Syntactic variation in English quantified all

Syntactic variation in English quantified all
Syntactic variation in English quantified
noun phrases with all, whole, both and half
Acta Wexionensia
Nr 38/2004
Humaniora
Syntactic variation in English
quantified noun phrases
with all, whole, both and half
Maria Estling Vannestål
Växjö University Press
Abstract
Estling Vannestål, Maria, 2004. Syntactic variation in English quantified noun phrases
with all, whole, both and half, Acta Wexionensia nr 38/2004. ISSN: 1404-4307,
ISBN: 91-7636-406-2. Written in English.
The overall aim of the present study is to investigate syntactic variation in certain
Present-day English noun phrase types including the quantifiers all, whole, both
and half (e.g. a half hour vs. half an hour). More specific research questions
concerns the overall frequency distribution of the variants, how they are distributed across regions and media and what linguistic factors influence the choice of
variant. The study is based on corpus material comprising three newspapers from
1995 (The Independent, The New York Times and The Sydney Morning Herald)
and two spoken corpora (the dialogue component of the BNC and the Longman
Spoken American Corpus).
The book presents a number of previously not discussed issues with respect to
all, whole, both and half. The study of distribution shows that one form often
predominated greatly over the other(s) and that there were several cases of regional variation. A number of linguistic factors further seem to be involved for
each of the variables analysed, such as the syntactic function of the noun phrase
and the presence of certain elements in the NP or its near co-text. For each of the
variables, all factors were ranked according to their strength of correlation with
particular variants. The study also discusses a possible grammaticalisation process concerning NPs with half and the possibility of all sometimes having another
function than expressing totality: to express large quantity.
The whole idea of grammatical synonymy has been questioned by some
scholars, but the conclusion drawn in the present study is that there are variables
that are at least very close to each other in meaning, and that a number of linguistic and non-linguistic factors influence our choices of variant. A great deal of the
information obtained was too detailed to be useful for pedagogical purposes, but
in several cases the results could clearly be used to improve school and reference
grammars.
Keywords: syntactic variation, quantifiers, all, whole, both, half, linguistic factors, British English, American English, Australian English, grammaticalisation,
totality, corpus, newspaper corpus
Akademisk avhandling för filosofie doktorsexamen vid Institutionen för
humaniora, Växjö universitet 2004
Skriftserieredaktörer: Tommy Book och Kerstin Brodén
ISSN: 1404-4307
ISBN: 91-7636-406-2
Tryck: Intellecta Docusys, Göteborg 2004
To my father Lars
Preface
Some time ago, I heard someone suggest that researchers seem to choose their
scientific approach according to their personalities. People who lead very organised lives tend to go for structuralism and other well-organised formal theories,
whereas people who live in chaos (with animals and children, for instance) opt
for messier theories. I found this quite interesting and realised that, considering
the fact that my life includes a large number of these chaotic elements (children,
dog, cats), I should go for a messy theory. On the other hand, I have another,
more organised side to my personality, a side that keeps all documents in neat
files and all photos in albums, so perhaps structuralism would suit me just as
well. Perhaps it is this combination of messiness and order that made me not
want to opt for a particular school or theory, but rather try to carry out a theoryneutral study.
What really set this study afloat is frustration at the shortage of information
about syntactic variation in many grammar books. This frustration, however,
turned into fascination as my research progressed and as I evolved into a fanatic
fan of authentic text corpora. Sometimes I wondered whether the writing of the
thesis would take half (of) my life, all (of) my life or perhaps even both (of) my
lives (provided I end up as an English linguist in my next incarnation as well).
But here it is.
Many people have been involved, in one way or another, in the process of
completing this book. I would like to thank…
¾… Hans Lindquist, my supervisor at Växjö University, for believing in me
from the very beginning, for letting me participate in his project GramTime
and for supporting me practically and mentally throughout.
¾… Karin Aijmer, my supervisor at Göteborg University, for reading my
manuscripts very conscientiously and for constantly providing me with insightful ideas and suggestions for improvement.
¾… Joakim Nivre, my supervisor at Växjö University, for always giving me
constructive criticism and clarifying answers to my many e-mail questions
on everything from semantics to statistics.
¾… all my colleagues at the School of Humanities at Växjö University,
especially Magnus Levin, for sharing my enthusiasm over new findings, for
providing me with useful references and for laughing with me at the many
absurdities of the academic world, Staffan Klintborg, for being an excellent
role model and source of inspiration, Barbro Lindhe, my great mental
support, with whom I have had many rewarding discussions concerning
77
both linguistic and non-linguistic matters, Eva Larsson Ringqvist and Olof
Eriksson, for supplying me with many insightful comments on my texts, and
Marianne Sandberg, who shared office, thoughts, tears and laughs with me
during the last hectic period of writing my dissertation.
¾…Jan Svartvik, yet another excellent role model and advisor of the GramTime project, with whom I have had many rewarding discussions on everything from quantifiers to sailing boats.
¾… my former colleagues at the English department at Göteborg University,
especially Anna-Lena Fredriksson, for (apart from sharing many laughs and
interesting discussions) showing great hospitality during a period when I regularly had to stay the night in Gothenburg to participate in PhD courses.
¾… Maria Gruvstad and Lena Rask at Växjö University for helping me come
back when times were rough.
¾… Satish Patel for reading my manuscript with his keen native speaker eyes
and spotting most if not all of my non-native mistakes. Those still there are
of course my own responsibility.
¾… my wonderful family: my husband Anders, my daughters Sanna and
Maya, my mother Kersti, my sister Helena and my friends and relatives (all
included) for always being there for me when I need you.
My final thanks go to my father Lars, who often talked about his disappointment
at not being able to undertake academic studies himself, but who always supported me in my own studies. Sadly, he is not with us physically anymore, but I
am convinced that, from wherever he is now, he shares my happiness about finishing the PhD project. I dedicate this book to him.
Målajord in March 2004
Maria Estling Vannestål
8
8
Table of contents
1. Introduction .................................................................................................... 13
1.1 Background ..................................................................................... 13
1.2 Aims ................................................................................................ 14
1.3 Outline of the thesis......................................................................... 18
2. Syntactic variation in English......................................................................... 19
2.1 Problems and perspectives .............................................................. 20
2.1.1 The meaning constant ............................................................. 20
2.1.2 Two forms = two meanings? ................................................... 20
2.1.3 Knock-out effects..................................................................... 23
2.1.4 Fixed expressions .................................................................... 25
2.1.5 The interplay of factors............................................................ 26
2.2 Non-linguistic factors: region and medium..................................... 27
2.2.1 American, British and Australian English ............................... 28
2.2.2 Spoken and written English ..................................................... 30
2.3 Linguistic factors............................................................................. 31
2.4 Language variation and change....................................................... 33
2.5 Summary ......................................................................................... 34
3. English quantified noun phrases..................................................................... 37
3.1 The English noun phrase ................................................................. 37
3.1.1 The parts of the NP .................................................................. 38
3.1.1.1 The head ............................................................................ 39
3.1.1.2 Determiners........................................................................ 40
3.1.1.3 Modifiers and complementation ........................................ 41
3.1.2 The syntactic functions of the NP ............................................ 42
3.1.3 NP or DP? ................................................................................ 43
3.1.4 Some semantic concepts pertaining to the NP ......................... 44
3.1.4.1 Reference........................................................................... 44
3.1.4.2 Definiteness ....................................................................... 45
3.2 Quantifiers and quantification ......................................................... 47
3.2.1 Quantification in theoretical linguistics ................................... 47
3.2.1.1 Formal semantics ............................................................... 47
3.2.1.2 Cognitive linguistics .......................................................... 48
3.2.1.3 Generative grammar .......................................................... 49
3.3 All, whole, both and half.................................................................. 49
3.3.1 Historical background.............................................................. 50
99
3.3.1.1 All and whole ..................................................................... 50
3.3.1.2 Both ................................................................................... 51
3.3.1.3 Half.................................................................................... 52
3.3.2 The syntax of all, whole, both and half.................................... 54
3.3.2.1 Syntactic classification ...................................................... 54
3.3.2.2 Syntactic differences between the quantifiers ................... 55
3.3.3 The semantics of all, whole, both and half .............................. 58
3.3.3.1 Classification in terms of definiteness ............................... 58
3.3.3.2 Totality .............................................................................. 60
Explaining totality..................................................................... 60
Totality vs. large quantity, intensity etc. ................................... 61
3.3.4 Variation patterns – syntactic and semantic aspects ................ 66
3.3.4.1 Variants with and without of ............................................. 66
3.3.4.2 Variants with and without the definite article.................... 70
3.3.4.3 Alternating positions of the quantifier............................... 71
A half vs. half a ......................................................................... 71
Floating quantifiers................................................................... 72
3.3.4.4 Issues pertaining to the choice between all and whole ...... 75
Type of quantity......................................................................... 76
Countability and divisibility...................................................... 77
Animacy..................................................................................... 79
Time division in NPs with temporal nouns ............................... 81
3.4 Summary ......................................................................................... 82
4. Method and material....................................................................................... 83
4.1 A corpus-based approach ................................................................ 83
4.2 Corpus material ............................................................................... 85
4.3 Procedure......................................................................................... 86
4.3.1 Corpus queries ......................................................................... 86
4.3.2 Exclusions................................................................................ 87
4.3.3 Sampling .................................................................................. 89
4.3.4 Data analysis ............................................................................ 91
4.4 Methodological problems................................................................ 95
4.5 Summary ......................................................................................... 96
5. Overall frequency distribution of variants...................................................... 99
5.1 Results ............................................................................................. 99
5.1.1 NPs with common nouns ......................................................... 99
5.1.1.1 NPs with all/whole and common nouns .......................... 100
5.1.1.2 NPs with both and common nouns .................................. 103
5.1.1.3 NPs with half and common nouns ................................... 104
5.1.2 NPs with demonstrative pronouns ......................................... 105
5.1.3 NPs with personal pronouns .................................................. 107
5.1.4 NPs with half, the indefinite article and a singular noun or
numeral.................................................................................... 108
5.1.5 NPs with all/whole and geographical names ......................... 109
10
10
5.1.6 Subgroups .............................................................................. 110
5.1.6.1 NPs with all/whole and temporal nouns .......................... 110
5.1.6.2 NPs with both and nouns for body parts or kinship ........ 111
5.2 Fixed expressions .......................................................................... 112
5.3 Summary ....................................................................................... 114
6. Non-linguistic factors ................................................................................... 115
6.1 Distribution according to region ................................................... 115
6.1.1 NPs with common nouns ....................................................... 115
6.1.2 NPs with demonstrative pronouns ......................................... 117
6.1.3 NPs with personal pronouns .................................................. 117
6.1.4 NPs with half, the indefinite article and a singular noun or
numeral.................................................................................... 118
6.1.5 NPs with all/whole and geographical names ......................... 119
6.1.6 Subgroups .............................................................................. 119
6.1.6.1 NPs with all/whole and temporal nouns .......................... 119
6.1.6.2 NPs with both and nouns for body parts or kinship ........ 120
6.1.6.3 NPs with all/whole and collective nouns......................... 121
6.17 General observations............................................................... 121
6.2 Distribution in spoken and written English................................... 122
6.3 Summary ....................................................................................... 127
7. Linguistic factors .......................................................................................... 129
7.1 Introduction ................................................................................... 129
7.2 Type of central determiner in the NP ............................................ 133
7.3 Factors relating to the NP head ..................................................... 136
7.3.1 Noun vs. demonstrative pronoun as head .............................. 136
7.3.2 Number and countability in NPs with common nouns .......... 137
7.3.3 Divisibility ............................................................................. 138
7.3.4 Animacy................................................................................. 141
7.3.4.1 NPs with all/whole and singular count nouns ................. 142
7.3.4.2 NPs with plural nouns ..................................................... 143
7.3.4.3 NPs with all/whole and geographical names ................... 143
7.3.4.4 NPs with both and nouns for body parts or kinship ........ 145
7.3.5 Natural vs. arbitrary time division ......................................... 145
7.3.6 Type of “head” in NPs with half, the indefinite article and a
singular noun or numeral ........................................................ 148
7.4 The presence of certain elements in the NP or its near co-text ..... 149
7.4.1 Modifiers................................................................................ 149
7.4.2 An adjacent of ........................................................................ 154
7.4.3 Focus markers ........................................................................ 157
7.4.3.1 NPs with all ..................................................................... 158
7.4.3.2 NPs with half ................................................................... 161
7.5 Syntactic function of the NP ......................................................... 163
7.5.1 NPs with common nouns ....................................................... 165
7.5.2 NPs with demonstrative pronouns ......................................... 168
11
11
7.5.3 NPs with personal pronouns .................................................. 169
7.5.4 NPs with half, the indefinite article and a singular noun or
numeral.................................................................................... 170
7.5.5 NPs with geographical names ................................................ 173
7.5.6 Subgroups .............................................................................. 174
7.5.6.1 NPs with all/whole and temporal nouns .......................... 174
7.5.6.2 NPs with both and nouns for body parts or kinship ........ 175
7.6 Summary ....................................................................................... 176
8. Conclusion.................................................................................................... 179
8.1 Summary and visualisation of results............................................ 179
8.1.1 NPs with common nouns ....................................................... 181
8.1.1.1 NPs with all/whole and common nouns .......................... 181
8.1.1.2 NPs with both and common nouns .................................. 184
8.1.1.3 NPs with half and common nouns ................................... 185
8.1.2 NPs with all and demonstrative pronouns ............................. 186
8.1.3 NPs with all and personal pronouns ...................................... 187
8.1.4 NPs with half, the indefinite article and a singular noun or
numeral.................................................................................... 188
8.1.8 NPs with all/whole and geographical names ......................... 189
8.1.6 Subgroups ............................................................................... 190
8.1.6.1 NPs with all/whole and temporal nouns .......................... 190
8.1.6.2 NPs with both and nouns for body parts or kinship ........ 191
8.2 Concluding remarks ...................................................................... 192
Appendices........................................................................................................ 197
A.
B.
C.
D.
E.
Variables ....................................................................................... 197
Exclusions ..................................................................................... 199
Search procedures used with the different corpora ....................... 203
Frequency tables for the analysis of linguistic factors .................. 205
Phi coefficients for the correlations............................................... 215
References......................................................................................................... 221
Index ................................................................................................................. 229
12
12
1. Introduction
1.1
Background
English, like many other languages, contains a large number of seemingly synonymous grammatical structures. Some linguists claim that grammatical synonymy does not exist (see Section 2.1.2), but this is still how language is presented in many school and reference grammars. Consider the following examples
from one of the most influential reference grammars of English: A Comprehensive Grammar of the English Language (Quirk et al 1985:381f).
(1:1) All (of)
the boys want to become football players.
Both (of)
(1:2) All of us
like Peter.
We all
The first example illustrates the fact that of is optional in quantified noun phrases
with all and both preceding the definite article. The second example shows a
case where all can occur in different positions in relation to a personal pronoun.
When there is more than one way of expressing something in a language, we
make a conscious or unconscious choice between variants. In some cases it is
obvious that the preference for one syntactic variant over another has to do with,
for instance, regional or register variation (Quirk et al 1985:16; Trudgill &
Hannah 2002:55ff; Biber et al 1998:5f). In other cases, no such factors seem to
be involved, and two syntactic variants are thus often considered interchangeable:
[…] we may not be able to account always for the choice of one
rather than another linguistic form; we sometimes find DIVIDED
USAGE, a choice between variants, the conditions for which cannot be attributed to the variety distinctions discussed in this
chapter [i.e. region, social group, field of discourse, medium and
attitude] […] Neither member of such pairs is necessarily linked to
any of the varieties that we have specified. (Quirk et al 1985:31)
13
13
School grammars are often frustrating for non-native speakers of English desiring to learn which variant to use in a particular situation. For the most part, little
or no information is provided concerning aspects such as frequency, regional
and stylistic variation, or linguistic factors influencing the variation. Comprehensive reference grammars like Quirk et al (1985) and Huddleston & Pullum
(2002) offer more than most school grammars do; nevertheless, there are cases
where the information is not sufficient. A likely reason for these shortcomings is
that authors are not aware of the details of the variation.
Corpora of authentic text are a convenient tool for the study of language
variation in cases where intuition does not give enough information1 (cf. Biber &
Finegan 1991 and Biber et al 1998). Clearly, corpora are useful for investigating
non-linguistic factors such as region, medium and register influencing the variation. Furthermore, several linguists have shown that they can also be used for
detecting more subtle linguistic patterning in language (e.g. Sinclair 1991 and
Biber et al 1998:5).
In recent years, several areas of syntactic variation have been studied by
means of text corpora.2 This study focuses on one small area of English
grammar: a number of noun phrase types with all, whole, both and half. These
words are generally referred to as “quantifiers”, and the area contains several
interesting types of syntactic variation. Quantification has often been explored
by theoretical linguists, but there are few empirical studies that describe how
these words are used in Present-day English. For two exceptions, see Kennedy
(1987) on quantification in relation to English language learning and J. Hudson
(1998) on fixed expressions with all. To my knowledge, there are no corpusbased variation studies involving these quantifiers.
Besides supplying quantitative data on frequency distribution overall and
across different dialects and media, the present study will also look at how various linguistic factors (located in the NP itself or in its co-text3) can be involved
in the choice between two or more different variants.
1.2
Aims
The overall aim of the study is to investigate syntactic variation in some Presentday English noun phrase types including all, whole, both and half. More specific
aims are accounted for below in this section (in the form of research questions).
An additional aim is to test a methodology for investigating syntactic variation.
The underlying assumption is that although the NPs in focus are near-synonymous grammatical structures, there are a number of linguistic and non-linguistic
factors which influence the variation.
1
2
3
14
The advantages and problems of corpus linguistics are discussed in Section 4.1.
A few examples are presented in Sections 2.2 and 2.3.
I prefer to use the word ‘co-text’ rather than ‘context’ (following Brown & Yule 1983:46), since
the latter can be ambiguous. ‘Co-text’ refers to words in the textual surroundings of a word or
structure, whereas ‘context’ is used by some linguists with this sense, by others to refer to a particular situation or culture.
14
Henceforth, we will follow the standard terminology used in syntactic variation research and speak about “variables” and “variants” to refer to the NPs investigated in the present study, as illustrated by Figure 1.1.
VARIABLE:
all + GEOGRAPHICAL NAME
VARIANT:
all (Cuba)
VARIANT:
all of (Cuba)
VARIANT:
the whole of (Cuba)
Figure 1.1. Example of variables and variants in the study
“Variant” is used to describe one of two or more alternative constructions, e.g.
[all + GEOGRAPHICAL NAME] (as opposed to [all of + GEOGRAPHICAL NAME] and
[the whole of + GEOGRAPHICAL NAME]), and “variable” refers to a group of such
variants.
The following four variation patterns were observed among the variables
investigated:
•
•
•
•
patterns resulting from the presence or absence of the preposition of 4:
e.g. all the children vs. all of the children
patterns resulting from the presence or absence of the definite article:
e.g. both children vs. both (of) the children
patterns resulting from alternative positions of the quantifier: e.g. half
an hour vs. a half hour
patterns resulting from the use of alternative lexical elements: e.g. all
the book vs. the whole book
The last of these four patterns is a case of lexical rather than syntactic variation.5
The inclusion of whole was necessary, however, so as to give a comprehensive
description of cases like all the/all of the/the whole book. A complete list of the
noun phrase types investigated can be found in Appendix A. It was compiled by
consulting a number of school and reference grammars as well as more popular
usage guides. The list includes cases where the head of the NP is a common
noun, a geographical name, a pronoun (demonstrative or personal) or a numeral.
4
5
Whether of is really a preposition or not has been discussed by, for instance, Sinclair (1991:82f).
See further Section 3.3.4.1.
Entire was excluded on the basis that it is less frequent, less neutral and seems to be less often
perceived as a function word, compared to whole. This decision was strengthened by the fact that
it is seldom described as a function-word alternative to all, whereas whole is often included in
grammars (see, for instance, Quirk et al 1985:259f).
15
15
The following research questions will be addressed:
1. What difference in overall frequency distribution can be found between the
variants?
2. Which of the following non-linguistic factors influence the variation and how
strong is their influence?
•
Regional variation: Is there a difference in the distribution of variants
between British, American and Australian English?
•
Medium: Is there a difference in the distribution of variants between
spoken and written English?
3. Which of the following linguistic factors influence the variation and how
strong is their influence?
•
Type of central determiner: Does the NP contain the definite article
(e.g. all the books), a demonstrative determiner (e.g. all these books) or
a possessive determiner (e.g. all my books)?
•
Factors relating to the NP head:
(a) Noun vs. demonstrative pronoun: Is the head of the NP a noun (e.g.
all these children) or a demonstrative pronoun (e.g. all these)?
(b) Number and countability: Is the head of the NP a singular count noun
(e.g. half the book), a mass noun (e.g. half the butter) or a plural noun
(e.g. half the children)?
(c) Divisibility of the noun in NPs with singular count nouns: Is the noun
divisible into equal parts (e.g. all the family) or indivisible (e.g. all the
valley)?
(d) Animacy of the noun: Is the head of the NP animate (e.g. all the family) or inanimate (e.g. all the book)?
(e) Natural vs. arbitrary time division in NPs with temporal nouns: Is the
noun a result of natural time division (e.g. the whole year) or arbitrary
(man-made) time division (e.g. the whole century)?
(f) Type of “head”6 in NPs with half and the indefinite article: Is the
“head” of the NP a noun expressing time or space (e.g. half a
day/mile), a noun for partitive relations (e.g. half a pint of beer), a
numeral or noun to do with figures (e.g. half a dozen) or another word
not regularly associated with measurement (e.g. half a victory)?
6
16
See Section 7.3.6 for an explanation of the inverted commas around “head”.
16
•
The presence of certain elements in the NP or its near co-text:
(a) Modifiers: Does the NP include a modifier (e.g. all the small children)?
(b) An adjacent of: Is there another of in the NP (e.g. all the different
kinds of people) or immediately preceding it (e.g. the importance of
all the regulations)?
(c) Focus markers7: Does the NP contain an element which gives more
focus to the totalising function of the quantifier (e.g. nearly all the
children)?
•
Syntactic function of the noun phrase: Is the NP a subject (e.g. All the
books were good), a direct object (e.g. I bought all the books), a prepositional complement (e.g. I gave extra handouts to all the students) etc.?
The linguistic factors analysed were arrived at partly by intuition, partly by consultation of secondary sources. We will gradually move from NP-internal to NPexternal factors. It should be observed that in some cases a linguistic factor is
relevant to all the variables under investigation, whereas in other cases only one
or a few variables are involved (see further Chapter 7).
The focus is on variants that are presented as synonymous or nearly synonymous in the literature. In other words, the choice between variants should not
cause any vital change in propositional content. The study could have been extended to include words like either, neither and each (which indeed exhibit
variation similar to all, whole, both and half: e.g. either boy – either of the boys).
Instead, at an early state I opted for greater detail by limiting the scope to a few
words. The variants studied are part of standard English, even though this concept is not unproblematic (see Section 2.2). Some non-standard variants will also
be accounted for briefly, but not included in the figures on distribution.
The study is mainly descriptive and aims at being relatively theory-neutral,
but not theory-independent. This implies that it does not adhere to one particular
theory, but, when useful, draws on previous work within different syntactic and
semantic traditions8. The study is located, however, within two methodological
frameworks: syntactic variation research and corpus linguistics. The study looks
at one particular language (English), even though it makes some comparisons
with other languages, mainly French and Swedish. Besides improving our general knowledge about syntactic variation in quantified noun phrases, the results of
my analyses will hopefully contribute to the improvement and updating of teaching and reference materials. Some teachers dismiss variation studies as some7
8
“Focus marker” is a term used in the present study to refer to elements, e.g. other quantifiers and
approximators, that tend to give more focus to the totality meaning of NPs with all and to the
measuring function of NPs with half. See further Section 7.4.3.
Besides the more scholarly based literature, a few more popularly written usage guides (e.g.
Swan 1995 and Berry 1997) proved useful since they are influential and contain some interesting
claims on usage.
17
17
thing which complicates their work by showing that everything is possible (cf.
