulence, such violent contradictions.
(40) (QUAL_ARC112) Camelot, which has run the National Lottery for the past six
years and has been criticised for profiteering, has won the right to be considered
for the next seven-year-licence.
(41) (RAD16_ARC3) I go outside and work in the workshop | in the garden | where
I cant be seen from the street.
(42) It s likely to make little difference to the big retail chains <,> who can negotiate
cheap terms with the card companies (ICE-GB)
(43) (FIC_ARC257) Its soothing to be among strangers, who require from her no
efforts, no explanations, no reassurances.
(38) a. The late Democratic governor of Missouri, Mel Carnahan, as/since/because
he died in a plane crash three weeks ago, also made history of a kind as the
first dead man to win election to the US Senate.
(39) a. Its the mix of domestic image and mass bloodshed that does it to them; a
mix that would have been appreciated by Zenia, as/since/because/for
enjoyed such turbulence, such violent contradictions.
(40) a. Camelot, although it has run the National Lottery for the past six years and
has been criticised for profiteering, has won the right to be considered for
the next seven-year-licence.
(41) a. I go outside and work in the workshop | in the garden | because this is a place
where I cant be seen from the street.
(42) a. It s likely to make little difference to the big retail chains <,> because they
can negotiate cheap terms with the card companies
(43) a. Its soothing to be among strangers, as they require from her no efforts, no
explanations, no reassurances.

A syntactic manipulation of this sort is not possible with other relevance ARCs
such as those aiming to level the shared cognitive space as illustrated in:
114 Appositive Relative Clauses in English

(18) b. #At the Dayton conference in 1995, as/since/because it ended the war in
Bosnia, he was still central to American diplomatic efforts, winning friends
with his focus on the big picture and his evocative language.
(19) a. #It is hard to square his action with the Energy Policy and Conservation
Act, as/since/because/although it authorizes the president to tap the
reserve.

As far as the legitimizing relevance ARCs are concerned, such a manipulation is

possible, but the causal link that is established is different: it does not provide an
explanation or a cause for the informational content of the MC, but rather explains
why a referent is in a legitimate and relevant position as the subject (or object) of
the verb in the main clause. This is how we could paraphrase (25) and (26):

(25) a. Tony Sewell, as he has just finished an inquiry into soaring levels of
exclusions among black pupils from a London school, claimed that too
much concern with money and consumer goods was almost as damaging
to black pupils chances as racism.
(26) a. The story of genetic research on humans, which took 10 years to uncover,
is likely to shake the world of anthropology to its core, according to professor
Terry Turner of Cornell University, and he can say that because he has read
the proofs.

The use of as in (25a) is different from the use of as in (40a) or (41a): the gloss
provided in (25a) explains why Tony Sewell is enough of an expert to say some-
thing about black pupils difficulties in school. The gloss of (26) clearly makes this
legitimization goal explicit.
We distinguish two types of relevance ARCs with the aim of explanation, jus-
tification or concession vis--vis the main clause, viz anticipatory and reparatory,
as illustrated by the following examples:

(44) (TABL_ARC364) On a pitch like that, where there is absolutely nothing for
fast bowlers, it is imperative to show some aggression, and that is what I shall
continue to do. = reparatory explanation
(45) (QUAL_ARC142) During the course of the 11th century, Sicily was unique in
being a Christian kingdom in which Muslims, who had previously ruled the
island, continued to hold important positions. = anticipatory explanation

In (44), the speaker explains in the ARC what he means by a pitch like that, the
antecedent of the ARC, whereas in (45) the speaker anticipates the use of the
process continued, the semantic features of which presuppose anteriority (see the
use of the past perfect) while no such anteriority has been mentioned in the prior
 Chapter 4. The discourse functions of ARCs 115

discourse. In the first example, the speaker aims to repair any lack of relevance
that is likely to appear for the addressee(s) if they do not share with the speaker
the information that what he means by a pitch like that is a pitch where there is
nothing for fast bowlers; in the second example, the speaker anticipates such a
lack of information by providing information before the problematic element oc-
curs in the sentence. These two examples could be rephrased as follows:

(44) a. On a pitch like that, and what I mean by a pitch like that is a pitch where
there is absolutely nothing for fast bowlers, it is imperative to show some
aggression.
(45) a. Sicily was unique in being a Christian kingdom in which Muslims, and I
remind you that they had previously ruled the island, continued to hold
important positions.

3.3.4 Metatextual function: Relevance through transitions

Finally, relevance ARCs can be used by speakers to make the organization of dis-
course relevant. The speaker can then make a transition to a new discourse topic
thanks to an ARC, a strategy that seems less abrupt than the use of an indepen-
dent clause (examples from the BNC):

(46) Since the publication of the proposal, the Government have taken the opportu-
nity to make two minor technical amendments, which I shall deal with now.
(47) In summary, Arup estimates that there is a gap of almost 500 million, which
could completely, or to a large extent alone, be compensated for by the additional
tourism benefits generated apart from the wider legacy, which I shall discuss
now.

If we adopt a more Gricean approcah to communication, we could say that these

examples abide by the Maxim of Manner: through the use of a transition between
two discourse topics, the speaker avoids a change that might be too abrupt and
signals the change in an explicit way.

. As far as the use of ARCs for transitions is concerned, we discovered and studied a specific
use of ARCs in broadcast news (both on television and the radio) in French. The use of an ARC
enables the newscaster to shift from one item of news to another in a smooth way by using an
element that is common to both items of news, as in the following example, which we provide
translated into English:
[Beginning of live report by journalist] () Last Sunday, Nicolas Sarkozyi
announced that he would use the new tax law to reduce inheritance taxes () The
reform should allow a majority of legatees to be exempt from paying any taxes and the
wealthiest to pay a little less. [end of live report on inheritance taxes]
116 Appositive Relative Clauses in English

This use of ARCs as a transition can also be found in fiction texts. The ARC
allows a shift in the point of view and thus in the organization of the text:

(48) (FIC_ARC244) She pauses at the coffee room, where two of her colleagues, both
dressed in fleecy jogging suits, are having milk and cookies.

The ARC in (48) allows a transition between the discovery of the room by one
of the characters and a discussion with two other characters. The ARC serves to
progress the narrative.
ARCs can also be used in live debates or interviews on the radio or on televi-
sion to bring back the addressee into the topic of the debate/interview in case
s/he shifts away from the discourse topic. ARCs then represent a polite and subtle
way to shift from a topic to another (the original one), as is illustrated by the fol-
lowing example, taken from CNN show Larry King Live:

(49) A: Theres something about being big on stage, where you have to project,
you have to get into the back row. Where a lot of times films if youve been
in films a long time you start modulating down low, and trying to be sincere
and all that. That doesnt work in the theater. You have to give it out.
B: And the camera has to like you, right?
A: I think so.
B: You cant make it like you.
A: I think that the camera knows, some way or another, it knows who you
are.
B: And separates the wheat from the chaff, do you think it does?
A: Yes, except that because the way they edit film a lot of times, you can get a
performance out of somebody. Maybe not for 20 years, but you know, once
in a while, you can get a great performance out of what I would call an
amateur actor.
B: Which you couldnt do in theater. Because theater is the actors art. Once
the curtain opens, the directors useless.
A: Youre up there where the winds blowing.

Newscaster: Nicolas Sarkozyi who was thus with his European colleagues, has claimed
that the French public deficit would not exceed 3% of the GDP.
For a detailed analysis (discourse-based and prosodic) of such examples, see Auran & Loock
(2007).
. The example has been taken directly from the CNN website. The transcription that is pro-
vided has to be considered with caution, as their first aim is to report the contents of interviews.
As a consequence, some repetitions, hesitations, etc. have very likely been removed.
 Chapter 4. The discourse functions of ARCs 117

The journalist (B) is interviewing an actor (A) on cinema actors performing on a

theatre stage. The topic of the interview is then more about theatres than the cin-
ema. And yet, after a while, the interview seems to shift gradually to the cinema.
The journalist then wants to re-center the debate through the use of an ARC
which ensures continuity and shifts the discourse topic back to the theatre.

3.4 From relevance to politeness

Leech (1983) has shown convincingly that Grices Cooperative Principle is not
sufficient to explain why a speaker can sometimes be so indirect in the way s/he
conveys a message, with consequences that are necessarily linguistic (lexical and
grammatical choices). He therefore defines the Politeness Principle (among other
additional principles) in order to explain the minimization of the expression of
impolite beliefs and the maximization of the expression of polite beliefs. Rel-
evance ARCs, which we just defined as a compromise oriented towards the ad-
dressees, informed or uninformed, present information that is potentially known
by at least some of those addressees and potentially unknown by others. As a
consequence, ARCs can be used to truly inform readers/listeners of something
they do not know while presenting the information as something they already
know, therefore making no presupposition as to their ignorance. To use Princes
terminology, relevance ARCs enable the presentation of hearer new information
as if it were hearer old, or at least indeterminable.

3.5 Relevance vs. relevance

This title might sound incongruous, but what is meant here is that a relevance
ARC can convey information that apparently is not relevant in the discourse, but
is actually (very) relevant for the addressees. In such cases, the ARC conveys the
information that the addressees are interested in knowing although it does not
belong to the discourse topic sufficiently to be conveyed as foreground informa-
tion (through the use of an independent clause, for instance, although this cor-
relation is not systematic, as discussed above). This is a strategy that is frequently
used by tabloid newspapers to convey spicy information that is not necessary to
understand the discourse but that will very certainly satisfy the readers curiosity,
as in the following examples:

(50) Next to them [Harry, William, Prince Charles and Queen Elizabeth] were
Dianas brother and two sisters, as well as artists close to the Princess like Elton
John who did not sing , Cliff Richard or photographer Mario Testino.
118 Appositive Relative Clauses in English

(51) Nicolas Sarkozy arrived yesterday with his wife Cecilia, whom he was holding
by the hand, in the seaside resort of Heiligendamm, Germany.
(52) (TABL_ARC344) Heres Mel C sticking to what she does best these days per-
forming away from the other spice Girls. () Wearing a Clash T-shirt Sporty
Spice, who we revealed was being treated for depression, also appeared to have
a new tattoo on her wrist which said cred something she is desperate to
achieve.
(53) (TABL_ARC329) Lisa admits she is grateful other people have shown more faith
in her beauty than she ever has. That was demonstrated when she was just 16.
Visiting her mum, who had throat cancer, in hospital, she was approached by a
woman from a model agency at a tube station. She gave me her card and asked
me to think it over, but I forgot all about it.

The first two examples, which are actually translations of sentences found in
French newspaper articles, provide examples of ARCs whose informational con-
tent appears to be at odds with the discourse topic: the 10th anniversary of Prin-
cess Dianas death for (50), the 2007 G8 summit in Germany for (51). However,
the information (that Sarkozy was holding his wife by the hand or that Elton John
did not sing) is relevant for the addressees as they correspond to details (however
trivial) they might be wondering about when reading the articles. In (50), the
information that Elton John did not sing might seem irrelevant supplementary
information but represents anecdotal information that the readers might want to
hear about (this refers to the fact that Elton John sang at Princess Dianas funeral
in 1997 and was mocked for it). In the same way, the article from which (51) has
been extracted was written at a time when rumors of divorce for Nicolas and
Cecilia Sarkozy were all over French newspapers. The information that the presi-
dent was holding his wife by the hand has nothing to do with the G8 summit but
interests the readers all the same. Examples (52) and (53), extracted from tabloid
newspapers, illustrate a journalistic device that is quite common for such press:
information that seems to be disconnected from the discourse topic and which
as a consequence might appear to violate the Maxim of Relevance, is conveyed by
ARCs, which take the information to the background as defined in this book. This
information sometimes bears no relation with the discourse as a whole, except
that it brings spicy details, details about celebrities private lives, which readers
are interested in. In (52), the fact that Sporty Spice is being treated for depression
is disconnected from the discourse topic of the article (tattoos on stars). Similarly,
the fact that Lisas mother has throat cancer in (53) has no link with the content
of the article, the topic of which is Lisas career as an actress. It simply brings extra
details on her private life.
 Chapter 4. The discourse functions of ARCs 119

We now turn to the description of the formal characteristics of relevance

ARCs that distinguish relevance ARCs from continuative and subjectivity ARCs.

3.6 Characteristics

3.6.1 Antecedent and type of information in the ARC

Relevance ARCs are most often inserted after an antecedent that is a proper name
(about three quarters of relevance ARCs modify a proper name, compared to one
third for the whole corpus). Similarly, half of the antecedents occupy the subject
position in the MC. At the same time, three fourths of these relevance ARCs are
introduced by a relative pronoun that is also the subject in the ARC itself.
As far as the informational status of antecedents is concerned, 70% of the
antecedents followed by a relevance ARC are discourse new, among which only
half are hearer old or inferrable, i.e. retrievable by the addressees. The content of
the ARC itself is indeterminable in half of our relevance ARCs while the other
half is hearer new.
These results provide us with quite a homogeneous configuration for the in-
sertion of relevance ARCs. The results show that a relevance ARC is typically in-
serted when a discrepancy exists between the antecedents syntactic position and
its informational status. In English, syntactic subjects are canonically sentence-
initial while the general progression in the sentence is from old to new informa-
tion (e.g. Chafe 1987, 1994; Prince 1981:247). The low accessibility of the referent
of the antecedent should not allow it to occupy the subject position. As a conse-
quence, the role of the ARC is to provide information that is necessary to ensure
satisfactory knowledge of the referent of the antecedent, i.e. all the information
considered to be relevant at that point. This information is typically a permanent
property as illustrated by the fact that most verbs are state verbs or process verbs
in the simple present, predicating a permanent property.

3.6.2 Suppressibility
In Section 2.2.4 dedicated to the suppressibility of continuative ARCs, we claimed
such ARCs could not be suppressed without important damages to the sentences.
As far as relevance ARCs are concerned, they seem to be suppressible under cer-
tain conditions. A study we conducted with native speaker consultants, in which
they were asked to provide judgments on sentences from which ARCs had been
removed, offered contrasting results. Because of the indeterminable informational
status of both the referents antecedents and the ARCs, the possibility of suppress-
ing the ARCs is clearly dependent on the speakers/writers states of knowledge.
120 Appositive Relative Clauses in English

4. Subjectivity ARCs

4.1 Definitions

Speakers can use an ARC to interrupt the main predication so as to insert an opin-
ion, a comment or a judgment on the antecedent and/or the predication itself. The
speaker then shifts away temporarily from the natural course of discourse. There
is thus a discrepancy, a break between two levels: the referential/objective level vs.
the commentary/subjective level (the ARC). The ARC contains markers of mo-
dality, such as modal auxiliaries, vocabulary denoting judgment or appreciation.
The following examples provide examples of subjectivity ARCs:

(54) (QUAL_ARC552) This incredible spirit which Chelsea so clearly lack is

summed up beautifully by Gemmill, who has been unable to command a regular
plane and has also been a target for some of the fans on his rare appearances.
(55) (QUAL_ARC117) The mens 4 x 100m team, who might not have qualified
anyway, went out in the heats when they bungled a change-over, straying out of
the prescribed area.
(56) (QUAL_ARC78) Such defensive reactions are, in part, a response to the growing
problem of litigation, which is creating a culture of punishment.
(57) (QUAL_ARC113) After Atlanta, where southern hospitality was killed off by
incompetence and corporate greed, the movement desperately needed a feel-
good Olympics.
(58) (RAD09_ARC1) Unfortunately | the Gilbert Library | which is a great favourite
of mine in Pearse Street | is closed for the moment | its refurb being refur-
bished
(59) a lot of men describe women as bitchy <,> and and some some women describe
men as bitchy which is interesting (ICE-GB)

In these ARCs, in which the speaker conveys his/her own judgment, parentheti-
cals like I think, in my opinion or to me can be inserted, while this is impossible
with continuative and relevance ARCs:

(54) a. This incredible spirit which to me Chelsea so clearly lack is summed up

beautifully by Gemmill, who has been unable to command a regular plane
and has also been a target for some of the fans on his rare appearances.

. This insertion is problematic for example (58), but this has nothing to do with the subjec-
tive character of the informational content of the ARC, which is clearly evaluative therefore
subjective.
 Chapter 4. The discourse functions of ARCs 121

(55) a. The mens 4 x 100m team, who I think might not have qualified anyway,
went out in the heats when they bungled a change-over, straying out of the
prescribed area.
(56) a. Such defensive reactions are, in part, a response to the growing problem of
litigation, which to me/in my opinion is creating a culture of punishment.
(57) a. After Atlanta, where in my opinion/I think southern hospitality was killed
off by incompetence and corporate greed, the movement desperately needed
a feel-good Olympics.
(59) a. a lot of men describe women as bitchy <,> and and some some women
describe men as bitchy which I think is interesting (ICE-GB)
(19) b. #It is hard to square his action with the Energy Policy and Conservation
Act, which to me/I think/in my opinion authorizes the president to tap the
reserve.
(25) b. #Tony Sewell, who to me/I think/in my opinion has just finished an
inquiry into soaring levels of exclusions among black pupils from a
London school, claimed that too much concern with money and consumer
goods was almost as damaging to black pupils chances as racism.
(38) b. #The late Democratic governor of Missouri, Mel Carnahan, who to me/I
think/in my opinion died in a plane crash three weeks ago, also made
history of a kind as the first dead man to win election to the US Senate.
(1) c. #So we asked a man, who in my opinion/to me shrugged his shoulders and
disappeared into a nearby shop.
(3) d. #She was found face down in the water and airlifted to hospital, where I
think/to me she died hours later.

The source of the subjectivity can be the speaker himself as in (54)(59) but also
somebody other than the speaker, whose opinion is relayed by the speaker:

(60) This years strawberry-seekers can look forward to sweeter and juicier berries,
which, according to Laurence Olins, chairman of British Summer Fruit, is a
result of the mild and warm spring weather, with no frosts and good pollina-
tion, lots of bees as well as a good amount of uninterrupted sun to ripen fruit.
(61) A semi-permanent state of crisis has been proclaimed the war on terror which
will not cease just because the name has now been abandoned. It was accom-
panied by global emergency powers (like our anti-terrorist and surveillance
legislation and the American Patriot Act which, according to Human Rights
Watch, dictators need do nothing more than photocopy).
122 Appositive Relative Clauses in English

(62) Members of two Norwegian environmental groups have staged spectacular pro-
tests against oil exploration in the Barents Sea, which they claim is too risky.
(63) (FIC_SR214) This guy was stuck in the same day, over and over again, although
they didnt really explain how that happened, which Marcus thought was
weak he liked to know how things worked.

4.1.1 Explicit and implicit subjectivity

We distinguish between explicit and implicit subjectivity in ARCs. Subjectivity
can be explicit through the use of clearly evaluative vocabulary, as in the examples
mentioned above, where the speakers express their (or others) opinions or judg-
ments in an explicit way. On the other hand, subjectivity can be less explicit, more
subtle, as in the following examples:

(64) (TABL_ARC343) The singer, who believes she is a rock diva trapped in the body
of a pop star, launched into her new single.
(65) (QUAL_ARC19) I tell this story to a bunch of parents at a grammar school
fundraiser, where everybody has paid $20 per head to sip cocktails in the school
lunch room

In (64), the speakers opinion on the singer is expressed in the ARC and is clearly
negative. The subjectivity is subtle, as the opinion is not expressed explicitly. Rath-
er, it is the use of the verb believe that seems to presuppose that she is not a rock
diva trapped in the body of a pop star. In (65), the content of the ARC appears to
be at odds with what is expected from a grammar school fundraiser and therefore
seems to be a negative comment on the fundraiser (here, of course, a larger con-
text is required for addressees to make the inferences).
Sometimes the subjectivity is so implicit that we need to create an extra sub-
category of ARCs between relevance and subjectivity ARCs. Speakers sometimes
exploit the relevance function which is the predominant reason why speakers
use ARCs in discourse to convey some comments, criticisms or opinions on the
antecedent and/or the content of the MC. We consider such ARCs as an exploi-
tation of the complex inter-clausal link we mentioned earlier. Such a technique
enables the speaker to say things without encoding them linguistically but paving
the way for inferences that readers are meant to make.
Below we provide two examples, which need to be given with the general
context of the article from which they have been extracted:

. For a distinction between which he said is interesting and which, he said, is interesting, see
Loock (2006).
 Chapter 4. The discourse functions of ARCs 123

(66) (QUAL_ARC169) Mr Blair, whose son attends a selective school, complained

that too many comprehensives had a one-size-fits-all mentality
Context: criticism of comprehensive schools and the British educational sys-
tem in general. Tony Blair is seen as not handling the problem of education
very well.
(67) (TABL_ARC538) Christopher, who was decorated for his service in Ulster, was
arrested for breach of the peace outside a hospital where he had been taken after
being hit in a scuffle.
Context: Christopher, a young boy, has been arrested although he had done
nothing wrong. The article insists on the protagonists identity, which is at
odds with what he has been reproached with.

Within the articles from which they have been extracted, (66) and (67) can be
rephrased as follows:

(66) a. Mr Blair complained about the one-size-fits-all mentality of too many

comprehensives schools, but his son does not have to face this problem
because he attends a selective school. So what does he know about it?
(67) a. Christopher was arrested for breach of the peace outside a hospital where
he had been taken after being hit in a scuffle and this is very surprising
because it does not match with his personality, in particular with the fact
that he was decorated for his service in Ulster.

In such examples, the content of the ARCs sheds new light on the informational
content of the MCs. The effect for the addressees is then the triggering of a sub-
jective reading: denial of the legitimacy of Tony Blairs comment on the British
educational system in (66), criticism of the polices action in (67). In neither case,
however, has the journalist explicitly expressed such judgments, which are left for
the readers to infer.

4.1.2 Polyphony and subjectivity ARCs

Dealing with subjectivity in literary prose could seem inappropriate, but in the
same way as for the examples above, an ARC in fiction can convey a point of
view that is different from that presented in the MC. There is then a point of view
shift, from the narrators to one of the characters, for instance, and subjectivity
is to be understood as a technique, among other literary devices, as a vehicle
to provide the reader with the point of view, more precisely the voice of one
of the characters or the narrator him/herself. This gap between the two clauses
gives rise to effects of polyphony as defined in Bakhtin (1929) for literature and
then Ducrot (1984) for linguistics. Ducrot rejects the idea that a speaker should
be considered unique, and distinguishes between the person actually speaking
124 Appositive Relative Clauses in English

(producer/talking subject), the person to whom the first person markers refer
(locutor/speaker) and finally the person responsible for the content of the utter-
ance (enunciator/utterer). Interestingly, one of the core examples Ducrot uses to
illustrate his analysis contains an ARC:

(68) And this same Nero, who is virtues model,

Makes night the screen to have her carried off!10

This sentence is extracted from the English translation of Racines Britannicus

(Act I, sc. 1). The speaker is Agrippina and makes an ironical comment on the
remarks of Albina, her confidant, as the latter considers that Neros so-called
independent behavior is the consequence of his virtue. Agrippina disagrees
with such a point of view. So when uttering the ARC, although she is the one
speaking, she is not the source for this part of the sentence. Two distinct voices
can be heard: Agrippinas and Albinas. The sentence could be more explicitly
glossed as follows:

(68) a. And this same Nero, who is virtues model as you say,
Makes night the screen to have her carried off!

The ARC therefore enables the superimposition of different voices, providing

irony and therefore subjectivity. Below are examples from our corpus (FIC):

(69) (FIC_ARC216) He had been to see two films with his mum, Home Alone 2,
which wasnt as good as Home Alone 1, and Honey, I Blew Up the Kid, which
wasnt as good as Honey, I Shrunk the Kids, and his mum had said that modern
films were too commercial, and that when she was his age
(70) (FIC_ARC221-5) Frances introduced him to each of them: Sally and Moira,
who looked tough, ignored him completely, helped themselves to a paper cup-
ful of white wine and disappeared off to the further corner of the room (Moira,
Will noted with interest, was wearing a Lorena Bobbit T-shirt); Lizzie, who was
small, sweet and scatty; Helen and Susannah, who obviously regarded SPAT as
beneath their dignity, and made rude comments about the wine and the loca-
tion; Saskia, who was ten years younger than anybody else in the room, and
looked more like somebodys daughter than somebodys mother; and Suzie, who
was tall, blond, pale, nervy-looking and beautiful. She would do, he thought,
and stopped looking at anyone else who came in.

10. Ducrot (1984) mentions the example in French:

Et ce mme Nron, que la vertu conduit,
Fait enlever Junie au milieu de la nuit.
 Chapter 4. The discourse functions of ARCs 125

In example (69), the two ARCs which wasnt as good as Home Alone 1 and which
wasnt as good as Honey, I Shrunk the Kids provide the readers with the opinion
of Marcus, the central character of the novel (Nick Hornbys About a Boy), which
corresponds to a shift in the point of view from the narrators to Marcuss, provid-
ing the young boys opinion on the two movies. In (70), where the character Will
watches women entering a room for a meeting and is clearly evaluating them,
there is a shift from the objective description of the womens entry and Wills
evaluations (ARCs). In both examples, the comment is made from the point of
view of the character (source) and is different from that of the narrator.
Below are two further examples, where the verbs seem and look are to be re-
lated to one of the characters, not the narrator, who is the source for the contents
of the main clauses:

(71) (FIC_ARC233) It [the kitchen] was much smaller and whiter, and it had loads
more of gadgets, all of which looked as though they had never been used.
(72) (FIC_ARC220) He never managed to strike up much of a rapport with Maisy,
Angies mysteriously sombre five-year-old, who seemed to regard him as frivo-
lous to his core.

The narrator him/herself can also be the source for the content of the ARC but,
in contrast with the more objective content of the MC, provides the reader with
information that is to be considered as an aside, evaluative and sometimes hu-
morous. Examples (73)(74) illustrate this phenomenon.

(73) (FIC_ARC201) The princess who after nine months of pregnancy had swollen
ankles, post-natal depression and a figure completely shot to hell and had turned
into just another tired old queen gave birth to her baby daughter.
(74) (FIC_ARC202) They decided to call her Little Red Riding Hood, which was
something they found in a book of baby names between Letitia and Lola.

4.2 The different discourse strategies

As for relevance ARCs, subjectivity ARCs as defined here can be used by speak-
ers for different discourse strategies, which we list and illustrate with examples
below.

4.2.1 Comment/judgment
Subjectivity ARCs can be used by speakers to evaluate and provide personal
opinion on the referent depicted by the antecedent and/or the subject-predicate
relation in the MC. The speakers opinion is made explicit through the use of
126 Appositive Relative Clauses in English

evaluative vocabulary. Examples (54)(59) above illustrate this discourse strategy.

Examples (75)(78) below also illustrate the same phenomenon:

(75) (QUAL_ARC26) Leaving aside the pashmina palaver of two years ago, which
was more of a fad than an interest in Indian fashion.
(76) (SPEC_SR401) In response to the tone, the autistic children exhibited slight
acceleration above baseline, which was not significant.
(77) (RAD03_ARC5) Continental has said that theyre going to lay off 12,000
people| which is a kind of astonishing er astonishingly quick action on their
part I think to announce that er yesterday
(78) (BNC_ARC7) Weve had ten meetings and only thirty cases, and thatll be the
three cases where each meeting spread over a five different locations, which er
doesnt seem a very good way to use our resources!

4.2.2 Correction
Subjectivity can also be understood as a corrective function: the speaker inter-
rupts the flow of discourse to rephrase, to correct, either the antecedent or the
informational content of the MC. This strategy is illustrated by the following
examples:

(79) (QUAL_ARC22) He took my ticket (which I meekly gave him), returned trium-
phant
(80) (FIC_ARC 294) West was taking Modern History which wasnt modern
history at all, it was simply not Ancient History, which ended with the fall of
Rome because he was interested in folk songs and ballads ().
(81) It is probably because of his criticism that his enemies (the Church) labeled him
[Voltaire] an atheist which he was not.

The subjectivity is not explicit in these examples, but the speakers intervention is
obvious. In (79), the same extra-linguistic event is first described through the rela-
tion [he/take my ticket], which is subsequently requalified/corrected by [I/meekly
give him my ticket]. In (80), a requalification is also provided: what is considered
as modern history is actually not modern history (hence the transformation of
Modern History into modern history). Finally, (81) cancels the presupposition Vol-
taire is an atheist that is created by the Churchs labeling of the writer.
In such examples, it is possible to insert phrases such as actually/in fact/to tell
the truth, for instance, which provides additional evidence for the analysis that a
reformulation considered to be more adequate is being used to replace the first
linguistic formulation:
 Chapter 4. The discourse functions of ARCs 127

(79) a. (QUAL_ARC22) He took my ticket (which actually/in fact/to tell the

truth I meekly gave him), returned triumphant
(80) a. West was taking Modern History which in fact/actually/to be honest
wasnt modern history at all, it was simply not Ancient History, which
ended with the fall of Rome because he was interested in folk songs and
ballads ().
(81) a. It is probably because of his criticism that his enemies (the Church) labeled
him [Voltaire] an atheist which actually/in fact he was not.

4.2.3 Assessment and conclusion

Subjectivity ARCs can also express a general assessment or conclusion by provid-
ing an interpretation of the antecedent (generally a sentential one). This is shown
by the frequent presence of verbs like mean or suggest.

(82) (QUAL_SR3) GM is offering other carmakers the chance to fit OnStar, which
suggests the firm will soon look to float off this business.
(83) (FIC_SR205) Macaulay Culkin probably got three hundred and fifty pounds a
week, maybe even more, which meant that if he were Macaulay Culkin he could
pay his mum to teach him.
(84) (RAD13_ARC1) And and the next sort of step for us was actually the er instal-
lation of the new 70mil projector | which means well we are effectively the only
70mil cinema or cinema thats capable of showing 70mil films in Ireland
(85) (QUAL_SR6) Addiction is a pathological relationship with an object or event,
which means it can happen in any job, he says.

4.3 Characteristics

4.3.1 Punctuation
From a typological point of view, it is interesting to note that the written ARCs
introduced with dashes or parentheses (6% and 3% of the written corpus, respec-
tively) are mostly subjectivity ARCs. The dash, which is the typical introducer of
parenthetical constructions, is particularly compatible with the function fulfilled
by the subjectivity ARCs in discourse:

(86) (QUAL_ARC6) A circular has been issued by the Chamber of Mines to mining
companies offering an interpretation of the law which some lawyers would
regard as tendentious prohibiting employers from HIV testing.
128 Appositive Relative Clauses in English

Example (80), which we repeat below, clearly shows the affinity between dash-in-
troduced subjective ARCs and parentheticals, as within the opening and closing
dashes the ARC is followed by a juxtaposed independent clause (sentential paren-
thetical) without any coordinator or connective:

(80) (FIC_ARC 294) West was taking Modern History which wasnt modern
history at all, it was simply not Ancient History, which ended with the fall of
Rome because he was interested in folk songs and ballads ().

4.3.2 Relative type

Another typical characteristic of subjectivity ARCs that distinguishes them from
continuative and relevance ARCs is the fact that sentential relatives (SRs) are
over-represented: 92.5% of the sentential relatives in our written corpus are sub-
jectivity ARCs, as is the totality of the sentential relatives in our spoken corpus. As
a consequence, subjectivity ARCs are mostly final (3 out of 4).

4.3.3 Antecedent
The informational status of the antecedents referent and of the ARCs content
is also specific: about 70% of antecedents have already been mentioned in the
prior discourse (discourse old) and are therefore known to the addressees (this
proportion was only one third for relevance ARCs). This means that a subjective
comment is added after an element that addressees are already familiar with. As
far as the informational content in the ARC is concerned, the content is naturally
discourse new, but also hearer new, as for continuatives. This result contrasts with
relevance ARCs: the subjective comments represent completely new information
for the addressees.

4.3.4 Suppressibility
We mentioned in Section 2 that continuative ARCs are impossible to suppress,
contrary to a widely-accepted claim (even a test in some grammars) that ARCs
can be suppressed without damaging the discourse. In Section 3 we mentioned
that the possibility to suppress a relevance ARC was clearly dependent on the
addressees state of knowledge. Subjectivity ARCs, on the other hand, can be sup-
pressed more easily, entailing no specific problems for the understanding of the
discourse. Addressees are then simply deprived of a personal point of view or
qualification. The survey we conducted with native speakers on the suppressibil-
ity of ARCs confirmed that discourses deprived of ARCs do not pose any specific
problems.
 Chapter 4. The discourse functions of ARCs 129

5. Register variation: Spoken vs. written ARCs

The taxonomy presented above holds for both written and spoken ARCs, as is
shown by the examples used in the definitions of the three main categories and
the associated discourse strategies. Below we discuss register variation according
to the production mode (speech vs. writing) in order to define the specificity of
spoken ARCs, which are not infrequent, contrary to what is generally claimed in
English grammars, and which play important functions as far as the organization
of discourse is concerned.

5.1 Frequency

To write their Longman Grammar of Spoken and Written English, Biber et al.
(1999) worked on a forty-million-word corpus and compiled statistics on many
grammatical phenomena. As far as relative clauses are concerned, they quanti-
fied phenomena such as the choice of relative pronoun or the determinative/ap-
positive (to use our terminology) status of the RCs. However, regarding the latter
question, it seems to us that only the quantitative data on written RCs can be
trusted. As we mentioned earlier, their corpus of spoken English is not prosodi-
cally labeled and Biber et al. did not discriminate between DRCs and ARCs. The
study of their different tables and figures11 allows a global contrastive evaluation
of the frequency of ARCs and DRCs in writing: 2.5 ARCs per thousand words for
ARCs vs. 9.7 DRCs per thousand words. DRCs are therefore about four times as
frequent as ARCs.
As far as speech is concerned, we faced the same problems as Biber et al. The
British National Corpus we used to compile our corpus12 does not provide such
systematic labeling. Consequently, we extracted a 46,000-word sample, which
we manually annotated. The figure 46,000 corresponds to the extraction of five
transcripts of spontaneous English. We collected the relative clauses introduced
by wh- only, as our primary aim was to determine the frequency of ARCs, only
marginally introduced by that and never by . In total, 126 wh-introduced RCs
were collected, 79 of them identified as DRCs and 38 of them as ARCs (9 of them
had a questionable status, which could not be disambiguated because of the

11. Obviously this cannot be rigorous and precise as we would like it to be, as Biber et al. (1999)
do not provide the reader with precise statistics but simply statistical graphs.
12. Recall that the ICE-GB, which does provide audio files alongside the electronic corpus, was
not available to us for this first part of our study but was used only later for the prosodic study
(see Section 6 below).
130 Appositive Relative Clauses in English

non-availability of audio files). These figures show that the frequency of ARCs
in spontaneous spoken English ranges from 0.82 to 0.95 per thousand words,
that is to say about three times less frequent than in writing. A study by Tao &
McCarthy provides a frequency of 0.59 per thousand words for a sample taken
from the five-million word CANCODE corpus (Cambridge and Nottingham
Corpus of Discourse in English), which consists in spontaneous speech recorded
across the British Isles, and a frequency of 0.61 per thousand words for the CSAE
corpus (Corpus of Spoken American English), which consists in recordings of
natural speech from all over the United States.13 These results show a slightly
lower frequency than ours.
In any case, what these figures show is that ARCs do exist in spoken English,
even spontaneous English, contrary to what some pedagogical grammars claim.

5.2 Discourse distribution

As shown in Sections 2, 3 and 4, spoken ARCs can fulfill the same three dis-
course functions that are fulfilled by written ARCs, namely continuative, rel-
evance or subjectivity functions. However, the written and spoken corpora di-
verge in the distribution of ARCs according to their functions in discourse. As
mentioned in Section2, continuatives seem to be rarer in speech than in writ-
ing. Also, spoken ARCs are mostly subjectivity ARCs, while written ARCs were
mostly relevance ARCs.

5.3 Specific (conversational) functions

In addition to these distribution-related differences, the specificity of oral com-

munication as in presentia communication addressees are present and have the
possibility to interrupt, answer or react to what the speaker says also engenders
other differences with respect to the role of ARCs in discourse. Spoken ARCs do
not fulfill new discourse functions per se, but play a significant role in the distri-
bution of turn-taking between speakers, who can use ARCs to take their turns or
leave their turns, inviting their addressee(s) to speak. ARCs therefore have spe-
cific conversational functions which combine with the continuative, relevance
and subjectivity functions defined in the previous sections. Below we distinguish

13. The CANCODE corpus was developed by the Department of English Studies at the Uni-
versity of Nottingham. The CSAE corpus was developed by the Linguistics Department of the
University of California, and is retrievable from http://projects.ldc.upenn.edu/SBCSAE/.
 Chapter 4. The discourse functions of ARCs 131

two possibilities: (i) to take an affiliative turn by the use of an ARC (interpola-
tion); (ii) to hand the speaking turn over to the addressee by the use of an ARC.

5.3.1 Interpolation
In spontaneous conversation, the addressee can take his/her turn, sometimes
interrupting the speaker, to insert a commentary on what has just been said
or provide extra information which s/he considers helpful to the speakers tell-
ing. The use of an ARC provides a smoother, less abrupt interruption. The com-
mentary conveyed by the ARC can then correspond to a relevance, a subjectivity
(most frequently) or a continuative ARC. Below we provide examples taken from
various sources:

(87) (TEL02_ARC2)
A: Because the thing is your fellow is the stepson of Barb Streisand
B: Which makes it sort of interesting I think to the press
(88) (TEL02_ARC4)
A: Were nearly waterproof now
B: Which you need to be cos its nearly winter.
(89) [Context: Speaker 1 is recounting some dramatic travel experiences she has
heard from someone else who had experienced them in Africa. She is doing
this in response to speaker 2s request. Speaker 2, apparently possessing the
knowledge of what speaker 1 is saying, nevertheless prefers speaker 1 to retell
it to speaker 2.]
{speaker 1} They went out, Actually they just went to Chisera, to go out to
the river.
{speaker 2} Which is a hundred miles in the bush.
{speaker 1} Its about a hundred miles away. (Tao & McCarthy 2001)
(90) B: Are we going to get that fan
A: The fan will be well Toni s put an order in <,> today as I said <,> oh
probably just when things cool down <,,> that s the trouble with this this
room we were sort of moaning in the winter that it was <,> cold and they
were saying wait until the summer and you ll get the benefit then <,>
because it s the coolest room in the building
B: Which means it ll come next month (ICE-GB)
(91) A: And that was the end of your marriage except for a brief period of
reconciliation
B: Which I d prefer not to talk about (ICE-GB)
132 Appositive Relative Clauses in English

In these examples, there are clear instances of interpolation of the first speaker by
the second speaker. In (87), the actress interrupts the journalist, who does not have
the possibility to finish her turn. This enables speaker B to put an end to a topic she
does not want to address during the interview. In (88), B interrupts A, who also
cannot finish her turn: in doing so B contributes his opinion that our being water-
proof should very soon be the case. These two ARCs represent subjectivity ARCs,
as is shown by the use of the adjective interesting (to be understood as ironical)
and the verb need, which both express the speakers point of view. The example in
(89) shows the use of a relevance ARC this time: the information conveyed by the
ARC which is a hundred miles away is important to explain the accidental death
of the people who had left for Chisera (context from Tao & McCarthy 2001:672).
Interestingly, the information is hearer old for both speakers, but it is somehow
reactivated, thanks to an ARC that contributes to the story that speaker 1 was
telling. Extra information is inserted to make the discourse as a whole maximally
relevant. Finally, examples (90) and (91) are supplementary examples of interpo-
lations to insert subjectivity ARCs with a metatextual function: in (90) speaker B
wants to go back to the fan issue (it) whereas A was somehow shifting away from
the fan issue, providing peripheral information that B does not seem to be inter-
ested in. Example (91) is somehow an explicit version of (87): speaker B clearly
indicates that she does not want to talk about the issue raised by speaker A.
The motivation for speakers to use an ARC, which is by definition embedded
into a main clause, rather than an independent clause seems clear. If we modify
some of our examples and replace the ARCs with independent clauses, here is
the result:

(87) a. A: Because the thing is your fellow is the stepson of Barb Streisand
B: (And) that makes it sort of interesting I think to the press
(88) a. A: Were nearly waterproof now
B: And (that) you need to be cos its nearly winter.

Compared to the use of an independent clause, the use of the relative pronoun sig-
nals continuity despite interpolation and legitimizes the turn-taking by a second
speaker at a moment that might otherwise seem inappropriate. Such a strategy
consists in a continuation of As turn in spite of its apparent completeness (incre-
ment) that is both other-initiated and other-completed, to use Lerner (2004)s
terminology for the analysis of the organization of talk-in-interaction. Speaker B
uses what Lerner calls an initiator (in this instance the relative pronoun which in
the five examples given above) to elaborate himself on As turn.
 Chapter 4. The discourse functions of ARCs 133

5.3.2 Handing over to the addressee

ARCs can also be used by speakers to hand the talk over to the addressee, asking
him/her a question or inviting him/her to take his/her turn. The prosodic con-
tour, but also the incomplete syntactic structure, signals this intention to hand
the talk over to the addressee, without rupture, inviting him/her to complete the
unfinished structure that has been initiated:

(92) (TEL02_ARC5)
A: Its cheap, compared to the price of Welsh slates | which is
B: I dont know | two pounds I think.
(93) A: Right and acid and perhaps the most awkward one the carbonate. Which
gave?
B: Right er hydrogen

In (92) the aborted ARC which is actually corresponds to a question: what is the
price of Welsh slates. Instead of using a direct interrogative clause that would lose
the grammatical continuity and therefore seem abrupt, the speaker initiates an
ARC that the addressee is invited to complete. In (93), the use of the ARC is
quite similar. The incomplete ARC corresponds to a question, uttered here by
a teacher, who invites his students to complete his sentence. This strategy cor-
responds to what Lerner (2004) identifies as an increment, i.e. a continuation of
the turn of talk, that is initiated by the speaker himself and completed by another
one. Speaker A uses a linguistic initiator (here, the use of an aborted ARC) to
prompt the co-participant (speaker B) to voice the remainder of the increment
(Lerner 2004:157). The increment is then self-initiated but other-completed to
use Lerners (2004) terminology.
This type of ARC is not necessarily uttered by the same speaker. A second
speaker can react to what has just been said because s/he wants to know more
information on what A has just uttered. Instead of using a direct interrogative
clause, the second speaker somehow ties his/her turn to that of the first speaker,
inviting him/her to complete an unfinished ARC. The following example illus-
trates this use of ARCs:

(94) A: Have you attended conventions, Senator?

B: I have. One, to be precise.
A: Which was?
B: In 2000.
A: The nomination of Al Gore?
B: Thats correct.
134 Appositive Relative Clauses in English

In this example the journalist is trying to find out which conventions the senator
attended. This type of configuration corresponds to what Lerner (2004) defines
as the use of a type-specific, syntactically incomplete, increment initiator (in this
instance the use of which was), showing Bs answer as being deficient although
apparently complete and inviting him to complete it. More specifically, A is
prompting a type-specific elaboration () but leaving it to [B] to substantively
complete the elaboration (Lerner 2004:153). The incompleteness of the ARC and
the prosody function as increment initiators designed to reopen the turn of a prior
speaker after it has apparently been completed (Lerner 2004:174). Speaker B then
completes the increment, constituting what Lerner calls a self-completed form of
other-initiated repair.

6. The correlation between the discourse functions of ARCs

and their prosodic realizations14

6.1 Introduction

In the previous section we established that the different discourse subcategories

of ARCs continuative, relevance and subjectivity can be illustrated with mor-
phosyntactic, semantic and above all pragmatic characteristics for those ARCs.
The availability of spoken corpora with audio files, the ICE-GB and Aix-Marsec
corpora in this instance, have enabled the investigation of another possible type of
correlation, viz between prosodic phenomena and the discourse functions of one
specific syntactic structure, in this instance ARCs. A study by Auran & Loock has
confirmed that the hypothesis according to which differences in discourse func-
tions might also correspond to differences in prosodic features (tonal, temporal
and intensity aspects) is in fact correct. The study is based on the analysis of 157
ARCs which have been extracted from the ICE-GB (109 ARCs) and Aix-Marsec
(48 ARCs) corpora.15 The ARCs were initially annotated according to their dis-
course functions (providing 102 relevance ARCs, 45 subjectivity ARCs and 10
continuative ARCs), after which a series of acoustic and prosodic parameters

14. This section could not have been written without a fruitful collaboration with Cyril Au-
ran, a phonetician from the University of Lille 3 and the Savoirs, Textes, Langage CNRS labo-
ratory. I am indebted to him for making this joint research program on prosody in discourse
possible.
15. Originally 243 ARCs were collected, but for technical reasons (background noise, poor re-
cording conditions, overlapping voices), only 157 ARCs could undergo an acoustic analysis.
For the description of the two corpora, see Chapter 3, Section 2.2.
 Chapter 4. The discourse functions of ARCs 135

were compiled through a semi-automatic analysis that used original Praat scripts
(cf. Boersma 2001; Boersma & Weenink 2006). These parameters cover up to 50
prosodic parameters in total concerning tonal, temporal and intensity features.
Formal statistical analyses were carried out using the R software (for a more de-
tailed analysis of the protocol and an exhaustive list of the prosodic annotation,
see Auran & Loock 2010, to appear).
This subsection is organised as follows. First we provide an overview of the
prosodic properties of ARCs as a whole, as opposed to parentheticals in general,
in which ARCs are traditionally included (see e.g. Bolinger in particular).16 Then,
to explain the ambivalent global results, we provide a differential analysis in rela-
tion to the discourse functions played by the ARCs and discuss the significance of
these results in terms of discourse continuity/discontinuity.

6.2 The prosody of ARCs vs. the prosody of parentheticals in general

There exists quite an extensive literature on the prosodic realization(s) of par-

enthetical structures (parentheticals for short), which generally include ARCs
alongside other structures like sentential parentheticals, sentence adverbials,
comment clauses and reporting verbs, for instance. It is traditionally assumed
that they are characterized by a lowering and a narrowing in pitch range, often
coupled with pauses before and after the parenthetical (e.g. Armstrong & Ward
1931; Crystal 1969 or Cruttenden 1997). However, more recent, corpus-based
studies have shown that parentheticals actually correspond to a variety of pro-
sodic configurations (e.g. Bolinger 1989), such diversity being possibly related
to differences in syntactic, semantic and pragmatic distinctions, as noted for
example by Wichmann (2000, 2001) for and-parentheticals, or Deh (2009) for
ARCs, sentential parentheticals and comment clauses.
What the study of our corpus shows is that ARCs display characteristics that
are both typical and atypical of parentheticals in general. In particular, the corpus
study has shown that the normalised pitch level in ARCs is significantly lower
than in the preceding and following intonation units,17 which is in line with tra-
ditional descriptions for parentheticals. However, other characteristics contrast
with what is traditionally assumed for ARCs: (i) the pitch span of ARCs is not

16. For a precise overview on the category of parentheticals from a syntactic, semantic, prag-
matic and prosodic point of view, see Deh & Kavalova (2007).
17. The segmentation into intonation units (IUs) refers to the identification of separate units
representing a single continuous prosodic contour that is characterized by the presence of a
nuclear accent and optionally pauses, onset pitch reset and final lengthening (Chafe 1987; Du
Bois et al. 1993; Wightman et al. 1992; Cruttenden 1997; Gussenhoven 2004).
136 Appositive Relative Clauses in English

Table 1. ARC pitch, intensity and onset values compared with preceding
and followingIUs
Prosodic feature Preceding IU ARCs Following IU
Pitch Level 0.3 ST 0.6 ST 0.07 ST
(KS: D = 0.3758, p < 0.001) (KS: D = 0.242, p < 0.001)
Span 10.44 ST 10.71 ST 10.57 ST
Intensity Level 68.29 dB 67.04 dB 68.29 dB
Span 8.095 dB 8.142 dB 8.134 dB
Onset Absolute 1.267 ST
Differential 1.539 ST

significantly different from that of the preceding or following intonation units,

which means that while it is true that a speaker uttering an ARC resorts to a lower
pitch level, s/he does not systematically use a narrower span as is traditionally ac-
cepted; (ii) there is no significant difference in intensity levels and spans between
the ARCs and their preceding and following intonation units; (iii) the onset value
in the ARCs is clearly positive and the onset differential displays unusual positive
value. Table 1, borrowed from Auran & Loock (to appear), provides figures to
support these claims.
What this global prosodic analysis shows is that results contrast partially with
traditional prosodic descriptions and necessarily questions the notion of dis-
course continuity/discontinuity: parentheticals are in essence discontinuous and
their prosodic characteristics are traditionally markers of discourse discontinuity.
ARCs, on the other hand, while retaining some of the traditional discontinuity
markers, also display markers of discourse continuity but only partially. These
results concern ARCs as a whole and do not reflect the variety of realisations that
can be found if we study ARCs differentially, that is by discriminating between
the different discourse categories as defined in the taxonomy suggested in this
chapter. This is what we turn to now.

6.3 The prosody of ARCs in relation to their functions in discourse

6.3.1 Results18
Statistical results show that no differences can be found between ARC types in
terms of pitch span: no category displays any significant difference with its sur-
rounding intonation units as well as in terms of intensity (levels and spans).

18. Below, we provide the general results only, without any specific figures or statistics. For
complete statistical results, see Auran & Loock (to appear).
 Chapter 4. The discourse functions of ARCs 137

Table 2. Continuity (C) and discontinuity (D) marking in relation to ARC type
and prosodic parameter
Parameters
ARC type Onset value Onset differential Pitch level Intensity level
Continuative C D C C
Relevance D C D C
Subjectivity D D D C

However, some important differences appear between relevance, subjectivity

and continuative ARCs as far as other prosodic parameters are concerned, which
allow a classification of the different ARC types in terms of continuity/disconti-
nuity. First of all, we noticed that, contrary to the global trend mentioned in Sec-
tion 6.2, the pitch level in continuative ARCs is not significantly lower than that of
preceding and following IUs, a typical cue of continuity in line with the absence of
difference in pitch and intensity levels with surrounding IUs. Continuative ARCs
therefore display three typical cues of continuity. On the other hand, subjectivity
ARCs are characterised by discontinuity cues: significant positive onset values,
higher onset differential values, as well as higher pitch level. Only the intensity
level, which remains constant as compared with the preceding intonation units,
represents a cue of continuity. As for relevance ARCs, they display both cues of
continuity and discontinuity: absence of significant onset differential values and
constant intensity level (continuity); significantly positive onset values and lower
pitch level (discontinuity). Table 2, also directly borrowed from Auran & Loock
(to appear) provides a summary of these results, indicating markers of continuity
(C) and discontinuity (D) for each of the three ARC types.

6.3.2 Discussion: A (dis)continuity gradient?

The prosodic data mentioned in Table 2 provide evidence of the existence of a
(dis)continuity continuum along which the three ARC types can be arranged, a
continuum which appears to be in line with the definitions we have provided for
the three ARC types. Continuative ARCs can be regarded as the most discourse
integrated clauses: out of the 4 parameters, they display 3 typical cues of continu-
ity. At the other end of the continuum, subjectivity ARCs, characterised mostly
by discontinuity cues, seem to be the least discourse integrated clauses: out of
the 4 parameters, they display 3 typical cues of discontinuity. In between these
categories, relevance ARCs share both continuity and discontinuity cues. Figure 2
schematically illustrates this continuum.
Interestingly, these results are in line with the definitions we have provided
for the different discourse functions fulfilled by ARCs. For instance, we have
138 Appositive Relative Clauses in English

Subjectivity Relevance Continuative

discontinuity continuity

Figure 2. (Dis)Continuity gradient of ARC types

shown that the information conveyed by a subjectivity ARC, systematically dis-

course new/hearer new, clearly departs from the discourse topic, with a rupture
(or disjunction) between the referential (MC) and interpretative (ARC) levels.
This idea of a strict disjunction is confirmed by the prosodic characteristics
established by the corpus study. Relevance-oriented information, on the other
hand, is more integrated with the discourse topic than subjective information.
As we have established, a relevance ARC does not provide narrative progres-
sion (continuative ARCs) nor departure from the discourse topic through the
insertion of subjective comments, but aims to optimize the relevance of the
information that is conveyed by the main clause (current discourse topic). As
such, relevance ARCs provide background information, often of indeterminable
information status, the role of which is to support foreground information in a
closer relationship with the discourse topic than that observed with subjectivity
ARCs. Finally, the prosodic characteristics of continuative ARCs, characterized
by cues of continuity, are in line with their role in discourse, which is to provide
smooth narrative progression.

7. Conclusion

The aim of the chapter was to provide a positive discursive definition of ARCs by
providing an answer to the question of what speakers use ARCs in (written and
spoken) discourse for. After detailing the formal characteristics of ARCs in both
writing and speech in the previous chapter, we have provided evidence for the
existence of three main functions that ARCs fulfill in discourse. Figure 3 sum-
marizes the suggested taxonomy.
The suggested taxonomy distinguishes between continuative, relevance and
subjectivity ARCs, each category being defined in terms of pragmatic character-
istics as well as formal characteristics. In addition to these three discourse func-
tions, ARCs also play a role in the conversational strategies used by speakers to
organize turn taking with their addressees. The taxonomy has been illustrated
with morphosyntactic, semantic and pragmatic characteristics. Complementa-
rily, we also established a correlation between discourse functions and prosodic
 Chapter 4. The discourse functions of ARCs 139

CONTINUATIVE ARC

The ARC supports the trajec-

tory of the narrative. The events
are shown in a sequence and a
causal link may be inferred.

RELEVANCE ARC SUBJECTIVITY ARC

The aim is to optimize the rel- The ARC conveys information

evance of the antecedent and/or that is explicitly subjective and
the subject-predicate relation results in a disjunction between
within the MC. two levels:
The antecedent, in spite of its ARC The referential level (MC)
referential stability, is not suffi- The interpretative level (ARC)
ciently determined, for at least
some of the addressees, to be
used alone in discourse.

EXPLOITATION OF THE
INTER-CLAUSAL LINK

The inter-clausal link between

the MC and the ARC is exploit-
ed to bring a new perspective to
the content of the MC.

Figure 3. ARCs and their discourse functions: A taxonomy

characteristics, the prosodic realizations of ARCs depending on the discourse

continuity/discontinuity provided by the different types of ARCs.
This chapter has therefore provided a positive definition of ARCs. However,
to understand fully what speakers use ARCs for in discourse, this definition is
not sufficient. If we know the discourse motivations and conversational strategies
which explain why speakers use ARCs, we still have no idea as to the linguistic
factors that lead speakers to choose an ARC rather than other syntactic structures
that fulfill similar, or closely related, discourse functions. This is what the third
and final part of this book will explore.
part iii

ARCs and their competing allostructures

 chapter 5

Definition of ARCs allostructures,

with particular emphasis on nominal
appositives

Beyond the essential question of knowing how speakers use ARCs in discourse,
another question needs to be addressed if one is to define the discourse func-
tions of a specific syntactic structure thoroughly. If it is to be comprehensive, a
discourse-based definition needs to specify the linguistic motivations underlying
speakers use of an ARC instead of another syntactic structure that could fulfill
similar or closely related functions such as, for instance, independent clauses, jux-
taposed or coordinated. In other words, we want our definition to be both posi-
tive and differential by completing the taxonomy suggested in the previous part.
To achieve this aim, we place our study within the information-packaging
theoretical framework as defined by Vallduv (1992) in particular, but also Prince
(1981) and other researchers who have proposed links between form and mean-
ing, including Chafe (1987, 1994) or Lambrecht (1996).
This chapter is organized as follows. First we provide a definition of informa-
tion packaging and of the structures which compete with ARCs, allostructures,
as we call them. We then define and make a list of these competitive structures,
among which are appositives in general, viz apposition, which is sometimes con-
sidered by linguists to include ARCs in addition to nominal appositives and sen-
tential parentheticals. We will show that appositive structures, which require a
precise definition, can be considered natural allostructures for ARCs.

1. Definitions

1.1 Information packaging

The aim of the information packaging theory is to explain why sentences that share
the same meaning and are considered paraphrases are not completely equivalent.
Having the same meaning but not the same interpretation, such sentences are not
144 Appositive Relative Clauses in English

interchangeable in context. The same propositional content can be conveyed by

different syntactic structures, i.e. packaged in different ways, providing different
interpretations depending on the syntactic vehicle chosen by the speaker. Packag-
ing, then, can be considered to be additional meaning, namely an instruction, to
the meaning of the content.
Information packaging was first defined by Prince as the tailoring of an ut-
terance by a sender to meet the particular assumed needs of the intended receiver
(Prince 1981:224). The speakers choice of a specific packaging reflects the send-
ers hypotheses about the receivers assumptions and beliefs and strategies. But it
is Vallduv (1992, 1993), in particular, who formalized the definition of infor-
mation packaging, the aim of which is to denote th[e] nonsyntactic non logico-
semantic structuring of sentences (Vallduv 1992:12). Here is the definition that
he provides:
A sentence, in one of its facets, may be viewed as a structural vehicle used to
transfer some piece of knowledge (a proposition) from speaker to hearer. Infor-
mation packaging is the speakers tailoring of this structural vehicle to suit some
communicative aspect of the transfer of knowledge (propositional content) to
the hearer. In other words, when communicating a proposition p a given speaker
may encode p in different sentential structures according to his/her beliefs about
the hearers knowledge and attentional state with respect to p. (Vallduv 1993:2)

The choice of a syntactic structure represents an instruction that tells the addressee(s)
how the information is to be received and processed. The global meaning of a sen-
tence is the sum of its meaning (propositional content) and the meaning/instruc-
tion provided by the syntactic structure chosen to convey this content.
Syntactic structures, as a consequence, are in competition with each other. To
express propositional content, the speaker necessarily has the choice between n
syntactic vehicles, n ways of packaging the information, among which s/he choos-
es the one that fits best his/her purpose and his/her perception of the addressee(s).
These different ways of packaging the information can differ syntactically, mor-
phologically or, for languages like English, intonationally, where prosodic con-
tours themselves convey instructions. The aim of an information-packaging ap-
proach, then, is to determine the speakers linguistic constraints and motivations
when opting for a specific syntactic vehicle at the expense of others.
One case in which such an approach is now commonly used is to distinguish
canonical sentences (SVO word order) from marked structures such as clefting,
topicalization, extraposition or locative inversion. Huddleston & Pullum (2002)
provide a section on information packaging and provide rules for the use of the
passive voice, for instance. Birner & Ward (1998) have studied the constraints that
explain the (in)felicitousness of sentences with non-canonical word orders, such
 Chapter 5. Definition of ARCs allostructures 145

as (left and right) dislocation or topicalization. The constraints that are defined are
related to the informational status of the elements: the felicity of topicalization for
example is related to the discourse-old status of the pre-posed constituent or to the
existence of a discourse-old link with the left co-text (Birner & Ward 1998).
Such an approach does not claim that syntactic phenomena are not to be
taken into account. Choices for the relevant packaging necessarily involve syntax-
related parameters, such as end-weight, for instance, or constrained word order.
Examples (1) and (2) below illustrate the two phenomena, where the ungram-
maticality of (1a) results from constrained word order and the questionable ac-
ceptability of (2a) is due to a complex NP acting as the beneficiary of gave:

(1) I gave a guy I met last night my telephone number.

a. *I gave my telephone number a guy I met last night.
(2) I gave my telephone number to a guy I was introduced to by one of Leos col-
leagues I had never seen before that party.
a. ??I gave a guy I was introduced to by one of Leos colleagues I had never
seen before that party my telephone number.

As concerns our study of ARCs and their competitive structures, some syntactic
phenomena also play a role. The syntactic weight of the antecedent, for instance,
which triggers retrievability problems for the relative pronoun, can explain why a
speaker might reject the ARC in (3) to convey propositional content and opt for
another syntactic structure such as an independent clause (3a), which enables an
anaphoric relationship that avoids ambiguity:

(3) They found the master of a dangerous starving dog, who had run away after
stealing meat from his neighbours kitchen.
a. They found the master of a dangerous starving dog. The man had run
away after stealing meat from his neighbours kitchen.

Using (3a) instead of (3) blocks the interpretation of who as standing for the dog
instead of his master.

. It is also important to mention that, concerning dative alternation, the choice between
the give NP1 NP2 construction and give NP2 to NP1 is also influenced by what Bybee (2002,
2006, and 2007) has called the frequency effect, the type frequency in this instance. As is
mentioned by Bybee & Thompson in Bybee (2007), for instance, only verbs that show a very
high frequency in language use accept the give NP1 NP2 construction, which is an older and
more conservative construction. The motivation is then not due to syntactic considerations
such as end-weight mentioned in (2)(2a), but to this specific effect of type frequency. This
effect can explain the impossibility of certain uses of the give NP1 NP2 (with the verb to
whisper for instance), in addition to pragmatics-related phenomena such as the fact that the
given referents generally precede non-given referents in the sentence (e.g. Thompson 1990).
146 Appositive Relative Clauses in English

As a consequence, the information-packaging function of syntactic structures

should not be given more importance than it really has: it is false to claim that the
formal structure of a sentence is solely dependent on the speakers communica-
tive intention or that the syntax of a specific language can be explained in terms
of information packaging. If this were the case, new syntactic structures would
be created to express new intentions, as Lambrecht (1996:25) notes. In addition,
the frequency effect as defined by Bybee (see Note 1) has also to be taken into ac-
count, to explain dative alternation for instance, which cannot be accounted for
by informational considerations only.
In this book, we reject the idea that word order is governed first and foremost
by syntactic considerations, as is claimed by Hawkins (1992, 1994), while infor-
mation-packaging considerations play only a subsidiary role (Hawkins 1994:111).
Following Birner & Ward (1998), we maintain that a compromise is the best solu-
tion to account for speakers choices. We therefore borrow and adapt our conclu-
sion from them for this subsection: Although we do not dispute that weight plays
an important role in certain word-order phenomena, it will be a major thesis of this
book that context and information status are crucial to the felicity of the [different
syntactic structures] we are considering (Birner & Ward 1998:2627).

1.2 Allostructures: Definitions and methodology

The studies mentioned in the previous section have focused on the distinction be-
tween a canonical word-order structure and what is considered a marked struc-
ture because it is deviant from the canonical word order for English sentences, e.g.
active vs. passive sentences (Huddleston & Pullum 2002), sentences with topical-
ized vs. non-topicalized elements, dislocation vs. non-dislocation (Birner & Ward
1998), or extraposition vs. non-extraposition (Miller 2001; Jugnet 2008). Here,
we extend such comparisons to considerations that go beyond a binary alternative
by providing accounts for the choice of speakers in favor of an ARC in opposi-
tion to n other syntactic vehicles, which are defined below. However, among the
n+ARC structures, none of them is to be considered marked in opposition to a
more canonical, standard sentence.