Aarts 1976:246; Rydén 1979:14). However, as pointed out by Rydén (1979:14):
“our aim is of course not to show that everything is possible, but to show when
and how often (in terms of relative frequencies) a specific variant is used, i.e. the
contextual significance or appropriateness of syntactic variants […]. Schoolgrammars must necessarily supply oversimplified rules […], but the formulation
of these rules must be founded on the delicacies of syntactic research.”
The corpus material comprises British, American and Australian newspaper
text and British and American conversation. For the study of overall frequency
distribution and the non-linguistic factors (region and medium), all of the material was used. For the more in-depth study of linguistic factors (such as the presence of a modifier in the NP or the syntactic function of the NP), samples were
extracted from the total corpus population. The tokens in these samples were
then categorised according to two or more categories for each linguistic factor.
All correlations (both with non-linguistic and linguistic factors) were tested for
significance and the results of this testing were also used for a ranking of the
strength of each factor category with respect to each of the variants. The material
and methodology are described in more detail in Chapter 4.
1.3
Outline of the thesis
This first chapter serves as an introduction to the study, providing a general
background, the overall aim and more specific research questions. The next
chapter offers a description of the area of syntactic variation in English. Chapter
3 gives an overview of how the English noun phrase and especially the quantifiers all, whole, both and half have been treated in the literature, including a brief
historical background to each of the words. Chapter 4 provides the prerequisites
for the corpus-based analysis by discussing the use of corpora, describing the
corpus material and accounting for the procedures used in the corpus searches,
sampling of data, analysis of factors and ranking of factor strength. The actual
results of the analysis are presented in Chapters 5 to 7. Chapter 5 gives the
overall frequency distribution of the variants for each variable and Chapter 6
describes their distribution according to region (British, American and
Australian English) and medium (speech and writing). Chapter 7 presents and
discusses a number of correlations between linguistic factors and variants.
Throughout Chapters 5 to 7, numerous tables, graphs and examples are used for
illustration of the result. A visualisation of the results including a ranking of the
correlations is presented in Chapter 8, together with some concluding remarks on
the usefulness of the study and ideas for future research. Finally, the appendices
provide (i) complete lists of the variants studied and (ii) complete lists of
exclusions, (iii) more specific information about the search techniques used for
each of the different corpora, (iv) tables showing the figures from the analyses of
linguistic factors and finally (v) tables showing the results of significance tests.
18
18
2. Syntactic variation in
English
Modern linguistics has often focused on the discreteness of linguistic categories
and claimed that variation and frequency of use can be discarded as irrelevant to
the study of linguistic competence (Bod et al 2003:1). However, language variation has been given a great deal of attention since the 1960s, especially through
the sociolinguistic research9 carried out by, among others, William Labov, David
and Gillian Sankoff, Henrietta Cedergren and Peter Trudgill (Rydén 1979:5;
Quirk 1995:1). In the beginning, the focus of variation research lay heavily on
phonology, but the area was later extended to include syntactic variation (cf. G.
Sankoff 1973). Many variation studies have been framed within sociolinguistic
theory, where the focus has been on social and stylistic factors. Social differentiation is in fact a frequently used criterion for defining appropriate variables
(Coveney 1996:53). During the 1970s and 1980s, a project initiated at Stockholm University, Syntactic variation in English, put the focus on quantitative linguistics and (to a great extent) on linguistic factors influencing variation (cf. Jacobson 1980b; 1982; 1983; 1986).
Coveney (1996:30) brings up three types of grammatical patterns of variability: “omissible items, alternating items and alternating structures”. The present
study involves all three of these:
(1) omissible items: of and the definite article
(2) alternating items: all vs. the whole
(3) alternating structures: the position of the quantifier (as in half a/an vs. a
half)
Two commonly used methods for variation research are elicitation tests (based
on performance or judgment) and investigations of authentic corpus material (Jacobson 1982). In recent years, the possibility to access corpora has greatly facilitated variation research (see further Chapter 4).
9
For more than thirty years now, the NWAV(E) association (standing for “New Ways of
Analysing Variation (in English)” ) has arranged sociolinguistic conferences on variation (for the
most recent proceedings, see Johnson & Sanchez 2002 and Sanchez & Horez 2003). There is
also a journal devoted specifically to variation study: Language Variation and Change.
19
19
2.1
Problems and perspectives
Variation research has encountered a number of different problems and various
perspectives have been taken. We will here look at some of them, starting with
perhaps the most widely debated issue, the syntactic variable and the meaning
constant associated with it.
2.1.1
The meaning constant
Some scholars, especially within sociolinguistics, have discussed the problems
of establishing a syntactic variable, since, as regards variation, syntax is much
more complicated than phonology (cf. Lavendera 1978). The most important
bone of contention is the nature of the meaning constant against which the variation is considered. Labov (1978:2) used formal semantic terms to refer to meaning, such as “same truth conditions”, but Jacobson (1980a:24), among others,
questions whether truth-conditional equivalence is a sufficient criterion for syntactic variation. Should perhaps two syntactic variants result in sentences having
not only the same propositional content10, but also sharing stylistic, connotative,
emotive and pragmatic meanings (Jacobson 1982:10f)? Romaine (1984:422ff)
has argued for a wider definition, suggesting that same pragmatic meaning (i.e.
two structures having “the same communicative intent”) could sometimes be regarded as the crucial constant. An example is two structures such as I’m cold and
Would you close the window?, which can in some circumstances have the same
communicative purpose, viz. to make someone close a window. Biber et al
(1999) recently defined the variants of syntactic variation as being “optional variants, in the sense that they are nearly equivalent in meaning” (ibid 14) and have
“roughly the same communicative effect” (ibid 6). This implies that two variants
should be near equivalents both from a semantic and a pragmatic point of view.
Jacobson (1980a:26f; 1989:382) argues that each researcher should be allowed
to choose a definition that best fits a certain purpose. The meaning constant issue
is certainly important within many areas of syntax, especially when analysing
fairly different grammatical structures. The starting-point here, however, is that
the variables investigated are presented as such in the grammatical literature.
The way these variables are presented in the grammars is likely to reflect the
way the authors see them as “nearly equivalent in meaning” and having “roughly
the same communicative function”.
2.1.2
Two forms = two meanings?
A related topic concerns the questioning by some scholars of the very existence
of two synonymous grammatical structures (two forms – one meaning) in a language. As Biber et al (1998:77) put it:
10
20
Other terms used to refer to the same thing are, for instance “descriptive synonymy” and same
“conceptual, cognitive or denotative meaning” (Coveney 1996:52).
20
An obvious question from a use perspective is why a language
should have structural alternatives with similar or equivalent
meaning. That is, what different purposes do they serve, and how
does a speaker know when to use each option?
Bolinger (1977:1ff, 19) declared that if two contrasting syntactic forms have survived in a language, they must have different semantic meanings11. What he
turned against most of all was the transformationalist view that two surface
structures, such as passive and active voice, have the same underlying deep
structure. Instead, he points to several facts that undermine the idea of synonymy
in the passive–active distinction and similarly with some other supposedly synonymous structures, e.g. relative that/which/zero, to-infinitive vs. -ing form and
the -one/-body ending in pronouns. Bolinger admits that not all cases of variation
are easily dismissed as “non-synonymous”, especially not ellipsis and pronominalisation. One example is the two utterances Yes, I would and Yes, I would like
to have some, used as replies to the utterance Would you like to have some tea.
In these cases the difference is pragmatic rather than relating to “the inner structure of the sentence” (Bolinger 1977:5).
In maintaining that no grammar lacks meaning, cognitive grammarians like
Langacker (1987:39, 1999:76) adhere to the principle that two different syntactic
structures have different meanings, even if they have the same truth condition.
For instance, since every grammatical item has meaning, there must be a difference between a structure with of and one without (Langacker 1991b:112f and
also Sapir 1930:11, see further Section 3.3.4.1). Langacker views syntactic
variation as “alternate ways of mentally construing the same objective circumstances” (Langacker 1988:7), i.e. situations being presented from different perspectives or with different foci (Langacker 1987a:39). Consider the following
examples, based on Langacker (1988:8):
(2:1) (a) All cats are
(b) Any cat is
(c) Every cat is
(d) Each cat is
playful.
In one way, these four sentences mean the same thing, the characteristic “playfulness” being applied to all members of the species “cats”. However, Langacker
observes subtle differences in what images are mentally accessed by the speaker
in the four cases:
All refers to the class collectively, as an undifferentiated mass.
[…] The other three quantifiers each refer to a single, arbitrary
class member, but this member is conceived as being selected in
such a fashion that the property attributed to it is similarly attributed to all the other members. The image conveyed by any is one
11
Bolinger (1977:2) uses a cognitive explanation (“the mind is freer than the tongue”) to account
for the fact that the opposite case (i.e. polysemy, several meanings attached to the same word or
structure) is much more frequent.
21
21
of random selection: if one chooses a member at random, it will
invariably display the property in question. Every and each are
alike […] in attributing the property to the full set of class members on an individual […] basis. The difference between them is
that each further suggests that the members are examined sequentially – one at a time – for this purpose. (Langacker 1988:8).
An interesting question arising from the idea that two grammatical structures are
never synonymous is why two different regional varieties can prefer different
grammatical structures (for examples, see Section 2.2.1). Does this mean that,
for instance, an American and a Briton conceptualise the world in different
ways? Langacker (1988:38) believes that individual speakers within each variety
may very well perceive a subtle meaning distinction between two forms and use
them under different circumstances. That there are differences in frequency distribution between varieties has to do with conventionalisation:
This merely reflects the imagery embodied by the symbolic
resources of a language: out of all the ways of constructing a give
type of situation, certain possibilities become conventionally
established (i.e. represented in the grammar by symbolic units) to
the exclusion of others. (ibid)
Therefore, even though two different structures are equally natural from a cognitive point of view, one form can, due to convention, become the preferred
alternative in one particular variety; consequently another form is dispreferred.
Still, Langacker argues that there are subtle meaning differences between structures: “speakers of the two dialects conventionally employ strictly different
images to construe the situation for expressive purposes” (ibid). Actually, this
should not be more surprising than the fact that two languages can have different
conceptualisations of such things as colours and words describing family members. An example of the latter is that English has one single word for the son of
either one’s sister or one’s brother, whereas these two concepts are expressed by
two different words in Swedish.
Goldberg (1995:3) mentions some other scholars within various linguistic traditions who have also stated that they are in favour of the “Principle of No Synonymy of Grammatical Forms”, e.g. Givón (within functional grammar), Wierzbicka (within cognitive semantics) and Goldberg herself (within construction
grammar).
The discussion of grammatical synonymy is relevant to my research, since the
study includes cases of two variants which have been claimed by some people to
be semantically different but which exhibit fairly great regional variation in the
material (e.g. a half vs. half a/an). We will look at these cases in Chapter 7 and
finally evaluate the idea of grammatical synonymy in Section 8.2. Also, how
should cases where there is a correlation between a variant and a factor category
in one regional variety but not in the other be dealt with? In the present study,
such cases are accounted for in tables in the presentation of speech vs. writing,
since here the whole corpus material was used and the correlations were considered relatively reliable. In the case of linguistic factors, however, only small
22
22
samples of the whole material were used. To ensure (to as large an extent as possible) that the correlations reported on had not occurred by chance, I present
only those found in both British and American English in the tables and graphs.
Correlations found in just one variety are only mentioned in the running text.
2.1.3
Knock-out effects
In a treatment of syntactic variation two important concepts that put syntactic variation out of play cannot be ignored: knock-out effects/constraints and fixedness. I will discuss these concepts in the following section. Knockout effects are
cases where a factor “knocks out” the variation, resulting in a 100–0 relationship
between variants and factors (Jacobson 1980a:28; Tottie 1991:62ff). A case of
variation where two knock-out factors (one in each direction) are involved is described by Jacobson (1980a:32). He looked at variation between pre- and postauxiliary placement of the adverbial probably and found that when a clause
included contracted not, as in (2:2), probably was always used in pre-auxiliary
position in his material. On the other hand, in the case of a contracted finite auxiliary, as in (2:3), post-auxiliary placement was the only variant occurring. The
examples are taken from my corpus material, since Jacobson does not provide
full example sentences.
(2:2) As a result, it probably can’t be done quickly. (NYT9512)
(2:3) As long as bank managers have confidence in you, they’ll probably be
pretty co-operative. (IND95)
Another example is taken from Tottie (1991:63) and concerns the choice between affixal and non-affixal negation (e.g. impossible vs. not possible). Here
she found that when the negation is located in a premodification, the affixal variant is the only possible one:
(2:4) At the core of the problem is a political question: how to make choices
within an imperfect society. (ibid)
Should tokens which are the result of a knock-out effect be included in a frequency study of syntactic variation? Jacobson (1980a:28) argues that perhaps
they should, since they represent, so to speak, one end of a continuum13. One
12
The abbreviations of corpora used in the study are explained in Section 4.2. In this and the following chapter, the illustrating examples are, for practical reasons, taken from two of the corpora
used, The Independent 1995 (henceforth IND95) and The New York Times 1995 (henceforth
NYT95). In the presentation of overall distribution (Chapter 5), all corpora used in the study are
represented in the examples, whereas in Chapter 7, all the examples are from IND95 and NYT95
(since only samples from those corpora were analysed in that chapter).
13
Jacobson (ibid) also observes that what the researcher presumes to be a clear knock-out factor at
the outset of an investigation may turn out not to be entirely watertight, i.e. may not give a 100–0
relationship.
23
23
could imagine a scale from necessity to impossibility, with different degrees of
probability constituting the intermediate area14 (see Figure 2.1).
high degree of
probability
of Variant X
low degree of
probability
of Variant X
necessity
of Variant X
Factor A
impossibility
of Variant X
Factor B
Factor C
Factor D
Figure 2.1. Variation pattern for an imagined variant (X).
What the figure shows is that in the presence of Factor A, the particular variant
at issue is necessary and alternative forms are impossible. At the other end, in
the presence of Factor D, Variant X is not possible. In between the two endpoints, in the presence of Factor B, the probability of Variant X being used is
lower than in the presence of Factor A, but higher than in the presence of Factor
C, and so forth. Factors A and D are knock-out factors here, the former in necessitating Variant X and obstructing another variant and the latter in obstructing
variant X and necessitating another variant.
Even though Jacobson (1980a:28) remarks that knock-out effects should be
visible in an analysis of different factors, he reports on excluding them from the
statistics in his own work (e.g. Jacobson 1982:7). This seems quite reasonable,
since the correlation percentage is always 100–0 in those cases. D. Sankoff
(1978:66) also writes that it is “necessary to identify knockouts and to remove
data pertaining to them from the data set, prior to further statistical analysis.”
There were a few instances of knock-out effects in my material. One clear case
appeared in the variable with all/both and personal pronouns (e.g. all of us vs.
we all). In cases when the NP was used in a minor clause (e.g. an apposition or
short answer), the form with all of, as in (2:5), was always used.
(2:5) Not, of course, that Sir Robin will be sent back to go 12 rounds with Lu
Ping again; but that he – indeed all of us – will one day soon have
undreamt-of opportunities to regret the deal that was done in 1984.
(IND95)
In the present study, all tokens resulting from knock-out effects will be accounted for and discussed, but they will be excluded from the statistics (see further Sections 4.3.2 and Appendix B).
14
24
For a thorough and recent discussion of probabilistic linguistics, see Bod et al (2003).
24
2.1.4
Fixed expressions
Another case where variation is obstructed is when particular phrases have frozen to become fixed idiomatic expressions. These expressions are often accounted for in dictionaries and comprise, for instance, phrasal verbs (to stick
out), conversational formulae (How do you do), similes (as good as gold) and
proverbs (a stitch in time saves nine) (Aijmer 1996:1ff; J. Hudson 1998:1515;
Moon 1998:2, 62).
J. Hudson (1998:8f) refers to four common ways of classifying “fixed expressions”16:
•
•
•
•
syntactic variability restrictions: e.g. the other day – *the other days
collocational variability restrictions: e.g. disaster area – *catastrophe
area
anomalous syntax or usage: e.g. all of a sudden (the adjective sudden
used as a noun)
figurative meaning: e.g. a hot potato
She also points out that the first two of these are the most generally applicable
criteria. Figurative meaning is also a very common characteristic of fixed expressions; however, J. Hudson regards this as an underlying conceptual phenomenon as opposed to the other criteria which are rather “symptoms” of fixedness.
In a study of syntactic variation, the first criterion is naturally the most relevant
one, even though collocational variability restrictions and figurative meaning
usually occur as well.
An example of a fixed expression from my material is the phrase all the rage,
as in (2:6), which fulfils at least three out of Hudson’s four criteria.
(2:6) Tax efficiency is all the rage when picking mutual funds. (NYT95)
There is no syntactic variability (*all of the rage), no collocational variability
(*half the rage) and the expression has figurative meaning (‘very fashionable’).
Furthermore, this NP has predicative function, which is not anomalous syntax
but an infrequent function of an NP with all (at least according to my material,
cf. Section 7.5.6.1). At an early stage, I decided to exclude such expressions
since they could contribute to a skewed picture of the variation. In contrast to the
case of knock-out factors, these expressions do not belong at the endpoint of a
variation continuum, since they concern particular expressions rather than whole
variants. They need not have been excluded, however, since they did not affect
the frequency distribution as much as expected (see Section 5.2).
Moon (1998:120ff) notes that fixed expressions are often not at all as fixed as
one might think; there is certainly variation here as well, which becomes very
15
Interestingly, from the perspective of this study, J. Hudson has looked at fixed expressions with
all. We will have reason to come back to her investigation on several occasions below. Moreover, Aijmer (1985) has studied conversational routines including all, such as and all that used
as a vague tag preceding a possible boundary in the discourse.
16
Cf. also Aijmer on fixedness criteria that particularly concern conversational routines, such as
How do you do (1996:12ff).
25
25
clear when corpus material is used17. She even found variation in such a textbook example of fixedness as the phrase kick the bucket (kick the pail and kick
the can). Moon (ibid 132ff) further shows that there is dialectal as well as Register variation and that some variation is more institutionalised than other. The
variation patterns exemplified by Moon mainly involve lexis. In many cases, the
variation is restricted to one or a few words (e.g. a skeleton in the closet/cupboard) (ibid 124f). In other cases, the fixedness rather concerns a kind of frame
with one or more context-dependent open word slot(s) which can be filled with
basically any word (or phrase) as long as it corresponds to a particular grammatical element, such as a noun phrase, as in X catches Y red-handed (ibid 145ff).
In a few cases, expressions that are presented as fixed in dictionaries turned
out to exhibit syntactic variation in my corpus material (see Appendix B). All the
time, for instance, is often considered to be fixed, but all of the time, as in (2:7),
and the whole time occurred several times in the material.
(2:7) They’re segregating us all of the time. (LSAC).
2.1.5
The interplay of factors
The perspective taken in the present study is (i) that there are grammatical structures which have nearly equivalent propositional content and roughly the same
communicative function, and (ii) that linguistic and non-linguistic factors influence the choice of variant. Rydén (1979:8) particularly emphasises the “intimate
interplay” of such factors. Jacobson (1982:6), among others, is of the opinion
that variation studies should not only aim at describing frequencies, but also “estimate the probability of different variants in new sentences resembling those of
the corpus” by means of “variable rules”. Labov coined this term and used it in
close adherence to Chomsky’s early generative grammar, in an attempt at explaining how such a model could account for different kinds of syntactic variability (Romaine 1984:414; Winford 1996:177; Bod 2003:106). D. Sankoff
(1988:984) provides the following outline:
Whenever a choice among two (or more) discrete alternatives can
be perceived as having been made in the course of linguistic performance, and where this choice may have been influenced by factors such as features in the phonological environment, the syntactic context, discursive function of the utterance, topic, style, interactional situation or personal or sociodemographic characteristics
of the speaker or other participants, then it is appropriate to invoke
the statistical notions and methods known […] as variable rules.
A variable rule is thus optional as opposed to ordinary syntactic rules, but generally the variation is not quite random since various factors “make the application of the rule more or less likely” (Paolillo 2002:33). One frequently used sta17
26
Of course, if we think of language in probabilistic terms, there is no reason why fixed expressions should be any different from other areas, where absolute categories are exceptions rather
than the norm.
26
tistical method for analysing the interplay and for estimating the degree of influence of individual factors is referred to as “factor analysis” (ibid 7f, 32ff).
Some scholars, especially within sociolinguistics, have performed advanced
factor analysis by means of a statistical computer program from the VARBRUL
family18, developed by Pascal Rousseau and David Sankoff (see D. Sankoff
1988, Bod 2003:107). Tottie (1991), for instance, used VARBRUL in her study
of negation in English speech and writing, which involved a number of syntactic
and lexical factors influencing the choice between using a no-negation and a notnegation. Another example is Gries (2001), who used factor analysis to explore
the interplay of linguistic and non-linguistic factors influencing the choice of
variant in English transitive phrasal verbs (e.g. to pick up the book vs. to pick the
book up).
The empirical analysis carried out in this study could be followed up by a
factor analysis of the VARBRUL kind. However, here, the interplay of factors
will only be presented in terms of the relative weight of the different correlations
between factors and variants. Such information was obtained from the chi-square
test used for estimating statistical significance of correlations (see further Sections 4.3.4 and 8.1). I will also present a few cases where two factors probably
interact, so that the correlation with one factor is related to the correlation with
another factor.
2.2
Non-linguistic factors: region and
medium
Several non-linguistic factors (e.g. region, medium and style) can influence syntactic variation. An important issue in the discussion of variation is standard versus substandard language, especially since the variables investigated here are
part of standard English. With a language as globally spread as English, it is difficult to clearly define standard English, but many people informally agree upon
the existence of standard BrE and AmE varieties (cf. Quirk 1995:24). It is also
possible to find references to (more or less formally defined) standard varieties
of, for instance, Australian and New Zealand English (cf. Quirk 1995:24; Hundt
1998b; Blair & Collins 2001).
As Bauer (1994:3) remarks, “it seems to be widely accepted that a standard
requires a certain amount of codification”. English spelling and grammar was
first codified in Samuel Johnson’s dictionary from 1755 (Melchers & Shaw
2003:5). However, attempts at establishing language academies in Britain and
America for “providing direction toward a standardized model and toward controlling language change” have not been successful (Kachru 1992b:49). Still,
widely consulted dictionaries, pronunciation guides, usage handbooks, descriptive grammars and teaching materials for EFL learners have served a purpose
similar to the French and Swedish academies, for instance, since many people
see them as authorities (Kachru 1992b:50, Bauer 1994:3). Interestingly, the concept of standard does not always correspond to frequency of use. For example,
18
The newest variable rule program, based on programs previously circulated by David Sankoff
and others is called Goldvarb (http://www.york.ac.uk/depts/lang/webstuff/goldvarb/).
27
27
the “standard” British dialect, referred to as Received Pronunciation (RP), is
only spoken by three to five per cent of the English population (Trudgill &
Hannah 2002:9). The concept of standard English has been much debated in
terms of power and attitude, and in association with models for the teaching of
English (Kachru 1992a:8, 1992b:48ff). In the present study, we will not go into
such discussions, but only concern ourselves with variation in standard English
as it is described in the above-mentioned influential comprehensive reference
grammars like Quirk et al (1985).
2.2.1
American, British and Australian English
From being an exclusive property of a few million people in the time of Shakespeare, English is today the most widely spread language in the world, not because of any intrinsic superiority, but rather because of extralinguistic factors
(Kachru 1992a:10f; Quirk 1995:3f). As a natural consequence of its wide distribution, the language is characterised by extensive regional variation. Kachru
(1985:12f) proposed a model of regional varieties of English consisting of three
circles, the inner circle (countries where English is the dominant language, e.g.
the U.S. and UK), the outer circle (former British colonies where English is a
second, official language, e.g. India and Singapore) and the expanding circle
(countries where English is taught as a foreign language, e.g. Sweden and the
Netherlands). Studies of regional variation in English have largely concentrated
on the two largest standard varieties from the inner circle: American English and
British English, henceforth AmE and BrE (cf., for instance, Kachru 1992a:3).
Dictionaries and other literature on differences between AmE and BrE often
give the impression that such differences are absolute (cf. Strevens 1978; Algeo
1988; Crystal 1995 and Modiano 1996). However, some authors, such as Görlach (1991:25) and Svartvik & Sager (1996:2), emphasise the fact that differences are questions of tendencies, frequencies and sometimes the level of formality. It is also important to remember that there is a great deal of variation
within the varieties, in fact more than between them (Quirk 1995:7).
Differences between AmE and BrE are often described as mainly concerning
phonetics, lexis and orthography. Grammatical differences are not very numerous and are often considered to be of little importance, since they seldom lead
to complications of understanding (Quirk et al. 1985:19; Biber 1987:99; Algeo
1988:2; Trudgill & Hannah 2002:55; Melchers & Shaw 2003:21). Nevertheless,
scholarly investigations have studied particular linguistic phenomena such as
verb complementation (Mair 1990 and 1995), NP modification (Jucker 1992)
and mandative subjunctives (Övergaard 1995; Hundt 1998a). Others take a more
general approach (e.g. Strevens 1978; Johansson 1980; Algeo 1988; Trudgill &
Hannah 2002 and Tottie 2002). This study goes slightly beyond the standard format, as it also comprises Australian English. The following list exemplifies some
typical grammatical differences between AmE and BrE:
28
28
•
•
•
•
•
the more frequent use of the modal auxiliary shall in BrE (Johansson
1980)
the more frequent use of mandative subjunctives in AmE (Övergaard
1995; Hundt 1998a)
the more frequent use of a bare infinitive (rather than a to infinitive) as
complement of the verb help in American English (Mair 1995)
the more frequent use of plural agreement with collective nouns in BrE
(Levin 2001).
the more frequent use of the definite article before words like university
and hospital in prepositional phrases like at /the/university in BrE
(Tottie 2002)
Tottie (2002) in her recent comprehensive description of American English devotes some thirty pages to grammatical differences between BrE and AmE.
Biber brings in a stylistic perspective, as illustrated by the following quote.
[…] American written genres are consistently more colloquial and
involved than British written genres, while at the same time American written genres are consistently more nominal and jargony
than British genres. (Biber 1988:201)
Lexis, pronunciation and spelling tend to be more dialectally consistent, with
one form often predominating in BrE and another one in AmE (e.g. the use of lift
in BrE and elevator in AmE, or the spelling -our in BrE vs. -or in AmE). With
syntactic variation, the same form is often predominant in both varieties, whereas the alternative form (or forms) mainly occurs in one of them. One example is
agreement with collective nouns, where singular agreement (the family is) is predominant in both BrE and AmE. Meanwhile, the alternative, plural agreement
(the family are), is more frequent in BrE than in AmE (cf. Levin 2001).
Australian English (henceforth AusE) has only been recognised as an independent variety for a few decades (Peters 2001:163). Australia was populated
mainly by people from the British Isles in successive waves of immigration.
Consequently, BrE, especially dialects spoken in the South-East of England
towards the end of the 18th century, constitutes the origin of AusE (Peters & Fee
1989:135, 143; Blair & Collins 2001b:1), even though today, AusE is not the
same as BrE (Peters & Fee 1989:135, 143). AusE has developed on its own and
has been influenced by AmE via military activity, tourism and media. Thus,
present-day AusE is a mixture of AmE and BrE features (ibid 136, 146), which
is particularly clear when it comes to spelling (Butler 2001:160). The same pattern is reflected in prepositional variation (Estling Vannestål 2001a).
AusE differs most from BrE and AmE in its pronunciation and vocabulary.
According to Newbrook (2001:113), AusE has virtually the same syntactic
norms as BrE. Some “informal” syntactic features are reported to be more readily accepted in AusE than in other standard varieties, e.g. contractions and get
passives in writing (Peters 2001:168ff, 175). The dialect also contains some
unique features, such as possessive pronouns in combination with some (e.g. my
some), some shared with AmE, such as epistemic have (got) to, and some shared
29
29
with South-East Asian English, such as scope of negation in clauses with all and
both (see further Section 3.2.1.1) (Newbrook 1992:15f; 2001:119ff, 128f).
The present study tests the validity of claims about regional variation in NPs
with all, whole, both and half. It also looks for hitherto undetected differences.
Finally, it seeks to find out whether AusE resembles BrE or AmE usage more
with respect to the variables in focus. The results are presented in Section 6.1.
2.2.2
Spoken and written English
Biber, in his influential, multi-dimensional description of spoken and written
English (1988:5f), notes that written language was long considered the “better”
form, and the only one worthy of being studied19. Not until the late 19th century
and the rise of phonetics as an independent branch of linguistics did speech become an accepted object of study among scholars such as Jacob Grimm, Henry
Sweet, Ferdinand de Saussure and Leonard Bloomfield (ibid; Chafe 1994:46).
From a historical-developmental point of view speech could easily be argued to
be primary to writing, and linguists like Sapir and Bloomfield emphasised this
aspect (Biber 1988:5f). Chafe (1994:41) regards the prototypical form of speech,
ordinary conversation as “a baseline from which all other uses [of language] are
deviations” (also cf. Halliday 1994:xxiii), even though many people tend to hold
the traditional opinion of writing as a superior form of language. From a
functional point of view, neither could be said to be primary: they are simply
used for different functions (Biber 1988:7; Chafe 1994:45).
Very often, especially before the 1960s, the focus in studies of speech and
writing has been on either spoken or written language, rather than on differences
between them (Biber 1988:6, 47). One way of comparing the two media, or
“modes” as Chafe (1994:41) prefers to call them, is to look at the very acts of
speaking and writing. Chafe (1994:42ff), Röhr (1994:30ff) and Cornbleet &
Carter (2001:74ff) exemplify a number of differences, such as:
(a) the evanescence of speech vs. the permanence of writing
(b) the differences in tempo, speech typically being produced at a higher rate
of speed
(c) the typically private character of speech vs. the typically public character
of writing
(d) the typical spontaneity of speech vs. the typical planning of writing
(e) the degree of interaction and co-operation, speech typically being characterised by the participation of more than one person
(f) the possibility of clarification in speech, usually lacking in writing
Of course, the relevance of these differences is also related to type or register of
speech and writing (see further below). A written note on a fridge door is not
particularly permanent, whereas a recorded political speech certainly is (Cornbleet & Carter 2001:81).
19
30
There are different ways of referring to the two physical properties of communication, e.g.
“medium”, “channel” and “mode”. In this study, we will use “medium” to refer to speech vs.
writing (Pettersson 1996:20).
30
Another way of comparing speech and writing is to find out how they differ
in form. Such studies have frequently used dichotomies, such as “formal” vs.
“informal”. Biber (1988:9ff) suggests using different continua (or “dimensions”)
of variation, rather than absolute dichotomies. Spoken language tends to use a
more informal, involved and interactive style (for instance through the use of
many pronouns), while writing is generally more formal and less involved and
interactive (for instance through the use of more passives).
A natural question arising is whether it is at all possible to define spoken and
written language as two subsets of language in general, as there are so many
different text types or registers within each medium. Biber (1988:24) shows that
differences among different types of spoken or written text, respectively, are
often just as great as differences between the two subsets. He concludes that no
real spoken-written dimension actually exists, even though there are more and
less “typical” or frequent/unmarked/characteristic forms of speech and writing,
such as face-to-face conversation and informational exposition (ibid 37). The
present study compares spoken and written English from a very coarse perspective, without distinguishing between different registers within them. The corpus
material consists of what Biber refers to as “typical” speech (mainly face-to-face
conversation) and writing (newspaper language, that is mainly informational
exposition)20. The results are presented in Section 6.2.
2.3
Linguistic factors
That linguistic factors should influence syntactic variation is perhaps not surprising. Nevertheless, a great deal of such, sometimes quite subtle, patterning tends
to go unnoticed when intuition is relied upon as the sole source of information.
With computerised corpora and concordancing programs, linguistic patterns can
become obvious to the naked eye more easily than before. Such observations
have been made by, among others, Sinclair (1991:6f) and Biber et al (1998:5).
Examples of studies that have reported on linguistic factors influencing variation
are Jacobson (1975) on the placement of English adverbs, Gustafsson (1983) on
English adverbials, Hargevik (1983) on the auxiliary need, Tottie (1991) on negation, Mair (1995) on complementation after the verb help, Biber et al (1998)
on that-clauses, Lindquist (2000) on the comparison of disyllabic adjectives and
Gries (2001) on transitive phrasal verbs.
These are the three most frequently investigated areas of linguistic factors:
(a) lexis: certain words, the presence of which favours one variant
(b) semantics: certain semantic qualities of the variable or something in its
co-text favouring one variant
(c) syntax: certain syntactic features within the variable or in its co-text
which favour one variant
Let us look at a few examples from the areas described above, starting with
lexis. Some of the studies consulted do not provide sentences illustrating the
20
A problem with using newspaper text, however, is that parts of it consist of spoken language (in
the form of interviews) written down.
31
31
phenomena described. Therefore, the exemplified sentences for this part of the
presentation are selected from my corpus material.
With collective nouns, the distribution of singular and plural agreement of
verbs and pronouns is related to the particular noun us. Thus plural agreement is
more frequent with, for instance, the word family, as in (2:8), than with the word
audience, as in (2:9), where singular concord is used (Levin 2001:131).
(2:8) The family hid their icons behind other pictures during the most difficult
years. (IND95)
(2:9) The audience, mostly women, got what it wanted, to a degree. (NYT95)
Similarly, Tottie (1991:250) in her study of not vs. no-negation found that certain high-frequency lexical verbs, viz. give, know, make and the main verb do, as
in (2:10), were used much more often with no-negation than other lexical verbs,
as exemplified by (2:11) where another lexical verb, conceal, is used.
(2:10) Mr Chirac, unlike his Socialist opponent, Lionel Jospin, made no
promise to shorten the presidential term.. (IND95)
(2:11) The company did not conceal any salary, bonus or any other benefit to
which Ms. Potter was entitled. (IND95)
The semantic factor type can be either internal to the variable itself or concern
external aspects, mainly other elements within the clause or sentence in which
the variable occurs. An example of an internal semantic factor is animacy in collective nouns. Levin (2001:128) found that plural agreement is more frequent in
nouns with animate reference, as in (2:12), than in inanimate nouns, as in (2:13),
if words like group and majority are excluded.
(2:12) The crowd, smartly dressed, watch eagerly as two Polish bands perform.
(IND95)
(2:13) This expanding army of barely potable frilly-drinks makes fermented
apple cider taste, by contrast, like ’75 Grange. (SMH95)
External semantic factors can be exemplified by a study of prepositional variation (cf. Estling Vannestål 2001:84). Here it was discovered that in the choice
between outside and outside of, outside of was more frequent if the noun in the
following NP was abstract, as in (2:14), than when it was concrete, as in (2:15),
where simple outside is used.
(2:14) Luckily, the UK’s 23 million volunteers do not appear to be put off by
the lack of participation, outside of politics, by the country’s political
leaders. (IND95)
(2:15) Outside the building Greenpeace demonstrators massed, rolling out a
dummy bomb. (IND95)
32
32
Just like semantic factors, syntactic ones can be divided into internal or external
factors. Variable-internal syntactic factors relate to elements within the variable
and can be illustrated by another example concerning prepositional variation
(Estling Vannestål 2001:81f). In the choice between the two expressions out X
window/door and out of X window/door, the simple out variant turned out to be
more frequent when the determiner in the prepositional phrase was the definite
article, as, in (2:16), rather than a demonstrative or possessive determiner, as in
(2:17), where the complex preposition is used.
(2:16) As it sped past the group, someone leaned out the window with a .45caliber semiautomatic pistol and opened fire into the group. (NYT95)
(2:17) The two convicts, both infected with HIV and described as dangerous,
had been driving around the town, tossing banknotes out of their car
window, before they were seized. (NYT95)
An external syntactic factor can concern (i) another element in the clause or
sentence in which the variable occurs or (ii) a relation between the variable and
the clause or sentence. One example of an external syntactic factor concerns prevs. post-auxiliary placement of the adverbial probably. Jacobson (1989:384) discovered that the structure of the NP in subject position influenced the variation
in that pre-auxiliary position was more frequent when the subject was a pronoun,
as in (2:18), than when it was a full NP, as in (2:19).
(2:18) But the chief thing that it probably has taught me is that we must deal
not only with the supply of drugs but with the demand for drugs.
(NYT95)
(2:19) The new magazine has probably received more prepublication coverage
than most magazines receive in their lifetimes. (NYT95)
A large part of the present study will be devoted to linguistic factors influencing
the variation between two or more syntactic variants. The results are presented in
Chapter 7.
2.4
Language variation and change
Every language spoken by human beings, its pronunciation, lexicon and
grammar, has been changing continuously since its birth, and language change
and variation go hand in hand (Bauer 1994:7, 11). As Coveney (1996:47) puts it,
“since it seems inconceivable that a grammatical change could […] run its
course overnight, from innovation to completion, it must be concluded that all
change is reflected in synchronic variation between two or more items or
structures.” The work of Labov and other sociolinguists has shown that we can
learn about language change by relating synchronic variation to, for instance, the
age of speakers (Bauer 1994:11ff) or region and media (Mair 1998:155). The
latter perspective was the starting-point for the Freiburg project, where Christian
33
33
Mair and his colleagues created two updated “clones” of the well-known British
and American LOB and Brown corpora (FLOB and Frown) with the objective of
finding out how synchronic regional and stylistic variation and diachronic
change interact (Mair 1998:140f, 155). One interesting aspect discussed by Mair
is that there are differences in the speed of change in different media, genres and
registers. Newspapers, for instance, belong to the “fast genres”, where changes
in a language are likely to be reflected before they are reflected in, for instance,
academic prose (ibid 155).
One area of linguistics concerned with language change is grammaticalisation
theory, where linguistic items go through a change from lexical to grammatical
form (function words, clitics and affixes) or from less grammatical to more
grammatical. In the functionally motivated grammaticalisation process (Hopper
& Traugott 2003:2ff), syntactic variation is an important ingredient. The co-existence of syntactic structures (referred to as “layering”) is one of the identifying
criteria for grammaticalisation (Hopper 1991:22f). Hopper (ibid 23) writes:
(…) when a form or set of forms emerges in a functional domain
[e.g. tense/aspect/modality], it does not immediately (and may
never) replace an already existing set of functionally equivalent
forms, but rather the two sets of forms co-exist. They may be specialized for particular lexical items, particular classes of constructions, or sociolinguistic registers; they may have slightly different
meanings, or simply be recognized as ‘stylistic’ alternatives.
One standard example of layering is the two forms going to and gonna
representing different stages of a grammaticalisation process (from the previous
and still existing function as an indicator of direction + purpose to its present
function as a tense marker) and also co-existing with other markers of future,
such as will and shall. The analysis in the present study is little concerned with
language change, hence the brevity of this section. However, Section 3.3.1 will
give a diachronic background to the establishment of the variants in the variables
and present a hypothesis concerning the grammaticalisation of half.
2.5
Summary
This chapter has tried to establish the frame within which the present study is
conducted. It first presented some important problems within variation research.
We saw that the most extensive theoretical discussion has concerned what
should constitute the meaning constant. Should it be “the same propositional
content”, “the same pragmatic function” or something else? The whole idea of
grammatical synonymy has been questioned by Bolinger and Langacker, among
others. My own starting-point for defining syntactic variables in the present
study is their presentation as such in the grammatical literature (esp. Quirk et al
1985). This is taken as an indication that they are “nearly equivalent in meaning”
and have “roughly the same communicative function”, which is Biber et al’s
(1999) definition of a syntactic variable. The variants of a few of the variables
occurring in the study have been claimed by some people to be semantically
34
34
different. This possibility will be further discussed in Chapters 5–7. Two
phenomena complicate the picture further: knock-out factors and fixed
expressions. The first part of the chapter ended with a brief note on the interplay
between different factors.
Section 2.2, on non-linguistic factors influencing the variation, described
regional variation in English and gave a brief outline of research into spoken vs.
written English. In Section 2.3, we turned to linguistic factors and saw some
examples from three areas: lexis, semantics and syntax. Finally, Section 2.4
brought up the correlation between language variation and change. The next
chapter will be concerned with the part of English grammar that is in focus in
this study of syntactic variation: the English noun phrase and especially the four
quantifiers all, whole, both and half.
35
35
36
3. English quantified noun
phrases
In this chapter we will deal with English NPs including the quantifiers all,
whole, both and half. The chapter starts with on outline of the English noun
phrase (Section 3.1) and goes on to briefly discuss research into quantification in
some formal linguistic theories (Section 3.2). Section 3.3 goes through a number
of issues relating to the four English quantifiers in focus, starting with a historical background before exploring the words syntactically and semantically.
Although this presentation is mainly descriptive, some claims will be questioned and also some new hypotheses will be suggested. The information was, to
a large extent, accessed from the three most recent comprehensive reference
grammars, especially for the section on the English noun phrase. A Comprehensive Grammar of the English Language (Quirk et al 1985) is the most widely
consulted descriptive grammar in recent years. The Longman Grammar of
Spoken and Written English (Biber et al 1999) is a corpus-based grammar which
in its framework mainly follows Quirk et al (1985). This grammar takes frequencies in actual texts into account and also brings up variation across registers
wherever such information is relevant. The most recent one, The Cambridge
Grammar of English (Huddleston & Pullum 2002), takes a somewhat more theoretical perspective than the other two, especially using ideas from generative linguistics. I will also, where appropriate, briefly refer to treatments in more theoretical branches of linguistics.
3.1
The English noun phrase
In his analysis of postmodifying clauses in the English noun phrase, de Haan
concisely defines the English noun phrase in the following way:
A noun phrase is a string of words which, syntactically, is a constituent with an internal structure containing a determiner, a modifier
and a head. The head (the only obligatory element in the structure
of the noun phrase) may be a noun or a noun equivalent. Semantically, a noun phrase can be used as a referring expression (de
Haan 1989:8).
Although subject to disagreement, most linguists would probably accept this definition as a useful starting point. One bone of contention concerns headedness,
where many linguists within the generative school argue that determiners rather
37
37
than nouns are the true heads of the kind of phrases usually referred to as noun
phrases (see Section 3.1.3).
The early descriptive grammarians paid rather little attention to units larger
than the word, such as the noun phrase. Instead they focused on individual
lexical items and, if mentioning word groups at all, defined relationships between their different constituents rather loosely (de Haan 1987:8, 24). The more
recent comprehensive reference grammars analyse the syntax of the English
noun phrase in greater detail (see Section 3.1.1).
According to Biber et al (1999:232), noun phrases can be considered to play
the most important parts in language because of their referential specification,
i.e. they “specify who and what the text is about”. This is easily illustrated if we
(i) leave out all noun phrases from a text, as in (3:1), and (ii) leave out all other
elements except the noun phrases, as in (3:2). The text is the introductory paragraph of this book.
(3:1) […] like […] contains […] claim that […] does not exist (see further […])
but still […] is how […] is presented in […]
(3:2) English […] many other languages […] a large number of seemingly synonymous grammatical structures. Some linguists […] grammatical synonymy […] Section 2.1.2 […] this […] language […] many school and
reference grammars.
It is only in the second case that some of the content would be at least possible to
grasp. Also, the NP is the phrase type that can take on the largest number of different grammatical functions (Quirk et al 1985:60ff, 657) (see Section 3.1.2).
3.1.1
The parts of the NP
In Quirk et al (1985:1238f), the internal structure of a prototypical English noun
phrase is defined as a cluster of words with a “head”, typically consisting of a
common noun, preceded by one or more “determiners” (also called determinatives) and preceded and/or followed by one or more “modifiers”. Huddleston &
Pullum (2002:326) collectively denote all elements that are not the head of the
NP as “NP dependents”. Using one of two structuralist approaches to the description of noun phrase, the slot-and-filler model21 (cf. de Haan 1987:11f), a
typical NP is illustrated in Figure 3.1.
21
38
In the slot-and-filler model, all elements in the noun phrase are assumed to be on the same level.
It is used in a large number of linguistic works within different traditions, e.g. Quirk et al (1985),
some of which are far from structuralism and its reluctance to take semantic aspects of grammar
into account (see, for instance, Givón 1993:248ff). Another structuralist approach is the binary
analysis, used by, for instance, N. Francis (1958:298ff) and Fries (1967:264ff). Here, the noun
phrase (just like the clause) is seen as a two-part group (two “immediate constituents”) and
further analysis is carried out in terms of different sublevels.
38
determiner
this
premodifier
exciting
head
book
postmodifier
by Ernest Hemingway
Figure 3.1. An example of a typical English noun phrase
Gleason (1965:409) distinguishes six possible slots before the NP head, noting
that all the modifier slots are rarely filled at the same time (Figure 3.2).
N–6
all
N–5
the
N–4
three
N–3
other
N–2
new
N–1
school
N
houses
Figure 3.2. Slot-and-filler analysis of a noun phrase (Gleason 1965:409)
In a model used by de Haan (1989:31), the determiner and premodifier slots can
each be filled with more than one item, and he remarks that is not quite clear
how many possible determiner or modifier slots there really are. He further distinguishes an optional introductory “limiter slot” (see Figure 3.3), which is
sometimes filled with a word or phrase modifying the determiner, typically an
approximating adverb, as in nearly all the children. This sentence element has
traditionally been seen as belonging to the noun phrase (de Haan 1989:32, cf.
also Börjars 1998:15).
limiter
nearly
determiner
all
premodifier
the
head
children
postmodifier
−
Figure 3.3. Alternative slot-and-filler model (based on de Haan 1989:31)
We will return to limiters in the chapter on linguistic factors (see Section 7.4.3)
but now look in more detail at the different parts of the NP.
3.1.1.1
The head
Some of the linguistic factors accounted for in Chapter 7 concern syntactic and
semantic aspects of the head of the noun phrase. This is the only obligatory element, and it is also the part of the NP which governs the form of the finite verb
in a finite clause. Although the head most frequently appears as a count or mass
noun, it can also consist of other word types, e.g. a proper noun, a pronoun22 or
an adjective (Quirk et al 1985:62ff, 288, 1238; Huddleston & Pullum 2002:
328ff). Corpus research has shown that different head types exhibit different
usage patterns. Pronoun heads are very frequent in conversation, whereas more
complex noun phrases mainly occur in written registers (Biber et al 1999:231).
Furthermore, nouns as heads are in the majority in object and predicative positions, while pronoun heads are particularly frequent in subject position. This has
to do with thematic structure, since pronouns are often used as themes of clauses
22
Huddleston & Pullum (2002:327) treat pronouns, which have traditionally formed a grammatical
class of their own, as a subclass of nouns, because they share several qualities with common and
proper nouns.
39
39
– for given information – whereas more complex noun phrases are typically used
to provide new information (Biber et al 1999:235f).
In line with the generative tradition, Huddleston & Pullum (2002:329) distinguish an intermediate rank between nouns and noun phrases, and refer to this
rank as “nominals”. A nominal could be described as a head plus modifiers but
without determiners (e.g. old man in the old man). In NPs where there is no modifier (such as the man), the noun coincides with the nominal. In some other
cases, the nominal coincides with the full noun phrase, as in “bare NPs” (which
are contrasted to “determined NPs”), where the head is an indefinite plural count
noun, as in (3:3) or a mass noun, as in (3:4)23.
(3:3) Bilingual teachers need to become part of the total school community –
coaching a sport, being a class advisor. (NYT95)
(3:4) And it sends a clear message to art complexes that jazz music is important
[…] (NYT95)
Another concept used by Huddleston & Pullum (2002:332) is “fused-head” constructions, which “are those where the head is realised jointly with a dependent
function”, usually a determiner or predeterminer, as in Many would agree with
you on that point. Fused-head constructions also occur with all, both and half
(see Section 3.3.4.1).
3.1.1.2
Determiners
Even though the head is the only obligatory, and perhaps the most important,
part of the noun phrase, the determiner has a key function, being the element that
makes a full NP out of a nominal (Huddleston & Pullum 2002:354). A determiner can be realised by a definite or indefinite article, a pronoun, a numeral or a
specifying genitive noun phrase24 (Quirk et al 1985:253ff). Huddleston &
Pullum use the term “determinatives” as a unifying category for articles and pronouns that can take on the determiner function, observing that most of them can
be used in other functions as well.