. It should be specified that canonical word order and unmarked structures do not go hand
in hand systematically. For instance, the extraposition of subject that-clauses for instance (It is
very strange that the huge difference in staff salaries between the NHS and Kaiser can be written
off) is considered unmarked, i.e. less constrained in terms of discourse requirements and
therefore more frequent while the SVO structure (That the huge difference in staff salaries
between the NHS and Kaiser can be written off it is very strange) is actually marked, i.e. more
constrained (see Miller 2001) and as a consequence less frequent in discourse.
 Chapter 5. Definition of ARCs allostructures 147

As previously stated, we use the term allostructures to label these competing

structures. Just as allophones represent the different phonetic realizations of the
same phoneme or allomorphs the different realizations of the same morpheme,
allostructures represent the different possible syntactic realizations for the same
informational content (Loock 2005). Instead of using the term allosentences,
which was introduced first by Dane (1966) and Chafe (1976), and specifical-
ly defined by Lambrecht (1996:35) as semantically equivalent but formally and
pragmatically divergent sentence pairs, we prefer the term allostructures, since
the choice here has to be made between different syntactic structures, not whole
sentences. ARCs and the other structures mentioned above therefore represent
competing syntactic structures, the choice between which can be accounted for
linguistically in terms of constraints. It is the nature of these constraints that this
third part aims to investigate.

1.3 Defining the allostructures of ARCs

Defining the list of the syntactic structures that compete with ARCs in discourse
is the first question that needs to be addressed. Our study of the discourse func-
tions of ARCs has provided the definition of the discourse functions ARCs have.
Studying their competing structures, then, means studying structures that fulfill
similar or related discourse functions as those defined in the previous part of this
book (continuative, relevance, subjectivity). The observation of written and spo-
ken English enabled us to draw a list of the syntactic structures that count as allo-
structures alongside ARCs. Spoken English, where ARCs are not as frequent as in
written English, provide interesting clues as to the types of structures that might
well be rendered with ARCs in a written, more formal register. Similarly, register
variation in written English also provides interesting information for the defini-
tion of allostructures. Below we provide the list of structures that fulfill discourse
functions that are similar or closely related to those of ARCs. These structures
could compete with ARCs in discourse and are labeled alloARC. Under each
example with the alloARC (taken, as for ARCs, from different registers: QUAL[ity
press], TABL[oid press], FIC[tion texts], SPEC[ialized texts]), we systematically
provide the same sentence with an ARC without (for the time being) any gram-
matical judgment:

Coordinated independent clauses:

(4) (QUAL_alloARC32) The most difficult thing about the EU, and Im sure this is
why the Swedes voted no, is that it has very little to do with the people. It is con-
structed to make sure that it is an undemocratic structure because the authors
of it were frightened of popular power.
148 Appositive Relative Clauses in English

a. The most difficult thing about the EU, which Im sure is why the Swedes
voted no, is that it has very little to do with the people. It is constructed to
make sure that it is an undemocratic structure because the authors of it
were frightened of popular power.
(5) (QUAL_alloARC26) This is practically a taboo area for government and policy
makers because the economy is based on getting people to consume more, but
that simply cannot go on.
a. This is practically a taboo area for government and policy makers because
the economy is based on getting people to consume more, which simply
cannot go on.

Independent juxtaposed clauses:

(6) (QUAL_alloARC38) For my starter I chose the milk-fed veal, spider crab puree
and caviar, which sells for 70. This was very good, but I dont know if it was
actually 10 times as good as a normal starter.
a. For my starter I chose the milk-fed veal, spider crab puree and caviar, which
sells for 70, which was very good, but I dont know if it was actually 10
times as good as a normal starter.
(7) (FIC_alloARC20) It was so lovely and quiet here during the war, says Pamela.
You should have been here, George. She says this accusingly, as if its his fault
he wasnt.
a. It was so lovely and quiet here during the war, says Pamela. You should
have been here, George. Which she says accusingly, as if its his fault he
wasnt.

Sentential parentheticals:
(8) (FIC_alloARC1) Because of her original awe of Connor he was very smart,
he knew a lot about ancient bones, about foreign travel, about how to mix
drinks she did not drive nearly as hard a bargain with him as she could
have.
a. Because of her original awe of Connor who was very smart and knew a
lot about ancient bones, about foreign travel, about how to mix drinks she
did not drive nearly as hard a bargain with him as she could have.

adverbials:
(9) (QUAL_alloARC49) [A]ll British-donated blood has been stripped of its white
cells, because they are thought most likely to carry the infectious prion par-
ticles.
a. [A]ll British-donated blood has been stripped of its white cells, which are
thought most likely to carry the infectious prion particles.
 Chapter 5. Definition of ARCs allostructures 149

Nominal appositives:
(10) (QUAL_alloARC102) Mullah Mohammed Omar, the reclusive spiritual leader
of the Taliban, broadcast to the nation that Afghanistan had nothing to fear
from the cowardly United States.
a. Mullah Mohammed Omar, who is the reclusive spiritual leader of the
Taliban, broadcast to the nation that Afghanistan had nothing to fear from
the cowardly United States.
(11) (FIC_alloARC144) On the way out of the city, I asked Walter to drive me past
my old house the house where Id once lived with Richard.
a. On the way out of the city, I asked Walter to drive me past my old house
which is the house where Id once lived with Richard.

-ing or -ed clauses (participle clauses):

(12) (QUAL_alloARC5) Facing a mountain of debt in the early 1980s, caused mostly
by an expensive office relocation, the board repeatedly bought cheaper blood
products from risky sources, the investigators reported.
a. Facing a mountain of debt in the early 1980s, which was caused mostly
by an expensive office relocation, the board repeatedly bought cheaper blood
products from risky sources, the investigators reported.
(13) (QUAL_alloARC19) John Kerry, running as a war veteran, also faces a loss of
support, as does John Edwards, the South Carolina senator whose own formal
candidacy declaration yesterday was overshadowed by Gen Clark.
a. John Kerry, who is running as a war veteran, also faces a loss of support,
as does John Edwards, the South Carolina senator whose own formal
candidacy declaration yesterday was overshadowed by Gen Clark.

Co-referential NPs:
(14) (QUAL_alloARC45) Schwarzenegger warned that painful spending cuts are
coming as he outlined his agenda for 2004 in the annual state of the state
speech in Sacramento. () The world-famous screen hero, a Republican born
in Austria, was elected Oct. 7 to replace Gray Davis.
a. Schwarzenegger warned that painful spending cuts are coming as he
outlined his agenda for 2004 in the annual state of the state speech
in Sacramento. () Schwarzenegger, who is a world-famous screen hero,
a Republican born in Austria, was elected Oct. 7 to replace Gray Davis.

Non-restrictive premodifiers of N:
(15) (QUAL_alloARC40) No. 1 long-distance carrier AT&T on Thursday will
announce plans to sell Internet phone service to consumers and to more busi-
nesses in 2004, say people familiar with the matter.
150 Appositive Relative Clauses in English

a. AT&T, which is the no. 1 long-distance carrier, on Thursday will announce

plans to sell Internet phone service to consumers and to more businesses in
2004, say people familiar with the matter.
(16) (QUAL_alloARC41) Former president of the United States Bill Clinton will
attend an international seminar on AIDS and SARS (severe acute respiratory
syndrome) and deliver a lecture on global AIDS prevention and control efforts,
a seminar official said Friday.
a. Bill Clinton, who is the former president of the United States, will attend
an international seminar on AIDS and SARS (severe acute respiratory
syndrome) and deliver a lecture on global AIDS prevention and control
efforts, a seminar official said Friday.

2. Apposition: The natural allostructure of ARCs?

ARCs have often been considered a subtype of apposition in general. The label
appositive that we have chosen to label RCs that do not intervene in the referen-
tial construction of their antecedents itself is a label that suggests that ARCs and
apposition have properties in common. Linguists who have tried to establish a
connection between ARCs and apposition, whether inclusive or exclusive, have
suggested different arguments and analyses that we provide and compare below.
However, the main problem to be faced when studying apposition is the defini-
tion of apposition itself. Linguists have provided so many different definitions and
conceptions of apposition that this category, if it is indeed a category, has become
an extremely fuzzy one.
In this second section of Chapter 5 on ARCs allostructures, we first pro-
vide what we consider an operational definition of apposition before studying
the types of link that can be established between apposition and ARCs. Finally,
we compare the use of ARCs and appositive structures in discourse, providing
constraints that guide speakers choices between these allostructures. In doing
so, we focus on the NP-NP type of apposition, for two reasons: (i) the antecedents
of ARCs are generally NPs and only rarely sentential antecedents (2 ARCs out
of ten, see PartII); and (ii) the analysis of ARCs we discuss below considers that
they represent an abridged version of a full NP.
Example (17) below provides an example of nominal appositives, which can
compete in discourse with ARCs (17a):

(17) Gordon Brown, Prime Minister of the United Kingdom, met Barack Obama,
President of the United States.
a. Gordon Brown, who is the Prime Minister of the United Kingdom, met
Barack Obama, who is the President of the United States.
 Chapter 5. Definition of ARCs allostructures 151

2.1 Apposition: An attempt at definition

The definitions that the literature has to offer for apposition, or appositive struc-
tures, are not interchangeable, as linguists vary greatly in their definition of the
structure, concerning the syntactic and semantic characteristics of the elements
that are in apposition, as well as to the kind of syntactic link that exists between
the two (coordination, subordination, or other type of link). Although apposition
has attracted the attention of linguists recently (Meyer 1987, 1989, 1991, 1992;
Acua-Faria 1996, 1998; McCawley 1996; Neveu 1998 and OConnor 2008
among others), there is no consensus on the delimitation of the structure in par-
ticular. If everyone (or almost: Acua-Faria stands apart, for instance) agrees
to classify examples like (17) as appositions, disagreements are many on what
should be included or not under the label apposition. Definitions have sometimes
been so liberal (Meyers for instance) that all sorts of constructions have been
considered appositions, making a precise and homogeneous definition of their
defining characteristics plainly impossible.
In this book, following Acua-Faria, we call units the two elements that
are in apposition and label them U1 and U2 respectively in relation to their linear
order in the sentence. Below we repeat example (17) with these labels that will be
used throughout this chapter:

(17) [U1 Gordon Brown], [U2 Prime Minister of the United Kingdom], met [U1Barack
Obama], [U2 President of the United States].

Below we provide our definition of apposition. We are not claiming to have a

definitive answer to this thorny question, but we suggest that, although the analy-
sis is restricted to the apposition of nominal units, the definition we provide is
operational for our study, the aim of which is to compare the use of appositives
and ARCs in discourse. In particular our analysis draws an important distinction
between semantic and informational content to describe the contribution of U2
to the sentence.

. For instance, the following constructions have been classified as appositions by some
linguists:
(i) [U1 The fiction] [U2 that they write] is interesting.
(ii) They went to [U1 the room] [U2 above].
(iii) [U1 He] came home [U2 sick].
(iv) [U1 He] sat [U2 smoking].
(v) [U1 The city] [U2 of New York] is fascinating.
152 Appositive Relative Clauses in English

2.2 Definitions

2.2.1 A secondary predication

Appositive structures consist in the (generally) asyndetic relationship between
two units U1 and U2, which can belong to various grammatical categories, al-
though we are to focus mainly on the juxtaposition of two noun phrases, as in
(17). We consider apposition a relationship, the bringing together of these two
units. Contrary to some grammars, we do not consider that U2 is an apposition;
we consider that U2 is a unit that is apposed to U1 and that between them a rela-
tionship of apposition exists, for which we must try and determine the syntactic,
semantic and pragmatic properties.
The units can belong to different categories, such as NP and NP (17), NP and
PP (18); NP and VP with V in a non-finite form ((19)(20)); NP and AdjP (21);
VP and NP (22); sentential unit and NP (23):

(18) (QUAL_APP16) The Guardian reported on September 15 that, according to a

Middle Eastern intelligence agency, one of the highjackers, [U1 Wail al-Shehri],
[U2 from Khalis in Saudi Arabia], had trained in Bin Ladens al-Farouq training
camp.
(19) (QUAL_APP138) [U1 John Kerry], [U2 running as a war veteran], also faces a
loss of support, as does John Edwards, the South Carolina senator whose own
formal candidacy declaration yesterday was overshadowed by Gen Clark.
(20) (QUAL_APP18) In the past, Bin laden has shown himself willing to use western
banks when necessary but has demonstrated a preference for Islamic banks,
which do not pay or charge [U1interests], [U2 banned under Islamic law].
(21) (QUAL_APP133) As [U1 Holly], [U2 tired to death], stumbled along the last long
stretch, a backpacked figure appeared in the distance which looked like his own
image coming to meet him.
(22) At the end of the year, the house with the most points [U1 is awarded the House
Cup], [U2 a great honour].
(23) [U1 She went to London to visit him], [U2 a big mistake.]

Even the prototypical apposition of two NPs conceals different types of relation-
ships, depending on whether it is introduced by a definite (see example (17)) or
indefinite article (see (24) below):

(24) (QUAL_APP19) [U1 Said Bahaji], [U2 a 26-year-old German of Moroccan

origin], is believed to have fled to Pakistan.

The relationship that is established through the plain juxtaposition of the two
units is a relationship that has to do with an implicit be in what we call a secondary
 Chapter 5. Definition of ARCs allostructures 153

predication, that is a predication of the type U1 be U2 that is implicit and depen-

dent on the main predication (or main clause). The existence of such a predication
is established by the following glosses of our examples:

(17) b. Gordon Brown is the Prime Minister of the United Kingdom/Barack

Obama is the President of the United States.
(18) a. Wail al-Sherhi is from Khalis in Saudi Arabia.
(19) a. John Kerry is running as a war veteran.
(20) a. Interests are banned under Islamic law.
(21) a. Holly was tired to death.
(22) a. To be awarded the House Cup is a great honour.
(23) a. Going to London to visit him/That she went to London to visit him was a
big mistake.
(24) a. Said Bahaji is a 26-year-old German of Moroccan origin.

The existence of such a secondary predication explains why some of these ex-
amples may, according to constraints we aim to establish here, be rephrased with
ARCs:

(18) b. The Guardian reported on September 15 that, according to a Middle Eastern

intelligence agency, one of the highjackers, Wail al-Shehri, who was from
Khalis in Saudi Arabia, had trained in Bin Ladens al-Farouq training
camp.
(19) b. John Kerry, who is running as a war veteran, also faces a loss of support,
as does John Edwards, the South Carolina senator whose own formal
candidacy declaration yesterday was overshadowed by Gen Clark.
(20) b. In the past, Bin Laden has shown himself willing to use western banks when
necessary but has demonstrated a preference for Islamic banks, which do
not pay or charge interests, which are banned under Islamic law.
(21) b. As Holly, who was tired to death, stumbled along the last long stretch, a
backpacked figure appeared in the distance which looked like his own image
coming to meet him.
(22) b. At the end of the year, the house with the most points is awarded the House
Cup, which is a great honour.
(23) b. She went to London to visit him, which was a big mistake.
(24) b. Said Bahaji, who is a 26-year-old German of Moroccan origin, is believed
to have fled to Pakistan.
154 Appositive Relative Clauses in English

2.2.2 Two types of semantic link

In spite of surface similarities, a closer look at appositive structures reveals that
two different U1-U2 relations can be distinguished. The classical test to distin-
guish ordinary copulative sentences and identificational sentences shows that ex-
amples such as (17b) on the one hand and examples such as (24a) on the other
hand have respectively identificational and (one of) class-membership interpreta-
tions. Consider (25) and (26):

(25) [U1 Robert Gates], [U2 the US defence secretary],

(26) [U1Dr Konig], [U2 a follower of Rudolph Steiner],

In (25), the two NPs refer to the same person in the extra-linguistic world: there is
only one US defense secretary, who is Robert Gates at the present time (i.e. when
this book is being written). In (26), however, the relationship between U1 and U2
is different: Dr Konig is being claimed to be a member of the class of disciples of
Steiner. In the first case, linguists consider that this is an identification/identity
relation while in the second case the relation is one of class membership relation
(or attributive/copular relation if we refer for instance to Burton Roberts 1975s
terminology), as is shown by the following manipulation:

(25) a. Robert Gates is the US defence secretary.

b. The US defence secretary is Robert Gates.
(26) a. Dr Konig is a follower of Rudolph Steiner.
b. *A follower of Rudolph Steiner is Dr Konig.

With what is traditionally called identificational, reversible BE, the insertion of

U2 in (25) establishes what seems to be a co-referential relation between the two
referents (we however argue below against a co-referential analysis for units in
apposition). In (26), however, the one of class membership relation (U2 predi-
cates something about U1) is asymmetrical, as is shown by the unacceptability
of (26b). In Section 3.1.1 we show that several types of identificational/identity
relations exist (following Higginss (1979) taxonomy), which accounts for some
(im)possibilities of rephrasing U2 with an ARC.

. The information status of the NP in U2 (through the choice of determiners) often reveals
the relationship between the two units: in (25) the use of the establishes an identificational
relation, while in (26) the use of a establishes a one of class membership relation. This is not
systematic, however, as is shown by the following examples taken from Burton Roberts (1975)
and the glosses using be below:
(vi) He introduced me to [U1 the young man], [U2 the heir to a fortune].
a. The young man is the heir to a fortune.
b. ??The heir to a fortune is the young man.
 Chapter 5. Definition of ARCs allostructures 155

2.2.3 Co-reference: An illusion

The two types of relation described in the previous paragraph cast serious doubts
on a traditional characteristic of apposition, viz the co-reference of U1 and U2
when these units are nominal, a property that has been traditional ever since
the structure was defined. It is only recently and without any real ramifica-
tions, since co-reference is still mentioned in most pedagogical grammars to-
day that this property has been questioned (e.g. Hockett 1955; Sopher 1971;
Burton Roberts 1975; Lago 1991; Quirk et al. 1985; Meyer 1992; Neveu 1998;
Acua-Faria 1998 inter alia). As co-reference, defined as the semantic relation
that exists between two (or more) linguistic expressions denoting the same refer-
ent in the extra-linguistic world, is a property that can apparently be applied to
prototypical examples like (17) but not to all examples of appositions, linguists
have either abandoned the criterion completely (Neveu 1998) or extended the
notion of co-reference to include hyponymy, synonymy, attributive relation, and
speaker co-reference (Meyer 1987, 1991). Quirk et al. (1985) suggest the existence
of a gradient that includes equivalence, reformulation, attribution and inclusion.
Neveu (1998) considers that the position occupied by U2 prevents the element
from having any referent and that co-reference is nothing but an illusion engen-
dered by prototypical examples. We adhere to this view, which should be gen-
eralized to all examples, including prototypical examples like (17) or (25). The
rejection of co-reference as a defining criterion will allow us to generalize the
description of appositive structures so as to include ARCs, which do not establish
a co-referential relation with their antecedent.

2.2.4 Semantic vs. informational relation

When studying apposition, it is important to distinguish between the semantic
relation between the two units, U1 and U2, on the one hand, and the informa-
tional relationship, which concerns the link between the main assertion and the
secondary predication, i.e. U1 be U2, on the other hand. This means for instance,
that in example (17), the relation between Barack Obama and President of the
United States is semantically identificational, but from an informational point of
view, the implicit relation with U1 corresponds to the information brought by the
second unit, not the second unit only. In other words, because U2 is semantically
linked to U1, it establishes an identification that corresponds to the new informa-
tion. This means that a hierarchy exists between U1 and U2, U1 belonging to the

(vii) I dont know what I would have done if I had seen him, exclaimed [U1 Mary], [U2 the
prey of violent and obscure emotions].
a. Mary is the prey of violent and obscure emotions.
b. *The prey of violent and obscure emotions is Mary.
156 Appositive Relative Clauses in English

foreground as defined in the previous chapters and U2 bringing supplementary

information through an implicit secondary predication as defined above. Natu-
rally, this can be compared to the role played by an ARC.
The existence of an informational hierarchy means that the order of the two
units is not random, since only U1 is part of the main predication. The revers-
ibility of the two units, another characteristic that has been often used to define
apposition (Biber et al. 1999:638; cf. also Quirk et al. 1985, who use the reversibil-
ity test to define whether an apposition is total or partial), is not possible: in the
case of nominal units, only U1 refers while U2 establishes an implicit secondary
predication with U1. Also, if reversibility were a defining criterion for units in ap-
position, then only noun phrases could be apposed as other types of constituents
in apposition cannot be reversed:

(27) (QUAL_APP7) [U1 A cross-party alliance], [U2 supported by sacked Home

Office Minister, Mike OBrien], has already been formed to resist the introduc-
tion of identity cards.
a. *[U2 Supported by sacked Home Office Minister, Mike OBrien], [U1 a cross-
party alliance], has already been formed to resist the introduction of
identity cards.
(28) (QUAL_APP33) [U1 Lt Patrick Scollan], [U2 from Engine 239 in Brooklyn], was
a friend of Lt Quiltys.
a. *[U2 From Engine 239 in Brooklyn], [U1 Lt Patrick Scollan], was a friend
of Lt Quiltys.

Even with noun phrases, reversibility is not systematic, as is shown by the follow-
ing examples:
(29) (QUAl_APP3) While in Sudan, Bin Laden bought [U1 his own bank], [U2 the al
Shamal Islamic Bank], with a handful of Sudanese partners linked to the Islamic
regime in Khartoum.
a. #/?While in Sudan, Bin Laden bought [U2 the al Shamal Islamic Bank],
[U1his own bank], with a handful of Sudanese partners linked to the Islamic
regime in Khartoum.

. The manipulation suggested in (27)(27a), (28), (28a) is possible if U2 becomes what some
linguists sometimes call a fronted apposition. No punctuation separates the second unit and
this is therefore no inversion of U1 and U2. It only consists in the fronting of U2, which does
not belong to the main assertion:
(27) b. Supported by sacked Home Office Minister, Mike OBrien, a cross-party alliance has
already been formed to resist the introduction of identity cards.
(28) b. From Engine 239 in Brooklyn, Lt Patrick Scollan was a friend of Lt Quiltys.
 Chapter 5. Definition of ARCs allostructures 157

(vi) c. #He introduced me to [U2 the heir to a fortune], [U1 the young man].
(vii) c. #I dont know what I would have done if I had seen him, exclaimed [U2the
prey of violent and obscure emotions], [U1 Mary].

In (29), U1 and U2 cannot be reversed as his own bank has an intensional mean-
ing: what matters is the fact that he bought his own bank, which became his only
once he bought it you cannot buy a bank that is already yours, hence the label
#/?. In (vic)(viic), the second units are introduced by the determiner the and
establish a class-membership relation between the two units, making reversibility
impossible.
As a consequence, labels like anchor for U1 vs. supplement for U2 (Huddleston
& Pullum 2002) or nucleus for U1 vs. modifier for U2 (Lago 1991) are relevant and
appropriate labels, as they show the existence of a hierarchy between U1 and U2.

2.2.5 Conclusion: The asymmetry of the two units

The apparent uniformity that is suggested by the linearity U1, U2 comprises two
types of semantic relation: (i) identificational and (ii) one of class membership; co-
reference, in the end, represents an illusion and can be applied to a limited number
of examples only. This distinction has direct consequences on the possible use of
an ARC instead of a nominal unit in U2 (see Section 3 below). In both configura-
tions, a hierarchy exists between U1 and U2, U2 bringing supplementary informa-
tion about U1. This supplementary information is predicative, namely U1 be U2.
The second unit is therefore detached from the main predication as is an ARC. In
the following sub-section we investigate the question of the syntactic link between
the two units, a natural corollary to the semantic and informational hierarchies.

2.2.6 What kind of syntactic link?

Linguists have come up with different accounts of the syntactic link between U1
(the anchor) and U2 (the supplement), ranging from coordination (the two units
are on the same level and as U1 and U2 have the same syntactic function, no sub-
ordinating link is possible) to subordination (U2 is dependent on U1 and the link
between the two units is asymmetrical). Other types of relation, halfway between
coordination and subordination, interdependence, or absence of relation, have
also been suggested.
The main argument of those who consider that U1 and U2 are coordinated is
the possibility, in some cases, to introduce U2 with a coordinator (Matthews 1981
in particular):

(30) [U1 Larus argentatus], or [U2 the herring gull],

(31) He got [U1 a cheque], and [U2 a large one].
158 Appositive Relative Clauses in English

Example (30) is an instance of apposition that is very productive in scientific texts:

the Latin/scientific name is followed by a more common noun in English. In (31),
U1 is qualified a posteriori as belonging to the class of large checks. According to
Huddleston & Pullum (2002), and and or, which they label indicators alongside
that is, that is to say, namely, to wit, are similar to traditional coordinators.
However, a series of counterarguments questions the coordinator status of
and/or in such examples and, as a consequence, the coordination link between
U1 and U2. First of all the coordinator can be omitted, which is not the case with
prototypical coordination:

(30) a. [U1 Larus argentatus], [U2 the herring gull], nests on the coast.
(31) a. He got [U1 a cheque], [U2 a large one].
(32) The herring gulli or the kittiwakej nests on this coast.
a. *The herring gull, the kittiwake nest on this coast
(33) He got a cheque and another present.
a. #He got a cheque, another present.

Also, verb agreement depends on U1 only, even when U1 and U2 are linked with
and, which is not the case for prototypical coordination:

(31) b. He got a cheque, and a large one, which was/*were given to him by his
grandmother.
(33) b. He got a cheque, and another present, which were/*was given to him by his
grandmother.

This is a very strong argument in favor of the existence of an asymmetrical

relation between the two units: only U1 belongs to the main predication and
triggers the agreement of the verb; in other words, only U1 is the subject of the
verb. This is also shown by the fact that when U1 comprises multiple elements
and U2 is singular, the verb is in the plural (examples from Acua Faria 1998),
and vice versa:

(34) Lands, brains, wealth, technology (in other words) everything we need are/
*is plentiful in our country.
(35) Everything we need (in other words) lands, brains, wealth, technology is/*are
plentiful in our country.

The same phenomenon applies for gender in languages where verb agreement is
also gender-dependent (examples in Spanish and Galician from Acua Faria
1998; examples in French from Loock 2005):
 Chapter 5. Definition of ARCs allostructures 159

Su velero la goleta que ves ahora en el dique seco de Nogueira est

(36) a.

siendo modificado/*modificada para harcelo capaz de resistir ese tipo de

traversia.
His/her sailing boat (masc) the little schooner (fem) you can see now
in Nogueiras dry dock is being completely modified (masc/*fem) to
make it able to go through that kind of crossing.
b. O seu veleiro a Goleta que ests a ver agora no dique seco de Nogueira
est a ser *modificado/modificada para face-lo quen de resistir ese tipo de
percorrido.
The little schooner (fem) you can see now in Nogueiras dry dock
(masc) his/her sailing boat is being completely modified (*masc/fem)
to make it able to go through that kind of crossing.
(37) a. Cette brasserie de campagne, (un) vritable boui-boui, est transmise/
*transmis de gnration en gnration.
This countryside brasserie (fem), a real greasy spoon (masc), is handed
(fem/*masc) over from one generation to the next.
b. Ce boui-boui, (une) vritable brasserie de campagne, est transmis/
*transmise de gnration en gnration.
This greasy spoon (masc), a real countryside brasserie (fem), is handed
over (masc/*fem) from one generation to the next.

Such an asymmetrical analysis of units in apposition has led some researchers to

argue in favor of the syntactic non integration of U2:
Apposition, not belonging stricto sensu to the syntactic structure of a sentence
and not semantically contributing to the proper assertion of the sentence in
which it occurs, should be considered as a pragmatic phenomenon of natural
language; it is supposed to express invariably a commentary of the speaker on
the proper assertion of a sentence, thus conveying a secondary information of a
sentence. (Koktov 1985:3940)

Placing her analysis within the Functional Generative Description (or FGD)
framework, Koktov argues in favor of syntactic independence, as well as for
semantic independence, for U2. However, such an analysis is also problematic,
as there is some sort of dependence between the two units, as is shown by case
languages like German, for instance. The second unit systematically bears the
same case as the first unit:

. In spoken German, however, many instances show an absence of agreement between U1

and U2, U2 bearing the nominative (or marginally dative) case (Schanen Confais 1989:562).
160 Appositive Relative Clauses in English

(38) Das ist [U1 mein Bruder Pel]NOM, [U2 der beste Fuballspieler von Brasi
lien]NOM.
That is Pel, the best football player from Brazil.
(39) Kennst Du [U1 meinen Bruder Pel]ACC, [U2 den besten Fuballspieler von
Brasilien]ACC?
Do you know Pel, the best football player from Brazil?
(40) Sie spielen mit [U1 meinem Bruder Pel]DAT, [U2 dem besten Fuballspieler von
Brasilien]DAT.
Theyre playing with Pel, the best football player from Brazil.
(41) Das ist ein Photo [U1 Pels]GEN, [U2 des besten Fuballspieles von Brasi
lien]GEN.
That is a picture of Pel, the best football player from Brazil.

The same phenomenon is illustrated in Modern Greek:

(42) , ,
O adhelfos sou o kaliteros podhosferistis
TheNOM brotherNOM thePOSS theNOM bestNOM football playerNOM
tis omadhas
thePOSS teamPOSS
[U1 your brother], [U2 the best player on the team],
(43) , ,
Sinantisa ton adhelfo sou ton kalitero podhosferisti
I met theACC brotherACC thePOSS theACC bestACC football playerACC
tis omadhas
thePOSS teamPOSS
I met [U1 your brother], [U2 the best player on the team],
(44) ,

Edhose tin mpala tou adhelfou sou tou kaliterou
he gave theACC ballACC theGEN brotherGEN thePOSS theGEN bestGEN
podhosferisti tis omadhas
football playerGEN thePOSS teamPOSS
He gave the ball to [U1 your brother], [U2 the best player on the team].

What the data from German and Modern Greek show is that U2 is not com-
pletely autonomous and syntactically independent of U1. To conclude, it seems
that a syntactic hierarchy exists between the two units, as is shown by agreement

. We must however specify that other case languages do not show the same agreement be-
tween U1 and U2. In Akkadian, a Semitic language with three cases (nominative, accusative
 Chapter 5. Definition of ARCs allostructures 161

constraints, but also by the fixed order of the two units. This dependency pre-
cludes any coordination link between the anchor and the supplement. We shall
not go into more detail here and we now move on to a comparison between ap-
position and ARCs.

2.3 Apposition and ARCs: Many similarities

If apposition seems to be the natural allostructure to compete with ARCs in

discourse, it is because both structures display many syntactic, semantic and
pragmatic similarities, which we list below. These similarities are such that some
researchers have posited that apposition and ARCs actually represent one and
the same structure, either because an ARC is a subcategory of apposition or be-
cause the second unit in an apposition represents some abridged version of an
ARC, from which the relative pronoun and the verb be have been removed. We
discuss these attempts at unification in the next section (2.4). For the time being,
we limit ourselves to a list of the similarities between ARCs and apposition.
First of all, the syntactic analyses of ARCs and apposition provided by lin-
guists and grammarians suggest a certain affinity between the two structures. Al-
though debates have been fierce and the question not entirely settled, what we
have noticed in our previous chapters is that both the ARC and the second unit
in an appositive structure have characteristics suggesting that they do not belong
to the main predication (or matrix) but are dependent on it. The existence of this
hierarchy is not solely syntactic; it is first and foremost informational (with the
exception of continuative ARCs; see previous chapter): only the antecedent of the
ARCs relative pronoun and U1 belong to the main clause/predication. It should
however be noted that for apposition, the informational hierarchy is between U1
and U1 be U2, not between U1 and U2.
The existence of such a hierarchy explains why ARCs and the second units in
appositions (more specifically the relation U1 BE U2) are not within the scope of
negation and interrogation, as is shown by the following examples:

(45) (QUAL_ARC169) Mr Blair, whose son attends a selective school, complained

that too many comprehensives had a one-size-fits-all mentality...
a. Mr Blair, whose son attends a selective school, did not complain that too
many comprehensives had a one-size-fits-all mentality...
b. Did Mr Blair, whose son attends a selective school, complain that too
many comprehensives had a one-size-fits-all mentality...?

and genitive: arrum, arrim, arram), the second unit is in the nominative case, by default,
even if U1 is in the accusative or genitive cases.
162 Appositive Relative Clauses in English

(46) (QUAL_APP49) [U1 The US defence secretary], [U2 Donald Rumsfeld] last night
said he had no reason to believe the claim by the Taliban that Bin Laden was
under their control.
a. [U1 The US defence secretary], [U2 Donald Rumsfeld] last night didnt say
he had no reason to believe the claim by the Taliban that Bin Laden was
under their control.
b. Did [U1 the US defence secretary], [U2 Donald Rumsfeld] last night say
(that) he had no reason to believe the claim by the Taliban that Bin Laden
was under their control?