The group of determiners can be further divided syntactically, according to
their relative positions before the NP head, into “predeterminers” (e.g. all, both,
double, such), “central determiners” (e.g. a, the, some) and “postdeterminers”
(e.g. numerals, few, many). There is often more than one determiner in a noun
phrase, but predeterminers are generally mutually exclusive, even though there
are a few exceptions, such as half such (Quirk et al 1985:257f; Huddleston &
Pullum 2002:436). From the point of view of the present study, determiners is a
very interesting group of elements in the NP, since in most of the variables the
quantifier has this function (see 3.3.2.1 and 3.3.2.2).
Semantically, determiners are used to specify (in)definiteness and reference of
the NP head (Quirk et al 1985:253; Huddleston & Pullum 2002:355) (see Sec23
24
40
There are also some exceptions, where singular count nouns occur in bare NPs, such as chairman
in […] Mr. Sheaff was elected chairman in 1960 (NYT95).
A classifying genitive (e.g. girls’ school) has a modifying rather than a determining function (cf.
Section 7.4.1).
40
tion 3.1.4). Furthermore, all determiners have more specialised semantic functions, such as various kinds of quantification.
3.1.1.3
Modifiers and complementation
The head of a noun phrase can be premodified by, for instance, an adjective, as
in (3:5), a noun25, as in (3:6), or a genitive noun phrase, as in (3:7) (Quirk et al
1985:1239, 1321ff).
(3:5) ‘By Friday, all the temporary workers will be gone’, he said. (NYT95)
(3:6) […] in both of the Dallas losses […] Aikman has been hurt early and did
not return. (NYT95)
(3:7) Half the hospital’s patients are covered by Medicaid […] (NYT95)
An NP head can also be postmodified by, for instance, a prepositional phrase, as
in (3:8) or a clause (finite26, non-finite or verbless), as in (3:9) (Quirk et al 1985:
1239, 1244ff).
(3:8) Partly because of all the activity in Bangalore, India’s tiny software Industry is growing rapidly. (NYT95)
(3:9) But all of the men, who have been called the Jenny Craig Eight, are
saying that they were fired, denied promotion or given unfavorable
assignments […] (NYT95)
Semantically, modifiers are used either to restrict the reference of the head or to
add extra information, thus having a function similar to that of adverbials in
clauses (Quirk et al 1985:65). A problematic distinction is the one between NP
modification and noun complementation, recently examined by Bowen (2003).
Quirk et al (1985:66) define the general difference between complementation
and modification (not specifically for nouns) as follows:
Although complementing elements may be optional, such elements
differ semantically from other optional elements (eg most modifiers) in that the omission of complementation […] implies that
some element of meaning in a preceding word is ‘unsatisfied’, and
therefore has to be provided through context.
Huddleston & Pullum (2002:436) point out, however, that the optional/obligatory distinction does not hold for NPs. Quirk et al (ibid) admit that the borderline between complements and modifiers is fuzzy and a matter of degree.
Examples of NP complementation in Bowen (2003:1) are the journey to Rome
25
26
This could also be regarded as an unmodified compound noun (see further below).
A finite relative clause can be either restrictive or non-restrictive, i.e. either necessary for the
identification of a referent or just adding extra information about the referent (Quirk et al
1985:1239).
41
41
and the decision to abandon the project. Modification is one of the linguistic
factors analysed in this study (see 7.4.1). To avoid dealing with the fuzziness of
the two categories, and since I was not interested in the semantic aspects of complementation/modification, all elements not being determiners, determiner modifiers (e.g. nearly and almost) or heads will be regarded as modifiers.
Another problematic case is that of NPs with two (or more) common case
nouns. Such an NP can either be classified as an unmodified compound noun or
as a premodified head. Traditionally, non-syntactic criteria, such as stress, orthography and meaning have been used to distinguish between the two (cf. Quirk et
al 1985:1330ff; Huddleston & Pullum 2002:451). Since these tests are anything
but watertight, Huddleston and Pullum (ibid) suggest that the distinction between compounds and premodified nouns be based on whether “the component
parts can enter separately into relations of coordination and modification”.
Blackcurrant sorbet, for instance, was classified as a premodified noun, since it
can be coordinated: a blackcurrant and a passion-fruit sorbet (Huddleston &
Pullum 2002:449). An example of a compound is backache, which cannot be coordinated (*back and toothache). The borderline is still fuzzy, and “the division
between a noun compound and a sequence of noun modifier + noun head is in
actuality a cline” (Biber et al 1999:589f). In my analysis, all cases of two juxtaposed nouns were classified as involving premodification. Again, the reason for
this decision is the fuzziness of categorisation and the fact that my main interest
is in the syntactic complexity of the noun phrases (see 7.4.1).
3.1.2
The syntactic functions of the NP
As described above, the NP is the most flexible phrase type as far as different
syntactic functions are concerned. It is typically used in one of the clausal functions of subject, object or predicative. However, it can also occur in many other
functions (most of them being part of other clausal elements), as indicated by the
following examples, accounted for by Huddleston & Pullum (2002:327):
•
•
•
•
•
•
•
•
subject-determiner in a noun phrase: I like Sue’s analysis of the passive
construction.
adjunct in a clause: Fred arrived the day before yesterday.
modifier in an adjective phrase: The nail was three inches long.
modifier in an adverb phrase: Fred arrived a whole day later.
modifier in a prepositional phrase: The wreck was discovered a mile
under the sea.
supplement: I finally met his wife, a distinguished anthropologist.
modifier in a noun phrase: She was writing a treatise on the opera
Carmen.
vocative: Elizabeth, your taxi is here.
These functions will be analysed in relation to the quantified NPs in focus in
Section 7.5.
42
42
3.1.3
NP or DP?
Many generativists (in particular) have argued that “determiner phrase” (DP) is a
more appropriate label for what is generally referred to as “noun phrase” (NP)
(Abney 1987; Payne 1994:2853f). It is remarked by de Haan (1989:24) that this
is one of the few areas within the generative framework where the noun phrase
has been discussed in its own right, in terms of its internal structure, rather than
in connection with certain more complex clausal phenomena, such as NP movement.
Until the mid-eighties, few people had questioned the noun as the typical head
of a noun phrase, except in noun-less NPs where a pronoun could have this function. Lyons (1977:464) is an exception, writing that determiners “despite their
conventional treatment as modifiers of the noun with which they occur, may
often be regarded, from a syntactic perspective, as heads rather than modifiers”.
The determiner head thus takes a nominal (cf. Section 3.1.1.1) as its complement. One argument for the hypothesis is that many determiners (e.g. some and
this) can stand on their own without a noun in noun phrases. Another is that
many important NP properties can be located in the determiner (e.g. negation
and definiteness). The DP hypothesis is the standard view, for instance, within
the Government & Binding theory and Minimalist theory (Börjars 1998:3,
119)27.
There have also been counter-arguments against the DP hypothesis, for instance, within Head-driven Phrase Structure Grammar (HPSG) (Börjars (1998:
130). The recent Cambridge grammar, though much influenced by generative
linguistics, refutes the DP-hypothesis for the following two reasons:
Firstly, it is the noun (or nominal) which defines the selectional
properties of the phrase. For example, a verb like assassinate selects a human NP as object […], whereas there is no verb in English which selects an object phrase determined by the as opposed
to no. This is because the basic semantic function of the determiner is to indicate whether the phrase is definite or indefinite
[…], and this is independent of the role the phrase otherwise plays
in the larger construction in which it occurs. The second reason
[…] is that while there is a wide range of ordinary NPs that contain no determiner28 […], NPs that do not have a noun as ultimate
head (NPs like both, several, the largest) are highly restricted in
their form and use (Huddleston & Pullum 2002:357f).
Without taking a stand as to whether NP or DP is the best alternative for classification, the traditional and more theory-neutral term NP will be used throughout
this study.
27
28
Moreover, R. Hudson (1990:271f) advocates the DP perspective in his Word Grammar theory.
What Huddleston & Pullum think of here are so-called bare NPs with plural or mass nouns (as in
I love cats/coffee).
43
43
3.1.4
Some semantic concepts pertaining to the NP
“Reference” and “definiteness” are both concepts central to descriptions of semantic aspects of the noun phrase. The concepts are complicated, and definiteness is difficult to distinguish from, for instance, specificity, one of the two basic
types of reference.
3.1.4.1
Reference
Reference is defined by Saeed (1997:12) as the phenomenon of linguistic expressions to “hook on to” entities around us. Similarly, Lyons defines it as the
relationship “between an expression and what the expression stands for on particular occasions of its utterance” (Lyons 1977:174). Language contains both referring (or “referential” as Huddleston & Pullum call them) expressions (mainly
noun phrases), which “enable the hearer, in the context in which the utterance is
made, to pick out the actual referent from the class of potential referents” (Saeed
1997:12), and non-referring (or “”non-referential”) expressions, which do not
identify particular entities in the real world, e.g. conjunctions, adjectives and
verbs (Saeed 1997:12, Huddleston & Pullum 2002:399f). According to Huddleston & Pullum (2002:400ff), however, not all noun phrases are referential expressions. An indefinite noun phrase, as in (3:10), is regarded as non-referential,
since the noun phrase does not refer to a particular entity.
(3:10) Did anyone make you a conditional offer that you refused? (NYT95)
“Referents” are the entities referred to, and these are usually defined with respect to the context (“variable” referents, e.g. the girl), but can also be contextindependent (“constant” referents, e.g. the Eiffel Tower 29) (Saeed 1997:26). The
most prototypical cases of reference concern concrete singular count nouns in
definite noun phrases and proper names (Chesterman 1991:188).
An important distinction is that between “specificity” or “specificness” on the
one hand and “genericity” or “genericness” on the other. When using specific reference, the speaker typically has a particular referent or referents in mind, while
generic reference refers in a more general way, typically to a whole class or
species (Lyons 1977:178). Besides the “speaker-having-a-certain-referent-inmind” definition, there is another way of defining specificity, often used within
logic and formal semantics in relation to quantification, which has to do with
“scope” of reference. Sentences such as (3:11) are ambiguous between (i) a
“wide/higher scope” reading, where someone refers to a specific person, the
same for everyone, i.e. a specific reading, and (ii) a “narrow/ lower scope” reading, where, for everyone, there is a person whom he/she loves, not the same for
everyone, i.e. a non-specific reading30 (Croft 1983:25; Lyons 1977:189,
Hawkins 1978:203f).
29
Constant (or “uniquely referring”) referents tend to be spelled with capital letters (i.e. treated as
proper names) (Lyons 1977:181). Note, however, that even “constant” NPs may have different
referents (e.g. the Eiffel Tower referring to a plastic replica).
30
Huddleston & Pullum (2002:405) prefer to relate scope to referentiality, rather than to specificity. They regard the wide scope interpretation as referential and the narrow scope interpretation as non-referential and “NP-bound”.
44
44
(3:11) Everyone loves someone. (Lyons 1977:189)
Another distinction between “distributive” and “collective” reference can be
made, the former focusing on individual members of a group, as in 3:12a, and
the latter focusing on the collective, as in 3:12b. (Lyons 1977:178):
(3:12) Those books cost £5.
(3:12a) ‘Each of the books costs £5.’
(3:12b) ‘The books cost £5 together.’
Huddleston (1984:255) suggests that the concept of reference be used about
whole sentences rather than about noun phrases, since “whether an NP is interpreted as specific or not depends in general on properties of the sentence containing it rather than being predictable from the form of the NP itself”. An interesting parallel can be drawn with French, where reference can sometimes be
expressed in the verb phrase rather than in the noun phrase. There, a sentence including a noun phrase which is ambiguous between a specific and a non-specific
reading can be disambiguated by choosing either the indicative or the subjunctive form of the predicate verb. The sentence in (3:13) could be translated either
into (3:13a) or into (3:13b) (cf. Grevisse 1993:1592).
(3:13) We are looking for a white cat.
(3:13a) On cherche un chat qui est blanc. (indicative, specific reference)
(3:13b) On cherche un chat qui soit blanc. (subjunctive, non-specific reference)
We will return to the phenomenon of reference in Section 3.3.4.2 below.
3.1.4.2
Definiteness
In his investigation of definiteness, Chesterman (1991:10) writes that, traditionally, a distinction has been made between “definite” and “indefinite” reference.
The former has a referent that the listener is supposed to be able to identify,
whereas this is not the case with the latter. Definiteness has also been explored,
especially within formal semantics, from Bertrand Russell’s “uniqueness condition” perspective, i.e. that a definite NP prototypically refers to something
unique. Russell’s proposal has mainly been used about singular noun phrases,
such as the The King of France, since it is “not so easily applicable to mass
nouns and plurals” (Chesterman 1991:11; see also von Heusinger 1997:11).
In English and many other languages, noun phrases are defined syntactically
as definite if, besides a head noun, they comprise a definite article, a demonstrative or possessive determiner, a specifying genitive or a quantifier such as
both. A definite NP can also lack such a determinative element, if the headword
is a proper name or a personal or demonstrative pronoun (Huddleston 1984:253).
In some languages, such as Swedish, definiteness is often marked by means of
45
45
an inflectional ending added to the noun rather than by a preceding determiner.
Hellberg (1992:36) argues that definiteness should always be ascribed to the
whole noun phrase in a particular context, even though particular words (articles
and pronouns) are sometimes described in terms of intrinsic (in)definiteness (cf.
Chesterman 1991 and see Section 3.3.3.1).
If a definite noun phrase is used, it means that the information given in the NP
is enough to identify its referent (Huddleston 1984:249f). Definite noun phrases
are therefore often called “definite descriptions”, a term coined by Russell who
showed that referents can be identified not only from names but also from descriptions, provided these are detailed enough. Previously, reference by naming
had been considered essential to language by many philosophers (Lyons
1977:180; von Heusinger 1997:10). von Heusinger (1997:8f) points to the following four different uses of definite descriptions (see Footnotes 32 to 34 for alternative terminology)31:
(a) “anaphorical linkage” (when the NP has a direct connection to something that
has been mentioned in the discourse context, e.g. Once upon a time there was
a king … and the king)
(b) “relational dependency” (when the NP is connected indirectly (via an association) to something else mentioned in the discourse, e.g. I read a book …
the author)32
(c) “situational salience” (when the NP is used because the situation or non-linguistic context gives enough information to identify the referent, e.g. The
boat is leaking, uttered by a person sitting in a leaking boat)33
(d) “uniques” (where only one referent is possible, e.g. the first man on the
moon)34.
Basing his argumentation on work by David Lewis and members of the Prague
School, among others, von Heusinger (1997:15) advocates a perspective in
which all these different categories come together under a concept of “salience”,
where “each context can be associated with an ordering among the elements of
subsets of the domain of discourse. The definite NP the F denotes the most salient F according to the situation”.
Definiteness will be further discussed with respect to the semantic classification of all, whole, both and half in Section 3.3.3.1, but let us first look briefly at
quantifiers and quantification in a few theoretical branches of linguistics.
31
Chesterman (1991:52f) provides two more uses of definite descriptions: (i) non-referential use
(as in John was the chairman) and (ii) generic use (as in The horse is a useful animal).
32
Chesterman (1991:52) uses the term “associative anaphoric use” for this type.
33
Chesterman (1991:52) distinguishes between “immediate-situation” use (as in Pass me the
bucket, please) and “larger-situation” use (as in Let’s go to the pub).
34
This is what Quirk et al (1985:270) refer to as “logical use” of the definite article (the first, the
only, the best etc.).
46
46
3.2
Quantifiers and quantification
As pointed out by Kennedy (1987:265), “one of the important things we do with
language is to measure or estimate quantity”, hence the interest in quantification
in both theoretical and applied branches of linguistics. The words analysed in the
present study (all, whole, both and half) are often termed “quantifiers” in the
literature. Examples of other English quantifiers are double, twice, each, every,
any, some and all numerals. The class of quantifier expressions in English is thus
not a uniform one, neither syntactically nor semantically, and there is no consensus among linguists as to what words belong to the class (Lehrer 1987:97).
There is, for instance, “no one-to-one relation between them and the syntactic
category of determinatives [i.e. determiners]” (ibid) since there are other word
classes that can express quantification as well: adverbs (e.g. always, very), adjectives (e.g. whole, numerous) and nouns (e.g. a lot, a number) (Huddleston &
Pullum 2002:358). Kennedy (1987:275) even suggests that “quantification is not
expressed mainly by numbers and grammatical quantifiers” but more frequently
by lexical means. Culicover (1982:90) draws the following conclusion: “it may
be that quantifier is not a true syntactic category, simply a traditional informal
one”. Similarly, Huddleston & Pullum (2002:358) emphasise that “quantification” and “quantifier” are indeed semantic terms.
3.2.1
Quantification in theoretical linguistics
3.2.1.1
Formal semantics
Within logic and formal semantics, quantification has long been a major area of
interest. ALL, often called “the universal quantifier”35 (e.g. van Eijck 1994:3423),
is one of the central concepts of logic, where it is analysed in relation to negation
(NO) and existential quantification (SOME), as in the following examples presented by Huddleston & Pullum (2002:359, 364):
(3:14) [All of the meat] was fresh. Ÿ [None of the meat] wasn’t fresh.
(3:15) [Some of the meat] was fresh. Ÿ [Not all of the meat] wasn’t fresh.
Issues concerning quantification within logic and formal semantics have mainly
focused on complicated phenomena such as scalar implicatures (e.g. Some of the
meat was fresh Ÿ Not all of the meat was fresh) and inferencing (Aristotles’ famous syllogism: All men are mortal – Socrates is a man – Therefore, Socrates
must be mortal etc.) (see e.g. Aldridge 1982:3ff; van Eijck 1994:3423). In “generalized quantifier theory”, a sub-theory within formal semantics, whole noun
phrases are referred to as “quantifiers” (see, e.g. Barwise & Cooper 1981). This
theory has been useful for the clarification of a number of complicated linguistic
phenomena, such as explaining why only certain types of noun phrases can
occur in existential there-clauses or in combination with “negative polarity
35
Within other branches of linguistics, the term “quantifier” is often used to refer to linguistic
manifestations rather than to the abstract concept.
47
47
items” (e.g. any and ever) (Bach 1989:56ff, see also Barwise & Cooper 1981).
Unfortunately, since it usually does not look into variation, none of this research
is of much relevance to the present study. Moreover, only a few of all quantifiers
have been examined extensively (esp. all, every, some, no), while “non-standard
quantifiers” such as half have generally been disregarded (van Eijck 1994:3423).
Still, one interesting area, also discussed within generative grammar, is quantifier scope (Carden 1976:17ff; Croft 1983:27;), which we touched upon in Section 3.1.4.1 (on reference). Quantifier scope is often analysed in relation to negation, which is involved in the linguistic factor presented in Section 7.4.3. Quantifier scope is further involved in patterns of regional variation, even though this
variation concerns meaning rather than form. Consider the following sentence
(from Newbrook 1992:16):
(3:16) All the students didn’t enrol.
From a logical point of view, there are two possible interpretations of (3:16),
either (3:16a), where the verb is not within the scope of the negation, or (3:16b),
where it is.
(3:16a) Not all of the students enrolled. [only some of them]
(3:16b) None of the students enrolled.
Tottie (1991:4) remarks that the second interpretation is extremely rare in natural
language. Newbrook (1992:15ff) observes a regional difference in interpretation.
In standard British English, a sentence such as (3:16) would unambiguously be
interpreted as in (3:16a). In elicitation tests, however, he found that many young
speakers of Australian English and especially South-East Asian speakers, interpreted it as in (3:16b). Many American informants found the sentence ambiguous between the two readings. He further emphasises that in spoken language intonation plays an important role in this case.
3.2.1.2
Cognitive linguistics
Another linguistic branch that has taken an interest in quantification is cognitive
linguistics. Within cognitive semantics, ALL is again a central concept, called a
“semantic primitive” (Wierzbicka 1996:47). In Langacker’s cognitive grammar,
quantification is realised by the number of the head and by quantifiers. Quantifiers can be absolute or relative. The former (e.g. many, five, much) only say
something about the size of a particular instance, whereas the latter (e.g. all,
some, no) also have a “grounding” function (Langacker 1991a:96ff). Grounding
is the establishment in the nominal of links to reality, related to the speaker, the
hearer, their interaction and the immediate situation. A grounding marker can
have either of two different foci: definiteness (e.g. the definite article) or quantification (e.g. all). A subgroup of the latter, “proportional quantifiers” (e.g. all,
both, most and some) “identify their referent as some proportion of the reference
mass [i.e. all possible referents]” (Langacker 1999:284). In the case of all, the
boundaries of the referent and the imagined reference mass are the same: “the restrictive nature of the subpart is vacuous […] the degree of restriction just
48
48
happens to be zero” (Langacker, 1991b: 112f). See Figure 3.4 for an illustration
of Langacker’s ideas.
‘all’
‘most’
RM
RM = R
‘some’
RM
R
R
R = referent; RM = reference mass
Figure 3.4. Referents in relation to a reference mass (based on Langacker
1999:255)
All/both marks that the instantiation in the actual usage event and the reference
mass (i.e. all possible referents), have the same “boundaries” (maximal extension). In contrast, most marks that these boundaries are close to each other and
some marks that the boundaries are relatively far away from each other. We will
return to Langacker and variation between all (of) in Section 3.3.4.1 below.
3.1.2.3
Generative grammar
Quantification has also played an important role within the development of generative linguistics (cf. Aldridge 1982:63f). Relevant to this investigation are two
aspects of quantification discussed within generative grammar: so-called “floating quantifiers” (as in The children were all happy) and possible transformations
in constructions with and without of. These will be dealt with in Sections 3.3.4.1
and 3.3.4.3 below.
3.3
All, whole, both and half
Even though it is convenient, there is no real justification for grouping all, both
and half together, since “each one has many uses not open to either of the
others” (Gleason 1965:411). Grammatically, all, both and half are defined as
pronouns or, together with words like double and twice, as predeterminers of
heads of noun phrases. Whole is generally categorised as an adjective, but is
often presented together with all, since these two words can have more or less
the same function since (e.g. all the book, the whole book)36. Besides expressing
quantity, all, whole, both and half relate to the concept of totality, but half has a
different relation to the concept than the other three words.
This section will bring up a number of issues relating to the syntax and semantics of the four quantifiers: syntactic classification, syntactic differences be36
Also, cf. Footnote 5 on the word entire.
49
49
tween the words, definiteness, totality and various issues pertaining to the variation patterns, e.g. floating quantifiers (see Sections 3.3.2 to 3.3.4). First, there
will be a short survey of the words from a diachronic perspective.
3.3.1
Historical background
Present-day English quantifiers have a long history. In Old English they were
used syntactically (e.g. had the same inflectional endings) as adjectives (Lehrer
1987:102). “Adjective” is also the grammatical term used in the Oxford English
Dictionary (henceforth OED) for the grammatical definitions of all, whole, both
and half. Since today these words are mostly classified as pronouns (see Section
3.3.2.1), it seems that grammaticalisation theory (see Section 2.4 and 3.3.1.3)
can offer at least part of the explanation of how some quantifiers developed from
lexical words (e.g. adjectives or “concrete and embodied nouns” (P. Svensson
(1998:204f)) to function words. Examples of the latter are a lot of and a couple
of. We will not go into a discussion of this process with respect to all the variables investigated in the present study, but in Section 3.3.1.3 I will present an
example hypothesis of a grammaticalisation process of half from noun to quantifying determiner.
The OED provides information on some of the relevant NP variants (cf.
Appendix A). We will look at their first attestations in order to find out which
variant in each variable occurred first in the English language. These diachronic
facts will serve as a background for the discussion of some statements in the
grammatical literature, concerning, for instance, markedness and transformations.
3.3.1.1
All and whole
The oldest attestation in the OED of all is on a runic inscription from around
700, although most instances in Old English texts are from the late 9th century
onwards. The word is described in the OED as “properly an adj. but passing on
one side into a n. [i.e. noun], on the other into an adv. [i.e. adverb]” (s.v. all a.,
n., adv.). J. Hudson (1998:59) regards the development of all, from an adjective
modifying nouns to its present-day use in a number of grammatical functions, as
a case of grammaticalisation, (see Section 3.3.1.3). She recognises that “a thorough investigation of the historical development of all is sadly lacking” (J. Hudson 1998:92). The word whole is first attested in a text from 888. Here, however,
whole does not express totality but is defined as “in good health; free from disease; healthy […]” (OED, s.v. whole adj. 3.a.). Table 3.1 illustrates some variables with all and whole.
50
50
Table 3.1. Variables with all and whole (based on OED, s.v. all and whole)
Variable
Modern example
First attestation
all + DETERMINER + SING. NOUN37
all the book
855
all of + DETERMINER + SING. NOUN: all of the book
1800
DETERMINER + whole + SING. NOUN
the whole book
900
the whole of the + SINGULAR NOUN
the whole of the book
1398
all + DETERMINER + PLURAL NOUN
all of + DETERMINER + PLUR. NOUN
all the books
all of the books
885
1903
+ all
all of + PERSONAL PRONOUN
we … all
all of us
1000
1593
PERSONAL PRONOUN
As the table shows, all the variants with of are much more recent than those
without. The of variant is “probably due to form-assoc. with none of, some of,
little of, much of, few of, many of ” (OED, s.v. all adj. 6). The construction is
said to be rare except when preceding a pronoun (as in all of us). As far as
all/whole in combination with a temporal noun is concerned, no examples are
given in the OED. In the variable with geographical names, only the simple all
variant (e.g. all England) is brought up (first attested in 886) (OED, s.v. all adj.
1.a). Similarly, the all + DEMONSTRATIVE PRONOUN variable was only represented by one of its variants: simple all, as in all these, first attested 700 (OED,
s.v. all adj. 1.c).
3.3.1.2
Both
The origin of both is believed to be an Old Norse word (ba&ar), where the suffix
&ar probably represented the definite article (OED, s.v. both adj., adv.). Table
3.2 illustrates the attestations of some variables including both.
Table 3.2. Variables with both (based on OED, s.v. both)
Variable
Modern example
both + PLURAL NOUN
both books
both + DETERMINER + PLURAL NOUN both the books
both
both of + PERSONAL PRONOUN
PERSONAL PRONOUN +
they … both
both of us
First attestation
1526
1297
1175
1590
Considering the possibility of both already containing a marker of definiteness, it
is surprising that the variant with a determiner (as in both the books) is attested
more than 200 years before the first attestation of the variant without a determiner (as in both books). A more natural development would have been that the
knowledge that both already includes a determiner was lost over time, and that
this resulted in a later “double” definiteness marking in both the/his/this etc.
(both the remnant of the definite article inside the quantifier and the overt defi37
In OED, no distinction is made between singular count nouns and mass nouns, as is done in the
present study.
51
51
nite determiner). Moreover, it is interesting to note the OED’s remark on both of
in NPs with a noun and a determiner, as in both of the children, both of these
arguments etc. The OED provides no first attestation of the both of variant, the
variant is dismissed as “colloquial” and it is claimed that such constructions
“scarcely ever occur in literature” (OED, s.v. both adj. 6).
No examples of the both + DEMONSTRATIVE PRONOUN variant are given in the
OED, and there were no instances in the Helsinki Corpus (1.5 million words of
Old and Middle English) either. Furthermore, the first attestation of both in a
floating quantifier construction with a personal pronoun, as in they were both…
(see Section 3.3.4.3), is more than three hundred years older than its alternative
(both of us). The floating quantifier with a personal pronoun and both is also
more than a hundred years older than the first attestation of both in a noun
phrase with a noun as head, as in both the books.
3.3.1.3
Half
The first attestation of half is as a noun. Only one of the variables with half investigated in this study was presented with both its variants in the OED (Table
3.3):
Table 3.3. Variable with half (based on OED, s.v. half, adj.)
Variable
Modern example
First attestation
half a/an
half an hour
1377
a half
a half hour
835
The variant with an indefinite article preceding the quantifier (as in a half hour)
was attested more than 500 years before the form where the quantifier comes
first (as in half an hour). The OED (s.v. half adj. 1.c) further adds that “when
these are viewed as independent numbers, amounts, coins etc., half is preceded
by a, the, etc. and hyphened to the n., as a half-dozen […].”
The construction half + DETERMINER + NOUN, as in half the book is first attested in the year 1000 (OED, s.v. half adj. 1.b). There are no attestations of half of
(as in half of the book) in the presentation of half in the OED38, even though the
construction type is mentioned in passing.
Let us now look at my grammaticalisation hypothesis. The assumption is that
a/one half of was originally used for concrete singular count nouns. The phrase
then extended its usage to more abstract mass and plural entities. Finally, a/one
and then of were omitted (see Figure 3.5).
38
52
Only one example occurred in the entire OED. It appears under another headword and originates
from as late as 1955: Surgery was performed on 70% of the hospitalized children, with tonsillectomies and adenoidectomies accounting for over half of the surgical cases (OED, s.v. adenoidectomy).
52
a/one half of + prototypical concrete object noun
a/one half of + ANY NOUN (SINGULAR, PLURAL, MASS NOUN)
half of + ANY NOUN
half + ANY NOUN
Figure 3.5. Model of the possible grammaticalisation process of half
The OED hints at such a development: “[…] in mod. use it [i.e. half] is sometimes viewed as a n. with of suppressed, as in ‘half (half of, one half of) the men
were sick, a quarter or a third of them seriously ill’” (s.v. half adj. 1.b). Similarly, Jespersen (1909–49:309) talks about a “hybrid between the sb and adj half)
and Poutsma (1914:215) explicitly declares that “half is a noun when followed
by of ”, as in half of his funds. My hypothesis is strengthened by the fact that the
OED’s first attested instance of half as a noun is from 700 (s.v. half n. 1.a),
whereas the first instance of half as a determiner/adjective is from 835 (s.v. half
adj. 1.a).
Although it is difficult to provide corroborating evidence for my hypothesis,
since the proposed grammaticalisation process must have taken place many years
ago, my hypothesis seems to fulfil several typical grammaticalisation criteria.
One is “reanalysis” (Hopper & Traugott 2003:3), by which [[a/one half] of the
apple] could have been rebracketed as [[a/one half of] the apple], so that the
noun serving as head of the original NP instead becomes part of a quantifier modifying the new NP head (the apple). “De-categorialisation”, i.e. the loss of
markers of grammatical category (Hopper (1991:30), would have occurred when
a/one half of lost its indefinite article or numeral, both typical noun markers.
In the process of “bleaching”, a word loses some of its meaning or changes
from one semantic domain to another, often involving a change from rather concrete to more abstract (Hopper & Traugott 2003:20; Haspelmath 1995:366).
Half, which was originally used for concrete objects, sides etc., changed its
meaning slightly when becoming a fuzzier quantifier, with the ‘side’ meaning
lost. Again, the OED (s.v. half n. 1.a) supports this reasoning, since there are attestations of half referring to concrete objects from 700 (“the right or left side,
the right or left ‘hand’). The earliest attestation of half in a more abstract sense
(“one of the sides of a conflict”) is from 885 (s.v. half n. 2).
When a construction is grammaticalised, the original form often lives on
alongside the new one (Hopper 1991:23), as mentioned in Section 2.5. This principle of “layering” allows for the co-existence of forms at different stages of the
grammaticalisation cline, as in the case of a/one half of, which coexists with half
of and half.
53
53
Frequency is further considered an important aspect (see, for instance,
Thompson & Mulac, 1991:314ff). As a rule, only frequent words are grammaticalised. There were around 600 tokens (400 per million words) of half in its
various spellings in the Helsinki Corpus and 30,000 tokens (300 per million
words) in the British National Corpus, so it is and has long been a frequent word
in English. Moreover, a new form typically becomes more frequent than the old
one(s) (cf. Hopper & Traugott 2003:67). A small investigation (Estling Vannestål 2001b) showed such a pattern for the three variants a/one half of the, half of
the and half the, as illustrated in Table 3.4.
Table 3.4. Distribution of half variants in three modern corpora39
half the
Total
a/one half
half of the 40
of the
N/M
%
N/M
%
N/M
%
N/M
%
BNC
1.1
3%
7.8
21%
29.0
77%
37.9 100%
Cobuild
0.5
1%
7.4
20%
28.4
78%
36.3 100%
LSAC
1.0
4%
4.0
16%
20.4
80%
25.4 100%
N/M = number of tokens per million words, Cobuild = the Cobuild Direct corpus of
English
A/one half of was less frequent than half of, which was less frequent than half 41.
The distribution was the same in all three corpora, even though there were differences between the corpora as regards the relative frequency of the variants.
Judging from the fact that several grammaticalisation criteria are fulfilled, it is
not unlikely that a/one half of the has been grammaticalised from a lexical
phrase into its present-day form and function. A problem is that no attestations
of the half of the stage were found in the OED and that there were no tokens in
the Helsinki corpus. The possible development of half could, however, be compared with other cases where nouns have turned up in partitive constructions,
lost some of their noun-like characteristics and acquired typical quantifier characteristics (cf. Lehrer 1987:102).
We will now leave history aside and examine the quantifiers from a presentday perspective, approaching a number of issues relating to the syntax and semantics of all, whole, both and half.
3.3.2
The syntax of all, whole, both and half
3.3.2.1
Syntactic classification
Since “determiner” is usually not recognised as a word class in its own right (cf.
Huddleston & Pullum’s (2002:355) use of “determinatives” as a unifying word
category), the words all, both and half are usually defined as pronouns which
39
The British National Corpus (BNC): 100 million words of spoken and written BrE; the Cobuild
Direct corpus: 60 million words of spoken and written BrE and AmE; the Longman Spoken
American Corpus (LSAC): 5 million words of spoken AmE
40
Instances where half was used as a noun preceded by the definite article (e.g. the bottom half
/of/) were excluded, since the purpose was to show the frequency of half as a quantifier only.
41
The corpus queries were restricted to half in NPs with the definite article.
54
54
can function syntactically as determiners in a noun phrase (cf., for instance,
Quirk et al 1985:381f). Historically they have been classified as adjectives, and
sometimes they still are (cf., for instance, Huddleston & Pullum 2002:356 on
both in both sides).
The words are often grouped together with words such as double and twice as
predeterminers, since each of them can precede the central determiner – the definite article or a possessive/demonstrative determiner – in a noun phrase (e.g.
Quirk et al 1985:257), as illustrated in examples (3:17) to (3:19).
(3:17) The answers to all the questions are inextricably intertwined. (IND95)
(3:18) Both my brothers were killed at Dunkirk and my husband died later from
the wounds he got when he was 18 in the First World War […] (IND95)
(3:19) Only half these managers could respond to these letters in the same
language. (IND95)
All and both are in fact unique among the determiners in that they can also occur
on their own, without another determiner, as in (3:20), (cf. Huddleston & Pullum
2002:376) (see Section 3.3.4.2).
(3:20) The Cranes argue that comprehensive schools are there to educate all
children. (IND95)
All, both and half can be used as pronominal heads, as in All/Both/Half were
happy (cf. Quirk et al 1985:25842). Similarly, when they are used in a construction with of (as in All of the people were happy), Quirk et al (1985:258) refer to
them as pronouns functioning as NP heads (see Section 3.3.4.1). Huddleston &
Pullum (2002:434) consider half to be a noun since it can take a preceding indefinite article or a cardinal number. They also observe that it “behaves somewhat
differently from other fractional nouns” (such as quarter and third) in that it can
be used without a determiner, as in half his share.
Whole can be either an adjective, in the DETERMINER + whole construction, as
in (3:21), or a noun, in the the whole of + DETERMINER construction, as in (3:22)
(Quirk et al 1985:259f, 381).
(3:21) The whole world is tremendously concerned, […] (IND95)
(3:22) The whole of the city was packed, […] (IND95)
3.3.2.2
Syntactic differences between the quantifiers
In the literature all, whole, both and half are often treated as if they were quite
similar. Leech (1989:66) for instance maintains that “all is very similar to both”.
However, besides the obvious syntactic difference between whole and the other
42
Confusingly, Quirk et al (1985:258, 377ff) discuss pronouns and determiners on the same level,
in terms of function. Since “pronouns” is a word-class rather than a syntactic function, I use the
term “pronominal head”, thereby remaining on the same level as “determiner”.
55
55
three quantifiers, there are also several differences between the words in the all–
both–half group. These differences, arrived at by means of intuition and various
grammatical sources, are summarised in Table 3.5 and explained below.
Table 3.5. Syntactic differences between all, whole, both and half
all
whole both
Can combine with singular count nouns
*
+
−
Can combine with mass nouns
+
*
−
Can combine with plural nouns
+
*
+
Can be used with the indefinite article
+
−
−
Can be used without a central determiner
+
+
−
Can be used as a floating quantifier
+
+
−
Can be used in predicative function
+
+
−
Can be used as a postdeterminer
−
−
−
Can be modified
+
+
−
Can be followed by a postdeterminer
+
−
−
half
+
+
+
+
−
−
+
+
+
+
* Only rarely or in certain cases or with a particular meaning (see further below)
From the table we can conclude, first of all, that all and half are the most versatile quantifiers, occurring without restrictions in the largest number of syntactic
construction types. Whereas both only combines with plural count nouns, all and
half are used with both mass nouns and singular and plural count nouns, even
though all is said to be rare with singular count nouns, as in (3:23) (Putseys
1985:388f) (see Section 3.3.4.4).
(3:23) A constitutional court responsible for all the country would be created.
(IND95)
Quirk et al (1985:381) consider all with singular count nouns to be “somewhat
formal43” and point out that a whole construction, either with whole as an adjective (e.g. the whole world) or as a noun (e.g. the whole of the world) is generally
preferred. The only case where all is preferred with a singular count noun is
when the noun expresses an established amount of time, as in (3:24) (Putseys
1985:389) (see Section 3.3.4.4).
(3:24) The children swam and played all day […] (IND95)
All is also quite common in NPs with a geographical name, as in (3:25).
(3:25) […], all Ireland had the stench of black, mushy tubers. (IND95)
Whole is not supposed to combine with mass nouns, as in (3:26) (Quirk et al
1985:260), but in my corpus material it sometimes does, hence the star in the
table.
43
56
Claims about differences in formality level between variants will be further discussed in Section
6.2.
56
(3:26) The whole mystique is perpetuated by newspapers and magazines who
repeat everything without criticism. (IND95)
Whole sometimes occurs in plural noun phrases, as in (3:27), but the meaning is
then different from plural NPs with all. Their whole lives does not refer to
number, but rather to amount (see 3.3.4.4).
(3:27) They were settlers; their whole lives were in the baggage car. (IND95)
Whole and half differ from the other quantifiers in that they can be combined
with the indefinite article in NPs like a whole apple, half an hour and a half(-)
hour. The a whole construction will not be examined further in this study since it
exhibits no syntactic variation. In Putseys’ (1985:379) view, a half, which is the
oldest variant (see 3.3.1.3), is more formal and correct than half a (see 3.3.4.3).
A third, non-standard, alternative, a half a, is mentioned by Jespersen (1909–
49:361) and Svartvik & Sager (1996:275). Huddleston & Pullum (2002:434)
also bring up the (marginal) construction one(-)half the, as in exactly one half
the amount. There was only one (questionable) occurrence in the whole corpus
material (and very few in other corpora consulted), which is why it was not
included as a variant to half (of) the in Appendix A.
All can be used both with and without the definite article or another determiner (e.g. all children/all the children, all morning/all the morning). Generally,
when the head is not a temporal noun, there is a difference in meaning from generic (with the article absent) to specific (with the article present), but this is not
always the case (see 3.3.4.2). Half and whole cannot be used without a determiner of some kind (half the children, the whole morning but not *half children,
*whole morning).
Both can be used with or without the definite article in NPs with a noun as
head44, as exemplified by (3:28) to (3:30), but without any obvious difference in
meaning.
(3:28) Both reports were unconfirmed […] (IND95)
(3:29) Damien Hirst, showing new work at White Cube, is now so popular that
both the major works were sold before they went on exhibition. (IND95)
(3:30) I click on a link and recent reviews for both of the productions come up.
(IND95)
There is also a non-standard form with the definite article preceding both in an
NP with a personal pronoun, as in the both of us. The pre-both construction is
not used when the NP head is a noun rather than a personal pronoun: *the both
of the students (Huddleston & Pullum 2002:377).
All and both can be grouped together on the basis of both of them occurring
as “floating quantifiers”, that is in a position after the NP (and sometimes after
44
Note that in cases where the head of the NP is a nominalised adjective, as in both (of) the latter/
others, the definite article is required.
57
57
the predicate verb as well) rather than at the beginning of it. This variant can be
used both with NPs with a nominal head, as in (3:31), and in NPs with a pronoun
as head, as in (3:32). Half and whole, on the other hand, cannot (Quirk et al
1985:126).
(3:31) The lights were all on for the first time in four years […] (IND95)
(3:32) They both love Richmond and are delighted that Freddy, seven, and
Tabitha, five, will be growing up there. (IND95)
Floating quantifiers are further discussed in Section 3.3.4.3 below.
Putseys (1985:386) writes that noun phrases with both cannot be used as predicatives in a clause. In such cases the two is used instead. Examples (3:33) and
(3:34) are both taken from Putseys.
(3:33) *These are both the things you want.
(3:34) These are the two things you want.
There is no such restriction for the other quantifiers.
Half is the only word that can occur in postdeterminer position, which it does
in the a half variant, as in (3:35).
(3:35) After a half hour or so, he moved on to Chinatown. (NYT95)
All, whole and half can all be modified by certain adverbs etc. expressing, for instance, approximation, comparison, exception etc., as in (3:36) and (3:37),
whereas both cannot (Huddleston & Pullum 2002:375) owing to its intrinsic exactitude (‘two, neither more nor less’).
(3:36) Nearly all the great museums of Europe have all-day Sunday opening.
(IND95)
(3:37) While millions starve, more than half the adult population of the
Western world is overweight, […] (IND95)
Finally, all and half are the only quantifiers that can be directly followed by certain postdeterminers, as in all last night, all three men, all the many boys
(Putseys 1985:382) and half last year. As for the other two quantifiers, words of
this type are only used if the noun phrase includes of, as in both of the two boys
and the whole of last night.
3.3.3
The semantics of all, whole, both and half
3.3.3.1
Classification in terms of definiteness
Even though definiteness is perhaps best analysed in whole NPs in context (cf.
Hellberg 1992:36), many school and reference grammars ascribe definiteness to
58
58
isolated elements. One area where there are problems in the semantic classification into definite, indefinite or neither is in the case of quantifiers. Traditionally,
words like all and both have been called “indefinite pronouns”, but, as Haspelmath (1997:11) observes, this category has served as a “waste-basket” for words
that do not fit in anywhere else, and have rather little in common. He claims that
some, any and no and their compounds are in fact the only real indefinite pronouns. Jespersen (1924:83) also notes that there is no consensus among linguists
as to what words should be included in the group.
In a survey of how all, whole, both and half and their Swedish equivalents are
classified in a number of grammar books (Estling Vannestål 2000), it was concluded that “indefinite pronoun” is a frequent label in both languages. Quirk et al
(1985), for instance, use the term on the grounds that that these words “lack the
element of definiteness which is found in the personal, reflexive, possessive and
demonstrative pronouns” (Quirk et al 1985:376). In spite of their indefiniteness,
these quantifiers can sometimes combine with elements of definite meaning,
such as the definite article (ibid). As far as Swedish is concerned, the most comprehensive Swedish reference grammar refrains from labelling the words indefinite, since these words cannot be said to be inherently indefinite (Teleman et al
1999:369). On the other hand, the grammar does not say explicitly that the
words are definite, that label being instead reserved for personal, demonstrative,
reflexive, reciprocal and relative pronouns and the definite article (ibid:236). Interestingly, Sapir (1930:15) takes a perspective that is entirely opposite to the
predominant tradition in remarking that “in a sense all totalizers are definite”.
Jörgensen & J. Svensson (1987:24) also designate the quantifiers at issue (e.g.
Swedish alla ‘all’) as definite pronouns – words which describe something that
is clearly defined in the speech situation or the linguistic context. The same view
is expressed by Huddleston & Pullum (2002:376): “Since a unique set is indicated and all indeed makes a stronger statement of totality than required for definiteness, all must be considered as a definite determiner.” The grammatical literature thus comprises a whole spectrum of labels, from indefinite to definite,
with some grammarians finding neither of the two appropriate. I agree with the
view that the totalizers all and both (but not half and whole) are intrinsically definite. Even in the case where, for instance, all X refers generically, the listener
will be able to identify the referent (since it is all imaginable X in the world),
which is the crucial criterion for a definite expression.
Sapir (1930:15), in spite of his suggestion that all totalizers are definite, distinguishes between indefinite and definite uses of all. In all the people in the
room, all is indefinite, since the number of items (people) does not have to be
known by the speaker and listener. In all the cardinal points, referring to ‘all
four of the cardinal points’, however, the number is known by the speaker and
listener, even if it need not be overtly expressed, and all is definite in such a
noun phrase. This supports Hellberg’s idea that the property of definiteness can
only be acquired in context, and should not be ascribed to isolated elements.
Both is treated by Sapir as an invariably definite totalizer, since the number of
items referred to is always two (Sapir 1930:15). It could be argued that the degree of intrinsic definiteness varies from one quantifier to the other.
Jespersen (1933:68) gets around the problem of the definite/indefinite distinction by dividing determiners into “pronouns of definite indication” (the, this
59
59
etc.), “pronouns of indefinite indication” (a, some etc.) and “pronouns of totality” (all, no etc.). Reed (1996:143ff, 173) distinguishes between indefinites (e.g.
a, some), definites (e.g. the, this) and partitives (e.g. all, both, half). She bases
her analysis on the different discourse functions of the NPs in which the words
are used: to evoke new discourse entities (indefinites), to access discourse
entities (definites) and to evoke subgroups of discourse entities (partitives).
3.3.3.2
Totality
Apart from the general quantifying aspect, there is also a more specific semantic
similarity between all, whole, both and half, in that they are all related to totality
in some respect. The first three (all, whole and both) express one extreme of a
scale with no at the other end45. Half, together with one third, three fourths etc.,
is situated in between the extremes (Sapir 1930:20) and referred to by Huddleston & Pullum (2002:433f) as a “fraction”. We will now look at various issues
pertaining to the totality concept, and focus on NPs with all and whole.
Explaining “totality”
Sapir explains totality from a cognitive point of view, relating to the perceptional
experience of human beings. The notion is derived from either of two types of
experience:
(1) “the feeling of rest or of inability to proceed after a count, formal or informal, has been made of a set or series or aggregation of objects”
(2) “the feeling of inability or unwillingness to break up an object into smaller
objects” (Sapir 1930:7)
The former is connected with the “all” feeling and the latter to the “whole”
feeling. Words like all and whole are called “totalizers”, i.e. quantifiers “whose
function is to emphasize that in the given context the quantifiable is not to be
thought of as capable of increase, e.g. all, the whole flock” (Sapir 1930:6).
Sapir (1930:8f) ventures into a not so easily comprehended division into six
groups of totalizers. The first type (“whole existent”) refers to items that are “divisible into parts but ‘resisting’ such division”, as in the whole table. The second
type (“summated existent”) refers to “an aggregate of parts derivable from a normally undivided existent”, as in all of the table. The third type (“persistently/reassertedly whole existent”) refers to items that are “thought of as divisible into parts but apprehended as persistently resisting deformation”, as in
the whole of the table. The fourth type (“aggregate”) refers to items that are an
“aggregate of existents, each of which is considered as having functional
reality”, as in all the tables. The fifth type (“whole aggregate”) refers to an “aggregate thought of as divisible into members or parts and as ‘not resisting’ such
a division”, as in the whole set of tables. Finally, the sixth type is called
“reasserted aggregate” or “relapsed collection”. It refers to an “aggregate apprehended as threatening, as it were, to fall apart into a simple aggregate”, as in all
45
60
Cf. Newbrook’s statement (Section 3.2.1.1) about regional differences in the interpretation of
negative clauses including all (1992:15f; 2001:128f).