The negation or questioning of the main clause has no consequence whatsoever

for the following relations:

(45) c. Mr Blairs son attends a selective school.

(46) c. The US defence secretary is Donald Rumsfeld.

Similarly, both the antecedent of an ARC and the first unit in an appositive struc-
ture are referentially independent: neither the ARC nor U2 play a role in the ref-
erential construction of the entity represented by the antecedent/U1, which are
either proper names, or NPs that find their reference independently of the em-
bedding of the RC or the juxtaposition of U2, contextually or situationally. This in
turn paves the way for the idea that U2, just like an ARC, can be omitted without
any syntactic and referential consequences on the sentence. The damage is infor-
mational and pragmatic.
Other syntactic phenomena suggest ARCs and the second units in apposi-
tions behave in similar ways. Stacking, for instance, is possible for ARCs (see
Chapter 1, Section 2.3.3) and appositive structures (all examples but (47) from
Meyer 1992):

(47) (QUAL_APP46) [U1 The party treasurer], [U2 Margaret Prosser], [U3 the deputy
head of the T&G transport union], is understood to have been kept in the dark
about Mr Shears involvement.
(48) The interview had taken place on 18 November; and the news had come [U1 one
week later to the day]: [U2 25 November 1949], [U3 a Friday].
(49) But in Goodrichs return there was really too much room for conjecture, too wide
a margin of uncertainty. After all, [U1 a figure] had to be arrived at, [U2 a definite
sum in pounds, shillings, and pence], [U3 a last chord in which all the conflicts
and problems of the return would be resolved].
(50) [U1 Lois Lane], [U2 Supermans girlfriend], [U3 enterprising female reporter].
 Chapter 5. Definition of ARCs allostructures 163

From a discourse point of view, ARCs and apposition (more specifically the rela-
tion U1 be U2) share similar characteristics and are used in discourse with the
same goals. In both cases, the structure brings supplementary, non-referential in-
formation through an embedded predication.
If we compare the information conveyed by U1 be U2 with the discourse
functions we defined for ARCs in our second part, it appears that the use of appo-
sition serves the two main functions of ARCs, viz relevance and subjectivity. The
second unit of an appositive structure cannot fulfill a continuative function, how-
ever: the absence of a finite verb makes this function inaccessible for apposition.
Through an implicit secondary predication, the role of U2 can be the optimi-
zation of the relevance of U1 in the discourse. Interestingly, statistics from Biber
at al. (1999:639) show that appositive structures appear mostly in journalistic
prose. The following examples illustrate this strategy:

(51) (QUAL_APP53) [U1 Zergul], [U2 Shirins mother], tried to raise money to treat
the infant by selling the familys winter stock of dried mullberries and borrowing
a few pennies from neighbours.
(52) (QUAL_APP1) [U1 Ossama bin Laden], [U2 the prime suspect in the September
11 attacks on New York and Washington], is being hidden at a secret location
inside Afghanistan for his safety and security.
(53) (QUAL_APP55) [U1 Martin Townsend], [U2 editor of the Sunday Express], has
made a personal appeal for her safe return, and said the paper had given its full
support to her decision to enter the country illegally.
(54) (QUAL_APP85) [U1 Ahmed Wali Massoud], [U2 charg daffaires in London
for the Afghan government ousted by the Taliban], is the brother of the rebel
Northern Alliances assassinated military leader Ahmed Shah Massoud.
(55) (QUAL_APP113) It was a measure of social control, says [U1 Penny Green],
[U2 a professor of law and criminology at Westminster University].

In (51), the implicit identification relation between Zergul and Shirins mother
optimizes the relevance of the proper name Zergul, which is discourse new and
hearer new whereas Shirin is discourse old/hearer old. Providing the family tie be-
tween the given element and the new element in discourse ensures the relevance
of the latter. In (52), the uninformed reader is provided with the identity of the
referent in U1. Given the political context at the time the article was written, this
information aims at levelling the shared cognitive space (see Chapter 4). In (53),

. One exception to this is complex nominal appositives defined in Section 3.2: Robinson
College asks candidates to complete a questionnaire on their education background, a question-
naire which is then used to help decide whether they are offered a place.
164 Appositive Relative Clauses in English

the information provided by the insertion of U2, i.e. M. Townsend is the editor of
the Sunday Times is less accessible for the addressees, the relation M. Townsend/be
editor of the Sunday Express being discourse new for all addressees and indeter-
minable, although very probably hearer new (hearer new+) for most of them.
This provides M. Townsend with a legitimate position as the syntactic subject of
make and say. In (54), the information provided by the insertion of U2, aims at
providing the addressees with information about U1s referent, in particular his
political role, optimizing its presence in the discourse: Ahmed Wali Massoud, in
addition to being Ahmed Shah Massouds brother, has a political role in the coun-
try. Finally, in (55), the legitimacy of U1s referent, Penny Green, discourse new
and hearer new for all the addressees, is ensured thanks to the insertion of U2 that
specifies the occupation (and expertise) of P. Green: it is because she is a professor
of law and criminology at Westminster University that she occupies a legitimate,
or privileged, position to provide her opinion of the situation.
Most of these relevance-oriented examples can easily be rephrased with
ARCs, which have all the characteristics of relevance ARCs as we defined them in
Part II. However, the second units of appositive structures and ARCs do not cor-
respond to similar predications: ARCs can resort to different types of predication
while the second unit in an apposition can only bring information with U1 be U2
relations. In particular, finite verbs do not appear in U2 with the exception of
[NP+RC] units, which are discussed in Section 3.2 below. As a consequence, the
type of relation established by U2 is necessarily linked to the time-frame to which
the discourse is anchored. In (53) for instance, the relation Martin Townsend/be
editor of the Sunday Express is only valid at the time the article is written. If the
speaker wants to refer to another time-frame (past or future), s/he has to resort
to an ARC with a verb in a finite form. Even the insertion of adverbs like former,
previous or earlier links the relation to the narrative time in the discourse.
The second units of appositive structures can also serve a subjectivity func-
tion, as subjectivity ARCs do. U2 then conveys an opinion, a judgment, an assess-
ment of the content in U1.

(56) (QUAL_APP121) [U1 I pray to you God to forgive me all my sins, to allow me
to glorify you in every possible way.] [U2 A hideous blasphemy of a prayer before
the holocaust of thousands of souls of all faiths].
(57) (QUAL_APP132) On Sunday, they had to contend with the singer Madonna
and the over-privileged offspring of 100 of her closest friends for [U1 the
launch of The English Roses], [U2 her attempt to reinvent herself as the new Enid
Blyton].
 Chapter 5. Definition of ARCs allostructures 165

(58) (FIC_APP30) Hes still smoking [U1 ready-mades], [U2 a luxury for him].
(59) (QUAL_APP41) What you are left with is [U1 a very heavily controlled econ-
omy, a big slice of state ownership] [U2 rather similar to old Labour].
(60) (QUAL_APP105) South African experts in medical ethics say it is questionable
whether [U1 informed consent] [U2 crucial for an ethically sound trial] can
be obtained from groups such as soldiers or prisoners.

In (56), a letter written by a university professor to the Guardian Europe in which

he analyses the motivations of the suicide bombers involved in the 2001 terrorist
attacks, the speaker qualifies with U2 a quotation that he has provided in U1 and
which supposedly reflects some inner thoughts the terrorists might have had. In
(57), the journalist qualifies the launch of Madonnas book for children: the vo-
cabulary s/he has chosen, in particular the noun attempt, which does not presup-
pose success, makes the qualification subjective, particularly negative the com-
parison Madonna-Enid Blyton being so incongruous that the addressees clearly
feel the journalists irony. In (58), U2 is clearly a qualification of U1, which could
be either the NP ready-mades (ready-mades are a luxury for him) or the relation
hes still smoking ready-mades, in which case U1 is a sentential unit. In both cases,
the word luxury is evaluative. In (59), the journalist compares the Labour Partys
policies at that time and those of the Old/pre-New Labour Party in order to ex-
press a personal judgment, according to which things have not really changed
with New Labour. Finally, the use of the adjective crucial in (60) is an explicit
marker of the journalists opinion.
From a discourse point of view, it is therefore clear that appositive structures
and ARCs share similar discourse functions, viz relevance and subjectivity.

2.4 A failed rendez-vous?

Because of the similarities we have discussed in the previous sections, we have

noted that researchers have repeatedly attempted to bring together the two struc-
tures, either considering that ARCs are a subcategory of appositive structures
or considering that appositive structures are some abridged version of ARCs.
Sometimes, the equivalence between the two structures is presented as obvious
and is not even discussed. For instance, Huddleston & Pullum (2002) classify
nominal appositives and ARCs in the same category, viz supplementation/supple-
ments without any more justification than claiming they are elements that are
not syntactically integrated into the main predication (Huddleston & Pullum
2002:1350). Meyer (1992) also brings ARCs and apposition together without
166 Appositive Relative Clauses in English

justification, which is not very surprising if we consider the liberal definition he

gives for apposition.
More formal attempts at unifying analyses for ARCs and appositive structures
are more convincing. Below we consider the two types of analyses: (i) apposition
as an abridged version of an ARC; (ii) an ARC as a subcategory of appositive
structures.

2.4.1 Apposition as abridged ARC

It is not rare to read in grammars that apposition and ARCs can be brought to-
gether because an apposition corresponds to an ARC from which the relative pro-
noun and the verb BE have been removed. We can read for instance that (61) is a
short version of (62):

(61) [U1 John], [U2 the president of the association], went to college in Mexico.
(62) John, who is the president of the association, went to college to Mexico.

This is the approach defended by McCawley (1996), for instance, who considers
that a certain number of appositions are in fact reduced non-restrictive relative
clauses. Burton Roberts (1975) also claims that traditional appositive structures
such as in [U1 Mrs Thatcher], [U2 one of the most important political figures in
England since WWII] are abridged ARCs. This manipulation is found in many
grammar textbooks, the aim being to explain the role of U2, which can indeed
frequently be rephrased with an ARC. This attractive hypothesis is sensible and
many examples can undergo the manipulation, but has found objections from
Quirk et al. (1985), Meyer (1987, 1991), and Burton Roberts (1975). Their argu-
ments are the following: (i) U2 can be rephrased with an ARC only if the rela-
tion between U1 and U2 is one of class membership ((63)(63a)). If the relation
is identificational ((64)(64a)), then the rephrasing is more difficult, sometimes
impossible (Burton Roberts 1975; Meyer 1992:55; Quirk et al. 1985:1313); (ii) if
U1 and U2 are not juxtaposed, then the reformulation with an ARC is blocked
((65)(65a)), even though extraposed ARCs do exist (see Chapter 1), (iii)finally
such reformulations are impossible for U2 starting with a present participle or
gerund for some grammarians ((66)(67) and (66a)(67a)):

. As a consequence, Burton Roberts (1975) chooses not to classify such examples as apposi-
tions. More generally, he considers that when the relation between the two units is not identifi-
cational but class membership, then the second unit is actually a reduced relative clause and the
structure has nothing to do with an apposition.
 Chapter 5. Definition of ARCs allostructures 167

(63) [U1 Captain Madison], [U2 a company commander], assembled his men and
announced their mission.
a. Captain Madison, who is a company commander, assembled his men and
announced their mission.
(64) [U1 Our neighbour], [U2 Fred Brick], is on the telephone.
a. ??Our neighbour, who is Fred Brick, is on the telephone.
(65) [U1 The man] is difficult to work with, [U2 an unsurly individual who scowls at
just about everyone he encounters.]
a. ??/*The man is difficult to work with, who is an unsurly individual who
scowls at just about everyone he encounters.
(66) The towns people, knowing legends of such creature (sic), gave about twenty
gallons of milk to the worm and he drank it.
a. *The towns people, who were knowing legends of such creature, gave about
twenty gallons of milk to the worm and he drank it.
(67) (QUAL_APP23) Theses forces, including Harrier bombers equipped with preci-
sion weapons, could be diverted to attack targets in Afghanistan, at a moments
notice.
a. *Theses forces, which were including Harrier bombers equipped with
precision weapons, could be diverted to attack targets in Afghanistan, at
a moments notice.10

What these arguments show is that U2 cannot be systematically replaced with

an ARC. The distinction identification vs. class membership requires tempering,
though. The following example shows that there is no systematic correlation be-
tween identification relation and impossibility to rephrase U2 with an ARC, on
the one hand, and class membership relation and possibility to rephrase U2 with
an ARC, on the other:

(68) [U1 Fred Brick], [U2 our neighbour], is on the telephone.

a. Fred Brick, who is our neighbour, is on the telephone.

While (64a) is unacceptable, (68a) is perfectly acceptable. It seems that it is the use
of a proper name in U2 that blocks the use of an ARC (see however Section3.1.1
below):

10. For such examples, a modification of the verbs aspect is necessary, which means we are not
facing a plain abridged version:
(66) b. The towns people, who knew legends of such creature, gave about twenty gallons of
milk to the worm and he drank it.
(67) b. Theses forces, which included Harrier bombers equipped with precision weapons,
could be diverted to attack targets in Afghanistan, at a moments notice.
168 Appositive Relative Clauses in English

(69) [U1 The Prime Minister of the United Kingdom], [U2 Gordon Brown], met [U1
the President of the United States], [U2 Barack Obama].
a. ??The Prime Minister of the United Kingdom, who is Gordon Brown, met
the President of the United States, who is Barack Obama.

2.4.2 ARC as abridged apposition

Other analyses, still in the minority, have treated the question in a reverse way:
Canac Marquis & Tremblay (1996) and Cotte (1998) have posited that an ARC
is an abridged version of the second unit of an appositive structure, a unit that
corresponds to an NP whose head is somehow missing but retrievable. What is
particularly interesting about these studies is that they come from very different
theoretical backgrounds but reach the same conclusion. This analysis of ARCs is
also in line with our conclusion in Chapter 1 of this book when we suggested that,
instead of two types/kinds of relative clauses in a language like English, one should
perhaps refer to two uses of one single type of relative clauses. Below we start with
Cottes (1998) analysis and then turn to Canac Marquis & Tremblay (1996).
Cotte (1998:275277), who adopts a French enunciative approach, suggests
that an ARC is an abridged version of an NP in which the real antecedent of the
relative pronoun has not been expressed. His analysis starts from the following
two examples:

(70) But when I began to consider the subject in this last way, ( ) which seemed the
most interesting, I soon saw that it had one fatal drawback.
(71) But when I began to consider the subject in this last way, [the way] which seemed
the most interesting, I soon saw that it had one fatal drawback.

More specifically, Cotte claims that the RC is appositive only because it appears
within an NP that is in apposition to a first nominal/sentential unit. On the sur-
face, this NP is composed of the RC only, the real antecedent not being expressed
because it has already been identified. The re-introduction of the real antecedent
is then possible (see (71)). As a consequence, what is traditionally called the ante-
cedent of the relative pronoun is not its real antecedent but a fake antecedent. The
RC is semantically linked to this fake antecedent, but not genetically linked, to
use Cottes term, by which he means not linked syntactically. In (71), this last way
has no syntactic link with the apposed RC. ARC is therefore a misleading label
and, according to Cotte, should be replaced with noun phrase in apposition that
contains an RC.
From a very different theoretical perspective, Canac Marquis & Tremblay
(1996) reach the same conclusion. They posit the existence of one single type
of relative clause, determinative (or restrictive) RCs. The differences that can be
 Chapter 5. Definition of ARCs allostructures 169

observed between what is traditionally seen as the two types of relative clause are
the result of independent constraints that govern the identification of null pro-
nominals and the syntax of unmerged objects. They posit the existence of a pro
element that is co-indexed with the first unit of the apposition, i.e. our U1:

(72) Johni, [DP proi [CP who Mary knows well]], came late.

The pro element is labeled a covert pronominal, which can be introduced:

(73) John, he who is usually not late, came late.

Like Cotte (1998), they consider that ARCs are in fact traditional apposed nomi-
nal units, an analysis that has the advantage of removing a certain number of
useless complexities in the grammar of English, in particular the intriguing dif-
ferences that exist between ARCs and DRCs (e.g. the facts that only an ARC can
be juxtaposed to a proper name, a sentential antecedent and must be introduced
by a wh- pronoun). According to their analysis, the differences between what is
traditionally known as the two types of RCs are due to constraints on the pro
element and have nothing to do with the relatives themselves. For instance, they
analyze the impossibility of replacing a wh- pronoun with in an ARC as due to
the non-instantiation of the pro element.11,12
One of the major problems with these analyses, though, is the theoretical issue
posed by the absence of the real antecedent for the relative pronoun, the head of
the NP that corresponds to U2: should we speak of the presence of an absence or
rather of the absence of a presence? Also, none of these analyses provides linguis-
tic constraints that govern the presence or absence of the real antecedent. Though
we cannot treat here the theoretical question of the real antecedents absence, as
this question goes far beyond the scope of our study, the second question will be
addressed in detail in Section 3.2.

11. Some of their arguments are not supported, though. They consider for instance that what
is traditionally known as an ARC can be neither extraposed nor stacked. As we have shown in
Chapter 1 of this book, this is perfectly possible.
12. Zhang (2001) provides the same type of analysis for a comparative study between English
and Chinese relative clauses. She starts from Canac Marquis & Tremblays (1996) analysis to
explain why ARCs apparently do not exist in Chinese (see Chapter 1, Section 3.3.2). She dis-
criminates between headed appositives (HAs, see (viii)), which correspond to noun phrases in
apposition, and null-head appositives (NHAs, see (viiia)), viz ARCs according to our terminol-
ogy. She suggests that, contrary to English, Chinese has only headed appositives while English
has a rule that allows the head not to appear.
( viii) Jane, a woman whom Mary met at the party, is brilliant. (HA)
a. Jane, whom Mary met at the party, is brilliant. (NHA)
170 Appositive Relative Clauses in English

To conclude, what these analyses show is that bringing together ARCs and
the second units in apposed structures is possible. The question is not settled
yet hence the title failed rendez-vous for this section but the second type
of analysis (ARCs are abridged versions of NPs) seems to be more promising.
Such a uniform treatment accounts for many similarities syntactic, semantic
and pragmatic between the two structures and simplifies the grammar of Eng-
lish in an Ockhamian way. Inevitably, appositive structures, in particular nominal
appositives, become allostructures that compete in discourse to convey the same
informational content. We now turn to the definition of the constraints that gov-
ern speakers choices between them.

3. Apposition and ARCs as allostructures

In this section we compare ARCs and the second unit in appositions (U2). We
focus primarily on nominal appositives but discriminate between simple and
complex nominal units. By complex we mean second units where NPs are
composed of a head noun followed by a relative clause, i.e. NPs that, in Cottes and
Canac Marquis & Tremblays analyses, are considered as RCs preceded by their
true antecedent. Such a distinction is made in order to address a flaw in these
studies, namely the definition of constraints that govern the presence/absence
of what is considered to be the real antecedent of the relative pronoun in these
analyses. The term simple refers to NPs in which no relative clauses appear. Ex-
amples (74) and (75) illustrate these two configurations:

(74) (QUAL_APP88) He coached [U1 his daughter Kerrys girls soccer team], [U2 the
Petrides Panthers], and each of them came to his funeral wearing their soccer
shirts and clutching red carnations.
(75) (QUAL_NP+RC12) Ill be playing the part of [U1 Danny Ocean], [U2 a gangster
who decides to hold up every casino in Las Vegas in a single night].

3.1 ARCs vs. simple nominal apposition

These are cases where U2 is an NP that contains no relative clause. Below we

provide examples, some of which have already been used above but are repeated
below for the sake of clarity:

(76) [U1 The King of Pop], [U2 Michael Jackson], has died today.
(77) He [U1 went to London], [U2 a big mistake].
 Chapter 5. Definition of ARCs allostructures 171

(78) (QUAL_APP56) [U1 The Palestinian leader], [U2 Yasser Arafat], is facing an
open revolt against his efforts to impose a ceasefire on the year-long uprising
after unruly mobs hurled hand grenades at two of their own police stations.
(79) (QUAL_APP63) I think people were scared, said [U1 Peter Prows], [U2 a poli-
tics student from Oberlin College, Ohio].

Such apposed units have the reputation of being easily glossed with an ARC, es-
pecially when the relation between the two units is one of class membership, but
with the exception of cases where U2 is a proper name. As the semantic link be-
tween U1 and U2 is an important factor when reformulation with an ARC is to
be considered (even with proper names), we distinguish between identificational
(Section 3.1.1) and class membership (Section 3.1.2) relations below.

3.1.1 Identificational relation between the two units

When the relation between the two units is identificational, the type of NP that
appears as the second unit plays a role in the competition with ARCs. In particu-
lar we must distinguish cases where U2 is a proper name from other cases (com-
mon nouns).

3.1.1.1 U2 is a proper name. If U2 is a proper name, then replacing the apposed

NP with an ARC is in general clearly impossible, as we suggested in the previous
section. The identification relation is too obvious to be made explicit and an ARC
is pragmatically unacceptable. Proper names then do not compete with ARCs
((76a)(78a)). However, it is interesting to note that if who is X is replaced with
whose name is X, the reformulation is acceptable ((78b)(27c)):

(76) a. ??/*The King of Pop, who is Michael Jackson, has died today.
(78) a. ??/*The Palestinian leader, who is Yasser Arafat, is facing an open revolt
against his efforts to impose a ceasefire on the year-long uprising after
unruly mobs hurled hand grenades at two of their own police stations.
b. The Palestinian leader, whose name is Yasser Arafat, is facing an open revolt
against his efforts to impose a ceasefire on the year-long uprising after
unruly mobs hurled hand grenades at two of their own police stations.
(27) c. A cross-party alliance, supported by sacked Home Office Minister, whose
name is Mike OBrien, has already been formed to resist the introduction of
identity cards.

This is due to the fact that U2 does not mention a particular property or charac-
teristic of U1s referent but provides addressees with its name. We can however
imagine one exception to this, in which a proper name can be introduced within
an ARC instead of a simple NP. Consider the following example:
172 Appositive Relative Clauses in English

(80) Barack Obama is already being criticised for his economic policy to tackle the
economic crisis. Some of his fiercest opponents even consider him to be incom-
petent as far as economic matters are concerned and even deny him his right
to the title of president. These people keep reminding us that the people that
really are in command of the countrys economy are industrialists, who know
what should be done to boost economic development. These industrialists are
sometimes called the real presidents of the United States. But the president of the
United States, who is Barack Obama, has repeatedly re-asserted that he is the
one in charge of the country, no matter what his opponents might say.

This time, it is the replacement of the ARC with a nominal appositive that is not
acceptable:

(80) a. ??But the president of the United States, Barack Obama, has repeatedly
re-asserted that he is the one in charge of the country, no matter what his
opponents might say.

What the speaker is doing is not identifying U1, but reasserting the relation
U1 be U2, which has been questioned in the preceding co-text. (80) could be
glossedas:
(80) b. But the president of the United States, who no matter what people say is
Barack Obama indeed, has repeatedly re-asserted that he is the person in
charge of the country.

Also, even though the impossibility of introducing a proper name with wh- +
be seems to be quite systematic, the existence of a specific semantic link between
U1 and U2 allows such reformulation. Consider (81) and (82) the first example
is taken from Mikkelsen (2004), whose use of Higginss (1979) distinction be-
tween specificational copular clauses and identificational copular clauses will
help us account for the existence of particular cases where U2 can be replaced
with an ARC:

(81) [U1 The doctor on call], [U2 Dr Jones], is to arrive at 9 p.m.

(82) [U1 The player serving], [U2 McNamara], is sick.

Replacing U2 with an ARC in these two examples is possible but requires a spe-
cific reading. In (81), if one considers a situation where two nurses know Dr. Jones
and work in his unit, then U2 does not identify the doctor but specifies who the
doctor on call is among n other doctors (The doctor on call is Dr Jones, not Dr
Smith or Dr Wilcockson for instance). In such a context, rephrasing U2 with an
ARC is possible (81a); on the other hand, if the role of U2 is to provide the name
of the doctor on call to a patient, for instance, then (81a) is not acceptable:
 Chapter 5. Definition of ARCs allostructures 173

(81) a. The doctor on call, who is Dr Jones, is to arrive at 9 p.m.

If we refer to Higginss (1979) taxonomy of copular relationships, we can say that

reformulation with an ARC is therefore possible for a specificational reading but
impossible for an identificational reading. Similarly, (82) can be rephrased with
an ARC if we imagine a context where the reading is specificational (McNamara,
not X or Y for instance, is the player serving), not identificational (the name of the
player who is serving is McNamara).
To conclude, we can say that when U2 is a proper name, discourse competi-
tion exists with ARCs within a specificational reading, or within contexts where
the identity link needs to be reasserted.

3.1.1.2 U2 is not a proper name. When U2 is not a proper name, rephrasing it

with an ARC is sometimes acceptable, sometimes clearly unacceptable, as is shown
by the following examples:

(83) (QUAL_APP139) Some expected [U1 Barack Obama], [U2 the (newly elected)
president of the United States], to appoint a completely new economic team so
as to implement another New Deal.
a. #Some expected Barack Obama, who is the (newly elected) president of
the United States, to appoint a completely new economic team so as to
implement another New Deal.
(84) (QUAL_APP141) [U1 Bill Clinton], [U2 the former president of the United
States], will attend an international seminar on AIDS and SARS (severe acute
respiratory syndrome) and deliver a lecture on global AIDS prevention and
control efforts, a seminar official said Friday.
a. #/??Bill Clinton, who is the former president of the United States,
will attend an international seminar on AIDS and SARS (severe acute
respiratory syndrome) and deliver a lecture on global AIDS prevention
and control efforts, a seminar official said Friday.
(85) (QUAL_APP138) [U1 Angela Merkel], [U2 the German chancellor], on Friday
described Barack Obamas presidency as a unique opportunity to revive the
Middle East peace process as the US leader continued his international tour with
a stop in the historic eastern city of Dresden.
a. #/??Angela Merkel, who is the German chancellor, on Friday described
Barack Obamas presidency as a unique opportunity to revive the Middle
East peace process as the US leader continued his international tour with a
stop in the historic eastern city of Dresden.
174 Appositive Relative Clauses in English

(86) (QUAL_APP142) [U1 Djokovic], [U2 the Australian Open champ], earned the
top spot based on his victory over Davydenko during the week.
a. ?Djokovic, who is the Australian Open champ, earned the top spot based
on his victory over Davydenko during the week.
(87) (QUAL_APP55) [U1 Martin Townsend], [U2 editor of the Sunday Express], has
made a personal appeal for her safe return, and said the paper had given its full
support to her decision to enter the country illegally.
a. Martin Townsend, who is the editor of the Sunday Express, has made a
personal appeal for her safe return, and said the paper had given its full
support to her decision to enter the country illegally.
(88) (QUAL_APP101) [U1 Edgar Griffin], [U2 the father of the BNP leader Nick
Griffin], was sacked as a vice-president of the Duncan Smith campaign in Wales
after he admitted answering a BNP telephone inquiry line.
a. Edgar Griffin, who is the father of the BNP leader Nick Griffin, was
sacked as a vice-president of the Duncan Smith campaign in Wales after
he admitted answering a BNP telephone inquiry line.

The question, then, is how to account for the speakers choice, sometimes highly
constrained, but sometimes rather free, for a nominal apposition as opposed to an
ARC, to express the following relations:

(83) b. Barack Obama is the (newly elected) president of the United States.
(84) b. Bill Clinton is the former president of the United States.
(85) b. Angela Merkel is the German Chancellor.
(86) b. Djokovic is the Australian Open champ.
(87) b. Martin Townsend is the editor of the Sunday Express.
(88) b. Edgar Griffin is the father of the BNP leader Nick Griffin.

What we posit to explain such a pragmatically constrained choice is that there

seems to be a link with the degree of familiarity/saliency of the relationship A is
B: the more familiar the relation, the more infelicitous the use of an ARC, which
makes the link explicit (relative pronoun + be). As the relationship in (83b) is
probably known by most (if not all) addressees, using an ARC makes the identifi-
cation relationship too explicit to be felicitous. This is why (83a)(85a) are clearly
infelicitous; (86a) questionable but not completely imaginable and (87a)(88a)
quite felicitous. Statistics in Loock (2005) show that the more familiar the infor-
mation (conveyed by the ARC or the antecedent itself), the more likely the use of
a noun modifier or a nominal appositive.
However, motivation based on familiarity is only a one-way constraint: it
explains why an ARC is infelicitous, but this does not mean that non-familiar/
 Chapter 5. Definition of ARCs allostructures 175

non-salient relationships require the use of an ARC. Consider (89)(91), as well

as the relationships (89a)(91a), which represent information that is most likely
unknown by most of the intended addressees (recall that our corpus comprises
journalistic prose from the United Kingdom and the United States only). How-
ever, this does not exclude the use of a nominal appositive:

(89) Mr Miliband is scheduled to hold bilateral talks with [U1 Mr A. Raja], [U2 the
Indian Environment Minister].
(90) [U1 AT&T], [U2 the No. 1 long-distance carrier] will announce Thursday plans
to sell Internet phone service to consumers and to more businesses.
(91) [U1 Hassan Rohani], [U2 the chief Iranian negotiator], said after three hours of
talks in Geneva yesterday that the Europeans had until the end of July to come
up with a more concrete offer.
(89) a. Mr A. Raja is the Indian Environment Minister.
(90) a. AT&T is the No. 1 long-distance carrier.
(91) a. Hassan Rohani is the chief Iranian negotiator.

The contrast with examples (83), (84) or (85) is that the use of an ARC is pos-
sible:

(89) b. Mr Miliband is scheduled to hold bilateral talks with Mr A. Raja, who is the
Indian Environment Minister.
(90) b. AT&T, which is the No. 1 long-distance carrier, will announce Thursday
plans to sell Internet phone service to consumers and to more businesses.
(91) b. Hassan Rohani, who is the chief Iranian negotiator, said after three hours of
talks in Geneva yesterday that the Europeans had until the end of July to
come up with a more concrete offer.

It also needs to be pointed out that the notion of familiarity is subject to varia-
tion, both geographically and in time. What is known (i.e. hearer-old) today in
a specific country might not be as well-known in a decade and/or on another
continent. This might explain the existence of example (92), which we already
mentioned in the previous section to illustrate the category of relevance ARCs,
while no such examples could be found about Barack Obama or George Bush for
instance, who are much more famous/salient today than William Howard Taft:

(92) The Governor is Bob Taft, fourth generation of a Republican dynasty founded
by his great grandfather, William Howard Taft, who was elected president in
1908.
176 Appositive Relative Clauses in English

To conclude, it is clear that the degree of familiarity between antecedent and in-
formation content in the ARC is a constraint that plays a role in the choice be-
tween an ARC and one of its competing allostructures: the more familiar the in-
formation, the harder it is to use an ARC. However, this factor is not crucial when
the information is clearly new information, for which an ARC or a competing
allostructure such as a nominal appositive are both acceptable.