60
of the tables. It is very difficult to grasp the difference between, for instance, a
“whole existent” and “persistently whole existent”. Furthermore, how does one
differentiate between an item that is “divisible into parts but ‘resisting’ such division” and one that is “thought of as divisible into parts but apprehended as persistently resisting deformation”? Still, this attempt at describing possible semantic differences between the words is quite interesting and therefore worth mentioning, especially since it is quite contrary to the traditional classification of all
and whole in terms of divisibility (cf. Section 3.3.4.4). There will be reason to
come back to Sapir’s presentation of a part–whole relationship and possible semantic differences between variants with and without of in Section 3.3.4.1 below.
Totality vs. large quantity, intensity etc.
Some scholars have implied that all and whole do not always contribute much
meaning to a noun phrase. For instance, Quine finds these words “logically
redundant” (cited in Aldridge 1982:16) and Sapir (1930:17) uses the term
“pseudo totalizer”, observing that, sometimes, all is an “explicitly definite totalizer”, whereas in other cases perhaps it is not. All men are mortal, for instance,
might just paraphrase Men are mortal, all in this generic sense being a classindicator rather than a true totalizer. Similarly, the definite article could,
according to Huddleston & Pullum (2002:376), replace all in sentences such as
(3:38) and (3:39), even though the totality concept is slightly weakened in the
latter:
(3:38) All students who have failed (= The students who…)
(3:39) We spent all day at the beach (= We spent the day…)46.
Huddleston (1984:253) writes that “putting all before the emphasises, reinforces
the meaning of totality, rather than introducing it”. Similarly, R. Hudson
(1990:286) suggests that the contribution of all to the noun phrase is here not
semantic, but rather pragmatic, stressing the completeness aspect of the NP.
I would like to take Huddleston’s and R. Hudson’s ideas one step further by
claiming that in some cases, all does not even express totality at all. In its original, prototypical function, all either introduces or emphasises the totality meaning; however, it can also fulfil another communicative function. Let us look at
the noun phrase all the flowers. In a sentence such as (3:40), all is clearly a
marker of totality, introducing the “completeness aspect” mentioned by R. Hudson. In this instance, we are talking about all the flowers, not just some of them.
(3:40) Are all the flowers perennial or just some of them?
Consider next the song title Where have all the flowers gone? Here, all does not
seem to express totality meaning, but indicates that we are thinking of a large
number of flowers. This nuance would not have been expressed had the quanti46
When the NP functions as an adverbial, however, this replacement is impossible: He worked at
home all day/*the day (Huddleston & Pullum 2002:376).
61
61
fier been left out (as in Where have the flowers gone?). Therefore, the quantifier
is not redundant, as suggested by, for instance, Quine and Huddleston & Pullum.
The hypothesis of totality meaning vs. large-quantity meaning can also be
connected to the fact that all functions as a plural marker – that is saying something about quantity in general rather than expressing totality – in informal
second person pronouns you all, you-all, y’all (Quirk et al 1985:344; Wales
1996:73, see Section 3.3.4.3). Stress can probably be at play in the difference between totality and large-quantity meaning, so that the quantifier does not carry
stress when used with large-quantity meaning, while it typically does when the
totality interpretation is intended (see further below)47. In many cases, a possible
test for checking whether an instance of all has totality or large-quantity meaning is the following:
(a) For totality meaning to add not just some etc.:
ING has in effect purchased the Barings name, and nearly all of the active
businesses […] (IND95) Æ ‘not just some (of the businesses)’
(b) For large-quantity meaning to replace all with many, extensive etc.:
[…] for all the fine words, this government would raise taxes after an election victory […]. (IND95) Æ ‘in spite of the many fine words’
The seemingly double function of all resembles two different structures expressing duality. A noun phrase such as both /the/ children emphasises totality, but a
noun phrase like the two children does not (Quirk et al 1985:259). This difference can be expressed in other languages as well, although they may have
other ways of making the distinction. In Swedish, for instance, the difference is
expressed by a combination of phonology and syntax. With totality meaning
proper, båda and bägge (‘both’) introduce the noun phrase and the pronoun
carries stress, as in 'båda de andra böckerna (‘both the other books’). On the
other hand, båda/bägge can also be used without stress and then occur after the
definite determiner, as in de båda 'andra böckerna (‘the two other books’48). In
this case the pronoun does not emphasise the totality of the noun phrase (Teleman et al 1999 II:377f). This can also be compared with the double function of
another quantifier, some. There is one function where some is stressed and the
pronunciation is [s8m], as in (3:41), and another where some [s"m] is unstressed, as in (3:42).
(3:41) There is 'some coffee [but not enough for everybody].
(3:42) Would you like some 'coffee? [rather than, for instance, tea]
These two functions resemble those of all described above.
47
48
62
Unfortunately, this phonological aspect could not be studied in the corpus material since the
spoken corpora used have not been annotated for prosodic features.
The English corresponding construction (the two) can also be directly translated into Swedish: de
två andra böckerna.
62
Teleman et al (1999:372f) bring up the totality meaning of all in the context
of plural determiner + possessive attribute in Swedish. In alla Svenssons böcker
(‘all Svensson’s books’), all has totality meaning proper49, whereas in Svenssons
alla böcker, the totality meaning is said to be “weakened” (what I refer to as
“large-quantity meaning”). Weak totality meaning is also at play in noun phrases
including certain adjectives that express multitude, as in alla möjliga böcker (‘all
possible books’), where the meaning is similar to diverse böcker (‘sundry
books’). The authors do not, however, mention the fact that much more frequent
constructions – without a possessive attribute or an adjective like möjlig (‘possible’) – can also have weak totality meaning in Swedish, as exemplified by the
Swedish translation of the song title quoted above (3:43):
(3:43) Vart har alla blommorna tagit vägen? (‘Where have all the flowers
gone?’)
Although the two duality constructions (both/two) are mentioned in most English
school and reference grammars, the meaning of all is seldom discussed in grammars of the English language, apart from the idea about all being redundant. One
exception is Jespersen (1933:186), who comments on a particular type of all
construction. He writes that “the meaning of all is weakened” in sentences such
as (3:44):
(3:44) They understand that good management makes all the difference 50 […].
(IND95)
Jespersen interprets the noun phrase as ‘a great deal of difference’ (my emphasis), which can be compared with my hypothesis of a large-quantity meaning.
Moreover, Putseys (1985:372) points to the meaning of all in sentences like
(3:45):
(3:45) That ugly little house was all the home (that) I ever had.
Putseys does not talk about this example in terms of weak totality meaning, but
defines its meaning as ‘exclusively, only’51.
In Section 7.4.3 I will suggest that totality meaning is more likely with certain
types of elements occurring in the NP or its near co-text. This group of elements
(referred to in this study as “focus markers”) comprises other quantifiers (all /of/
the many/ten books), negations (not all /of/ the books), approximators (nearly all
/of/ the books), exceptions (all the books except one) plus a number of other,
more or less subtle words signalling totality.
Is the difference between totality meaning and large-quantity meaning reflected in the choice of variant? Compare Are all the flowers perennial? and
Where have all the flowers gone? with corresponding noun phrases including of.
49
However, it would also be possible to use this word order in large-quantity meaning, if the stress
were on the noun (böcker) rather than on the quantifier (alla).
50
The phrase make all the difference was excluded from my study as a fixed expression.
51
Similarly, Long (1961:336) writes that in sentences such as That’s all we ask, all “implies that
the totality it represents is small”.
63
63
Where have all of the flowers gone? sounds unidiomatic, whereas Are all of the
flowers perennial? is not. The of variant seems to be a more natural alternative
when the quantified noun phrase expresses totality meaning rather than largequantity meaning, as illustrated in Figure 3.6. It is also possible that there is a
meaning scale, rather than an absolute dichotomy, thus the arrows at the
extremes of the line.
Totality meaning
Large-quantity meaning
Are all the flowers
perennial?
Where have all the flowers gone?
Are all of the flowers
perennial?
?Where have all of the flowers gone?
Figure 3.6. The meaning of quantifiers in NPs with all
We will return to the hypothesis about totality meaning and the all of variant in
Section 7.4.3.1.
In some contexts, all might also add an emotional aspect to the meaning of
the NP. In (3:46) all expresses totality rather objectively:
(3:46) Clearly, not all these problems have gone away. [only some of them]
(IND95)
In (3:47) the speaker expresses a more subjective attitude, in this case worry,
sadness or the like.
(3:47) It was only when she arrived – she was so small and weak that it felt like
we had another baby – that suddenly it hit me. All those little mouths to
feed! All those responsibilities! (IND95)
Labov (1984:48ff) suggests that all and a few other quantifiers (e.g. no and
everybody) can sometimes be used pragmatically for expressing intensity, rather
than totality. Consider the use of all in (3:48).
(3:48) I didn’t bring none of my clothes back … I left ‘em all down there.
(Labov 1984: 48, my emphasis)
Here the quantifier clearly refers to my clothes in the previous sentence, so it is
not an intensifying adverbial as in I knocked him all out in the street. Looking at
a wider context, however, it is clear that the speaker does not refer to all her
clothes, since in the interview she is not naked. Therefore, the sentence involves
“cognitive contradictions” (Labov 1984:ibid). Labov explains this paradox in the
following way:
64
64
All universal quantifiers are bound in an implicit set of ‘things that
count’ or ‘things that are worthy of mention’. […] To deal with
(14c) [example 3:48 above] we would have to interpret ‘things that
count’ as ‘clothes that are needed for more than a few days’. The
clothes that Dolly Ripley [the speaker] is wearing, and the few that
she needs for a short stay in New York City, are perhaps not
‘worth counting’ in evaluating the statement, I left all my clothes
down there. […] This pragmatic approach accepts the meaning of
universal quantifiers that is conventional for sentence grammar,
and tries to show how the apparent illogic of usage is the result of
interaction with a larger context52. […] I will use the term ‘loose
interpretation’ to mean a sense of a universal quantifier that
focuses only on the whole and makes no division among the members of the class, with no attention to possible exceptions. The
term ‘strict interpretation’ will apply to a sense of universal quantifiers that conforms to logical practice and specifically rules out
exceptions. (Labov 1984:49f).
Labov also brings up phrases like all the time, where all would normally be
interpreted “loosely”, since a sentence like (3:49) is logically impossible (cf.
Section 7.3.2):
(3:49) He eats all the time and don’t even get fat. (Labov 1984:53)
Similarly, in cases like The baby was crying all night, the noun phrase really
means something like “what seemed like all night’ or ‘most of the night’53, with
all used in a kind of intensifying rather than true totalising function (Chesterman
1991:67). In many cases such as these, intensification is not an appropriate label
for the meaning of all, and perhaps the hypothesised large-quantity meaning
would be a better description. Interestingly, all the time and the whole time are
often classified as fixed expressions in dictionaries, having a rather metaphorical
meaning of ‘continuously, very often’, sometimes with a negative connotation of
irritation (cf., for instance, Longman 2003, s.v. all determiner, predeterminer 1,
and whole adj 1). Still, it is probably quite possible to use all the time with totality meaning proper, which is strengthened by the fact that the corpus included
cases with of, as in (3:50).
(3:50) On the other hand he knows that […] he wasted his talent in his teens
and early twenties by wanting to have a good time all of the time.
(IND95)
52
Another example given by Labov (1984:48f) is you ain’t never been no place. Here, the negative
quantifier no is used, like all above, in a hyperbolic way. It is not to be interpreted as if the person has not been anywhere at all (which is a logical impossibility) but simply that he or she has
not been to “any place that counts as far as places worth being to are concerned”.
53
Referring to Lyons (1977:456), Chesterman remarks that all can sometimes have a more determinative than quantitative function, as in Which sweets do you want? All of them. (1991:4, my emphasis).
65
65
In light of the discussion of different meanings of all 54, it is obvious that the semantic concept of ALL is a very complex one. Therefore the idea of both being a
mere dual variant of all in semantic terms, as expressed by, for instance, Sapir
(1930:16) and Jespersen (1933:186), can clearly be refuted. It could be argued,
however, that if all occurs in various functions, all1, all2, all3 etc., then both is a
dual variant of one of these. Not only the semantics, but also the syntax of all is
in some respects quite different from that of both.
3.3.4
Variation patterns – syntactic and semantic aspects
As mentioned in Section 1.2, this investigation takes four different variation patterns into account, repeated here in a shorter form:
•
•
•
•
presence or absence of the preposition of
presence or absence of the definite article
alternating positions of the quantifier
alternating lexical elements (all/whole)
In this section we will look more closely at these four patterns.
3.3.4.1
Variants with and without of
All, both and half differ from other quantifiers in that of is optional in definite
noun phrases55. Whereas we can make a choice between saying all/both/half the
children or all/both/half of the children, we cannot choose between saying
*most/*many/*some the children and most/many/some of the children. Only the
latter construction is possible with such quantifiers (Langacker 1999:75). All,
both and half cannot be used in of constructions lacking a definite determiner
(all of the/these/my children but not *all of children). This restriction is often
called “the Partitive Constraint” (Reed 1996:149).
As illustrated in Figure 3.7, there are (at least) two ways of analysing quantified NPs including of, referred to as “fractional partition” in Quirk et al
(1985:258) and as a “(partitive) fused-head construction” (see 3.1.1.1) in
Huddleston & Pullum (2002:333, 376). In the first analysis, used by both Quirk
and Huddleston & Pullum, the quantifier is the head of the NP and the of phrase
is some kind of postmodification or complement. In the second analysis, the
noun in the of phrase is the head.
54
It is also possible that whole can have a more pragmatic function. In phrases such as the whole
point and the whole matter, where the stress is on the noun rather than on whole, it is sometimes
difficult to perceive a true totality meaning.
55
According to Quirk et al (1985:373), the of-phrase is preferred to the of-less one when another
quantifier precedes the noun in the NP (e.g. all of the many boys), especially in American
English (see Section 7.4.3).
66
66
(a) quantifier = head
all
of the children
(b) noun = head
all of
the children
Figure 3.7. Two ways of analysing quantified NPs including of
Reed (1996:147) uses the terms “matrix NP” or “full partitive NP” when
referring to noun phrases with of. Furthermore, she regards them as consisting of
two noun phrases. The whole NP is the matrix NP and the part following of is
embedded, in line with analysis (a). Previously, Jackendoff (1968:422f) expressed the same view, in distinguishing between words like some and all in pronominal use, when the quantifier occurs (i) alone or with a following of phrase,
as in some /of the men/, and (ii) in adjectival use, when the quantifier directly
precedes a noun, as in some men. The idea of the quantifier being the head of the
noun phrase seems reasonable, since, as Jackendoff shows, it can actually stand
on its own, as in (3:51) and (3:52), as long as it is specified from the context.
(3:51) Two path-breaking icons of the post-war theatre were produced in
Sarajevo during the siege. Both had their own particular contextual
ironies, […] (IND95)
(3:52) Six million of Poland’s pre-war population of 32 million died. Almost
half were Jews. (IND95)
Within generative grammar of constructions have been explored in terms of possible transformational rules (cf. Aldridge 1982:82). Seppänen & Seppänen
(1986:169f, 175) use a rule of of dropping to explain why both the children and
both of the children can mean the same thing. They refer to the possibility that
an NP like three children is as a transformation from [THREE OF CHILDREN]. According to this rule, both /the/ children would be a transformation from [BOTH OF
CHILDREN]. However, considering the fact that the all and both variants including of are attested much later than those without, a rule involving an of-less form
as the underlying structure is more motivated.
In fact, 90 years ago, Poutsma (1914)56 expressed the possibility of the quantifier rather than the quantified noun being the headword, not only in NPs with
all of, but also in of-less noun phrases.
When all and both stand before a plural noun modified by the definite article, or an adnominal (pro)noun, as in All (both) the (my,
these, my brother’s) children sat at the table, they may be
considered as substantival pronouns or numerals, to which the following noun is related as a kind of apposition. Apposition may be
considered to represent genitive inflection, as will be seen from
56
Poutsma (1914:217) also suggests the possibility of all and both in fact being adverbials, since
these words can also appear in adverbial positions, as in the children have all (both) come home,
what is today referred to as floating quantifiers (see Section 3.3.4.3).
67
67
the fact that the preposition of is occasionally placed before the
plural noun: All of these tribes had the same language. (Poutsma
1914:216f)
Similarly, Allan (1999:26) regards not only NPs with of, but also of-less constructions, such as all the children, as consisting of two noun phrases rather than
one. This disregards the function of of, an idea which can be questioned, especially in light of the possibility of a distinction between totality meaning and
large-quantity meaning of all being connected to all of vs. all.
The other approach to defining the syntax of of phrases is advocated by
Sinclair (1991:82f), who rejects the traditional classification of of as a preposition, since of phrases are used in so many functions that differ from typical
prepositional phrases. When occurring in an NP, a prepositional phrase is
typically a postmodifier (or complement) of the preceding noun. Of phrases are
seldom described in the literature as behaving any differently. Sinclair (ibid)
argues, however, that very often, an of phrase is used in an NP where the noun
following of could be regarded as the most important word. Examples of such
phrases are the bottle of port, millions of cats and both of them. Sinclair’s classification is quite convincing, since the “the omission of N2 [i.e. the noun following the preposition] does the greatest damage to coherence”. For instance, if, in
some cases, all and whole are more or less redundant (cf. 3.3.3.2), the noun following of is certainly the most natural headword, at least from a semantic point
of view. On the other hand, all, both and half can also stand on their own, so that
Sinclair’s example both of them could be ellipted to just both, in which case the
quantifier constitutes the head of the NP. Headedness is indeed a complicated
issue, and it may be that both these analyses can be applicable depending on the
circumstance. We will return to this topic in Section 7.3.3.
Another important issue pertaining to the choice between variants with and
without of is grammatical synonymy (see Section 2.1.2). In constructions like all
of the children, of has often been regarded as a meaningless grammatical particle, especially within most generative approaches. Generativists have frequently
supported their arguments with examples of seemingly synonymous pairs like
the machine’s humming/the humming of the machine (Langacker 1999:73), disregarding that of can have different functions. Reed (1996:165) also believes that
variants with and without of mean the same thing, so that words like all and both
have a partitive interpretation whether the partitive structure is overt, as in both
of the children, or not, as in both /the/ children 57.
The meaninglessness of of has been questioned. Aldridge (1982:204), for instance, points to its semantic function as ”linking the overt quantifier with the
limited universe of discourse”. Admittedly, the multitude of functions that of has
(cf. Sinclair’s discussion above) leads him to confess that he is sometimes
“tempted to conclude that it does no more than establish the broadest possible relationship between items” (ibid:205). Still, he tentatively suggests a difference
between variants with and without of, relating it to the fact that other quantifiers
(like many, each etc.) require of in combination with definite noun phrases. He
57
68
Apart from Sapir’s brief comment (see Section 3.3.3.2), no accounts of possible semantic differences between the whole vs. the whole of have been found in the literature.
68
hypothesises that in noun phrases without of (as in all the children), we are
imagining “an undifferentiated whole, whereas if we retain it [as in all of the
children], we are conceptualising the set as made up of distinct individuals”
(ibid:208, see also J. Hudson 1998:54). Support is taken, inter alia, from the fact
that of cannot be deleted after numerals (e.g. two of the children), which are inherent markers of individuality. The argumentation is, as the author admits,
somewhat vague and would probably require some further consideration. Also, it
only concerns plural noun phrases.
Huddleston & Pullum (2002:434) find a possible subtle difference in meaning
between half and half of, since “the plural is possible, however, in the partitive
construction: both halves of the apple. This indicates that the two variants are
not absolutely equivalent: the partitive allows half to be interpreted in a more
physical sense as ‘half-portion’” (cf. 3.3.1.3).
Langacker, in line with his general cognitive grammar principles, also argues
against the meaninglessness of of. In his view, the difference between all and all
of is that with all of, the notion of subpart–whole is more salient in the speaker’s
mind than otherwise (Langacker 1991b:113). Langacker makes another reflection with respect to NPs with all vs. NPs with all of:
Unlike the other quantifiers, all profiles a mass whose relation to
the reference mass is one of coincidence58, i.e. it exhausts the
reference mass. Thus for all, but not for other quantifiers, the subpart relation between the two masses fails to qualify as a proper
one […] It is hardly surprising, then, that for this quantifier in particular the language might evolve an alternative construction
lacking of […] (Langacker 1999:75)
The idea of an of-less variant “evolving” can be questioned, however, since, as
we saw in Section 3.3.1.1, all of is a much more recent variant than that without
of. Another objection could be that half, which can be used just like all and both
without an of construction, does not fulfil the totalising function that all and both
do, so half does not profile a mass that corresponds to the entire reference mass,
but rather resembles constructions like some of and much of. But if the hypothesis that half the has developed from the nominal construction a/one half of the
(see Section 3.3.1.3) is correct, half constitutes a special case, not really comparable with all and both. It is unfortunate that Langacker does not even mention
half.
Sapir (1930:11), although observing semantic differences between syntactic
variants in the case of all and whole (see Section 3.3.3.2) concludes:
It is not claimed for a moment that the ordinary English uses of
‘the whole’, ‘all of’, ‘the whole of’, and ‘all’ necessarily correspond to our exacting distinctions, merely that they tend to do so.
In actual practice there is considerable confusion.
58
The same applies to both, which Langacker does not mention.
69
69
3.3.4.2
Variants with and without the definite article
There are variables where a quantifier can be used either with or without the
definite article without any obvious difference in meaning:
•
•
•
all men/activity vs. all the men/activity (with specific reference)
all day vs. all the day
both men vs. both (of) the men
Noun phrases including all in combination with a plural or mass noun without a
definite article or other determiner typically have generic reference, as in (3:53).
(3:53) All men are created equal. (Quirk et al 1985:259)
Givón (1993:251) states that “unless used with a definite determiner […] it is inherently generic, or non-referring,”. Halliday (1994:182) also considers all to be
non-specific. The literature does, however, provide examples of all in clearly
specific situations, as illustrated by the following examples.
(3:54) All men in this room are secret policemen. (Allan 1999:10)
(3:55) All activity in the saloon area stops dead. (Putseys 1985:382)
Both these examples include postmodification and it is possible that a specific
interpretation of an NP with simple all (without a determiner) is more likely
when the NP is postmodified. Similarly, in NPs with temporal nouns, there is
one variant with and one without the definite article, even though the latter is infrequent (Putseys (1985:382; Swan 1995:69). According to Quirk et al
(1985:69), “in the negated expression I haven’t seen him all day, only the zero
form is used”.
An NP with all and a plural or mass noun and no determiner can be either
specific or generic, and is generic in the normal case. In contrast, both is always
specific, regardless of whether it is combined with a definite determiner (as in
both the children) or not (as in both children). Besides the of-dropping transformation (cf. Section 3.3.4.1), Seppänen & Seppänen (1986:175, 183) introduce
a rule of the-dropping which is “optional after both [as in both children],
blocked after all [since the dropping of the generally implies a change in meaning from specific to generic, as in all children] and obligatory after the other
quantifiers [e.g. *any the boys].”
In the grammatical literature, the simple both variant (as in both children) is
generally introduced as an alternative to the variant with the definite article.
Leech (1989:66), for instance, writes that “we can omit the after both” (my emphasis, see also, for instance, Seppänen & Seppänen 1986:169f; Berry
1997:138). From a diachronic perspective, this is a natural approach, as simple
both is a later variant than both the. From a synchronic, quantitative and pedagogical perspective, however, both the would best be presented as an alternative
to simple both, since the latter is so much more frequent than the former (see
Section 5.1.1.2).
70
70
In certain cases a knock-out effect (see Section 2.1.3) affects the variation in
NPs with both. This happens when the NP has restricted reference, that is when
there are only two possible alternatives in the world (as with twins, halves etc.).