3.1.2 Class membership relation between U1 and U2

Following Burton-Robertss (1975) distinction, the existence of a class member-
ship relation between U1 and U2 generally makes the reformulation by an ARC
possible, even though sometimes it can be somewhat artificial:

(93) (QUAL_APP63) I think people were scared, said [U1 Peter Prows], [U2 a poli-
tics student form Oberlin College, Ohio].
a. I think people were scared, said Peter Prows, who is a politics student form
Oberlin College, Ohio.
(94) (QUAL_APP67) Mouhamedou Ould Slahi is a relative of [U1 Mahfouz Ould-
al-Walid], [U2 a close associate of Bin Laden].
a. Mouhamedou Ould Slahi is a relative of Mahfouz Ould-al-Walid, who is a
close associate of Bin Laden.
(95) (QUAL_APP85) [U1 Ahmed Wali Massoud], [U2 charg daffaires in London
for the Afghan government ousted by the Taliban], is the brother of the rebel
Northern Alliances assassinated military leader Ahmed Shah Massoud.
a. Ahmed Wali Massoud, who is charg daffaires in London for the Afghan
government ousted by the Taliban, is the brother of the rebel Northern
Alliances assassinated military leader Ahmed Shah Massoud.
(96) (QUAL_APP89) There is no coffin, but the lieutenants 15-year-old son Danny
is wearing his uniform and [U1 his wife Susan], [U2 a psychiatric nurse], is car-
rying a new helmet given to her by the company because Lt Quiltys helmet is
lost somewhere in Ground Zero.
a. There is no coffin, but the lieutenants 15-year-old son Danny is wearing
his uniform and his wife Susan, who is a psychiatric nurse, is carrying a
new helmet given to her by the company because Lt Quiltys helmet is lost
somewhere in Ground Zero.
(97) (QUAL_APP95) [U1 Lord Levy], [U2 a multimillionaire businessman and tennis
partner of the prime minister], has extended his role as chief fundraiser as the
party becomes more dependent on donations from wealthy figures.
a. Lord Levy, who is a multimillionaire businessman and tennis partner of
the prime minister, has extended his role as chief fundraiser as the party
becomes more dependent on donations from wealthy figures.
 Chapter 5. Definition of ARCs allostructures 177

(98) (QUAL_APP100) [U1 Mr Binding], [U2 a retired train driver], has admitted
dressing in a klansmans white robe that was sent from the US for use in cer-
emonies.
a. Mr Binding, who is a retired train driver, has admitted dressing in a
klansmans white robe that was sent from the US for use in ceremonies.

The heavier the NP, be it syntactically or informationally, the easier it is to use an

ARC that makes the link between U1 and U2 explicit. Examples (96) and (98),
where U2 is relatively light (a retired train driver and a psychiatric nurse), become
less natural when U2 is transformed into an ARC.
Although this possibility exists in a systematic way, journalistic prose in par-
ticular shows a more extensive use of nominal apposition, which is shorter and
responds to the economy principle. As the maximally abbreviated form of post-
modifier (Biber et al. 1999:639), apposition represents the favored noun post-
modifier in highly informational texts. The constraint explaining the choice be-
tween nominal apposition and ARC seems to be register-related, and therefore
relates to stylistic variation.
We have thus shown that the competition between simple nominal apposi-
tion (i.e. U2 is a simple NP that contains no relative clause) and ARCs was gov-
erned by the type of link between the U1/antecedent and U2/the ARC (which
can be treated as a case of nominal apposition if we consider the existence of an
unexpressed antecedent following Cottes or Canac Marquis & Tremblays analy-
ses). We now turn to complex nominal apposition, i.e. cases where the second
unit that can compete with an ARC is not a simple NP like President of the United
States or a psychiatrist nurse, but a complex NP constituted of a nominal element
that is modified by a relative clause.

3.2 ARCs vs. complex nominal apposition

3.2.1 Definitions
The second unit of an appositive structure can consist of a complex noun phrase,
i.e. an NP that contains a relative clause the determinative/appositive status of
which is problematic (see below) and which is juxtaposed to a nominal ele-
ment. To refer to such complex nominal units we use the label [U2 NP+RC]: the
unit is called U2 because it is the second unit in an appositive structure, and is
constituted of a noun phrase (NP) followed by a relative clause (RC). However,
this label needs to be used with caution, as it is U2 as a whole that constitutes an
NP, not the nominal element placed before the relative clause only. However, the
problematic status of the RC and as a consequence that of the nominal element
preceding it justifies the use of such the label NP. Below are a few examples:
178 Appositive Relative Clauses in English

(99) (QUAL_NP+RC1) And [U1 Charlene Smith], [U2 a rape victim who identified
herself and told her story to the press], was denounced as racist.
(100) (QUAL_NP+RC2) There isnt a single country that can handle such an opera-
tion on its own, said [U1 Ilya Klebanov], [U2 the deputy prime minister who
heads the government inquiry into the disaster.]
(101) (QUAL_NP+RC7) Theyve picked [U1 a man]. [U2 A 6-foot-4 German basket-
ball player, no less, who does the boys laundry, makes their lunches, helps with
their dinner and kisses them goodnight.]
(102) [U1 Kamlesh Bahl], [U2 the ousted vice-president of the Law Society who success-
fully took it to court for discrimination], has lodged an appeal against its finding
that she lied on oath.
(103) (QUAL_NP+RC100) The listing was published in [U1 the New Musical Express
(NME)], [U2 the weekly that Dickins co-founded in 1952], and his idea changed
the British music business forever.
(104) (QUAL_NP+RC99) It was the idea of Percy Dickins, who has died aged 80
of heart failure, to supplement this system with [U1 figures for weekly record
sales] [U2 a chart that was to become the Top 20].

In each example, the second unit of the appositive structure (U2) is made up of
a nominal element (a rape victim in (99), the deputy prime minister in (100), a
6-foot-4 German basketball player in (101), the ousted vice-president of the Law
Society in (102), the New Musical Express in (103), and a chart in (104)) that is fol-
lowed by a relative clause (RC), whose determinative/appositive status is unclear.
As mentioned above, it would be incorrect to qualify the nominal elements as
NPs themselves, as for some of the examples (see below) the relative clause is part
of the NP represented by U2 as a whole. Hence the use of the vague, somehow
unsatisfying term nominal element symbolized as NP.
The only difference between these examples and an ARC is the presence of
what in Cottes (1998) analysis, described above, is the relative pronouns real an-
tecedent. These structures are unquestionably instances of apposition, where U2
is an NP in which the head noun is postmodified by a relative clause.
While (99)(104) seem a priori to be similar on the surface, a more detailed
analysis shows that the link between the NP and the RC does not correspond to
one strategy only. This has direct consequences on the possibility of removing the
NP, and thus on the linguistic motivations for using an ARC rather than a nomi-
nal appositive. Punctuation separates U1 from U2, suggesting that the NP and RC
in U2 are to be treated as a block. If we consider (103), U2 identifies U1, the New
Musical Express, as being a specific, uniquely identifiable newspaper, viz the one
that Dickins co-founded in 1952. The content of the RC is necessary to identify
 Chapter 5. Definition of ARCs allostructures 179

the type of newspaper and is traditionally labeled a DRC. Similarly, in (104), the
content of the RC cannot be dissociated from a chart.
This is not systematic, however. If we consider (100) and (102) for instance,
we can see that the relation between the NP in U2 and the RC is different: there is
only one vice-president of the Law Society that has been ousted; similarly, there is
only one Russian deputy minister. As a consequence, the RC plays no role in the
referential definition of these nouns. The NP and the RC do not constitute a block,
and punctuation can be inserted between them without changing the meaning of
the sentences, which is not possible with (103) and (104) mentioned previously:

(100) a. There isnt a single country that can handle such an operation on its own,
said Ilya Klebanov, the deputy prime minister, who heads the government
inquiry into the disaster.
(102) a. Kamlesh Bahl, the ousted vice-president of the Law Society, who successfully
took it to court for discrimination, has lodged an appeal against its finding
that she lied on oath.
(103) a. *The listing was published in the New Musical Express (NME), the weekly,
that/which Dickins co-founded in 1952, and his idea changed the British
music business forever.
(104) a. *It was the idea of Percy Dickins, who has died aged 80 of heart failure, to
supplement this system with figures for weekly record sales a chart, that/
which was to become the Top 20.

The differences in the contribution of the RC correlate with the (im)possibility of

removing the NP. If the RC behaves like a DRC, then the NP cannot be removed,
whereas if the RC behaves like an ARC, then the NP can be removed more easily,
posing only informational problems.

3.2.2 Presence/absence of the NP in U2: What constraints?

Overall, the presence/absence of the NP in U2 is governed by different types of
constraints. In our 100-token corpus for the structure [U1 NP], [U2 NP+ RC], we
found that only one quarter of the examples could undergo the suppression of the
NP in U2 without any grammatical or semantic or informational change, as the
following two examples show:

(105) (QUAL_NP+RC55) They even took women prisoner and took them away,
said [U1 Moel], [U2 a villager who returned the day the Taliban left].
a. They even took women prisoner and took them away, said Moel, who
returned the day the Taliban left.
180 Appositive Relative Clauses in English

(106) (QUAL_NP+RC20) In [U1 France], [U2 a nation that hosts the world champion-
ships in 2003 and is bidding to stage the Olympics in Paris in 2008], the lack of
success is being treated as a national disaster.
a. In France, which hosts the world championships in 2003 and is bidding
to stage the Olympics in Paris in 2008, the lack of success is being treated as
a national disaster.

This contrasts with the other three quarters of our corpus that do not allow the
suppression of the NP in U2:

(107) (QUAL_NP+RC26) The museum staff and city council in Banyoles refused
repeated pleas for [U1 the return of the body], [U2 an idea that was first broached
in 1992 by a Haitian-born medical doctor who was outraged by the display].
a. #The museum staff and city council in Banyoles refused repeated pleas
for the return of the body, which was first broached in 1992 by a Haitian-
born medical doctor who was outraged by the display.
(108) (QUAL_NP+RC21) Audley Harrison [U1 d just had his hair braided and dotted
with coloured rubber bands], [U2 the sort of extrovert gesture that boxers new to
the big time often inflict on themselves and their friends].
a. ??[Audley Harrison] d just had his hair braided and dotted with coloured
rubber bands, which boxers new to the big time often inflict on themselves
and their friends.
(109) (QUAL_NP+RC44) They started arriving literally before the dust had settled.
The Washington Post dubbed them [U1 hitchhikers] [U2 people who want to
use the patriotic bandwagon as a vehicle for their favourite policy proposals].
a. ??/*They started arriving literally before the dust had settled. The
Washington Post dubbed them hitchhikers who want to use the patriotic
bandwagon as a vehicle for their favourite policy proposals.
(110) (FIC_NP+RC13) We can only applaud [U1 Captain Chase]s efforts, [U2 a man
who holds to his word, unlike the strikebreaking lockout tactics in centres such
as Winnipeg and Montreal ]
a. ???/*We can only applaud Captain Chases efforts, who holds to his word,
unlike the strikebreaking lockout tactics in centres such as Winnipeg and
Montreal

Different types of constraints govern the possibility of removing the NP be-

fore the relative clause in the second unit of an appositive structure. Below we
distinguish between syntactic constraints (the suppression of the NP makes
the sentence ungrammatical), semantic constraints (mainly for metalinguistic
comments on U1), and, of course, pragmatic, discourse-based constraints (e.g.
the loss of information is judged too important for the addressees).
 Chapter 5. Definition of ARCs allostructures 181

3.2.2.1 Syntactic constraints. In addition to the status and role of the RC that has
an influence on the possible removal of the NP before the relative clause, the only
syntactic phenomenon that can constrain speakers choice for such a structure
instead of an ARC is the retrievability of the relative pronouns antecedent. The
insertion of an NP, similar or not to the NP in U1, ensures the correct retriev-
ability of the antecedent, which is made necessary for instance by the important
syntactic weight of U1 or a potential ambiguity.

(111) (FIC_NP+RC27) I knew they were making [U1 jokes about me behind my
back] [U2 jokes that had to do with my deportment (the women) and my body
(the men)], and that this was their way of getting even.
a. ?I knew they were making jokes about me behind my back which had to
do with my deportment (the women) and my body (the men), and that this
was their way of getting even.
(112) (QUAL_NP+RC21) Audley Harrison [U1 d just had his hair braided and dotted
with coloured rubber bands], [U2 the sort of extrovert gesture that boxers new to
the big time often inflict on themselves and their friends].
a. ??Audley Harrisond just had his hair braided and dotted with coloured
rubber bands, which boxers new to the big time often inflict on themselves
and their friends.
(113) (QUAL_NP+RC49) Third, there is the danger of [U1 allowing police checks],
once cards have been introduced, [U2 a move which could prove a massive waste
of police time and do enduring damage to police-community relations].
a. #/?Third, there is the danger of allowing police checks, once cards have
been introduced, which could prove a massive waste of police time and do
enduring damage to police-community relations.
(114) (QUAL_NP+RC26) The museum staff and city council in Banyoles refused
repeated pleas for [U1 the return of the body], [U2 an idea that was first broached
in 1992 by a Haitian-born medical doctor who was outraged by the display].
a. ??The museum staff and city council in Banyoles refused repeated pleas
for the return of the body, which was first broached in 1992 by a Haitian-
born medical doctor who was outraged by the display.

In (111), the absence of the NP could potentially lead to an erroneous antecedent

retrieval, my back instead of jokes. Even if the risk is not great, there is potential
for a garden-path sentence. In (112), coloured rubber bands could be recognized
as the antecedent instead of have ones hair braided and dotted with coloured rub-
ber bands. In (113), the separation between the relative pronoun which and the
antecedent in U1 might lead to similar problems, with addressees considering
cards or cards have been introduced as being the antecedent of which. Finally, the
absence of the NP in (114) before the RC creates ambiguity for the antecedent
182 Appositive Relative Clauses in English

between the body and the return of the body; the use of an idea can only refer to
the return of the body and prevents any potential ambiguity.
Similarly, with multiple antecedents (a list, for example), the use of a syn-
thetic noun before the RC ensures the antecedents correct retrieval:

(115) (FIC_NP+RC43) She was influenced by the likes of [U1 Djuna Barnes,
Elizabeth mart, Carson McCullers] [U2 authors I know for a fact that Laura
never read].

The ellipsis/removal of authors could lead to the interpretation that only the last
item on the list is the antecedent of the relative pronoun:

(115) a. She was influenced by the likes of Djuna Barnes, Elizabeth Smart, Carson
McCullers who(m) I know for a fact that Laura never read.

3.2.2.2 Semantic constraints. A semantic constraint intervenes in particular

when the NP before the RC, and in fact U2 as a whole, has a metalinguistic func-
tion, i.e. when the role of U2 is to comment on the use of a word or an expression
in U1. The comment is on the linguistic sign, not its referential content, the NP
in U1 being sometimes typographically marked with italics or quotation marks.
In such cases, the NP before the RC is generally word, expression, sentence, com-
mentary etc.:

(116) (FIC_NP+RC20) It was a case of [U1 sauve qui peut], [U2 an expression which,
due to Mr. Erskines diligence, I could now translate].
(117) (QUAL_NP+RC44) They started arriving literally before the dust had settled.
The Washington Post dubbed them [U1 hitchhikers] [U2 people who want to
use the patriotic bandwagon as a vehicle for their favourite policy proposals].
(118) (QUAL_NP+RC67) From the [U1 existentialist] philosopher [U2 (a tag that
Camus always disliked)], people awaited a polished worldview; but Camus had
none to offer.
(119) (QUAL_NP+RC69) And with Hannah Arendt we have been introduced to a
further complication: the notion of the [U1 banality of evil], [U2 (a formulation
that Camus himself would probably have taken care to avoid)], the idea that
unspeakable crimes can be committed by very unremarkable men with clear
consciences.

Removing the NP before the RC in U2 is clearly infelicitous here, as beyond the

doubt on the correct antecedent retrieval, addressees might be misled to believe
that the content of the ARC modifies/comments on the referent rather than the
linguistic sign mentioned in U1:
 Chapter 5. Definition of ARCs allostructures 183

(116) a. ?It was a case of sauve qui peut, which, due to Mr. Erskines diligence, I
could now translate.
(117) a. *They started arriving literally before the dust had settled. The Washington
Post dubbed them hitchhikers who want to use the patriotic bandwagon
as a vehicle for their favourite policy proposals.
(118) a. #/??From the existentialist philosopher (which Camus always disliked),
people awaited a polished worldview; but Camus had none to offer.
(119) a. ??And with Hannah Arendt we have been introduced to a further
complication: the notion of the banality of evil (which Camus himself
would probably have taken care to avoid), the idea that unspeakable crimes
can be committed by very unremarkable men with clear consciences.

Beyond metalinguistic-related phenomena, another type of semantic constraint

governs speakers choice between complex U2s and ARCs depending on the se-
mantic relation between the NP in U1 and the NP before the RC in U2. As for
the relation between U1 and U2 for appositive structures in general, the rela-
tion between the NP in U1 and the NP before the RC in U2 can be one of class
membership (120) or identificational (121), as is shown by the manipulations in
(120 a/b)(121a/b):

(120) (QUAL_NP+RC89) They were strange, wicked people, said [U1 Monir
Mohammed Hosien], 37, [U2 a mechanic whose business went bankrupt when
the Islamic militia arrived four years ago].
(121) (QUAL_NP+RC71) [U1 Frederick Law Olmsted], [U2 the father of American
landscape architecture who designed Central Park and chose the land for
McLean, also died there].
(120) a. Monir Mohammed Hosien is a mechanic.
b. *A mechanic is Monir Mohammed Hosien.
(121) a. Frederick Law Olmsted is the father of American landscape architecture.
b. The father of American landscape architecture is Frederick Law Olmsted.

When the relation is class membership, U2 provides a double predication: in (120),

Monir Mohammed Hosien, which is discourse new/hearer new for the addressees,
is associated with a mechanic (first predication); extra information is then provid-
ed about the referent (his company went bankrupt). The result is a concatenation
of information in one complex NP. As a result, removing the second NP is seman-
tically possible in spite of the natural consequent informational loss.
184 Appositive Relative Clauses in English

(120) c. They were strange, wicked people, said Monir Mohammed Hosien, 37,
whose business went bankrupt when the Islamic militia arrived four
years ago.

Similarly (122) can become (122a) and remain acceptable:

(122) (QUAL_NP+RC65) By then, the obscure Algerian novelist had become a

national figure: a hero of the intellectual resistance, editor of [U1 Combat] [U2(a
daily paper that was hugely influential in the postwar years)], and an icon to a
new generation of French men and women hungry for ideas and idols.
a. By then, the obscure Algerian novelist had become a national figure: a
hero of the intellectual resistance, editor of Combat (which was hugely
influential in the postwar years), and an icon to a new generation of
French men and women hungry for ideas and idols.

Another configuration where U1 and the NP before the RC in U2 are in a class

membership relation, making the removal of the NP in U2 possible, is the con-
figuration where U1 is a sentential unit. In (123) and (124) below, the only poten-
tial problem concerning the removal of the NP in U2 ((123a)(124a)) is related
to antecedent retrieval phenomena (see previous subsection), not to a semantic
problem (bailing out the private company is a move and careful monitoring is a
process):

(123) (QUAL_NP+RC82) The government has already taken steps to alert the
European commission that it has been forced to [U1 bail out the private com-
pany] [U2 a move that would normally be against EU rules ] or face the
collapse of Britains railway network.
(124) (QUAL_NP+RC13) These patients will then [U1 be carefully monitored to see if
their conditions improve], [U2 a process that will take more than a year of careful
assessment].
(123) a. The government has already taken steps to alert the European commission
that it has been forced to bail out the private company which would
normally be against EU rules or face the collapse of Britains railway
network.
(124) a. These patients will then be carefully monitored to see if their conditions
improve, which will take more than a year of careful assessment.

However, when the relation between U1 and the NP before the RC in U2 is iden-
tificational, the removal of the NP in U2 is not systematically possible. It is of
course possible when the RC does not play a determinative role, i.e. only when
 Chapter 5. Definition of ARCs allostructures 185

the referent of this NP can find its reference independently of the embedding
of the RC in U2. The configuration is then quite similar to the concatenation
of information discussed above, as we could consider that U2 actually consists
in two synthesized second units. More formally, the punctuation test, viz the
insertion of a comma between the NP and the RC a test we have used earlier
in this section can be used to determine the possibility of removing the NP in
U2: only if a comma can be inserted can the NP be removed. As a consequence,
the examples provided in (121) and (125) below can undergo the manipulation
(121a)(125a), as they can also undergo the comma test (121b)(125b):

(121) a. Frederick Law Olmsted, who designed Central Park and chose the land
for McLean, also died there.
b. Frederick Law Olmsted, the father of American landscape architecture,
who designed Central Park and chose the land for McLean, also died
there.
(125) (QUAL_NP+RC2) There isnt a single country that can handle such an opera-
tion on its own, said [U1 Ilya Klebanov], [U2 the deputy prime minister who
heads the government inquiry into the disaster.]
a. There isnt a single country that can handle such an operation on its own,
said Ilya Klebanov, who heads the government inquiry into the disaster.
b. There isnt a single country that can handle such an operation on its own,
said Ilya Klebanov, the deputy prime minister, who heads the government
inquiry into the disaster.

It is when the RC has a determinative role that the NP in U2 is not systemati-

cally suppressible. Examples (126)(127) below accept the manipulation, while
(128)(130) do not:13

(126) (QUAL_NP+RC53) Atef, who is thought to have been in his late 50s, is a former
member of [U1 Egyptian Islamic Jihad], [U2 the group which killed President
Anwar Sadat in 1981 and has been a close aide to Bin Laden for more than 10
years].
a. Atef, who is thought to have been in his late 50s, is a former member of
Egyptian Islamic Jihad, which killed President Anwar Sadat in 1981 and
has been a close aide to Bin Laden for more than 10 years.

13. Naturally the relative pronoun, quite often that or because of the clearly determinative
status of the RC, needs to be changed into a wh- pronoun so that the judgment concerns only
the removal of the NP.
186 Appositive Relative Clauses in English

(127) (QUAL_NP+RC46) [U1 Enerkom], [U2 the state-owned company that devel-
oped oxihumate-k], nevertheless says that its trials, which have been carried out
under the auspices of the University of Pretoria, have been running since 1999.
a. Enerkom, which developed oxihumate-k, nevertheless says that its trials,
which have been carried out under the auspices of the University of Pretoria,
have been running since 1999.
(128) (FIC_NP+RC29) Miss Laura Chase had been seen around town said
Elwood in the company of a young man, the very same young man shed been
photographed with at the button factory picnic.
a. #Miss Laura Chase had been seen around town said Elwood in the
company of a young man, whom shed been photographed with at the
button factory picnic.
(129) (FIC_NP+RC37) Id worn my best daytime outfit, the only possible outfit I had
for such an occasion: a navy-blue suit with a pleated skirt, a white blouse
a. ??Id worn my best daytime outfit, which I had for such an occasion: a
navy-blue suit with a pleated skirt, a white blouse
(130) (FIC_NP+RC15) Soon Laura began to ask where the baby had gone, the one
that didnt look like a kitten.
a. ??Soon Laura began to ask where the baby had gone, which didnt look like
a kitten.

In these cases, it is the semantic contribution of the NP itself as well as the

strongly determinative status of the RC which makes the suppression impossible.
The use of very same, only and one require from the RC a crucial identifying role:
the speaker does not mention any young man, any possible outfit or any baby but
mentions specific elements which find their reference thanks to the embedding of
the RC. As a consequence, removing the NP damages the referential construc-
tion established by U2 as a whole. ARCs are then semantically infelicitous, the
change being referential rather than informational.

3.2.2.3 Pragmatic constraints. If, as we have just seen in Sections 3.2.2.1 and
3.2.2.2, the removal of the NP in U2 has consequences on a relative small num-
ber of sentences, it is because the presence of this NP is first and foremost due to
pragmatic/discursive considerations. More specifically, the function of a complex
U2 is to ensure the relevance of U1 in discourse: the more discourse new/hearer
new the referent of U1, the more important the role of the information conveyed
in U2. If we compare the informational status of the antecedents in our corpus of
ARCs and the informational status of the first units in our [U2 NP+RC] corpus,
here are the results that we obtain:
 Chapter 5. Definition of ARCs allostructures 187

Table 1. Informational status of antecedent vs. U1

Antecedent in ARCs corpus U1 in [U2 NP+RC]
Discourse new 68.5% 92%
Hearer new 20.90% 31.50%
both dn/hn 20.90% 31.50%

The proportion of discourse new/hearer new referents for the first unit of
[U2NP+RC] constructions is more important than the proportion of discourse
new/hearer new antecedents for the corpus of ARCs. The difference is much
higher if we consider journalistic prose only, a written register for which the het-
erogeneity of the addressees/readers states of knowledge accounts for the need
for relevance-oriented strategies: 16.5% of antecedents in the corpus of ARCs
are discourse new/hearer new, as opposed to 49% of U1s. The concatenation of
information in U2 maximally and in a synthesized way optimizes the relevance
of a referent entirely new for the reader, generally denoted by a proper name.
The concatenated aspect of the conveyed information complies with the Gricean
Maxim of Manner, Be brief.
As a consequence, removing the NP in U2 removes information that is impor-
tant for the addressees to identify and understand the relevance of U1s referent.
If from a purely syntactic and semantic point of view, the removal is not problem-
atic, the change is on the level of discourse and the constraint purely pragmatic.
Other constraints concern a level that is more discourse-organizational than
purely informational. The presence of the NP may then seem optional, as it is
not justified by the speakers will to provide his/her addressees with specific in-
formational content (to ensure the relevance of U1s referent, for instance, or to
convey a subjective comment), but it can nevertheless be explained by linguistic
motivations. Consider the following examples:

(131) (QUAL_NP+RC20) In [U1 France], [U2 a nation that hosts the world champion-
ships in 2003 and is bidding to stage the Olympics in Paris in 2008], the lack of
success is being treated as a national disaster.
(132) (FIC_NP+RC18) Or I read [U1 Alfred, Lord Tennyson], [U2 a man whose maj-
esty was second only to Gods], in the opinion of Miss Violence.
(133) (QUAL_NP+RC24) [U1 Vojislav Kostunica], [U2 the man who appears on the
verge of recognized power], faces overwhelming problems, not least the fact that
the disintegration of Yugoslavia may not be over.

The NPs in the second units do not provide extra information one could even
question the relevance of the informational content conveyed by these NPs. In
each of these examples, the relation established between the NP in U1 and the NP
188 Appositive Relative Clauses in English

in U2 is tautological, completely hearer old for all addressees, and could be per-
ceived as violating the Gricean Maxims of Manner (be brief) and Quantity (do not
make your contribution to the conversation more informative than necessary):

(134) France is a nation.

(135) Alfred Lord Tennyson is a man.
(136) Vojislav Kostunica is a man.

Naturally, removing the NPs provides examples that are completely acceptable
modulo adjusting the relative pronoun:

(131) a. In France, which hosts the world championships in 2003 and is bidding
to stage the Olympics in Paris in 2008, the lack of success is being treated as
a national disaster.
(132) a. Or I read Alfred, Lord Tennyson, whose majesty was second only to Gods,
in the opinion of Miss Violence.
(133) a. Vojislav Kostunica, who appears on the verge of recognized power, faces
overwhelming problems, not least the fact that the disintegration of
Yugoslavia may not be over.

Nevertheless, the presence of the NPs a nation, a man and the man corresponds
to a specific conversational choice by the speakers. In (131), the use of the word
nation is in direct connection with the end of the main clause, namely the lack of
success is being treated as a national disaster. When categorizing France as a na-
tion (rather than a country for instance), the speaker emphasizes the solemn and
institutional character of the country, whereas (131a) is much more neutral. The
solemnity is to be contrasted with the equation lack of success = national disaster.
In (132), the presence of a man is to be considered in opposition with God a few
words later: the speaker, specifying that Lord Tennyson belongs to the category
of men establishes a contrast between the referent of U1 and God. Similarly in
(133), the presence of the man, which in itself does not provide any extra relevant
information, corresponds to a specific motivation from the speaker: by classifying
V.Kostunica as the man who, the speaker singularizes the referent as the agent
of the verb appear, in opposition to other men that could have fulfilled this the-
matic role. In such examples, the motivations are linguistic, but in a more subtle
way, playing with connotations engendered by the use these NPs before the RCs
rather than pure truth-conditional semantics and informational contributions.
 Chapter 5. Definition of ARCs allostructures 189

4. Conclusion

The aim of this chapter was to elaborate, within Vallduvs information-packag-

ing theoretical framework, a list of syntactic structures with similar or closely-
related discourse functions, so as to explain, through the definition of specific
constraints, why speakers use an ARC instead of one of these other allostructures.
We have shown that appositives, in particular nominal appositives, stand out as
the natural allostructure for ARCs: syntactic, semantic and pragmatic similarities
explain why many linguists have attempted to unify their analyses to encompass
both appositive structures and ARCs. Although one specific type of analysis, that
is ARCs represent a nominal appositive whose head noun is absent or has been
ellipted, may seem reasonable (Cotte; Canac Marquis & Tremblay), we have nev-
ertheless shown that speakers choice between nominal appositives and ARCs is
not random, but rather is governed by morphosyntactic constraints as well as
semantic and pragmatic constraints, and that nominal appositives and ARCs are
not interchangeable in context. These constraints range from the type of noun in
U2, accessibility/ambiguity problems concerning the relative pronouns anteced-
ent to the informational status of the antecedents referent and the role of the RC
in complex nominal appositives.
In the next chapter we investigate other allostructures, such as independent
clauses and adverbials, and define constraints that account for speakers choice
between these structures.
 chapter 6

Appositive relative clauses and their other

competing allostructures

The previous chapter has defined apposition, in particular nominal appositives,

as a structure that competes in discourse with ARCs. Because of their numer-
ous similarities with ARCs, appositives have been treated as the natural competi-
tor for ARCs. The aim of this chapter is to investigate other syntactic structures
that compete in discourse with ARCs and provide constraints that account for
speakers choices between different types of packaging. Section 1 provides a list
of these different possibilities and provides various examples (or rather reminds
the reader of them, as a list was already provided in Chapter 5). Section 2 distin-
guishes between syntactic phenomena and semantic/pragmatic phenomena since
syntactic phenomena, which represent constraints governing speakers choice to
package the information they aim to convey, need to be put aside for the defi-
nition of purely semantic and pragmatic constraints. Section 3 is dedicated to
the definition of such constraints and takes into consideration intonation-related
phenomena, as intonation can also serve a crucial information-packaging func-
tion. Finally, we tackle the question of register and stylistic variation.

1. What (other) allostructures for ARCs (alloARCs)?

If appositive structures appear as the natural competitor for ARCs, it is in particu-

lar because of the hierarchy that these allostructures establish between the infor-
mational foreground and the informational background as defined in Chapter 4.
While the foreground represents the backbone or skeleton of the text, forming its
basic structure, the background put[s] flesh on the skeleton, but [is] extraneous to
its structural coherence (Hopper & Thompson 1980:281). To quote Cornilescus
(1981) terminology (see Chapter 1, Chapter4), ARCs and nominal appositives
draw a distinction between the hot news (the main predication) and the not
so hot news (the ARC with the exception of continuative ARCs or the sec-
ond unit of an appositive structure). Other syntactic structures, however, are not
192 Appositive Relative Clauses in English

meant to establish such a hierarchy, in particular independent clauses, whether

coordinated or juxtaposed (see examples (1) and (2) respectively, which are fol-
lowed by a reformulation with ARCs in (1a) and (2a)). Although we have already
mentioned that foregrounded information and independent clauses do not sys-
tematically go hand in hand (see Thompsons (1987) work on narrative subordi-
nation, for instance, or the definition of continuative ARCs in this book), a certain
affinity does exist between independent clauses and foregrounded information.
Just as syntactic embedding represents a powerful means to background infor-
mation (Reinhart 1984:796), main clauses are a powerful means to foreground
information. Naturally, we show below that the use of independent clauses as sen-
tential parentheticals (Section 3.1.2) and/or with specific prosodic characteristics
(Section 3.1.3) tempers such a claim.
The affinity between independent clauses and ARCs can be explained by the
fact that ARCs, from a semantic, informational, but also syntactic point of view,
have characteristics that are reminiscent of independent clauses: e.g. independent
speech act, autonomy of a second predication integrated into the main predi-
cation, syntactic characteristics that are generally typical of independent clauses
and which has led some researchers to adopt a radical orphanage approach (see
Chapter 1). Continuative ARCs in particular, which correspond to a case of nar-
rative subordination as defined by Thompson (1987) (see Chapter4), are quite
similar to independent clauses, as (3) vs. (3a) show.

(1) (QUAL_alloARC26) This is practically a taboo area for government and policy
makers because the economy is based on getting people to consume more, but
that simply cannot go on.
a. This is practically a taboo area for government and policy makers because
the economy is based on getting people to consume more, which simply
cannot go on.
(2) (QUAL_alloARC38) For my starter I chose the milk-fed veal, spider crab puree
and caviar, which sells for 70. This was very good, but I dont know if it was
actually 10 times as good as a normal starter.
a. For my starter I chose the milk-fed veal, spider crab puree and caviar, which
sells for 70, which was very good, but I dont know if it was actually 10
times as good as a normal starter.
(3) And the students also keep diaries of their clinical experiences which must
include ethical issues which are then brought back into the college and discussed
<,> in seminar groups again and in in big uh plenary sessions. (ICE-GB)
a. And the students also keep diaries of their clinical experiences which must
include ethical issues. These are then brought back into the college and
discussed <,> in seminar groups again and in in big uh plenary sessions.
 Chapter 6. Appositive relative clauses 193

In addition, as a predication that takes information into the background of the

sentence (as defined in Chapter 4), away from the main predication that repre-
sents the topic of the discourse, ARCs (with the exception of continuative ARCs,
see Section 3.1.6) naturally share some affinity with sentential parentheticals, as
in the following example:

(4) (FIC_alloARC1) Because of her original awe of Connor he was very smart, he
knew a lot about ancient bones, about foreign travel, about how to mix drinks
she did not drive nearly as hard a bargain with him as she could have.
a. Because of her original awe of Connor who was very smart and knew a
lot about ancient bones, about foreign travel, about how to mix drinks she
did not drive nearly as hard a bargain with him as she could have.

Also, because of the possible existence of an inter-clausal semantic link to be in-

ferentially reconstructed by addressees, ARCs sometimes fulfill discourse func-
tions that are closely related to adverbials:

(5) (QUAL_alloARC49) [A]ll British-donated blood has been stripped of its

white cells, because they are thought most likely to carry the infectious prion
particles.
a. [A]ll British-donated blood has been stripped of its white cells, which are
thought most likely to carry the infectious prion particles.

Finally, ARCs also compete with other forms of noun modifiers, such as premodi-
fiers, which, like nominal appositives, bring extra information in an asyndetic
way, but this time as a modifier appearing before the noun it modifies ((6)(7)),
while an ARC and an apposed unit are necessarily post-nominal (with the excep-
tion of fronted appositions). A closely-related strategy, which consequently also
competes with ARCs in discourse, is the use of co-referential NPs, as in (6b) and
(7b), adapted from (6) and (7), a larger context being required.

(6) (QUAL_alloARC40) No. 1 long-distance carrier AT&T on Thursday will

announce plans to sell Internet phone service to consumers and to more busi-
nesses in 2004, say people familiar with the matter.
a. AT&T, which is the no. 1 long-distance carrier, on Thursday will announce
plans to sell Internet phone service to consumers and to more businesses in
2004, say people familiar with the matter.
b. AT&Ti will announce today plans to sell Internet phone service to consumers
and to more businesses in 2004, say people familiar with the matter. David
Dorman, the CEO of the No. 1 long-distance carrieri, speaking at an
investor conference, will detail the opening phase in what would be the first
national rollout of voice over the Internet calling (VoIP). It will be the most
194 Appositive Relative Clauses in English

concrete indication yet that the promised cheaper alternative to regular

phone networks is ready for prime time.
(7) (QUAL_alloARC41) Former president of the United States Bill Clinton will
attend an international seminar on AIDS and SARS (severe acute respiratory
syndrome) and deliver a lecture on global AIDS prevention and control efforts,
a seminar official said Friday.
a. Bill Clinton, who is the former president of the United States, will attend
an international seminar on AIDS and SARS (severe acute respiratory
syndrome) and deliver a lecture on global AIDS prevention and control
efforts, a seminar official said Friday.
b. Bill Clintoni will attend an international seminar on AIDS and SARS (severe
acute respiratory syndrome) and deliver a lecture on global AIDS prevention
and control efforts, a seminar official said Friday. Hu Dongcheng, chairman
of the organizing committee of the Qinghua University International
Seminar on AIDS and SARS, told a press conference that the former
president of the United Statesi would lecture at the seminar which will open
on Nov. 10.

2. Syntactic phenomena

As for apposition, it is sometimes impossible for a speaker to use an ARC because

of the syntactic organization of the sentence, in which case the constraints that
govern speakers choices have nothing to do with the semantics and/or pragmat-
ics of the utterance within its context. The speaker then will use another structure
that fits in the syntactic organization of the discourse, even though the resulting
utterance might not be the best stylistically:
(8) (TABL_alloARC1) [Prince] Charles was amazed and shocked by Bollands
comments yesterday to a Sunday newspaper. He is the princes former deputy
private secretary and press adviser.

While supplementary information is needed about the referent of Bolland (dis-

course new in the article and possibly hearer new for some of the addresses) in
order that the relevance of the discourse as a whole be optimized for all the read-
ers which in itself is a motivation for the use of a relevance ARC the writer uses
another syntactic structure, in this instance an independent juxtaposed clause.
An ARC cannot be inserted because of the use of the genitive construction:

. In the newspaper article from which this sequence has been extracted, devoted to some
comments made by Bolland about Princess Diana, the use of a periphrastic genitive (the com-
ments of Bolland) is not felicitous.
 Chapter 6. Appositive relative clauses 195

(8) a. *Charles was amazed and shocked by Bolland, who is the princes
former deputy private secretary and press advisers,s comments yesterday
to a Sunday newspaper.
b. *Charles was amazed and shocked by Bollands, who is the princes
former deputy private secretary and press advisers, comments yesterday to
a Sunday newspaper.

As was discussed in the previous chapter concerning [U2 NP+RC] structures,

difficulty in retrieving the antecedent might be another explanation for speakers
choice of packaging. This is not a syntactic constraint per se, as the problem is
linked to the identification of a correct pronoun for an antecedent, but the ante-
cedent-retrieval difficulty is linked to the syntactic organization of the sentence,
which is why we consider this type of constraint a syntactic constraint. For in-
stance, in the following example, the use of an ARC instead of an independent
clause (beyond other problems linked to the hierarchization of informational
contents, see Section 3.1 below.) suppresses any referential problems:

(9) (QUAL_ARC67) The Commission had to withdraw the Branson offer after
Camelot successfully challenged their decision in the High Court, which said
that the Commission had been conspicuously unfair.
a. ?The Commission had to withdraw the Branson offer after Camelot
successfully challenged their decision in the High Court. It said that the
Commission had been conspicuously unfair.

3. Semantic and pragmatic phenomena

This section focuses on constraints that are linked with discourse organization
in relation to semantic and pragmatic factors such as information status, thus
identifying the specific roles an ARC plays in opposition to its competing allos-
tructures. Such constraints are at the core of our study on the discourse functions
of ARCs and determine what can account for a speakers choice of this structure
instead of the others when no syntactic, grammatical or stylistic constraint seems
to account for it.

. Some linguists, like Doug Arnold, do not agree with the ungrammaticality of such exam-
ples (see Arnold 2004 for instance, where he gives the following example: Prince Alphonso, who
ruined the partys, mother left early). In our different corpora of written and spoken English, we
have not encountered such examples. The very negative reactions of native speakers cast serious
doubt on their grammaticality.
196 Appositive Relative Clauses in English

3.1 The hierarchization of information

3.1.1 Hot news vs. not so hot news information

One of the main roles of relevance and subjectivity ARCs (continuatives need to
be set apart as far as the hierarchization of information contents is concerned;
see Section 3.1.6) is to place the conveyed informational contents into the back-
ground, away from the hot news, i.e. the information in the foreground (main
clause). It is therefore not surprising that relevance and subjectivity ARCs, not
only in middle but also final positions, cannot be replaced with an independent
clause, whether juxtaposed or coordinated. Consider on the one hand (10), (11)
and (12), the original attested examples, and (10a/b), (11a/b) and (12a/b) adapted
from them:

(10) (QUAL_ARC184) Olivier, who died in 1989 at the age of 82, co-founded the
National theatre and was famous the world over both as a stage performer and
a film star.
a. #Olivier died in 1989 at the age of 82. He co-founded the National theatre
and was famous the world over both as a stage performer and a film star.
b. #Olivier died in 1989 at the age of 82 and co-founded the National theatre
and was famous the world over both as a stage performer and a film star.
(11) (TABL_ARC343) The singer, who believes she is a rock diva trapped in the body
of a pop star, launched into her new single.
a. #The singer believes she is a rock diva trapped in the body of a pop star and
launched into her new single.
b. #The singer believes she is a rock diva trapped in the body of a pop star. She
launched into her new single.
(12) (QUAL_ARC39) The Governor is Bob Taft, fourth generation of a Republican
dynasty founded by his great grandfather, William Howard Taft, who was
elected president in 1908.
a. #The Governor is Bob Taft, fourth generation of a Republican dynasty
founded by his great grandfather, William Howard Tafti, and hei was elected
president in 1908.
b. #The Governor is Bob Taft, fourth generation of a Republican dynasty
founded by his great grandfather, William Howard Tafti. Hei was elected
president in 1908.

The information originally conveyed by the ARCs does not belong to the hot
news but represents background information. In (10), the information conveyed
by the ARC is information that is not part of the main message but aims to level

. Reversing the information contents has no effect on the pragmatic felicity of the examples.
 Chapter 6. Appositive relative clauses 197

the shared cognitive space, as the information related to the death of actor Lau-
rence Olivier might not be known by all the intended addressees. The function of
the ARC in the article from which the sentence has been extracted is clearly to po-
sition Olivier in time: the topic of the article is the acquisition of his archived let-
ters, or papers as the journalist puts it, by the British Library, papers which reveal
the love affair he had with Jack Grimston Fairbanks. The information in the ARC
is not directly linked to the topic of the article but complements the information
by providing readers with the date of Oliviers death as well as his age when it
happened. Conveying this relevance-oriented information through the use of an
independent clause would somehow promote it to the foreground of the sentence
(foreground promotion), and of the discourse as a whole, while the article actu-
ally deals with Oliviers letters. Similarly in (11), already discussed in the previ-
ous part to illustrate the subjectivity discourse function of ARCs, the content of
the ARC corresponds to a subjective, sarcastic comment from the speakers/jour-
nalists part. Such an aside does not belong to the informational foreground and
the use of an independent clause would somehow promote it to the informational
foreground, as is the case for the informational content conveyed by the relevance
ARC in (10). Such information needs to be conveyed by a syntactic structure that
provides the instruction (to use Vallduvs terminology) that it represents an aside
that is in disjunction with the informational foreground and ARCs do fulfill that
very discourse function (among other discourse functions). Finally, example (12),
also discussed in the previous part to illustrate the category of relevance ARCs,
cannot be reformulated with an independent clause. The topic of the article is
not William Taft but Bob Taft, and the aim of the insertion of the informational
content William H. Taft was elected president in 1908 only aims at specifying that
Bob Taft (the topic of the article) has famous ancestors in American politics. Here
again, the use of an independent clause seems to be promoting this information
to the informational foreground, which is infelicitous because of the disjunction
with the topic of the discourse (see definition of relevance ARCs).
To conclude on these examples, the information conveyed by the relevance/
subjectivity ARCs in (10)(12), represents background information, either to
optimize the relevance of the antecedent and/or the discourse as a whole or to
provide the addressee(s) with a subjective aside. This explains why using an in-
dependent clause, which brings the information to the foreground (foreground
promotion), is infelicitous.

. This must be tempered, as independent clauses do not systematically convey foregrounded

information (see above and below).
198 Appositive Relative Clauses in English

However, the use of independent clauses as sentential parentheticals and/

or with specific prosodic characteristics represents an exception to this general
claim. Such configurations are dealt with in the next two subsections.

3.1.2 Sentential parentheticals

Parenthetical constructions represent a syntactic configuration where the use of
an independent clause instead of a relevance or subjectivity ARC is felicitous. The
independent clause then appears as a parenthetical clause, introduced in writing
by dashes, brackets or parentheses. In (13) and (14), we provide two such ex-
amples of independent clauses, introduced by and for (13), used as parentheticals,
alongside (13a) and (14a), which provide felicitous reformulations of the original
examples with ARCs, with dashes or commas. We then provide examples with
the reverse manipulation, i.e. the ARCs in examples (10)(11) which can easily
be turned into independent clauses used parenthetically. Note that this remains
infelicitous for (12), however; we address why in Section 3.1.3:

(13) (FIC_alloARC1) Because of her original awe of Connor he was very smart, he
knew a lot about ancient bones, about foreign travel, about how to mix drinks
she did not drive nearly as hard a bargain with him as she could have.
a. Because of her original awe of Connor /, who was very smart, (who)he
knew a lot about ancient bones, about foreign travel, about how to mix
drinks /, she did not drive nearly as hard a bargain with him as she could
have.
(14) (FIC_alloARC50) If youre right and Im not saying you are these are deeper
waters than you should be swimming in alone.
a. If youre right /, which Im not saying you are /, these are deeper waters
than you should be swimming in alone.
(10) c. Olivier he died in 1989 at the age of 82 co-founded the National theatre
and was famous the world over both as a stage performer and a film star.
(11) c. The singer she believes she is a rock diva trapped in the body of a pop
star launched into her new single.
(12) c. ??The Governor is Bob Taft, fourth generation of a Republican dynasty
founded by his great grandfather, William Howard Taft he was elected
president in 1908.

The difference between (13), (14), (10c) and (11c) on the one hand and (13a), (14a),
(10) and (11) on the other, seems related to style. It is generally assumed that
parentheticals do not belong to the syntax of the sentence (see Deh & Kavalova
2007 for a collection of studies on parenthetical constructions). Parentheti-
cals thus correspond to what Biber (1988) defines as the fragmented aspect of
 Chapter 6. Appositive relative clauses 199

discourse, typical of spontaneous, unplanned speech but also other production

modes, including written discourse when unplanned (e.g. writing an informal
letter). On the other hand, the examples with ARCs, which show some syntactic
integration with the rest of the sentence (the relative pronoun agrees with its
antecedent in the main clause, for instance), are typical of more formal, planned
registers. They thus correspond to what Biber (1988) refers to as the integrated
aspect of discourse. This is why sentential parentheticals are quite rare in formal
registers; examples typically belong to informal writing or spontaneous speech,
the parenthetical construction then interrupting the sentence and showing par-
ticular prosodic characteristics. The examples provided in (13) and (14) have
been extracted from the FIC subcorpus, which does not consist in spoken data,
but in Chapter 3 dedicated to the study of our spoken corpus, we have shown,
adopting a multi-register approach such as Bibers, that the FIC written and the
oral subcorpora show a certain affinity as far as the characteristics of ARCs are
concerned. It is not completely surprising, then, to find parentheticals in writ-
ten fiction, all the more so as the affinity between fiction writing and speech it-
self as communication modes can be explained by the speakers communicative
goals: for both modes, the speaker generally has a precise idea of who his/her
addresse(s) is/are, and therefore a good representation of what the addresse(s)
know(s) or do(es) not know (see Chapter 3, Section 2.4).

3.1.3 The crucial role of intonation

In addition to the parenthetical use of independent clauses, another configuration
allows the use of independent clauses instead of appositive relative clauses in spo-
ken English. To signal the backgrounded character of the information, speakers
can use a specific prosodic coding. This is true for sentential parentheticals that
interrupt the flow of discourse (see previous subsection), but also for juxtaposed
independent clauses. This indubitably shows that independent clauses and fore-
grounded information cannot systematically go hand in hand.

. As mentioned in Chapter 3, ARCs are three times less frequent in speech than in writing.
. Integration is defined in Bibers work as the way in which a large amount of information is
packed into relatively few words in typical writing, because the writer operates under few time con-
straints and can therefore construct a carefully packaged text. Similarly, the reader, who can read
as quickly as (s)he pleases, is able to take advantage of a highly integrated text (Biber 1988:43).
On the other hand, fragmentation is defined as the linguistic characteristics of texts produced
under severe time constraints, the case for typical speech. Under these conditions, information
cannot be carefully incorporated into the text, and the resulting structure is much looser, or frag-
mented (Biber 1988:43).
200 Appositive Relative Clauses in English

As is known by any linguist working on spoken data, the role of intonation

is crucial as far as the organization of discourse is concerned. Like a syntactic
structure, the intonation contour of a clause can serve an information-packaging
function (Vallduv 1993:3336). A particular intonation pattern can signal the
addressee(s) that the information does not belong to the foreground but repre-
sents either a subjective aside or relevance-oriented supplementary information
(to use Huddleston & Pullums 2002 term). Intonation then provides addressees
with the same instruction as is provided by an ARC: put the information into the
background, away from the foreground. ARCs and independent clauses are then
in discourse competition. Naturally, this kind of competition is only valid for the
spoken production mode.
Consider the following pair, for which the reformulation with an indepen-
dent clause seems infelicitous in writing but is totally acceptable in speech with
a specific intonation contour on the independent clause (which explains why we
use the label % and not #):

(15) (QUAL_ARC170) Killing one Siamese twin to try to save the other might
amount to unlawful killing, a Court of Appeal judge said last week. The court is
ruling on whether the sisters known as Jodie and Mary should be separated.
They were born at St Marys Hospital, Manchester, in August, joined at the lower
abdomen and sharing one working heart and a set of lungs. Their parents are
appealing against a High Court ruling that an operation to separate them should
go ahead, even though it would lead to the death of Mary, who relies on her sister
for blood and oxygen.
a. %Their parents are appealing against a High Court ruling that an operation
to separate them should go ahead, even though it would lead to the death
of Mary. She relies on her sister for blood and oxygen.

In writing, especially in formal genres like the quality press, such sequences tend
to be avoided. However, in other, less formal, genres, such sequences are possible
(e.g. BBC News or Yahoo websites, where the articles are written in quite a frag-
mented way, closer to the speech production mode, as opposed to The Times, The
Guardian, or The New York Times, even in their electronic versions).
It is also important to note that such a possibility is restricted to cases where
there exists a causal link between the independent clause and its preceding con-
text: (12b) for instance, repeated below, cannot be realized with an intonation
contour that would make it pragmatically acceptable.

(12) b. #The Governor is Bob Taft, fourth generation of a Republican dynasty

founded by his great grandfather, William Howard Taft. He was elected
president in 1908.
 Chapter 6. Appositive relative clauses 201

3.1.4 The insertion of irrelevant information

The use of an ARC instead of an independent clause can also be explained by the
speakers intention to insert information that seems irrelevant if we consider the
topic of the discourse only, but is actually relevant if we consider the addressees
informational needs as evaluated by the speaker. Consider the following example,
already used in the previous part (Chapter 4, Section 3.5):

(16) Nicolas Sarkozy arrived yesterday with his wife Cecilia, whom he was holding
by the hand, in the seaside resort of Heiligendamm, Germany.

The information conveyed by the ARC consists in spicy information that the ad-
dressees are judged by the speaker to be eager to know, although the article deals
with the G8 summit and the decisions that were taken during this summit (topic
of the discourse). The use of the ARC is then not to optimize the relevance of
discourse or provide a subjective comment as is the case when the informational
contents are hierarchized according to their representing hot news or not so
hot news. The information is irrelevant to the topic of the article (which is the
G8 summit) and cannot be conveyed by an independent clause, as such a syntac-
tic configuration would show the irrelevance of the information in too blatant a
way. The speaker then uses an ARC, which somehow takes the information into
the background, as the structure generally establishes a disjunction either for rel-
evance or subjectivity motives.
Because of this, ARCs can also be the perfect syntactic vehicle to insert adver-
tising or commercial messages:

(17) (QUAL_ARC41) The Queen herself is believed to have condoned publication

[of a book about the Princess of Wales by her former private secretary], saying
that the palace should check that it was all right. At no stage did they tell
Jephson that he should not go ahead with the book, which is to be serialised in
The Sunday Times, until the Prince of Wales intervened.

3.1.5 Temporariness vs. elaboration

As a means of backgrounding information, an ARC represents a digression and
is thus, by definition, temporary. This is obvious when the ARC is in middle posi-
tion, inserted within the syntactic organization of the sentence and therefore dif-
ficult to reformulate with another syntactic structure, but this is also true of ARCs
in final position. Therefore, one constraint that can explain the choice of an ARC
over one of its competing allostructures is the speakers intention to elaborate
or not on the element about which some extra information is provided. In the
article from which (12) has been extracted, for instance, William Howard Taft is
mentioned only once; the only thing the journalist wants to let readers know is
202 Appositive Relative Clauses in English

that he was elected president of the US in 1908. Nothing else will be said about
him in the rest of the article, hence the use of an ARC, the perfect vehicle for a
temporary digression.
Should the speaker want to elaborate further, then s/he needs to resort to
a structure that will promote it to the foreground, away from the background,
a structure such as an independent clause (not used parenthetically, however).
Consider (18):

(18) The boy went into the living room, where his father lay face down. Mr Hunt was
fully clothed, but with the plastic bag clinging to his face. He was still holding the
plug to the vacuum cleaner, which was still running. Alastair slapped his fathers
face to try to rouse him then opened the back door to let his mother into their
home in Swanwick, Derbys.

The information was still running is given as a temporary digression with an ARC
(here for plain precision in the description of the scene, the fact that the vacuum
cleaner was still running being judged relevant by the speaker in his/her descrip-
tion of the crime scene). No elaboration on the vacuum cleaner is possible, and an
independent clause is infelicitous for the reasons mentioned in Section 3.1.1.

(18) a. #The boy went into the living room, where his father lay face down. Mr
Hunt was fully clothed, but with the plastic bag clinging to his face. He was
1still holding the plug to the vacuum cleaner. It was still running/and it was
still running. Alastair slapped his fathers face to try to rouse him then
opened the back door to let his mother into their home in Swanwick,
Derbys.

However, if the speaker has more information to provide about the vacuum clean-
er, then the use of an independent clause, promoting it to discourse topic posi-
tion, becomes acceptable:

(18) b. The boy went into the living room, where his father lay face down. Mr Hunt
was fully clothed, but with the plastic bag clinging to his face. He was still
holding the plug to the vacuum cleaner. It was still running in spite of the
fact that the power cord was completely tight: the electrical socket had
started to loosen from the wall. Alastair slapped his fathers face to try
to rouse him then opened the back door to let his mother into their home
in Swanwick, Derbys.

3.1.6 Continuative ARCs and ARCs with sentential antecedents

From the very beginning of this section dedicated to the opposition between hot
news and not so hot news informational contents, we have set apart continuative
 Chapter 6. Appositive relative clauses 203

ARCs. In previous sections, we have shown that this specific type of ARCs dis-
plays independent characteristics and behaves, semantically but also syntactically,
in a way that is reminiscent of independent clauses. As a consequence, replacing
continuative ARCs with independent clauses, whether juxtaposed or coordinat-
ed, does not provide unacceptable results (such manipulations are to be found in
most, if not all, pedagogical grammars). This is not surprising, as the function of
depicting a movement through narrative time is in general specific to indepen-
dent clauses, not subordinate clauses. In this respect, we know that continuative
ARCs are exceptional (Depraetere 1995). The difference between an ARC and an
independent clause (juxtaposed or coordinated) is the explicitness of a chrono-
logical and/or causal link between the event in the MC and the event in the RC.
We repeat example (3) here and add another example, (19):

(3) And the students also keep diaries of their clinical experiences which must
include ethical issues which are then brought back into the college and discussed
<,> in seminar groups again and in in big uh plenary sessions. (ICE-GB)
a. And the students also keep diaries of their clinical experiences which must
include ethical issues. These are then brought back into the college and
discussed <,> in seminar groups again and in in big uh plenary sessions.
b. And the students also keep diaries of their clinical experiences which must
include ethical issues, and these are then brought back into the college and
discussed <,> in seminar groups again and in in big uh plenary sessions.
(19) (QUAL_ARC183) In the book the consul is portrayed as a rather sad man who
drinks too much and marries an Indian girl out of loneliness, who eventually
leaves him.
a. In the book the consul is portrayed as a rather sad man who drinks too
much and marries an Indian girl out of loneliness and she eventually
leaves him.
b. In the book the consul is portrayed as a rather sad man who drinks too
much and marries an Indian girl out of loneliness. She eventually leaves
him.

The manipulated examples (3a/b) and (19a/b) are semantically and pragmatically
acceptable. The only missing element is the narrative dynamism (one event lead-
ing to or triggering another), which sometimes includes causal relationship. The
fact that continuative ARCs behave differently from relevance and subjectivity
ARCs is in line with the fact that continuative ARCs, as an instance of narrative
subordination, convey information that is very different from that conveyed by
other ARCs. In addition to this fundamental semantic difference, we also have
shown that from a syntactic but also prosodic point of view, continuative ARCs
have characteristics that are very different from relevance and subjectivity ARCs.
204 Appositive Relative Clauses in English

Continuative ARCs therefore represent an exception to the constraints defined

in the previous subsections both on the distinction between informational fore-
ground (hot news) and background (not so hot news), and on the temporary
dimension of the ARCs informational content, as (19c) illustrates:

(19) c. In the book the consul is portrayed as a rather sad man who drinks too
much and marries an Indian girl out of loneliness, who eventually leaves
him. One morning, she simply decides to pack all her things and flies to
Mumbai.

In a similar way, ARCs with sentential, non-nominal antecedents stand apart.

Such ARCs, traditionally in final position, can also be easily reformulated with
independent clauses ((20)(21)) and can even represent independent sentences
(22), which can even more easily be rephrased with ARCs starting with this/that
or it depending on the type of sentential antecedent:

(20) (FIC_SR205) Macaulay Culkin probably got three hundred and fifty pounds a
week, maybe even more, which meant that if he were Macaulay Culkin he could
pay his mum to teach him.
a. Macaulay Culkin probably got three hundred and fifty pounds a week,
maybe even more. This meant that if he were Macaulay Culkin he could
pay his mum to teach him.
(21) (SPEC_SR306) Too much water dilutes the bodys sodium, which is critical for
the functioning of nerve cells.
a. Too much water dilutes the bodys sodium. This is critical for the functioning
of nerve cells.
(22) (FIC_SR206-7-8) But if being Macaulay Culkin meant being good at drama,
then forget it: he was crap at drama, because he hated standing up in front
of people. Which was why he hated school. Which was why he wanted to be
Macaulay Culkin. Which was why he was never going to be Macaulay Culkin
in a thousand years, let alone in the next few days. He was going to have to go
to school tomorrow.
a. But if being Macaulay Culkin meant being good at drama, then forget it:
he was crap at drama, because he hated standing up in front of people. This
was why he hated school. This was why he wanted to be Macaulay Culkin.
This was why he was never going to be Macaulay Culkin in a thousand
years, let alone in the next few days. He was going to have to go to school
tomorrow.

The fact that such ARCs, because of a looser link with the main predication, seem
to have semantic affinities with independent clauses in the same way as continua-
tive ARCs explains why grammars have frequently used the label continuative to
 Chapter 6. Appositive relative clauses 205

refer to both types of ARCs. However, as we explained in Chapter 1 of this book,

this would result in definitions which mix discourse and syntax phenomena.
Like continuative ARCs, ARCs with sentential antecedents represent an ex-
ception to the hierarchization and elaboration phenomena: it is perfectly pos-
sible to elaborate on the informational contents conveyed by those ARCs. This is
shown by the sentence following the three stacked independent ARCs in (22), He
was going to have to go to school tomorrow, which comes as a conclusion to the
succession of arguments provided in the three ARCs, as well as by the existence
of example (23) below, where the speaker does elaborate on the ARCs content in
the right co-text without any acceptability problems. Naturally, the ARC can be
rephrased with an independent juxtaposed clause:

(23) (QUAL_SR6) Adrian Cole is senior addiction therapist at the Priory Clinic,
Bristol, and regularly treats work addicts. Addiction is a pathological relation-
ship with an object or event, which means it can happen in any job, he says.
Although we see more men than women, its not related to occupation or how
high up the ladder you are we see the full range, from managing directors to
technicians.
a. Addiction is a pathological relationship with an object or event. It means it
can happen in any job, he says. Although we see more men than women,
its not related to occupation or how high up the ladder you are we see the
full range, from managing directors to technicians.

This is also shown by the possibility of rephrasing an independent juxtaposed

clause; in this case, the speaker elaborates on the content in the following sentence
(examples (24) and (25)). The ARC does not damage the acceptability of the dis-
course as a whole ((24a)(25a)):

(24) (QUAL_alloARC49) Heads are unrepentant. If schools didnt sell the stuff, the
kids would just go down to the local shop and get their sugar fix there. This offers
some interesting possibilities. Schools with a drug problem could invite the deal-
ers in, take a rake-off, and invest the money in a just say no programme. Fights
in the playground could be dealt with by rearranging the bout in the school gym
at 4pm. Or heads could decide that some children have no chance of a decent
grade at GCSE and assign their worst teachers to those groups
a. Heads are unrepentant. If schools didnt sell the stuff, the kids would just
go down to the local shop and get their sugar fix there, which offers some
interesting possibilities. Schools with a drug problem could invite the dealers
in, take a rake-off, and invest the money in a just say no programme. Fights
in the playground could be dealt with by rearranging the bout in the school
gym at 4pm. Or heads could decide that some children have no chance of a
decent grade at GCSE and assign their worst teachers to those groups
206 Appositive Relative Clauses in English

(25) (FIC_alloARC2) Knitting is something she has just taken up again, having
learned it as a child from a mother who believed in the female domestic virtues.
She was also taught to crochet, to set in zippers, to polish silverware, to produce
a gleaming toilet. This was baggage shed discarded as soon as she hit Spinoza;
two years, a year ago, she would have despised knitting. But there is not a lot to
do in this town when Connor is not here
a. Knitting is something she has just taken up again, having learned it as a
child from a mother who believed in the female domestic virtues. She was
also taught to crochet, to set in zippers, to polish silverware, to produce
a gleaming toilet, which was baggage shed discarded as soon as she hit
Spinoza; two years, a year ago, she would have despised knitting. But there
is not a lot to do in this town when Connor is not here.