Both dictionaries and the corpus material suggests that the definite article is not
used in those cases; consequently the only valid alternative is simple both, as in
both halves. Two types of noun tend to have a special status in this case. When
the NP head refers to a body part or is a word expressing kinship (e.g. parents
59
), then a possessive determiner, as in both my eyes/parents, is sometimes a
valid alternative (cf. Quirk et al 1985:258)60). Heine (1997:85f) refers to these
two noun types as being “inalienable possessions”, since they “cannot normally
be separated from their owners”. So, consequently there are four different groups
of NPs with both and a plural noun:
•
•
•
•
3.3.4.3
both – both the – both of the (in NPs with non-restricted reference)
both these/my – both of these/my (in NPs with non-restricted reference)
both – both my – both of my (in NPs with restricted reference where the
noun denotes body parts or kinship)
both (in NPs with restricted reference, where the noun does not denote
body parts or kinship)
Alternating positions of the quantifier
There are two cases of alternating positions of the quantifier in the study:
•
•
a half vs. half a
all of us/them vs. we/they…all
After a few words on a half and half a, the main part of this section will be
devoted to the phenomenon called “floating quantifiers”.
A half vs. half a
As we saw in 3.3.1.3., a half is the oldest variant in this variable. In Jespersen’s
(1909–49:308) view, the half a variant may have originated from cases where
half is an adverb, comparing such noun phrases with those including quite (quite
a…).
Berry (1997:70f) believes that the two variants in fact have slightly different
meanings: half a is used for when an amount is seen as part of a whole (half a
bottle out of a whole bottle). In contrast, a half is used to indicate an established
unit (e.g. a half-bottle). It seems that this distinction holds for some nouns, e.g.
bottle, as in Berry’s example, whereas a semantic distinction is more problematic
with nouns like hour. Is a half-hour a more established unit than half an hour?
59
Some words can both denote kinship and have a more general meaning, such as children. It is
only when the context involves parents and their children that possessive determiners, as in both
(of) their children, can be used. In another context, where parents are not in focus, both (of) the
children would be a more natural alternative to both children.
60
There must be some kind of restriction in terms of definiteness and salience here, since the
simple both form is only natural when the “owner” is situationally salient in the discourse (cf.
von Heusinger (1997:8f) referred to in Section 3.1.4.2).
71
71
This possible semantic distinction will be discussed in Sections 7.3.6 and 7.4.3.2
below.
Floating quantifiers
A particularly interesting type of quantifier construction is the phenomenon
called floating quantifiers. This term is used for cases where a quantifier (all,
both or each) modifying the plural subject of a clause “floats away”. It occurs
either between an auxiliary and the main verb, as in (3:56), or directly precedes a
main verb, if there is no auxiliary, as in (3:57) (cf., for instance, Berry 1997:83;
Quirk et al 1985:126, 258).
(3:56) […], if we keep talking while you keep fishing, the fish will all be gone
and there will be nothing left to talk about’. (IND95) (cf. All /of/ the fish
will be gone…)
(3:57) The banks all said: “But Mr Lloyd, nothing like this has been tried
before.” (IND95) (cf. All /of/ the banks said…)
The quantifier is sometimes placed between the noun phrase and the auxiliary
(as in We both were working late), but this usage is rare and perhaps not entirely
accepted (Quirk et al 1985:126). Hoeksema (1996:58) suggests a third variant,
when a clause contains more than two auxiliary verbs. So, in fact there are three
different positions of the quantifier, as in (3:58a-c):
(3:58a) We all should have been drinking tea.
(3:58b) We should all have been drinking tea.
(3:58c) We should have all been drinking tea.
There are a number of syntactic constraints on how floating quantifiers can be
used. They cannot occur between a main verb and a direct object (*The men saw
all the accident), they cannot float from the main clause to the subordinate
clause of a sentence (*The boys said that it was all raining), and so forth
(O’Grady 1982:520f)61. It should further be remembered that the construction
with a floating quantifier is mainly used in plural noun phrases, even though
O’Grady (1982:538) points out that all can also float when the head of the noun
phrase is a mass noun, as in The linen has all been washed. The phenomenon of
floating quantifiers used with singular NPs is seldom discussed in the literature.
Within generative grammar, a floating quantifier construction has traditionally been regarded as the result of a transformation from a deep structure where
the quantifier is contained within the subject noun phrase, as in [ALL THE FISH
WILL BE GONE] (cf., for instance, O’Grady 1982:522 and Hoeksema 1996:64).
This perspective has influenced the terminology and descriptions in other approaches to grammar as well. Quirk et al (1985:382) write that “When all, both
61
72
The floating quantifier construction must not be confused with cases where all functions as an
intensifying adverb (as in He is all upset.) (Quirk et al 1985:260).
72
and each are postposed (my emphasis) [...] they appear in the position of a medial adverb.” Aldridge (1982:209) uses the term “post-positioning” and Berry
(1998:32) the term “delayed pronouns”. All this terminology (‘floating’, ‘postposed’, ‘delayed’ etc.62) implies that the construction where the quantifier introduces the noun phrase is the unmarked variant and the floating construction the
marked one. Berry (ibid), however, recognises that this might be an erroneous
interpretation. Hoeksema (1996:64f) claims that a number of problems arise with
the traditional generative view. It is also rejected by the generativists Haegeman
& Guéron (1999:231). They assume instead that the underlying structure is
something like [HAVE + [ALL THE GIRLS] BOUGHT THE BOOK], and that in the ordinary case (all the girls have bought the book) the whole NP ([ALL THE GIRLS])
is moved while in the case of a floating quantifier construction (the girls have all
bought the book), only [THE GIRLS] is moved.
In my opinion, two facts contradict the view of the floating quantifier as the
marked variant. First, this form is attested much earlier than the of form, both
with all and with both. Second, the floating quantifier is far more frequent overall than the other variant. Frequency is not the only criterion relevant to markedness (rather a symptom of it), but it is often discussed in relation to this complicated notion (cf. Comrie 1996:6ff).
Another approach is the one where floating quantifiers are analysed as adverbials “which serve as operators on the verb phrase or parts thereof” (Hoeksema
1996:74). This would explain why the quantifiers can be used in sentencemedial position and some other characteristics typical of floating quantifiers (cf.
O’Grady 1982:523; 535ff, Hoeksema 1996:74). In my view, however, this analysis of all is not convincing, since all seems to be quite tightly connected to the
NP even though it is placed in another part of the clause. Also cf. Quirk’s observation in Footnote 61, where adverbial all is contrasted with the quantifying
pronoun all.
Jespersen (1933:185), Reed (1996:167), Huddleston & Pullum (2002:376)
and others offer yet an alternative interpretation: they claim that floating quantifiers are a mere reduction of a partitive construction with redundant material, as
in (3:59). The longer variant exists in Present-day English, but is rare (cf.
Section 5.1.3).
(3:59) The children are both (of them) hungry.
In the literature, the syntactic option of floating is usually only treated in relation
to the three words all, both and each, and not with other quantifiers. According
to Aldridge (1982:209), this can be related to the fact that these three quantifiers
can all be used in appositional constructions of the type illustrated in (3:60). This
is generally not possible with other quantifiers without a semantic change.
(3:60) The men, all of them, carried shotguns. (my emphasis)
62
Long before the generativists, Poutsma (1914:336) talked about all being shifted to another place
(thus expressing the same idea: that the floating quantifier is the marked type).
73
73
Floating quantifiers is a phenomenon that occurs in other languages as well. The
leading Swedish reference grammar (Teleman et al 1999), for instance, refers to
floating quantifiers in terms of their clausal function as “free predicatives” (Teleman et al 1999 II:249). This term is fairly free of associations of markedness.
The authors also comment on the classification as floating quantifiers, suggesting that this term might be less appropriate since there is not always a corresponding variant with a noun phrase introduced by all or both, cf. example
(3:61):
(3:61) Kalle, Stina and Lisa have all left for China. (*All Kalle, Stina and Lisa
have left for China.)
So far, only NPs with a noun as head have been accounted for. The NP head can,
however, also be a personal pronoun, as in (3:62) and (3:63). The alternative is a
noun phrase including of 63.
(3:62) We are all black but we are not one people. (IND95) (cf. All of us are…)
(3:63) We all know what is acceptable. (IND95) (cf. All of us know…)
As mentioned above, floating is seldom discussed with respect to singular NPs.
With personal pronouns, however, floating is frequently used with the third
person singular pronoun it, as in (3:64)64.
(3:64) I realise now that it all seems very naive. (IND95) (cf. All of it seems… )
When the floating construction includes a personal pronoun, this may function
as a grammatical object, as in (3:65), or a prepositional complement, as in (3:66).
This is not possible when all is used with a noun (cf. 3:67 and 3:68).
(3:65) The police gave us all little slips of green paper. (IND95) (cf. The police
gave all of us…)
(3:66) He is the master of us all. (IND95) (cf. … the master of all of us.)
(3:67) *The police gave the people all little slips of green paper. (cf. The police
gave all of the people…)
(3:68) *He is the master of the people all. (cf. …the master of all of the people.)
63
In earlier forms of the English language, a construction with all preceding the pronoun was used,
as in All we are Australians (Jespersen (1933:185). This construction does not, however, seem to
be used in Present-day English.
64
Some NPs (e.g. It all depends how you define major (IND95)) can be ambiguous between an adverbial all reading (‘it depends entirely…’) and an NP quantifying all reading (‘all of it depends’) (cf. Footnote 61). In such cases we need more contextual information to decide which interpretation is the best one, and sometimes not even that may be enough.
74
74
The NP type where the quantifier is connected to a personal pronoun (we all)
rather than a noun (the children all) is seldom dealt with in the literature on
floating quantifiers; an exception is Quirk et al (1985:382). R. Hudson does not
even regard it as a type of floating quantifier, since “it is clear that these words
[personal pronoun + all, both etc.] constitute a phrase” (R. Hudson 1990:283).
As far as the us/you/them all construction (i.e. the one used in object and prepositional complement function) is concerned, R.Hudson’s interpretation seems to
be correct. When used in the subject function, however, we/you/they … all can
indeed occur in different positions with respect to the auxiliary, just like full NPs
(with a noun as head).
A knock-out effect of a syntactic character (cf. 2.1.3) is involved in the choice
between a floating quantifier and the of variant. In a minor clause such as a short
answer, as in (3:69), only the variant with of is possible (Swan 1995:35, 111):
(3:69) ‘How many of her customers’, I wondered, ‘acquired their funds
illicitly’? ‘All of them,’ she replied without hesitation. (IND95) (*They
all.)
Investigations of quantifier floating have mainly concerned their syntax rather
than their semantics. J. Hudson suggests that floating all is used for emphasis or
intensification of a noun phrase, rather than for expressing quantification, as in I
think the boys had all an advantage in that… (J. Hudson 1998:55). This remark
could be compared with Labov’s ideas about all sometimes having an intensifying rather than quantifying function (see Section 3.3.3.2 above). It could
further be hypothesised that with generic reference, the floating construction is
more likely with a predicative verb that expresses a subjective attitude, as in
(3:70) rather than an objective fact, as in (3:71).
(3:70) Men are all the same.
(3:71) ?Men are all mortal.
3.3.4.4
Issues pertaining to the choice between all and whole
This final section brings up various issues pertaining to the choice between all
and whole, both universal quantifiers used (mainly) to express totality. Haspelmath (1995:379) shows their close relationship cross-linguistically: “‘all’ frequently comes from ‘whole’ by way of a kind of metaphor by which a plural aggregate is treated conceptually like a single object”. We will first look at three
concepts, “type of quantity”, “countability” and “divisibility”, all three of which
relate to NPs with plural, mass and singular count nouns in general. We will then
proceed to the topic of animacy in NPs with collective nouns and geographical
names before ending with a brief note on NPs with temporal nouns65.
65
In the discussion of number, countability, divisibility and animacy of the noun in the noun
phrase, it should be noted that it is the combination of word and referent that is intended, even
though either of these components may be more important than the other under certain circumstances. As Quine (1960:90ff) observes, qualities of this kind cannot be considered merely in
75
75
Type of quantity
It is important to remember that all (and indeed also half) has in fact two clearly
different semantic functions, even though both relate to totality. These functions
are manifested syntactically in the ability of all (and half) to combine with plural, singular and mass66 nouns. In contrast, both only combines with plural
nouns67 and whole usually with singular nouns. Sweet (1898:85) makes an important distinction between “continuous quantity”, expressed by words such as
size, big and much, and “discrete/broken quantity”, expressed by, for instance,
number, numerous and many. The continuous interpretation (what Aldridge
1982:233 calls “unitary” meaning) of all is at play in noun phrases where the
quantifier is combined with a mass noun, as in (3:72), or a singular count noun,
as in (3:73). Here, all refers to the complete amount (cf. Putseys 1985:372).
(3:72) Mr Dalyell and other critics contend the Government is engaged in a
cover-up to put all the blame for Lockerbie on Libya. (IND95)
(3:73) They are a mortal danger for Algeria and for all the Arab world.
(IND95)
However, in plural noun phrases, as in (3:74), all generally refers to number
rather than to amount. Here, we talk about discrete/broken quantity (what
Aldridge 1982:232 calls “distributive” meaning68), i.e. ‘every member or separate part of’ (Putseys 1985:372).
(3:74) Here the cultural snobbery was confined to homosexuals, and all the
heterosexuals were philistines. (IND95)
Sometimes all in a plural NP has unitary meaning, as in (3:75), where all does
not refer to number of lives but rather to the “wholeness” of them.
(3:75) Many of these patrons will have known each other all their lives[….
(IND95)
It is only when all has unitary meaning that an NP with whole is a possible alternative (e.g. their whole lives). Whole always has unitary meaning and can therefore never be used in a plural NP with distributive meaning. All in plural noun
terms of either words or referents, since different types of words (e.g. shoes, footwear) can be
used to refer to the same “stuff” in the world.
66
“Mass noun” is only one of several different terms which have been used in the literature to refer
to the group of nouns that are not countable. Other terms are “noncounts”, “uncountables” and
“unbounded nouns” (Svensson 1998:12). This is perhaps not the best term, since not all uncountable nouns are masses (e.g. furniture) (ibid). However, since it is well established in the
literature and less awkward than many of the other terms, it will be used throughout this study.
67
Some language systems make a further distinction between “dual” (‘two’) and “plural” (‘more
than two’), as mentioned by Huddleston & Pullum (2002:334). They also declare that “there are
few places in English where a feature dual is relevant (e.g. in both and either), but the number
system itself simply contrasts singular and plural (‘more than one’).
68
The unitary-distributive distinction can be at play in NPs with both as well, even though the distributive meaning tends to be used more frequently (Huddleston & Pullum 2002:377).
76
76
phrases can be ambiguous, as in (3:76), in that it can either mean ‘all taken together’ (unitary meaning) or ‘all taken separately’ (distributive meaning69). This
is problem discussed in logic and formal linguistics (Jespersen 1933:185).
(3:76) All the boys are stronger than their teacher. (Jespersen 1933:185).
Countability and divisibility
There is another important distinction relating to quantifiers, the one between
count and mass nouns, even though many words can be used with both interpretations, depending on the context (Huddleston & Pullum 2002:334). Syntactically, mass nouns are mainly used with singular agreement in verbs (The equipment is…), determiners (this equipment) and pronouns (the equipment … it).
Quantifiers like all and some are, however, used with both plural and mass
nouns, (all books – all equipment, some books – some equipment), probably
owing to the semantic correspondences between them70. Barrett (1953:137), for
instance, observes that both plurals and mass nouns are “divisible but not multiplicable, the only difference being that plural concepts can be divided exactly
into complete units of themselves, while mass concepts can only be divided into
fractions of themselves”. Similarly, what people perceive as mass nouns can in
fact be interpreted as blurred aggregations of individual items, which become indistinguishable to our eyes or minds (P. Svensson 1998:107, 127).
Apart from temporal NPs, all is quite rare with singular count nouns, as in
(3:77) (Quirk et al 1985:259f; Huddleston & Pullum 2002:375).
(3:77) For the first time ever, it did not buy all the new government debt […]
(IND95)
Whole is clearly the preferred alternative, whereas in combination with mass
nouns “the whole of is less preferred and the whole is unacceptable” (Quirk et al
1985:260). Similarly, Sapir writes that:
In English, totality of an individualized object tends to be expressed as in ‘the whole table’, ‘the whole land was annexed’; and totality of an indefinitely massed object as in ‘all the milk has turned
sour’, ‘all the land was inundated’ (1932:11).
Aldridge (1982:230) states that “whole cannot be used with mass nouns, except
in the special sense paraphrasable as unmodified, pure”.
When all is used with a singular count noun, the noun tends to be divisible,
just like plurals and mass nouns (Quirk et al 1985:259f; Swan 1995:35; Huddleston & Pullum 2002:375). Klégr (1987:28) describe divisible nouns as referring
to “formless, extended and homogeneous” entities, typically used with all,
whereas indivisible ones refer to “something viewed as having definite form,
69
70
It is only in this meaning that all can be substituted by each or every (Each of the boys was
stronger…).
Another similarity is that both can occur as the head of noun phrases without modification
(Svensson 1998:19).
77
77
limits or articulation”, typically used with whole 71. Huddleston & Pullum
(2002:375) provide the following reflection in relation to all and singular nouns:
Universal quantification with count singulars involves quantification over parts, just as in the non-count case. The morning [in all
morning] is a period of time that can be subdivided into smaller
periods, just as sugar can be subdivided into smaller quantities of
sugar. The count singular use is therefore restricted to cases where
there is some relevant subdivision.
One typical group of divisible singular nouns is collective nouns, such as family
and government. All can be used with family, as in (3:78), since family can be
thought of in terms of individual family members. (Quirk et al 1985:259).
(3:78) After the funerals, all the surviving family met for a meal. (IND95)
Another group is temporal nouns, which are divisible into smaller units, as in the
example described by Huddleston & Pullum above. An NP such as *all her
brother, however, would be ungrammatical (Huddleston & Pullum 2002: 375).
The concept of divisibility is not, however, entirely straightforward. A problematic issue is what can be regarded as divisible and what cannot. Collective
and temporal nouns are clearly divisible, and a noun like book is quite easily
imagined as divisible into chapters. On the other hand, most things can be regarded as divisible in some way; a chair, for instance, though generally thought
of as indivisible, could be divided into the parts from which it has been constructed. One possible definition is that the noun should be divisible into similar
parts72 (e.g. the chapters of a book or the tracks of a CD album) rather than into
different parts (e.g. the body parts of a human being or the different tools in a
tool kit). This is the definition adopted in the analysis of divisibility in the singular count nouns occurring in the corpus material (see Section 7.3.3). Quirk et al
(1985:260) further claim that the possibility of combining all with divisible singular count nouns mainly concerns abstract nouns. They are more sceptical towards constructions including concrete nouns (e.g. ?all the book 73), except (possibly) if the quantifier rather than the noun is stressed (as in I haven’t read 'all
the book). Quirk et al (ibid) give all of the book as a perfectly natural alternative
to the whole book, so the all of variant is obviously quite different from the one
without of (Quirk et al 1985:260). We will come back to this in Section 7.3.3.
As mentioned above, whole is the dispreferred alternative with mass nouns,
since this quantifier generally is used for an indivisible unit (Quirk et al
1985:260). There are exceptions, however, such as the more or less fixed phrase
the whole truth (Huddleston & Pullum 2002:375, see Section 7.3.2). Here, one
71
Cf. Sapir’s (1930:8f) classification of constructions with all and whole in terms of divisibility,
where he gives all the table as an example of “a normally undivided existent”, whereas the whole
table is regarded as “divisible into parts but resisting such division” (see Section 3.3.3.2).
72
Just as in the case of mass nouns, this similarity concerns kind rather than size.
73
Interestingly, Quirk et al (1985:260) categorise the noun book as a divisible noun, although regarding it as not compatible with all, because of its concrete nature. Klégr (1987:28), however,
gives book as an example of an indivisible noun, “having definite form, limits or articulation”.
78
78
could perhaps argue that truth is seen as a definable, indivisible unit (generally
the truth of statements concerning a particular case in court) rather than an unlimited mass. This is in line with Huddleston & Pullum’s (2002:337) explanation
in terms of abstract vs. event instantiation of certain nouns. Some nouns, e.g.
discussion, can have either an abstract mass interpretation, as in (3:79), or be
used with an individuating function to refer to a particular event, i.e. a count
noun, as in (3:80).
(3:79) Labour likes all the discussion of community – it’s territory the Tories do
not know how to fight on. (IND95)
(3:80) The whole discussion about the exact amount cannot be concluded unless
we have the next volume of the report […] (IND95)
Animacy
According to Dahl & Fraurud (1996:47), the systematic exploration of animacy
and its relation to grammatical systems is a fairly recent research area. One example of this relation relevant to the present study is the link between syntactic
function and animacy. Subjects (especially in transitive constructions) are more
likely to be animate than objects, and indirect objects are more likely to be animate than direct objects (ibid:47ff).
Animacy is usually discussed either in terms of an animate–inanimate dichotomy or a hierarchy of degree of animacy, from HUMAN to ANIMAL to INANIMATE
(ibid:47; Yamamoto 1999:2). Yamamoto (1999:14, 22) stresses that the hierarchy is best analysed as a prototype system, where “(individual) human beings”
are at the core. The distinction between animate and inanimate is therefore
fuzzy, and one reason, observed by Dahl & Fraurud (1996:62), is that “we have
the possibility of sometimes treating inanimate entities as persons”. Typical examples are metonymical extensions of geographical names, as in (3:81), and collective nouns, as in (3:82).
(3:81) Now, all America hopes painful chemotherapy needed for the lung
cancer will yield a similar result. (IND95)
(3:82) Not all the Shadow Cabinet were watching on Sunday evening. (IND95)
A noun of this type can sometimes be conceptualised as denoting a body of individuals that can initiate actions (e.g. make decisions), have emotions etc.
Yamamoto (1999:16ff) refers to such cases as having “inferred animacy”. In his
corpus investigation of animacy in English and Japanese, however, the tokens
were mainly classified as inanimate, since “organisations and geographical entities/local communities are not living things but social units, and, in this sense,
they are not ‘animate’ in the light of animacy in the literal sense or ‘animacy per
se’” (ibid 138f). The only instance he analyses them as animate is when some individuals speak for their companies, countries etc. (ibid 139). In contrast, Levin
(2001:126) classified all his tokens of collective nouns as animate, regardless of
their syntax, i.e. if they occurred with singular or plural agreement in verbs and
pronouns. In the literature on collective nouns, the difference between choosing
79
79
singular agreement, as in (3:83), and plural agreement, as in (3:84), is often described to be a difference in the speaker’s perspective, i.e. whether the focus is
on the collective as a (typically administrative, thus inanimate) entity or the individuals in the group 74.
(3:83) Critics argue that the Government is not a neutral umpire, but is tied to a
number of controversial policies […] (IND95)
(3:84) He says straight out that the government are all mafia. (IND95)
Similarly, a geographical name can either refer to a geographical or political
area, as in (3:85), or to its inhabitants (or representatives of its inhabitants, such
as a government or a football team)75, as in (3:86).
(3:85) To make matters more confusing, this brand of Smirnoff, unlike all the
others, is actually distilled in Russia.(IND95)
(3:86) Russia has offered up to 20,000 troops but has insisted that they will not
come under direct Nato control. (IND95)
In my analyses of animacy in quantified NPs, I have followed neither Yamamoto
nor Levin. Instead, I distinguish between animate and inanimate cases by
looking at markers of animacy, such as verbs for thinking, feeling and acting
(see Section 7.3.4). I also distinguish between “animacy proper” and “inferred
animacy”.
Neither Dahl & Fraurud nor Yamamoto bring up syntactic variation in quantified NPs in relation to animacy, but both geographical names and collective
nouns are brought up by the comprehensive reference grammars. There are three
ways of forming NPs with geographical names and a totaliser: all Finland, all of
Finland and the whole of Finland 76 (Quirk et al 1985:260). Quirk et al (ibid)
further remark that the zero construction is rather formal and is used for the animate interpretation, where the NP referent is seen as the population, as in All
Paris welcomed the General. Swan (1995:38) writes that “we can use all with
place names […] to mean ‘every part of’, ‘the whole of’”, as in “All London was
74
The grammar of collective nouns is not a universal grammatical feature. In Swedish, for instance,
where verbal inflection according to number is not used nowadays, the difference between the
two perspectives can instead be expressed syntactically by means of inflection in a predicative
adjective, as in Familjen var intresserad (‘The family was/were interested-SING’) vs. Familjen
var intresserade (‘The family was/were interested-PLUR). In many languages plural agreement is
never or very seldom used (Levin 2001:39f).