Using ARCs as independent clauses is yet another option, although this is con-
demned by prescriptive usage guides. Here the use of colons seems necessary,
however:

(24) b. Heads are unrepentant. If schools didnt sell the stuff, the kids would just go
down to the local shop and get their sugar fix there. Which offers some inter-
esting possibilities: (i) schools with a drug problem could invite the dealers
in, take a rake-off, and invest the money in a just say no programme, (ii)
fights in the playground could be dealt with by rearranging the bout in the
school gym at 4pm, or (iii) heads could decide that some children have no
chance of a decent grade at GCSE and assign their worst teachers to those
groups
(25) b. Knitting is something she has just taken up again, having learned it as a
child from a mother who believed in the female domestic virtues. She was
also taught to crochet, to set in zippers, to polish silverware, to produce
a gleaming toilet. Which was baggage shed discarded as soon as she hit
Spinoza:two years, a year ago, she would have despised knitting. But there
is not a lot to do in this town when Connor is not here.

These two types of exception to the constraints related to the hierarchization of

the informational contents and on the speakers (non-) intention to elaborate fur-
ther on the content s/he is about to express leave room for stylistic variation only,
which we discuss in Section 4.

3.2 Thematic progression: ARCs vs. preposing

Another allostructure competing with ARCs is the independent clause with what
is traditionally called preposing, be it focus preposing or topicalization (Birner &
 Chapter 6. Appositive relative clauses 207

Ward 1998). In such cases the word order is the same as for an ARC, non-canoni-
cal with regards to the canonical SVO word order: wh- movement for the relative
clause, NP-movement for the preposed/topicalized element, both to the front of
the clause. Birner & Ward (1998) have shown that the requirements for a topical-
ization or a focus preposing to be felicitous are (i) the existence of a discourse-old
link between the preposed element and the previous discourse; (ii) the existence
of a salient or inferable presupposed open proposition (OP) as defined in Prince
(1986), for instance. A presupposed OP is a proposition which contains a vari-
able (hence the term open) that represents the new information of the utterance,
while the rest of the proposition is presupposed. For instance, in a context where
we know that Judith gave an identified/specific shirt to someone, the proposition
Judith gave the shirt to X is an open proposition, in which X is a variable that rep-
resents a specific beneficiary in opposition to other variables that could be the re-
cipients of the shirt. The fact that Judith gave the shirt to someone is presupposed,
while the identity of that someone is the new information. The existence of such
an open proposition is then the trigger for the use of specific syntactic structures
like a cleft structure (It is to Harry that Judith gave the shirt) or a topicalization (To
Harry Judith gave the shirt), or for specific prosodic characteristics (She gave the
shirt to HARRY).
With an ARC, the first constraint is obviously fulfilled the relative pro-
noun stands for an element in the immediate prior discourse, i.e. the antecedent,
de facto creating a discourse-old link. However, the second requirement (exis-
tence of a presupposed OP) is not fulfilled. Such a difference seems to explain
the acceptability of (26) and (27) while (26a) and (27a) are infelicitous within the
same context:
(26) The Kerrys had invited so many people that it was impossible to meet everybody
during the party. They had invited their family, friends and colleagues from
work. One of the first persons to arrive was Bob. Him I managed to talk to.
a. #The Kerrys had invited so many people that it was impossible to meet
everybody during the party. They had invited their family, friends and
colleagues from work. One of the first persons to arrive was Bob, whom I
managed to talk to.
(27) I entered the shop and saw the salesman, to whom I asked where I could find
screwdrivers.
a. #I entered the shop and saw the salesman. To him I asked where I could find
screwdrivers.

In these two examples and their reformulations, the existence of a discourse-old

link is obvious. However, what differentiates the two sequences is that in (26), a
presupposed OP exists (I did not manage to talk to X that night, where X is the
208 Appositive Relative Clauses in English

variable), whereas such a presupposed OP does not exist in example (27). As a

consequence, preposing is felicitous in (26) while the use of an ARC is not (26a);
for (27) the exact opposite is true.
ARCs, then, seem to obey the same constraints as independent clauses with
standard word order regarding that matter. Consider (28) below, which is taken
from our fiction corpus: the use of an ARC in (28a) to replace the underlined
independent clause in (28) is felicitous; on the other hand, the use of an inde-
pendent clause with preposing in (28b) is infelicitous. This is due to the fact that
the context presupposes no open proposition I reply to X, where this would be in
opposition with another variable as defined above:

(28) Rupa and Piu are eager to see the Aquarium and the Science Museum, my
father said. Perhaps you can take them one day, Kaushik.
I didnt reply to this. Very tasty, I said in Bengali, referring to the food,
something my mother had taught me to say after eating in the homes of other
people.
a. Rupa and Piu are eager to see the Aquarium and the Science Museum, my
father said. Perhaps you can take them one day, Kaushik.
To which I didnt reply. Very tasty, I said in Bengali, referring to the food,
something my mother had taught me to say after eating in the homes of
other people.
b. #Rupa and Piu are eager to see the Aquarium and the Science Museum,
my father said. Perhaps you can take them one day, Kaushik.
To this I didnt reply. Very tasty, I said in Bengali, referring to the food,
something my mother had taught me to say after eating in the homes of
other people.

The conclusion to this subsection is thus that the choice between an ARC and an
independent sentence with a preposed element as competing allostructures to
convey the same informational content is governed by the (in)existence of a pre-
supposed OP: a sentence with a preposed element requires the existence of such
an OP, while an ARC does not.

3.3 The explicitness/implicitness of the link between

the information contents

ARCs can also be in discursive competition with adverbial clauses introduced by

a subordinator that indicates cause, consequence, or concession. ARCs can also
introduce such semantic relationships with the main clause (MC), but this link
is implicit and has to be inferred by the addressees, based on their encyclopedic
 Chapter 6. Appositive relative clauses 209

knowledge and the context/co-text. Consider (29)(32), already used in Chap-

ter1 of this book:
(29) Salim, who passed all his final exams, is very happy.
(30) Salim, who failed all his final exams, is very disappointed.
(31) Salim, who passed all his final exams, is very disappointed.
(32) Salim, who failed all his final exams, is very happy.

In the world as we know it, where success is a source of happiness and failure a
source of disappointment, most addresses would infer the existence of a causal
semantic link between the ARC and the MC (29)(30), and a concession link in
(31)(32). The speaker could make these links explicit:

(29) a. Salim, since/as he passed all his final exams, is very happy.
(30) a. Salim, since/as he failed all his final exams, is very disappointed.
(31) a. Salim, although he passed all his final exams, is very disappointed.
(32) a. Salim, although he failed all his final exams, is very happy.

This might seem anecdotal and purely stylistic. However, using an ARC and thus
making the semantic link between two clauses implicit through the use of a rela-
tive pronoun (which in itself establishes no specific semantic link), enables speak-
ers to imply meanings without really saying them, as in the following example
(see Chapter 4, Section 4.1.1. on implicit subjectivity, as well as the definition
provided in the diagram at the end of Part II, exploitation of the inter-clausal
link, Section 7 of the same chapter):
(33) Mr Blair, whose son attends a selective school, complained that too many com-
prehensives had a one-size fits-all mentality...

As we explained in the definition of subjectivity in Chapter 4, in the article from

which this sentence is extracted, the information conveyed by the ARC is clearly
a means for the journalist to question the legitimacy of Tony Blairs complaint,
since he is not personally concerned with the situation and therefore does not
really know what he is talking about. The semantic link between the ARC and
the MC should clearly be understood as a concessive link, but this is left for the
addressees to infer. What looks like pure supplementary information about the
ex-Prime Minister is actually used as a means to convey an opinion (hence the
subjective character of apparently relevance-oriented information) about the con-
tents of his statement.
210 Appositive Relative Clauses in English

The following three examples, which are extracted from the International Ac-
tion Center press release following Bushs press conference in May 2002 about
US-Cuba relations, also provide an interesting use of such a strategy:

(34) George Bush, who spent $1 billion on his election campaign, calls for free and
fair elections ... in Florida? No, Cuba.
(35) George Bush, who presided over the execution of more people than any other
person in the U.S.[,] calls for rights for political prisoners ... in Texas? No,
Cuba.
(36) George Bush, who lost the 2000 election by 500,000 votes and had ultra-con-
servative Supreme Court justices step in to deny a re-count, calls for fair elections
in 2003 ... in Florida? No, Cuba.

The irony is obvious, and is particularly conveyed by the speakers comparison

between elections that were to be held in Cuba in 2003 and the 2000 American
presidential elections, particularly the endless Florida ballot recounts. In (34)
(36), the speaker uses three relevance ARCs, providing information about Bushs
life before his presidency and during the 2000 presidential campaign. These ARCs
aim at leveling the shared cognitive space: their informational status is indeter-
minable, as defined in our extension of Princes taxonomy for given/new infor-
mation (for some, even many addressees, the information is hearer old, while for
others, it is either hearer new or needs to be reactivated). This indeterminable
informational status represents a natural trigger for the use of relevance ARCs. In
our examples, the three ARCs are to be understood in opposition to the contents
of the main clauses in which they have been embedded and could be glossed as
(34a)(36a), provided below. The inter-clausal semantic links are to be recon-
structed inferentially and add to the incongruity of Bushs comments or opinions
conveyed in the MCs. Making the links explicit through the use of although for
instance, transforming the ARCs into adverbial clauses, is possible, although the
tone seems to lose most of the irony in the process. The explicitness engenders a
decrease in the subtlety, and thus the irony, of the comments.

(34) a. George Bush, although he spent $1 billion on his election campaign, calls
for free and fair elections ... in Florida? No, Cuba.
(35) a. George Bush, although he presided over the execution of more people than
any other person in the U.S.[,] calls for rights for political prisoners ... in
Texas? No, Cuba.
(36) a. George Bush, although he lost the 2000 election by 500,000 votes and had
ultra-conservative Supreme Court justices step in to deny a re-count, calls
for fair elections in 2003 ... in Florida? No, Cuba.
 Chapter 6. Appositive relative clauses 211

ARCs can then be in competition with adverbials introduced by subordinators

(such as although, because, since, while, or whereas) but also with coordinated
independent clauses introduced by but, for instance, the use of an ARC instead of
these structures enabling the speaker to put two propositions into perspective in
an implicit way, as examples (37) and its manipulation in (37a) show:

(37) (QUAL_alloARC26) This is practically a taboo area for government and policy
makers because the economy is based on getting people to consume more, but
that simply cannot go on. The social and environmental impact of over-con-
sumption will overwhelm the benefit.
a. This is practically a taboo area for government and policy makers because
the economy is based on getting people to consume more, which simply can-
not go on. The social and environmental impact of over-consumption will
overwhelm the benefit.

4. Other influencing parameters

4.1 Register variation

As hinted at in Section 3.1.2, dedicated to sentential parentheticals as alternatives

to ARCs in discourse, register necessarily plays a role. Not all choices between
ARCs and their competitive allostructures can be explained by the constraints
linked to discourse and pragmatic considerations, or even syntactic constraints.
In particular, we already mentioned that a spontaneous, informal register could
explain the choice of a sentential parenthetical instead of an ARC. This is due
to the fact that ARCs belong to more formal registers, characterized by integra-
tion as opposed to fragmentation, to quote Bibers (1988) terminology for his
multi-dimensional approach of the different registers in English. Examples (13)
and (13a), repeated below, illustrate the influence of register on speakers choices.
Both utterances are acceptable, (13a) seeming more formal than (13).

(13) (FIC_alloARC1) Because of her original awe of Connor he was very smart, he
knew a lot about ancient bones, about foreign travel, about how to mix drinks
she did not drive nearly as hard a bargain with him as she could have.
a. Because of her original awe of Connor /, who was very smart, (who)he
knew a lot about ancient bones, about foreign travel, about how to mix
drinks /, she did not drive nearly as hard a bargain with him as she could
have.

The following examples, extracted from the spoken section of our corpus of al-
lostructures (spontaneous English), illustrate the same phenomenon:
212 Appositive Relative Clauses in English

(38) (RAD05_alloARC1) So we borrowed a car | and it was an electric 225 | I dont

know if such a car exists today.
(39) (TEL03_alloARC2) Weve seen a couple of them in practice and they do tend to
involve the Procrustean bed | you know... everybody who doesnt fit gets a part
dropped off
(40) (RAD5_alloARC2) Then I lived in Washington DC for five years | and parking
in Washington DC is just impossible

In written, more formal English, these allostructures a coordinated indepen-

dent clause in (38) and (40), and a juxtaposed independent clause introduced
by you know, signaling the relevant character of the informational content, in
(39) would very likely be rephrased with ARCs, which require planning because
of the complexity of the structure:

(38) a. So we borrowed a car, which was an electric 225, I dont know if such a car
exists today.
(39) a. Weve seen a couple of them in practice and they do tend to involve the
Procrustean bed, where everybody who doesnt fit gets a part dropped off
(40) a. Then I lived in Washington DC for five years, where parking is just impos-
sible.

While one must be careful not to account for all choices by the register param-
eter, which naturally goes beyond a plain opposition between written vs. spoken
English, it is unquestionable that the type of register has an influence on speak-
ers choices. The use of an ARC requires planning and displays, except when the
relative pronoun occupies the subject position in the RC, a non-canonical word
order. Middle ARCs and complex pied-piping also represent elements that can
explain the use of another syntactic structure such as a sentential parenthetical or
an independent clause to convey information the discourse function of which is
similar or closely related to those of ARCs.

4.2 Stylistic variation

All things related to syntactic phenomena, register, discourse and pragmatic phe-
nomena being equal, speakers choices can be accounted for by stylistic variation
as well. Variation among speakers is the result of the interaction of many different
parameters such as sex, education, social background, experience, personality
(Crystal 1995:394). Speakers can choose to use one of ARCs allostructures in-
stead of an ARC because s/he considers that the sentence would be too complex
or long, deciding to cut the sentence in two parts, for instance. The notion of
 Chapter 6. Appositive relative clauses 213

syntactic length/complexity is probably what accounts for the existence of (41)

and not (41a), which definitely sounds awkward because of the length of both the
MC and the ARC:

(41) (QUAL_alloARC53) Wells, a talented all-rounder, was offered terms but

joined Leicestershire in anticipation of greater first-team opportunity.
Batsman Hinks went to Gloucestershire, and Chris Cowdrey as a part-timer
to Glamorgan, while Kelleher, Dobson and Merrick were resting between
engagements. That still leaves a healthy blend of seasoned hands and deter-
mined younger players, many of whom are happy to be regarded as all-round-
ers rather than specialists.
a. ?Wells, a talented all-rounder, was offered terms but joined Leicestershire
in anticipation of greater first-team opportunity. Batsman Hinks went to
Gloucestershire, and Chris Cowdrey as a part-timer to Glamorgan, while
Kelleher, Dobson and Merrick were resting between engagements, which
still leaves a healthy blend of seasoned hands and determined younger
players, many of whom are happy to be regarded as all-rounders rather
than specialists.

Other choices can even be more personal, corresponding to the speakers own
style. The use of co-referring NPs (see (6b) and (7b) at the beginning of this chap-
ter), not uncommon in English journalistic prose, as opposed to ARCs, for in-
stance, corresponds to a strategy that aims at providing extra information on a
referent, generally animate, without repetitions that might seem stylistically awk-
ward. Consider the following two extracts belonging to journalistic prose:

(42) (QUAL_alloARC45) In his first speech to Californias Legislature, Gov. Arnold

Schwarzeneggeri on Tuesday touted March ballot measures aimed at solving
the states budget problems and demanded an overhaul of worker compensa-
tion laws to cut employers costs and create more jobs. Schwarzeneggeri warned
that painful spending cuts are coming as he outlined his agenda for 2004 in the
annual state of the state speech in Sacramento. A report on state finances is a
routine January chore for the USAs 50 governors. But few governors draw 250
reporters, including journalists from Austria and Japan, to these appearances.
Schwarzenegger did. The world-famous screen heroi, a Republican born in
Austria, was elected Oct. 7 to replace Gray Davis.
(43) (QUAL_alloARC54) The ground was thick with snow in the early hours of
Christmas Eve 1993 when takeaway driver Paul Logani left the Golden Flower
Chinese restaurant in Shotley Bridge to deliver an order.
The 25-year-oldi had been called to Blue House Farm, Shotley Bridge, on
Northumberlands border with County Durham, by a call from a telephone kiosk
at the junction of Snows Green Road and Benfieldside Road in the village.
214 Appositive Relative Clauses in English

When the father-of-twoi arrived at the farmhouse just before 10pm, hei was told
by the householders that no meal had been ordered. Hei left, but the household-
ers later noticed his cream Peugeot car was still at the end of the lane leading up
to their home.

In (42), the NP the world-famous screen hero represents an alternative way of men-
tioning the proper name (Arnold) Schwarzenegger. The information conveyed by
the NP could be seen as belonging to a relevance strategy: the speaker is remind-
ing readers and perhaps informing some of them within a leveling of the shared
cognitive space strategy through the implicit relation Arnold Schwarzenegger be
a world-famous screen hero, that the newly-elected Governor of California is also
an actor who is famous all over the world. The use of such a periphrasis is then
triggered by informational considerations, not by accessibility problems for the
referent. Simultaneously, the use of the NP also provides stylistic variety, avoiding
the repetition of the name Schwarzenegger, whose status as international actor is
hearer old for most addressees anyway.
The informational, relevance-oriented strategy can be predominant as in
(43), which represents the beginning on an article on unsolved murders in the
Northumbria Police area. The referent of Paul Logan is discourse new and hearer
new, i.e. completely new for the addressees. The use of co-referring NPs, the 25-
year-old and the father-of-two, enables the speaker to provide extra information
to describe the victim without resorting to more complex structures like nominal
appositives or ARCs, which require the repetition of Paul Logan (or the use of the
the victim for instance). While providing information in quite a synthetized way,
the use of such NPs also allows for stylistic variation for the speaker.

5. Conclusion

The aim of Chapter 6 was to investigate ARCs competing allostructures other than
appositive structures studied in Chapter 5. After teasing out syntactic constraints,
we have studied independent clauses and have shown that speakers choices be-
tween ARCs and independent clauses were constrained by informational con-
siderations, with sentential parentheticals, continuative ARCs, and ARCs with
sentential antecedents representing exceptions. What is meant by informational
considerations is the hierarchization of informational contents depending on
their informational status (hearer new, old or indeterminable), as well as the deci-
sion whether to elaborate on the informational content of the allostructure. We
have also mentioned that ARCs could compete with independent clauses with
preposed elements and that the constraint was then related to the (in)existence of
 Chapter 6. Appositive relative clauses 215

a presupposed open proposition. The speakers decision to keep implicit or make

explicit the semantic link between a first clause and the allostructure (adverbial
clause or independent coordinated clause) represents another constraint that gov-
erns speakers choices to package the information they want to convey. Finally,
we have mentioned register and stylistic variation, both parameters playing a role
and accounting for speakers choices when no other constraint is present.
Together with Chapter 5, Chapter 6 aimed at complementing the definition
of ARCs discourse functions by providing motivations for speakers to choose an
ARC instead of other competing structures (allostructures). As continuative, rel-
evance and subjectivity functions as defined in the previous part of this book are
not the prerogative of ARCs only, the differential definition in terms of syntactic,
semantic, pragmatic and also register-related and stylistic constraints, provides in
fine a better definition of the use of ARCs in discourse.
 Conclusion

This book follows an abundant literature on the definition of ARCs in terms of

syntactic representation published in the course of the 20th century within de-
bate on the DRC-ARC opposition. Its aim was therefore to try and offer a dis-
course-based definition of this non-restricting type of relative clause as defined
by scholars ranging from logicians to generativists. Drawing on the information-
packaging theory as elaborated by Vallduv, pragmatic concepts such as Sperber
& Wilsons relevance theory or the informational status of the different elements
in the sentence, and a range of different corpora of attested examples, we pro-
vided a list of the different discourse functions of ARCs, offering an answer to
the question of why speakers use an ARC in discourse. The corollary question
was why speakers use an ARC and not another syntactic structure when possi-
ble. Parts two and three have answered these two questions respectively. We have
shown that ARCs fulfill specific functions namely continuative, relevance and
subjectivity and that a pragmatic-based positive definition could be provided
for ARCs, which had been defined mainly in opposition with DRCs. We have
also shown that the use of an ARC instead of other syntactic structures, including
nominal appositives, independent clauses, sentential parentheticals, or adverbials,
was not random. The choice between such allostructures is governed by specific
constraints, which go beyond syntactic constraints or considerations of style and
register: some constraints are based on considerations related to the organization
of discourse and include parameters such as the hierarchization of informational
contents, the (non-) existence of a presupposed open proposition or the explicit/
implicit character of a semantic link between the ARC and the MC.
The definitions of the discourse functions of ARCs and of the constraints
governing their use in discourse provide a description of the type of instruction
provided to addressees, as defined in Vallduvs work. The meaning of a sentence
in which an ARC appears corresponds to the addition of the informational con-
tents conveyed by the main clause and the ARC, to which must be added the
additional meaning that corresponds to the speakers syntactic choice to pack-
age the information s/he aims to convey, namely the instruction that informs the
addressees on how the information should be processed.
218 Appositive Relative Clauses in English

The aim of this book was also to provide a precise picture of the use of ARCs
in contemporary English, far from the artificial, constructed examples provided
by grammars. The use of corpora from various registers, ranging from quality
press to spontaneous conversation, and obtained through various methods of
compilation, has enabled us to specify the morphosyntactic, semantic and prag-
matic characteristics of ARCs as used by speakers in actual discourse. This calls
into question long-established preconceived intuitions such as the impossibility
of extraposing or stacking ARCs, or the non-existence of root transformations
such as subject-auxiliary inversion in what are traditionally considered embed-
ded clauses. In particular, working on attested examples has enabled us to better
apprehend atypical data (gap-filled and gapless ARCs) in spontaneous spoken
English, data which can also be found in fiction and informal writing. We hope
this specific chapter (Chapter 2) has shown that gap-filled and gapless ARCs are
not random dysfluencies but represent an orderly phenomenon of the English
language.
Together, the definition of discourse functions and the study of the linguistic
characteristics for this specific structure have provided new elements to feed the
debate on the syntactic status of ARCs: the existence of continuative ARCs, which
fulfill a discourse function (narrative dynamism) that is traditionally the preroga-
tive of main clauses; the existence of atypical ARCs, which are introduced by a
relative pronoun used as a plain connective sometimes similar to a coordinator
introducing a complete clause; and syntactic properties such as the use of senten-
tial ARCs as independent sentences or the existence of root transformations. All
these features provide arguments in favor of treating ARCs as main clauses rather
than embedded clauses. However, the use of ARCs to organize the informational
content conveyed by the MC and the relevance/subjectivity ARCs, establishing
a rupture between what is at issue and what is not, still shows that not all ARCs
can be considered to be on a par with MCs. Syntactically, this is echoed by word
order in V2 languages like Dutch or German, where the verb in ARCs is in final
position. The ambiguous syntactic behavior of ARCs, which has been puzzling
syntacticians for several decades now, could then be related to their functions in
discourse as defined in this book. It might therefore be interesting, in future re-
search, to conduct a study on the syntactic properties and representation of ARCs
with respect to the specific function they fulfill in discourse. This would not only
further strengthen the discourse functions that have been defined in this book but
would also show that discourse phenomena can provide crucial insights into the
syntactic properties of a specific structure.
 References

ACUA-FARIA, J. C. (1996). The Puzzle of Apposition. Santiago: Universidade de Santiago

de Compostela publications.
ACUA-FARIA, J. C. (1998). On Apposition. English Language and Linguistics, 3.1, 5981.
ARMSTRONG, L. & WARD, I. (1931). A Handbook of English intonation. Cambridge: Heffer.
ARNAULD, A. & LANCELOT, C. (1993). Grammaire gnrale et raisonne de Port-Royal.
Genve: Slatkine Reprints.
ARNOLD, D. (2004). Non-Restrictive Relative Clauses in Construction-Based HPSG. In Ml-
ler, S. (Ed.), Proceedings of the 11th International Conference on Head-Driven Phrase Struc-
ture Grammar (pp. 2747). Stanford, CA: CSLI Publications.
ARNOLD, D. (2007). Non-Restrictive Relatives are not Orphans. Journal of Linguistics, 43, 2,
272309.
ARNOLD, D. & BORSLEY, R. D. (2008). Non-restrictive Relative Clauses, Ellipsis and Anaphora.
In Mller, S. (Ed.), The Proceedings of the 15th International Conference on Head-Driven
Phrase Structure Grammar (pp. 325345). Stanford, CA: CSLI Publications.
AURAN, C., BOUZON, C. & HIRST, D. J. (2004). The Aix-MARSEC project: an evolutive data-
base of spoken British English. Proceedings of Speech Prosody 2004, Nara, Japan, 561564.
AURAN, C. & LOOCK, R. (2007). Fonctions discursives et formes prosodiques des proposi-
tions relatives appositives : le cas de la relative appositive de transition dans le discours
journalistique en franais. Nouveaux cahiers de linguistique franaise 28/2007, 177185.
AURAN, C. & LOOCK, R. (2010). Appositive Relative Clauses and their Prosodic Realization
in Spoken Discourse: a Corpus Study of Phonetic Aspects in British English. In Khnlein,
P. et al. (Eds.), Constraints in Discourse 2 (pp. 163178). Amsterdam/Philadelphia: John
Benjamins.
AURAN, C. & LOOCK, R. (submitted, to appear). The prosody of discourse functions: the case
of Appositive Relative Clauses in spoken British English.
BAKHTIN, M. M. ([1929] 1984). Problems of Dostoevskys Poetics. Edited and translated by
Emerson, C. Minneapolis: University of Minnesota Press.
BIANCHI, V. (1999). Consequences of Antisymmetry: Headed Relative Clauses. Berlin: Mouton
de Gruyter.
BIBER, D. (1988). Variation across Speech and Writing. Cambridge: CUP.
BIBER, D. (1995). Dimensions of Register Variation: a Cross-Linguistic Comparison. Cambridge:
CUP.
BIBER, D., JOHANSSON, S., LEECH, G., CONRAD, S. & FINEGAN, E. (1999). Grammar of
Spoken and Written English. London: Longman.
BIRNER, B. J. & WARD, G. (1998). Information Status and Noncanonical Word Order. Amster-
dam/Philadelphia: John Benjamins.
220 Appositive Relative Clauses in English

BLOOMFIELD, L. (1933). Language. London: Allen & Unwin.

BOERSMA, P. (2001). Praat, a system for doing phonetics by computer. Glot International,
5:9/10, 341345.
BOERSMA, P. & WEENINK, D. (2006). Praat: doing phonetics by computer (Version 4.4.17)
[Computer program]. Retrieved April 19, 2006, from http://www.praat.org/.
BOLINGER, D. (1989). Intonation and its Uses: Melody in Grammar and Discourse. Stanford:
Stanford University Press.
BURTON-ROBERTS, N. (1975). Nominal Apposition. Foundations of Language, 13, 391419.
BYBEE, J. (2002). Mechanisms of change in grammaticization: The role of repetition. In Janda,
R. and Joseph, B. (Eds.), Handbook of historical linguistics (pp. 602623). Oxford: Black-
well.
BYBEE, J. (2006). From usage to grammar: the minds response to repetition. Language, 82.4,
529551.
BYBEE, J. (2007). Frequency of use and the organization of language. Oxford: Oxford University
Press.
CANAC MARQUIS, R. & TREMBLAY, M. (1996). The wh-feature and the syntax of restrictive
and non-restrictive relatives in French and English. In J. Lema & E. Trevio (Eds.), Theo-
retical Analyses of Romance Languages (pp. 127141). Amsterdam: John Benjamins.
CANN, R., KAPLAN, T. & KEMPSON, R. (2004). Data at the Grammar-Pragmatics Interface:
the case of resumptive pronouns in English. In Borsley, R. (Ed.), Lingua: Linguistic Data
(pp. 15511577). Amsterdam: Elsevier.
CARABULEA, E. (1959). Attributivele circumstantiale. Scandinavian Conference of Linguistics,
10, 343354.
CARLSON, G. N. (1977). Reference to Kinds in English. New York/London: Garland Publish-
ing.
CHAFE, W. L. (1976). Givenness, contrastiveness, definiteness, subjects, topics and point of
view. In Li, C. (Ed.), Subject and Topic (pp. 2555). New York: Academic Press.
CHAFE, W. L. (1987). Cognitive constraints on information flow. In Tomlin, R. (Ed.), Coher-
ence and grounding in discourse (pp. 2151). Amsterdam/Philadelphia: John Benjamins.
CHAFE, W. L. (1994). Discourse, consciousness, and time: the flow and displacement of conscious
experience in speaking and writing. Chicago: University of Chicago Press.
CHEVILLET, F. (1991). Les Varits de langlais. Paris: Nathan Universit.
CHOMSKY, N. (1966). Cartesian Linguistics. New York: Harper & Row.
CHOMSKY, N. (1995). The Minimalist Program. Cambridge: MIT Press.
CINQUE, G. (2008). Two types of non-restrictive relatives. In Bonami, O. & Cabredo, P. Hof-
herr (Eds.), Empirical Issues in Syntax and Semantics 7 (pp. 99137).
COOK, V. J. & NEWSON, M. (2007). Chomskys Universal Grammar: an Introduction. Malden,
MA/Oxford: Blackwell.
CORNILESCU, A. (1981). Non-restrictive Relative Clauses, an Essay in Semantic Description.
Revue roumaine de linguistique, 26.1, 4167.
COTTE, P. (1997). Grammaire linguistique. Paris: CNED Didier Erudition.
COTTE, P. (1998). LExplication grammaticale de textes anglais. Paris: Presses Universitaires de
France.
COTTE, P. (2000). Les Propositions relatives et lnonciation. Cycnos, 17, 4757.
CRUTTENDEN, A. (1997). Intonation (2nd edition). Cambridge: CUP.
CRYSTAL, D. (1969). Prosodic systems and intonation in English. Cambridge: CUP.
CRYSTAL, D. (1995). The Cambridge Encyclopedia oft he English Language. Cambridge: CUP.
 References 221

CRYSTAL, D. (1997). A Dictionary of Linguistics and Phonetics (4th Ed.). Cambridge, MA:
Blackwell.
DAALDER, S. (1989). Continuative Relative Clauses. In Reiter, N. (Ed.), Sprechen und Hren,
Akten des 23. Linguistischen Kollokwiums Berlin (pp. 195207).
DANE, F. (1966). A Three-Level Approach to Syntax. In Dane, F. et al. (Eds.), Travaux linguis-
tiques de Prague Volume 1 (pp. 225240). Alabama: University of Alabama.
DAVISON, A. (1975). Indirect Speech Acts and What To Do With Them. Syntax and Semantics,
3, 4386.
DE VRIES, M. (2002). The Syntax of Relativization. Utrecht: LOT.
DE VRIES, M. (2006). The Syntax of Appositive Relativization: On Specifying Coordination,
False Free Relatives, and Promotion. Linguistic Inquiry, 37, 229270.
DEHE, N. & KAVALOVA, Y. (2007). Parentheticals: An introduction. In Deh, N. & Y. Kav-
alova, Y. (Eds.), Parentheticals [Linguistik aktuell/ Linguistics today, 106] (pp. 122). Am-
sterdam/Philadelphia: John Benjamins.
DEHE, N. (2009). Clausal parentheticals, intonational phrasing, and prosodic theory. Journal
of Linguistics, 45(3), 569615.
DEL GOBBO, F. (2001). Appositives Schmappositives in Chinese. Retrieved from http://www.
cbs.polyu.edu.hk/tang/PAPER%20DelGobbo%20RC.doc
DEPRAETERE, I. (1995). The Tense System in English Relative Clauses. Berlin: Mouton de Gru-
yter.
DEPRAETERE, I. (1996). Foregrounding in English Relative Clauses. Linguistics, 34, 699731.
DORON, E. (1982). On the syntax and semantics of resumptive pronouns. Linguistic Forum,
19, 148.
DU BOIS, J. W. (1991). Transcription design principles for spoken discourse research. Prag
matics, 1, 71106.
DU BOIS, J. W., SCHUETZE-COBURN, S., CUMMING, S. & PAOLINO, D. (1993). Outline of
discourse transcription. In Edwards, J. E. and Lampert, M. D. (Eds.), Talking Data: Tran-
scription and Coding in Discourse Research (pp. 4591). Hillsdale, NJ: Lawrence Erlbaum.
DUCROT, O. (1984). Le Dire et le dit. Paris: Les Editions de Minuit.
EMONDS, J. (1979). Appositive Relatives Have No Properties. Linguistic Inquiry, 10.2, 241
243.
ERNOUT, A. & THOMAS, F. (2002). Syntaxe Latine. Paris: Librairie Klincksieck.
ERTESCHIK-SHIR, N. (1992). Resumptive pronouns in islands. In Goodluck, H. and Roche-
mont, M. (Eds.), Island Constraints: Theory, Acquisition, and Processing (pp. 89108). Dor-
drecht: Kluwer.
FABB, N. (1990). The Difference between English Restrictive and Nonrestrictive Relative
Clauses. Linguistics, 26, 5778.
FOWLER, H. W. & FOWLER, F. G. (1949). The Kings English. Oxford: Clarendon Press.
FUCHS, C., RIVIRE, N. & RIVIRE, C. (1987). Les relatives et la construction de linter
prtation. Langages 88, Les Types de Relatives, 94127.
GARRO, L. & PARKER, F. (1982). Some Suprasegmental Characteristics of Relative Clauses in
English. Journal of Phonetics, 10, 149161.
GIVON, T. (1984). Syntax: A functional-typological introduction. Amsterdam/Philadelphia:
John Benjamins.
GRESSET, S. (1984). Which/That marqueurs de relatives. Cahiers de recherches en linguistique
anglaise, 2, 201268.
222 Appositive Relative Clauses in English

GREENBAUM, S. (Ed.). (1996). Comparing English Worldwide: the International Corpus of

English. Oxford: Clarendon Press.
GRICE, H. P. (1975). Logic and conversation. In Cole, P. and J. Morgan (Ed.), Studies in Syn-
tax and Semantics III: Speech Acts (pp. 4158). New York: Academic Press. Reprinted in
Grice, H. P. (Ed.). (1989). Studies in the Way of Words. Cambridge, MA: Harvard Univer-
sity Press.
GUSSENHOVEN, C. (2004). The Phonology of Tone and Intonation. Cambridge: Cambridge
University Press.
HAEGEMAN, L. (1991). Parenthetical Adverbials: the radical orphanage approach. In Chiba,
S. (Ed.), Aspects of Modern linguistics (pp. 232254). Tokyo: Kaitakushi.
HAEGEMAN, L. (2000). Extraction du sujet, rallocation de cas et localit. Cycnos, 17, 3945.
HAGGE, C. (1982). La Structure des langues. Paris: PUF, Collection Que sais-je?
HAWKINS, J. A. (1978). Definiteness and Indefiniteness: a Study in Reference and Grammatical-
ity Prediction. London: Croom Helm.
HAWKINS, J. A. (1992). Syntactic Weight vs. Information Structure in Word Order Varia-
tion. In Jacobs, J. (Ed.), Informationsstruktur und Grammatik (Linguistische Berichte 4)
(pp.196219). Opladen: Westdeutscher Verlag.
HAWKINS, J. A. (1994). A Performance Theory of Order and Constituency. Cambridge: CUP.
HIGGINS, R. F. (1979). The Pseudo-cleft Construction in English. New York: Garland Publish-
ing.
HOCKETT, Ch. F. (1955). Attribution and Apposition in English. American Speech, 30, 99
102.
HOOPER, J. B. & THOMPSON, S. A. (1973). On the Applicability of Root Transformations.
Linguistic Inquiry, 4.4, 465497.
HOPPER, P. & THOMPSON, S. A. (1980). Transitivity in Grammar and Discourse. Language
56, 251299.
HUDDLESTON, R. (1984). Introduction to the Grammar of English. Cambridge: CUP.
HUDDLESTON, R. & PULLUM, G. K. (2002). The Cambridge Grammar of the English Lan-
guage. Cambridge: CUP.
ISHIZUKA, T. (2008). Restrictive and Non-restrictive relative clauses in Japanese: Antisym-
metric approach. Retrieved from http://ling.auf.net/lingBuzz/000808.
JACKENDOFF, R. (1977). X-bar Syntax: A Study of Phrase Structure. Cambridge, MA: MIT
Press.
JESPERSEN, O. (1970). A Modern English Grammar on Historical Principles, Vol. III. London:
George Allen & Unwin.
JUGNET, A. (2008). Lextraposition des compltives objet en anglais. University of Lille 3, Ph.
D. dissertation.
KHALIFA, J. C. (1999). A Propos des relatives appositives. Syntaxe, smantique, pragmatique.
Anglophonia, 6, 729.
KJELLMER, G. (1988). Conjunctional/Adverbial which in Substandard English. Studia Anglica
Posnaniensia, 21, 125137.
KLEIBER, G. (1987). Relatives restrictives et relatives appositives: une opposition introuvable?.
Tbingen: Niemayer.
KNOWLES, G., WICHMANN, A. & ALDERSON, P. (1996). Working with Speech: perspectives
on research into the Lancaster/IBM Spoken English Corpus. London: Longman.
KOKTOV, E. (1985). Apposition as a Pragmatic Phenomenon in a Functional Description.
University of East Anglia Papers in Linguistics, 3979.
 References 223

KROCH, A. (1981). On the role of resumptive pronouns in amnestying island constraints vio-
lations. CLS, 17, 125135.
KUHA, M. (1994). Attitudes towards users of coordinate which. Ms.
KUNO, S. (1973). The Structure of the Japanese Language. Cambridge, Mass.: MIT Press.
KUNTSMAN, P. (1994). Relatif et liaison: le cas du relatif dit de liaison. In Combettes, B. and
Monsonego, S. (Eds.), Le Moyen Franais. Philologie et linguistique. Approches du texte et
du discours: Actes du VIII colloque international sur le moyen franais (pp. 517527). Paris:
Didier Erudition.
LABOV, W. (1972). The transformation of experience in narrative syntax. Language in the Inner
City. Philadelphia: University of Pennsylvania Press.
LAGO, J. (1991). Un Type spcial de modification non-restrictive: lapposition. Verba, 18, 487
519.
LAMBRECHT, K. (1996). Information Structure and Sentence Form. Cambridge: CUP.
LAMBRECHT, K. (2000). Prdication seconde et structure informationnelle: la relative de per-
ception comme structure prsentative. Langue Franaise, 127, 4966.
LANCRI, A. (1999). A Contrastive Analysis of Relative Clauses in Old English and Middle
Egyptian. In Cotte, P. (Ed.), Contrast, Comparison and Communication, English with Spe-
cial Reference to some Languages of the Mediterranean Shores (pp. 1132). Saint-Etienne:
Publications de lUniversit de Saint-Etienne.
LAVENCY, M. (1998). Grammaire fondamentale du latin, Tome V, Volume 2: La Proposition
relative. Leuven/Paris: Peeters.
LEECH, G. (1983). Principles of Pragmatics. London/New York: Longman.
LEECH, G. & SVARTVIK, J. (1994). A Communicative Grammar of English. London/New York:
Longman.
LERNER, G. H. (2004). On the place of linguistic resources in the organization of talk-in-inter-
action: grammar as action in prompting a speaker to elaborate. Research on Language and
Social Interaction 37, 2, 151184.
LOETSCHER, A. (1973). On the Role of Nonrestrictive Relative Clauses in Discourse. Papers
from the Ninth Regional Meeting, Chicago Linguistic Society (April 1315), 356368.
LOOCK, R. (2003). Les Fonctions discursives des propositions subordonnes relatives apposi-
tives en discours. Anglophonia, 12, 113131.
LOOCK, R. (2005). Appositive relative clauses in contemporary written and spoken English: dis-
course functions and competitive structures. Lille III University, France, and UMR 8528
SILEX, CNRS, Ph.D. Dissertation.
LOOCK, R. (2006). Propositions relatives appositives et discours rapport. In Ballard, M. (Ed.),
Le discours rapport: approche linguistique et/ou traductologique, Actes du Colloque du
CERTA (pp. 7793). Arras: Presses Universitaire de lArtois.
LOOCK, R. (2007). Appositive Relative Clauses and their Functions in Discourse. Journal of
Pragmatics, 39, 336362.
LOOCK, R. (2007). Are you a good which or a bad which ?: the relative pronoun as a plain con-
nective. In Celle, A. & Huart, R. (Eds.), Connectives as Discourse Landmarks, Pragmatics
and Beyond Series (pp. 7178). Amsterdam/Philadelphia: John Benjamins.
McCAWLEY, J. D. (1988). The Syntactic Phenomena of English, Volume 2, Chapter 13. Chicago:
CUP.
McCAWLEY, J. D. (1996). An Overview of Appositive Constructions in English. In Przez
dziecki, M. & Whaley, L. (Eds.), ESCOL 95 (pp. 195211). Cornell University Press.
224 Appositive Relative Clauses in English

McCLOSKEY, J. (1990). Resumptive pronouns, A binding and levels of representation in Irish.

In Randall, H. (Ed.), Syntax of the Modern Celtic Languages, volume 23 of Syntax and Se-
mantics (pp. 199248). San Diego, CA: Academic Press.
McDAVID, V. (1977). Which in Relative Clauses. American Speech: a Quarterly of Linguistic
Usage, 52, 12, 7683.
MALAN, N. (1999). La Proposition relative en anglais contemporain, une approche pragmatique.
Paris: Ophrys.
MATTHEWS, P. H. (1981). Syntax. Cambridge: CUP.
MLIS, G. (1999). Propositions relatives et niveau phrastique. Anglophonia, 6, 4359.
MLIS, G. (2000). Relatives et types de qualification. Cycnos, 17, 120.
MEYER, C. F. (1987). Apposition in English. Journal of English Linguistics, 20, 101121.
MEYER, C. F. (1989). Restrictive Apposition: an Indeterminate Category. English Studies, 70,
147166.
MEYER, C. F. (1991). A Corpus-Based Study of Apposition in English. In Aijmer, K. and Alten-
berg, B. (Eds.), English Corpus Linguistics (pp. 166181). London: Longman.
MEYER, C. F. (1992). Apposition in Contemporary English. Cambridge: CUP.
MIKKELSEN, L. (2004). Reexamining Higgins Taxonomy: a Split in the Identificational Class.
Paper presented at the 78th Linguistics Society of America meeting. Retrieved from http://
ist-socrates.berkeley.edu/~mikkelse/papers/LSA_04_ho.pdf
MILLER, J. (1988). That: A Relative Pronoun? Sociolinguistics and Syntactic Analysis. In An-
derson, J. M. and Macleod, N. (Eds.), Edinburgh Studies in the English Language (pp. 113
119). Edinburgh: John Donald Publishers Ltd.
MILLER, Ph. (1992). Clitics and Constituents in Phrase Structure Grammar. New York: Garland
Publishing.
MILLER, Ph. (2001). Discourse Constraints on (Non)-Extraposition from Subject in English.
Linguistics, 39.4, 683701.
MILROY, J. & MILROY, L. (Eds.). (1993). Real English: The Grammar of English Dialects in the
British Isles. London/New York: Longman.
MOREL, M. A. & DANON-BOILEAU, L. (1998). Grammaire de lintonation. Lexemple du fran-
ais oral. Paris/Gap: Ophrys.
NEVEU, F. (1998). tudes sur lapposition. Paris: Honor Champion diteur.
OCONNOR, K. (2008). Aspects de la syntaxe et de linterprtation de lapposition antcdent
nominal. University of Lille 3, Ph. D. dissertation.
POTTS, C. (2005). The Logic of Conventional Implicatures. Oxford: OUP.
PRINCE, E. F. (1981). Toward a Taxonomy of Given/New Information. In Cole, P. (Ed.), Radi-
cal Pragmatics (pp. 223254). New York: Academic Press.
PRINCE, E. F. (1986). On the Syntactic Marking of Presupposed Open Propositions. In Farley,
A. et al. (Eds.), Papers from the Parasession on Pragmatics and Grammatical Theory, 22nd
regional meeting of the Chicago Linguistics Society (pp. 208222).
PRINCE, E. F. (1990). Syntax and discourse: a look at resumptive pronouns. In Hall, K. et al.
(Eds.), Proceedings of the Sixteenth Annual Meeting of the Berkeley Society (pp. 482497).
PRINCE, E. F. (1992). The ZPG Letter: Subjects, Definiteness, and Information-Status. In
Thompson, S. & Mann, W. (Eds.), Discourse Description: Diverse Analyses of a Fund Rais-
ing Text (pp. 295325). Amsterdam/Philadelphia: John Benjamins.
PRINCE, E. F. (1995). On kind-sentences, resumptive pronouns, and relative clauses. In Guy,
G., Baugh, J. & Schiffrin, D. (Eds.), Towards a social science of language a festschrift for
William Labov (pp. 223235). Cambridge: CUP.
 References 225

QUIRK, R. (1957). Relative Clauses in Educated Spoken English. English Studies, 38, 97109.
QUIRK, R., GREENBAUM, S., LEECH, G. & SVARTVIK, J. (1985). A Comprehensive Gram-
mar of the English Language. London: Longman.
REINHART, T. (1984). Principles of Gestalt Perception in the Temporal Organization of Nar-
rative Texts. Linguistics, 22, 779809.
RIVIRE, C. & RIVIRE, N. (2000). Sur lopposition relative restrictive/relative appositive.
Cycnos 17, 2178.
ROACH, P., KNOWLES, G., VARADI, T. & ARNFIELD, S. (1993). MARSEC: A machine
readable Spoken English corpus. Journal of the International Phonetic Association, 23(2),
4753.
ROSS, J. (1967). Constraints on Variables in Syntax. MIT, Ph.D. dissertation.
SAUSY, L. (1977). Grammaire latine complte. Paris: Librairie Fernand Lanore.
SCHANEN, F. & CONFAIS, J. P. (1989). Grammaire de lallemand: formes et fonctions. Paris:
Nathan Universit.
SCHLEPPEGRELL, M. J. (1991). Paratactic because. Journal of Pragmatics, 16, 323337.
SELLS, P. (1984). Syntax and Semantics of Resumptive Pronouns. Ph.D. thesis, University of
Massachusetts, Amherst.
SOPHER, H. (1971). Apposition. English Studies, 52, 401412.
SOUESME, J. C. (1992). Grammaire anglaise en contexte. Paris/Gap: Ophrys.
SPERBER, D. & WILSON, D. (1986, 1995). Relevance: Communication and Cognition. Oxford:
Blackwell.
SPERBER, D. & WILSON, D. (1993). Linguistic Form and Relevance. Lingua, 90, 125.
SPERBER, D. & WILSON, D. (2002). Relevance Theory. In UCL Working Papers in Linguistics,
14, 249287.
TAGHVAIPOUR, M. (2003). Resumptive Pronouns in Persian Relative Clauses: an HPSG
Analysis. Retrieved from: privatewww.essex.ac.uk/~matagh/index_files/RPs_in_Persian_
RCs.pdf
TAO, H. & McCARTHY, M. J. (2001). Understanding Non-Restrictive which-Clauses in Spo-
ken English, which is not an Easy Thing. Language Sciences, 23, 651677.
THOMPSON, S. A. (1987). Subordination and Narrative Event Structure. In Tomlin, R. (Ed.),
Coherence and Grounding in Discourse (pp. 435454). Amsterdam/Philadelphia: John
Benjamins.
THOMPSON, S. A. (1990). Information Flow and Dative Shift in English. In Edmondson,
J., Feagin, K. and Mhlhusler, P. (Eds.), Development and Diversity: Linguistic Variation
across Time and Space (pp. 239253). Dallas: Summer Institute of Linguistics.
THORNE, J. P. (1972). On Nonrestricitive Relative Clauses. Linguistic Inquiry, 3, 552556.
VALLDUV, E. (1992). The Informational Component. University of Pennsylvania, Ph.D. dis-
sertation.
VALLDUV, E. (1993). Information packaging: a survey. Report prepared for WOPIS (Word
Order, Prosody, and Information Structure), Centre for Cognitive Science and Human
Communication Research, University of Edinburgh. Retrieved from: http://www.hcrc.
ed.ac.uk/publications/rp44.ps.gz.
VAN DEN EYDN, N. (1993). Syntactic Variation and Unconscious Linguistic Change: a Study of
Adjectival Relative Clauses in the Dialect of Dorset. Frankfurt am Main: Peter Lang.
VAN DER AUWERA, J. (1985). Relative that a Centennial Dispute. Journal of Linguistics, 21,
149179.
226 Appositive Relative Clauses in English

WENNERSTROM, A. (2001). The Music of Everyday Speech: Prosody and Discourse Analysis.
Oxford: OUP.
WICHMANN, A. (2000). Intonation in Text and Discourse: Beginnings, Middles and Ends. Edin
burgh: Pearson Education Limited.
WICHMANN, A. (2001). Spoken parentheticals. In K. Aijmer (Ed.), A Wealth of English: Studies
in Honour of Goran Kjellmer (pp. 171193). Gothenburg: Gothenburg University Press.
WIGHTMAN, C. W., SHATTUCK-HUFNAGEL, S., OSTENDORF, M. & PRICE, P. (1992).
Segmental durations in the vicinity of prosodic phrase boundaries. Journal of Acoustical
Society of America, 91(3), 17071717.
ZHANG, N. (2001). Sell Nonrestrictive Relatives. Retrieved from: http://www.usc.edu/dept/
LAS/ealc/chinling/articles/appositivies.pdf
ZIV, Y. & COLE, P. (1974). Relative Extraposition and the Scope of Definite Descriptions in
Hebrew and English. CLS, 10, 772787.
 Appendix

QUAL TABL FIC SPEC TOTAL Spoken

for ARCs
written
ARCs
Punctuation Comma 92,50% 83% 81% 94% 87,60%
Dash 3,50% 13% 6% 2% 6,10%
Brackets 1,50% 1% 6% 4% 3,20%
Full stop 1,50% 3% 6% 0% 2,60%
Square backets 0,50% 0% 0% 0% 0,10%
Semi-colon 0,50% 0% 1% 0% 0,40%
Relative which 44% 25% 62% 72% 50,70% 66,70%
pronoun who 33,50% 64% 26% 22% 36,40% 13,30%
where 12,50% 4% 9% 2% 6,90% 16,70%
when 3% 0% 0% 2% 1,20% 3,30%
whose 5,50% 6% 2% 2% 3,90% 0%
whom 1% 1% 0% 0% 0,50% 0%
that 0,50% 0% 1% 0% 0,40% 0%
Syntactic Subject 72,50% 87% 68% 86% 78,40% 68,30%
function/ Direct object 4% 2% 16% 2% 6% 10%
pronoun Indirect object 0% 0% 4% 0% 1% 0%
Adjunct 18% 5% 10% 8% 10,20% 21,70%
Noun 5,50% 6% 2% 4% 4,40% 0%
complement
Middle/final Middle 40,50% 66% 11% 26% 35,90% 11,70%
position Final 59,50% 34% 88% 72% 63,40% 88,30%
? 0% 0% 1% 2% 0,70% 0%
(Non-) Non-sentential 94,50% 94% 72% 96% 89,10% 71,70%
sentential Sentential 5,50% 6% 27% 4% 10,60% 28,30%
antecedent ? 0% 0% 1% 0% 0,30% 0%
228 Appositive Relative Clauses in English

QUAL TABL FIC SPEC TOTAL Spoken

for ARCs
written
ARCs
Type of Proper name 38% 62% 22% 26% 37% 23,40%
antecedent Definite NP 29,50% 25% 27% 34% 28,90% 23,30%
Indefinite NP 20% 3% 12% 34% 17,20% 25%
PP 7% 4% 9% 2% 5,50% 0%
AdvP 0% 0% 2% 0% 0,50% 0%
Sentential 5,50% 6% 28% 4% 10,90% 28,30%
antecedent
Syntactic Subject 32% 63% 15% 31% 35,20% 11,60%
function/ Direct object 16% 7% 35,50% 22% 20,10% 39,50%
antecedent Indirect object 12% 12% 9,50% 14% 11,90% 13,90%
(non-
Adjunct 23% 9% 26% 12% 17,50% 13,90%
sentential
antecedents) by-phrase 4% 4% 3% 9% 5% 2,30%
Noun 8% 5% 7% 12% 8% 4,60%
complement
Other 5% 0% 4% 0% 2,30% 14,20%
Verb type in Process 85% 82% 58% 84% 77,20% 38,30%
the ARC State 13,50% 18% 42% 16% 22,40% 61,70%
? 1,50% 0% 0% 0% 0,40% 0%
Verb tense in Simple present 36% 36% 25% 32% 32,30% 53,30%
the ARC Present perfect 9% 6% 3% 8% 6,50% 1,70%
Preterit 42,50% 49% 49% 48% 47,10% 31,70%
Past perfect 4% 0% 6% 0% 2,50% 1,70%
Modal 5,50% 2% 13% 12% 8,10% 6,70%
auxiliaries
-ing forms 3% 7% 4% 0% 3,50% 4,90%
Subjectivity S+ 22,50% 27% 41% 22% 28,10% 45%
S 14% 61% 30% 12% 29,30% 36,70%
S 61% 12% 29% 66% 42% 18,30%
? 2,50% 0% 0% 0% 0,60% 0%
Informational Discourse New 78% 47% 73% 64% 65,50% 86,70%
status/ Discourse Old 22% 53% 27% 36% 34,50% 13,30%
antecedent Hearer New 16,50% 20% 31% 16% 20,90% 31,70%
Hearer Old 44% 60% 44% 58% 51,50% 18%
Inferrable 30,50% 9% 24% 14% 19,40% 35%
? New+ 2% 2% 0% 2% 1,50% 6,70%
Old+ 1,50% 1% 0% 2% 1,10% 3,30%
New/Old 5,50% 8% 1% 8% 5,60% 5,30%
 Appendix 229

QUAL TABL FIC SPEC TOTAL Spoken

for ARCs
written
ARCs
Informational Discourse New 99% 99% 99% 100% 99,25% 100%
status/ARC Discourse Old 1% 1% 1% 0% 0,75% 0%
Hearer New 60% 55% 100% 64% 69,80% 81,40%
Hearer Old 2,50% 1% 0% 2% 1,40% 0%
Inferrable 0,50% 0% 0% 0% 0,10% 0%
? New+ 8,50% 17% 0% 10% 8,90% 6,80%
Old+ 4% 7% 0% 2% 3,20% 3,40%
New/Old 24,50% 20% 0% 22% 16,60% 8,40%
 Index

A C End-weight 145146, 181

Acua-Faria 151, 155, 158 Canac Marquis & Tremblay Explanation 89, 49, 111115
Adverbs 10, 27, 43, 44, 92, 164 168169 Extraposition
Adverbials 10, 49, 51, 69, 98, Case reassignment 80 Of relative clauses 2426,
148, 193, 208211 Chinese 4345, 169 51, 9192, 146, 169
Agreement phenomena Class (one of) membership Of that-clauses 25, 86
158161 relation 154, 157, 166167,
Aix-Marsec corpus 90 176177, 183184 F
Anaphora 2627, 34, 59, 65, Co-reference 53, 149, 155, Familiarity 85, 174176
6667, 70, 145 193194, 213214 Floor-keeping 7172
Antecedent Comma intonation 24 Foregrounding 48, 9697,
Identification 910, 168169 Constituency tests 21, 38 100, 117, 138, 155157, 191192,
Informational status 8687, Coordinated clauses 196204
186187 As allostructures 31, 68, 70, French 8, 3637, 4041, 115
Referential independence 147148, 192, 196197, 203, 116, 124, 139
12, 57, 77, 8889, 162 211, 212 Frequency 54, 129130
Retrievability 5859, 145, As origin of ARCs 24, 51, 69 Frequency effect 145, 146
181182, 184, 195 Coordination 24, 31, 3233, 34,
Sentential antecedent 16, 51, 6970, 100101, 157158, G
28, 4041, 46, 50, 68, 92, 203, 211, 212 Galician 159
204205 Cornilescu 48, 5052, 105 Gap 5354
Type 1517, 3739, 68, 80, 92 Copy see Trace German 4243, 47, 159160
Arabic 4546 Cotte 15, 37, 168 Greek 61, 160
Arnold 3334, 101
Auran 134 D H
De Vries 16, 17, 19, 31, 34, 69 Haegeman 21, 80
B Deh 135, 198 Higgins 172173
Backgrounding 4849, 9697, Detachment 2231, 69, 100 Huddleston & Pullum 12,
105, 192193, 196206 101, 157 1415, 17, 2425, 60, 100, 157,
Because 72 Determinative Relative Clauses 158, 165
Biber 60, 72, 88, 89, 129130, 747, 179
177, 198199 Dutch 31, 34 I
Birner & Ward 86, 144145, Dysfluency 54, 5965, 66, 87, 91 ICE-GB corpus 90
146, 206207 Identificational relation 154,
British National Corpus 90, E 166, 171176, 184186
129 Emonds 10, 19, 22, 23, 24, 26, Idiomatic expressions 1920
Burton-Roberts 154155, 166 27, 2930, 33, 49, 100 Increment 132134
232 Appositive Relative Clauses in English

Independent clauses 10, 2830, Syntax 157161, 168169 Relative pronoun

3334, 49, 57, 59, 67, 68, Type of identification Choice 1415, 20, 4041, 46,
7071, 97, 98, 100101, 132, 154157 68, 90
145, 147148, 192, 196206, Role 6572, 100101, 207,
206208, 211, 212 O 209
Inferential inter-clausal link OConnor 10, 16, 34 Relevance theory 104
4849, 69, 111115, 122123, Open proposition 207208 Resumptive phrase 56
139, 193, 210211 Orphanage 2131, 49, 192 Resumptive pronoun 5465
Information packaging Reversibility 156157
143146, 200 P Root transformations 2930
Information(al) status 52, Parentheticals 14, 2224, 29,
8487, 146 71, 78, 120, 127128, 135136, S
Instruction 144, 197, 200 148, 193, 198199, 211212 Sentential antecedent
Integration 2122, 31, 199 Participle clauses 149, 166167 see Antecedent
Interrogative clauses 19, 133 Performance error Spanish 41, 159
see Dysfluency Specificational relation 172173
J Performatives 11 Speech act 1011, 33, 192
Jackendoff 2122, 33 Persian 4142 Sperber & Wilson 104, 105, 111
Jespersen 4950, 65, 95 Pied-piping 19, 212 Stacking 1719, 162
Politeness 116, 117, 131132 Stylistic variation see Variation
L Polyphony 84, 123125 Subjectivity 8284, 92, 120127,
Lambrecht 3637, 146, 147 Potts 11, 19 164165
Lerner 132134 Preposing 206208 Subordination 48, 6970,
Latin 68, 112, 158 Presentative relative clauses 37 9697, 101, 157161
Presupposition 24, 30, 51, 126, Suppressibility 1112, 4748,
M 207208 87, 103, 119, 128
Main clauses Prince 57, 119, 144, 207
As origin of ARCs 2231 Princes taxonomy of given/new T
Link with ARCs 32, 37, 48, information 8587 Thompson 30, 9697, 145,
51, 68, 6970, 83, 100101, Prosody 12, 7172, 8889, 191, 192
111115 133134, 134138, 144, 199 Trace 5354, 58
McCawley 11, 23, 27, 166 200, 207
Metatextuality 115, 132 Proper names 10, 13, 38, 44, 80, V
Meyer 151, 155, 162, 165166 92, 109, 119, 167, 171173 Vallduv 144, 200
Punctuation 1214, 38, 50, Variation
N 7879, 127128, 156, 179, 185 Register/Stylistic 31, 88,
Narrative subordination 97, 9293, 129134, 147, 199,
192, 203 Q 206, 211214
Negation 16, 99100, 161162 Quantifiers 1617, 20, 3334, 44 Syntactic 5965
Neveu 155 Question-tag 11
Nominal appositives Quirk 10, 31, 69, 71, 155, 166 W
As allostructures 149150, Wh-movement 19, 53, 56, 80,
170188, 193 R 207
Definition 152157 Referential act 910, 155, 162,
Discourse functions 179
162165 Register variation see Variation
In the series Studies in Discourse and Grammar the following titles have been published thus
far or are scheduled for publication:

23 Barth-Weingarten, Dagmar, Elisabeth Reber and Margret Selting (eds.): Prosody in

Interaction. Expected Forthcoming
22 Loock, Rudy: Appositive Relative Clauses in English. Discourse functions and competing structures.
2010. xiii,232pp.
21 Betz, Emma: Grammar and Interaction. Pivots in German conversation. 2008. xiii,208pp.
20 Sadler, Misumi: Grammar in Use across Time and Space. Deconstructing the Japanese dative subject
construction. 2007. xiv,212pp.
19 Taleghani-Nikazm, Carmen: Request Sequences. The intersection of grammar, interaction and social
context. 2006. x,125pp.
18 Ewing, Michael C.: Grammar and Inference in Conversation. Identifying clause structure in spoken
Javanese. 2005. x,276pp.
17 Hakulinen, Auli and Margret Selting (eds.): Syntax and Lexis in Conversation. Studies on the use of
linguistic resources in talk-in-interaction. 2005. viii,408pp.
16 Koshik, Irene: Beyond Rhetorical Questions. Assertive questions in everyday interaction. 2005. x,183pp.
15 Golato, Andrea: Compliments and Compliment Responses. Grammatical structure and sequential
organization. 2005. xii,249pp.
14 Du Bois, John W., Lorraine E. Kumpf and William J. Ashby (eds.): Preferred Argument Structure.
Grammar as architecture for function. 2003. x,459pp.
13 Englebretson, Robert: Searching for Structure. The problem of complementation in colloquial
Indonesian conversation. 2003. x,206pp.
12 Hayashi, Makoto: Joint Utterance Construction in Japanese Conversation. 2003. xii,250pp.
11 Scheibman, Joanne: Point of View and Grammar. Structural patterns of subjectivity in American
English conversation. 2002. xiv,188pp.
10 Selting, Margret and Elizabeth Couper-Kuhlen (eds.): Studies in Interactional Linguistics. 2001.
viii,438pp.
9 Helasvuo, Marja-Liisa: Syntax in the Making. The emergence of syntactic units in Finnish conversation.
2001. xiv,176pp.
8 Mori, Junko: Negotiating Agreement and Disagreement in Japanese. Connective expressions and turn
construction. 1999. xii,240pp.
7 Laury, Ritva: Demonstratives in Interaction. The emergence of a definite article in Finnish. 1997.
viii,294pp.
6 Dorgeloh, Heidrun: Inversion in Modern English. Form and function. 1997. x,236pp.
5 Tao, Hongyin: Units in Mandarin Conversation. Prosody, discourse, and grammar. 1996. xvi,226pp.
4 Downing, Pamela A.: Numeral Classifier Systems: The Case of Japanese. 1996. xx,336pp.
3 Weber, Elizabeth G.: Varieties of Questions in English Conversation. 1993. x,252pp.
2 Iwasaki, Shoichi: Subjectivity in Grammar and Discourse. Theoretical considerations and a case study of
Japanese spoken discourse. 1992. xii,152pp.
1 Geluykens, Ronald: From Discourse Process to Grammatical Construction. On Left-Dislocation in
English. 1992. xii,182pp.