75
Yamamoto (1999:139) suggests that there is a difference between this construction and an NP
with a nominalised adjective, as in after the Russians invaded Afghanistan (IND95), which has a
higher degree of individuation than the NP in (3:86). Judging from the corpus examples presented here, this difference does not seem to be absolute, since there seems to be little difference
between the two examples as far as individuation is concerned.
76
Note that whole can only be used in its nominal function (not *the whole Finland). A plausible
explanation for this is that geographical names normally do not take a definite determiner when
preceded by a modifying adjective (cf. *the beautiful Italy etc.), and whole is never used without
a determiner.
80
80
talking about her affairs”. This definition is fairly vague, but the example implies that, like Quirk et al, he is, in fact, thinking of the population rather than
the geographical area. The whole of Paris can be used in not so formal situations
about the population, but more often is used for the area indicated by the geographical name (Quirk et al 1985:260).
We saw above that another area where animacy and grammar interact is in
agreement with collective nouns. Agreement is usually examined with respect to
verbs, personal pronoun and relative pronouns, whereas quantifiers are less seldom mentioned as agreement markers. Huddleston & Pullum (2002:375), however, suggest that all foregrounds the individual people in a collective, while
whole has more of a unifying effect; therefore, it is more likely when the NP referent is regarded as a more abstract, inanimate entity. As a consequence, all
should be more compatible with plural agreement of the predicate verb than
whole. This is corroborated by Levin (2001:124) (see also Section 6.1).
Time division in NPs with temporal nouns
J. Hudson (1998:119) makes an interesting attempt at explaining the use of all
vs. the whole in NPs with temporal nouns, viz. that all is only a valid alternative
to the whole when the noun refers to a natural category of time (see Table 3.6).
Table 3.6. Quantifiers in NPs with temporal nouns (based on J. Hudson
1998:119)
the whole
all/the whole
the whole
second, minute, hour
morning, evening, afternoon
decade, century
day, night
week
month77
season, winter, summer, term
year
“Days, nights, mornings and evenings, months, seasons, and years can be
directly perceived without measuring instruments since they are non-arbitrary
measurements of time defined according to the movements of the planets” (J.
Hudson 1998:119). Although these words and concepts are perhaps “natural” to
a higher extent, a human is needed to impose a somewhat arbitrary limit: for example, when does a morning begin and end? This problem is not dealt with by J.
Hudson. J. Hudson further assumes that the “natural” nouns have a longer history than the man-made ones. Why all should be more natural with the natural
time expressions is not quite clear from her argumentation (J. Hudson
1998:120). The explanation of the fact that week and term can be used with all
is, in Hudson’s view: that week is based on two other natural categories, days
and nights, and that term is based on the natural seasons (J. Hudson 1998:119).
My viewpoint is that decade and century are based on natural categories in the
same way as year. However, nouns based on another natural category, such as
century, decade, term and week, were categorised as arbitrary.
77
Month is a borderline case, but since it is related to the rotation of the moon, it is categorised as
natural by J. Hudson.
81
81
3.4
Summary
This chapter has provided an overview of how the English noun phrase, quantifiers in general, and (with more specification) all, whole, both and half in
particular have been described and discussed in the linguistic literature. This has
been done both from a syntactic and a semantic point of view. The first section
focused on the noun phrase: its different parts (head, determiners and modifiers)
and the many syntactic functions that the NP can take on, e.g. subject, object,
prepositional complement and adverbial. The section also presented an
alternative analysis of the noun phrase, as a determiner phrase (the DP-analysis),
whereby the determiner is seen as the most important part. The final sections
discuss some central concepts associated with the noun phrase: reference (specific and generic), definiteness and animacy. Section 3.2 gave a brief outline of
various aspects of quantification within three theoretical branches of linguistics:
formal semantics, cognitive grammar and generative grammar.
Section 3.3 examined all, whole, both and half from various perspectives. We
first looked at the words diachronically and established the first attestations of
the variants in the OED. The historical part also proposed a hypothesis for a
grammaticalisation process for half the. Next, some syntactic aspects, classification problems, and syntactic differences between the words were considered.
I concluded that in spite of all, both and half often being treated as if they were
very similar, there are, in fact, several syntactic differences between them.
We then moved to semantic aspects, the first of which was definiteness. All,
whole, both and half are sometimes labelled “indefinite”, sometimes “definite”
and sometimes neither in school and reference grammars. It is possible that definiteness is not intrinsic to the quantifiers themselves, but rather a quality that depends on the context. The section on totality showed that there may be other semantic and pragmatic functions for all besides the totality meaning proper, such
as a large-quantity meaning (my suggestion) and an intensifying meaning
(Labov and others). The third part brought up the different variation patterns analysed in the present study and discussed them from syntactic and semantics perspectives. The topics here concerned, among other things, variation between NPs
with of and NPs without of, floating quantifiers, countability, and animacy.
Quantifiers in general and all in particular involve many interesting aspects.
We will have reason to return to the topics raised in this chapter when analysing
the linguistic factors in Chapter 7. Before the presentation of the results of the
corpus investigation, the next chapter provides an overview of the material and
method used for the study.
82
82
4. Method and material
4.1
A corpus-based approach
Corpora, that is collections of authentic text, have a long and established history
of use in linguistic research. In the sixties some pioneers started to use computers for the storage of corpora. It is only in recent decades, however, that computerised corpora have become widely available. Corpus linguistics has experienced a veritable boom since then; more and larger corpora have been created
and search software has become more sophisticated yet easier to use. As a direct
consequence, more and more linguists within various traditions use computerised corpora. The common denominator is that corpora help us detect facts about
language use that our intuition fails to notice (cf. Sinclair 1991:39; Biber et al
1998:3). This is particularly important when the researcher’s native tongue differs from the language being investigated.
There is a plethora of areas within linguistics where corpora can be a useful
tool: syntactic and lexical variation and change, lexicography, text linguistics
and many more. A practical application is the improvement of reference works
and teaching materials. The pedagogical aspect has been emphasised by e.g.
Mindt (1997:50), who states that ”corpus-based studies of grammar geared to foreign language teaching can do much to bring the teaching of English into better
accordance with actual language usage”. Frequency analyses will, for instance,
help writers of materials for lower levels decide what features should be given
more focus than others (Kennedy 1992:365f; 1998:290; Sinclair 1997:31) and
corpus-based information about, for instance, language variation can help advanced learners towards near-native proficiency.
There can also be dangers in using corpora. The speed with which computers
produce thousands of examples, the enormous number of words behind the
figures and the neat concordance lines provided may give the incorrect impression that what appears on the computer screen is the whole truth. However, we
must not simply replace one authority (a dictionary or a grammar) with another
(the corpus); results should always be interpreted with caution and a critical
mind. The question of representativeness, for instance, is an important and difficult one (cf. Kennedy 1998:72). Chomsky certainly had a point when, in Syntactic Structures, he criticised corpora for not being representative: “any natural
corpus will be skewed” (1957:159). Indeed the corpora of today are much larger
and more complex than they were at that time, but it is still impossible to claim
that a corpus is representative, since the very concept of representativeness is so
complicated to define. Who, for instance, speaks a kind of English that is
83
83
representative of the rest of the English-speaking population? Do we produce
and receive more spoken language or more written language (Kennedy
1998:68ff)? However, for the same reason, it is equally impossible to claim that
a corpus is not representative. The choice that is made when a corpus is created,
or when a particular corpus is chosen for a particular study, is always to some
extent a subjective choice from the total (and endless) population of linguistic
utterances. Still, in a sense, a corpus at least yields more objective results than if
we rely exclusively on intuition.
Ever since Chomsky’s Syntactic Structures, corpus linguists and so-called
“armchair linguists” (cf. Fillmore 1992:35), those who rely on introspective data
only, have argued about which method is more useful. Another issue of debate
concerns whether corpus linguistics involves a theory in its own right, or
whether it is just another linguistic method. Kennedy (1998:7) is one of those
who hold that corpus linguistics is not a theory, but rather a tool that can be used
by scholars within many different schools of linguistics. This view is questioned
by Tognini-Bonelli (2001:1), who argues that the use of corpora has offered linguists a quite new understanding of language. She distinguishes between two
types of corpus linguistics: corpus-based research and corpus-driven research(cf.
G. Francis 1993). Within the first type, towards which Tognini-Bonelli (2001:
65ff) is critical, corpora are mainly used for corroborating pre-existing theories,
simply replacing introspective data with something else. In studies of this kind,
she maintains, corpus data that do not fit the researcher’s hypotheses are often
ignored. Tognini-Bonelli (2001:84ff) instead advocates the use of corpus-driven
research, where the researcher starts out without pre-existing theories (i.e. an
entirely inductive method). The corpus is then in focus from the start and the
researcher is committed to the integrity of the data. If the data does not fit one’s
hypothesis, then the hypothesis must be changed. In practice, an entirely
inductive corpus approach is virtually impossible, since each question one asks
the corpus must originate in some previous ideas (Stubbs 1996:46f). Most
linguists (including myself), however, tend to use the term “corpus-based” with
less negative connotations than Tognini-Bonelli does, regarding a study as
corpus-based if corpus data are used to either corroborate or refute a hypothesis.
In this investigation corpora are used in two different ways. In some cases, I
start out with previous claims made by other linguistic scholars (e.g. Quirk et al
1985) or writers of usage guides (e.g. Swan 1995), or I have a hypothesis based
on my own intuition. The corpus data are then used to corroborate or refute these
claims and hypotheses. In other cases, I go direct to the corpus data to see
whether previously not discovered patterns can be found. The second method of
working with the material is obviously more corpus-driven (in Tognini-Bonelli’s
terminology). However, also in those cases where there is a pre-existing hypothesis, I am highly committed to the integrity of the data. Consequently, if the
data do not fit the hypothesis, this is accounted for and discussed.
84
84
4.2
Corpus material
Five different corpora were used, for two reasons: (i) in order to compare regional varieties and spoken versus written English, and (ii) in order to minimise
the risk of obtaining skewed results with respect to overall distribution, regardless of region or medium, owing to bias in a particular corpus. Table 4.1 shows
the different corpora used.
Table 4.1. Corpora used in the study
American English
Australian English
Corpus
Size
Corpus
Size
ConversaLSAC
5M
tion
53 M SMH95
37 M
Newspaper NYT95
text78
British English
Corpus
Size
BNC
8M
IND95
43 M
Abbreviations: LSAC = Longman Spoken American Corpus; NYT95 = The New York
Times on CD-ROM, 1995; SMH95 = The Sydney Morning Herald on CD-ROM, 1995;
BNC = The British National Corpus (the dialogue component); IND95 = The Independent on CD-ROM, 1995; M = million words.
There are several reasons for using newspaper corpora (or archives as, for instance, Kennedy (1998:57) prefers to call databases that were not compiled for
linguistic purposes). First, I was interested in looking not just at the two most
widely explored regional varieties of English (British and American), but also at
another variety. At the time of the material collection, The Sydney Morning
Herald was the only non-British/American English material available to me. A
second reason was that newspapers, being read by so many people, are likely to
be influential language-wise. As Bell (1991:1) writes: “Media are dominating
presenters of language in society. Within the media, news is the primary language genre.” Biber et al (1999:16) point to the usefulness of newspaper corpora
in the research into regional variation, since they “provide one of the best reflections of American English v. British English dialect differences79 in writing”.
They are therefore more suitable than, for instance, academic prose, which is
“typically written for an international audience with relatively little influence
from the national dialect of the author”. A further reason was that the two existing comparable corpora of recent British and American written text (FLOB and
Frown80) were too small to yield enough tokens for some of the variables of interest in this study.
78
The figures on number of words in the newspaper corpora are from Minugh (2000:63).
It is hoped that IND95, although published in England, will give a sufficiently fair reflection of
the use of quantifiers in the whole of Britain, not just in England, and that, similarly, NYT95 and
SMH95 will give a fair picture of quantifier use in the U.S. and Australia.
80
These two corpora were compiled according to the same principles as the well-known LOB and
Brown corpora, one-million-word corpora with material from the 1960s. FLOB and Frown therefore comprise the same text categories but with material from the 1990s (also see 2.4).
79
85
85
Two corpora of spoken English were used for comparison between the
written and spoken media: the Longman Spoken American Corpus (LSAC) and
the greater part of the spoken component of the British National Corpus (BNC).
The spoken component of the BNC is divided in two different ways. First, the
corpus contains one “demographic” part, which mainly consists of every-day
conversation, and one “context-governed” part, which is comprised of more
planned talk in the form of lectures, sermons, business meetings etc. Second, the
corpus is divided into “dialogue” (both natural conversation and context-governed conversation) and “monologue” (Aston & Burnard 1998:31). LSAC contains only dialogue material, both natural conversation and more context-governed text types, which cannot be separated by the researcher. As a consequence, I chose to use the dialogue part of the BNC, thus excluding monologue, rather than using the demographic part, which would have resulted in all
context-governed conversation having been excluded. This decision is an
attempt at using corpora of spoken British and American English that are as
comparable as possible, given the material available. Unfortunately, no corpus of
spoken Australian English was available at the time of the material collection.
There are a great number of different registers of both written and spoken language (cf. e.g. Biber 1988) that could have been used as material in the present
study. Nevertheless, conversation and newspaper language are both, as pointed
out by Biber et al (1999:9), among the most familiar kinds of speech and writing,
since most people regularly talk spontaneously to other people and (at least in
the western world) read newspapers. In many ways, Biber suggests, they are
each other’s opposites, newspapers being carefully edited, relatively objective
presenters of information81 and conversation being spontaneous, unedited, personal and interactive (ibid). On the other hand, newspaper text is generally more
similar to conversation than some other written channels, and considered a “fast”
genre, where linguistic changes are reflected relatively quickly (Mair (1998:155,
see Section 2.4). Furthermore, newspapers contain a great deal of “spoken”
(although often edited) language in the form of dialogue from interviews. One
should also remember that both conversation and newspaper text contain a
number of different subregisters with different language styles.
4.3
Procedure
4.3.1
Corpus queries
The first step was to find quantified noun phrase types exhibiting syntactic variation. A number of school and reference grammars were consulted, and Quirk et
al (1985: 253f, 376f, 388) proved to be the most useful source in this respect.
After an initial pilot study, a list of corpus queries was compiled. The complete
list of variables is to be found in Appendix A. The following list is a less finegrained presentation of the different types of variables analysed.
81
86
Another objective and carefully edited register is of course academic prose, but no large corpus
of written American English is yet available.
86
•
•
•
•
•
•
•
all/whole, both and half in combination with a common noun (singular
count, mass or plural) (e.g. all the children vs. all of the children, both
boys vs. both the boys vs. both of the boys, half my book vs. half of my
book)
all, both and half in combination with a demonstrative pronoun (e.g. all
these vs. all of these)
all and both in combination with a personal pronoun (e.g. we (…) both
vs. both of us)
half in combination with the indefinite article and a singular noun or
numeral (e.g. half a bottle vs. a half bottle)
all/whole in combination with a geographical name (e.g. all China vs.
all of China vs. the whole of China)
all/whole in combination with a temporal noun (e.g. all morning vs. the
whole morning 82)
both in combination with a noun expressing body parts or kinship (e.g.
both (/of/ my) eyes
The group of NPs with temporal nouns is a subgroup of the variable with
all/whole and singular count nouns. They were separated from the larger group
owing to their showing a different variation pattern, including a variant without
the definite article, as in all morning. The same applies to both with a noun for
body parts or kinship, since in this case a possessive determiner is the only possible alternative, not the definite article.
Searches were carried out by means of the concordance program Wordsmith
and/or by means of special software coming with certain corpora (see Appendix
C for details about the different search procedures). Since all the corpora used
(except the BNC) are syntactically untagged, the searches were made for lexical
words (e.g. all + of + the). Virtually all standard-English variables involving the
quantifiers all, whole, both and half accounted for by Quirk et al (1985) will be
covered in the present study. Exceptions are described in Section 4.3.2 and Appendix B.
4.3.2
Exclusions
Some variables, variants and tokens were excluded for various reasons. First,
one whole variable was excluded:
•
all, both, half /of/ + PROPER NOUN IN THE GENITIVE + COMMON NOUN (as
in all /of/ John’s friends)
The reason is that, in the case of all, it would be very difficult to extract tokens
of the first of the variants (simple all) from the corpus material, since all is such
a frequent word and there is no other keyword to use in a corpus query. This
problem would not have existed, had the corpora used been tagged for syntax. In
82
These are words that express particular units of time, so the group does not include all the time.
87
consistency with the all variable, instances of both and half in NPs with proper
nouns in the genitive were also excluded from the study.
Second, in one of the variables (all /of/ + PLURAL NOUN), one possible variant
was excluded for the same reason:
•
simple all + PLURAL NOUN (as in All children came to school today)
Most frequently, this construction has generic reference (cf. Quirk et al
1985:259), but it can also be used with specific reference, as in the example
above. In those cases, it is an alternative to the all /of/ the variants. One possible
variant was also excluded in two other variables:
•
•
+ PLURAL NOUN … all (floating quantifier) in subject
function (as in The children were all happy), as an alternative to all +
DETERMINER + PLURAL NOUN and all of + DETERMINER + PLURAL NOUN
the + PLURAL NOUN … both, (floating quantifier) in subject function (as
in The children were both happy), as an alternative to both + PLURAL
NOUN, both the + plural NOUN and both of the + PLURAL NOUN
DETERMINER
In all the cases above, including them in the study would have entailed too much
manual work with semantic analysis of the co-text. For this reason, they were
excluded. The fact that not all alternative variants have been included need be
taken into account in the interpretation of the results.
Third, a number of tokens generated by the corpus queries had to be excluded. because their syntactic structure did not correspond to the structure investigated, as exemplified in (4:1):
(4:1) Last week it happened, but [not at all] [the way] I had expected. (SMH95)
Similarly, in the case of half a/an vs. a half, those tokens were excluded where
the noun phrase included a noun in the genitive (not a measurement word) and
the quantifier determined the head noun of the matrix NP rather than the genitive
noun, as in (4:2)
(4:2) But in the experiment, adolescent hamsters were placed in the cage of a
mature one, thus violating its territory, for an hour a day over a week’s
time – about half a hamster’s adolescence. (NYT95)
The reason for the exclusion is that the a half variant cannot be used in this case
(*a half hamster’s adolescence). Cases of a half X were excluded in constructions like pounds 3.50 a half-pint, since the indefinite article is here used in the
same way as an in 20 miles an hour etc. The half a/an variant is out of the question and competition cannot occur.
Fourth, tokens where competition was ruled out by a knock-out factor were
also excluded from the statistics, since the relationship between variants is always 100–0 in such cases (see Section 2.1.3). All of these cases are accounted
for in Appendix B. One of these is exemplified in (4:3). Here, some requires an
of construction, with the consequence that all the victims cannot be used.
88
88
(4:3) Fire experts say that if the door had been accessible or the sprinkler
system had worked, some or all of the victims might have escaped before
being overcome by smoke and carbon monoxide. (NYT95)
Fifth, a number of tokens were excluded from the statistics as fixed expressions
(see further Section 2.1.4), as in examples (4:4) and (4:5).
(4:4) Walter and his crew, featuring Augustin Rodriguez as his ultra-smooth
sidekick, start searching for Gwen all the while investigating a case of
international computer bank fraud. (SMH95)
(4:5) Getting you into the shop is half the battle. (SMH95)
The basis for the classification as fixed expressions was (i) that they could be
found in one or more dictionaries consulted, and (ii) that there was no variation
in the corpus material. All fixed expressions are accounted for in Appendix B.
As mentioned in Section 2.1.4, some expressions, even though regarded as fixed
expressions in dictionaries, were included since they showed variation in the
corpus material. NPs that belong to this group are also accounted for in Appendix B.
Finally in the variable with all and a personal pronoun (e.g. all of us vs. we …
all), a number of tokens were excluded for practical reasons. This was the case
when the NP included a postmodifier, since, in the floating quantifier variant
(we… all), the search procedure would be complicated owing to the fact that a
postmodifier can be infinitely long. Therefore, all tokens (of both variants) were
excluded when containing a postmodifier.
4.3.3
Sampling
For the comparison of frequencies of syntactic variants overall, regionally and
according to (written or spoken) medium in Chapters 5 and 6, all tokens found
and considered relevant were used. In the majority of the linguistic factors presented in Chapter 7, however, samples from the total “population”83 of the
tokens found in the British and American newspaper corpora were used in all
cases where a variable yielded more than 100 tokens per corpus and variant. The
greater part of the study of linguistic factors required manual analysis of each
individual token, and it would have taken far to long to go through all these
thousands of tokens.
Bell, when sampling newspaper language, used a method referred to as “the
constructed” week:
[…] the technique involves selecting days by a random process
from several weeks to make up a composite week of days from
83
In order to avoid using the word sample ambiguously, I use “population” to refer to the entire
material (of a particular variable) from the two newspaper corpora, fully aware that this is a
sample itself, i.e. a sample of the total population of all linguistic utterances.
89
89
five (for a Monday to Friday universe) or six (Monday–Saturday)
different weeks (Bell 1991:23).
Bell’s method could have been used with the newspaper corpora. For the research conducted in this study, however, it would have been unnecessarily complicated; words like all, both etc. are not likely to be affected by the time of the
year, whereas certain vocabulary (e.g. relating to Christmas or concerning sports
practised only during parts of the year) definitely can be. Instead the following
sampling method was used:
•
For each variable (e.g. [all + PLURAL NOUN]), a stratified84 sample of
200 tokens per variant85 was created. That is, if a variable comprised
two variants (e.g. all and all of), there were 400 tokens in the samples,
if the variant comprised three variants (e.g. all, all of and whole), 600
tokens were sampled and so on.
•
For practical reasons, not all the five corpora were used for the samples.
However, half of the tokens (100 per variant) were taken from The New
York Times and half from The Independent. Factors that seem to influence the choice of variant were discernible in more than one corpus.
The reason for choosing the newspapers rather than the spoken corpora
was that spoken text is often more complicated to analyse, especially
from a syntactic point of view.
•
In order to randomly select tokens for each variant (e.g. [all + PLURAL
NOUN]), all tokens (the keyword plus its surrounding text) from each of
the two newspaper archives were collected in one file each. These files
were then resorted (in the Wordsmith search program) in alphabetical
order after the fifth word to the left of the keyword. Thereafter, the first
100 tokens of each concordance list were extracted
The reason for using stratified samples rather than proportional samples is that
for many of the variables one of the variants was in an overwhelming majority.
Had a proportional sample (for instance 400 tokens per variable) been used,
there would have been very few tokens of some of the variants (e.g. 390 all and
10 all of). This would probably have obstructed observations about the possible
influence of linguistic factors on the choice of variant.
Besides the sampling of tokens from the entire corpus material, another kind
of exclusion was carried out for the analysis of linguistic factors. In order to be
able to take as many factors as possible into account, a selected choice of variables from Chapters 5 and 6 was made. For instance, in the case of NPs with personal pronouns (all of us vs. we/us all), only the tokens with all (not those with
84
In a stratified sample the number of tokens is the same for each variant, regardless of how many
tokens there are in the total population. The opposite is a proportional sample, where the proportions between variants are the same as in the total population.
85
In cases where a certain variant occurred less than 200 times, all tokens were used, and the analysis was adapted to this circumstance.
90
90
both) were included. Similarly, in NPs with demonstrative pronouns as heads,
only the tokens with all were included. All excluded cases are accounted for in
Section 7.1.
4.3.4
Data analysis
The overall frequency distribution is presented in Chapter 5. Here, the main purpose was to discover the predominant form of each variable, and to assess
whether some variants are very rarely used. As mentioned above, a large number
of examples were excluded as fixed expressions. In order to ensure that these deletions did not affect the results, I conducted a few frequency counts where the
fixed expressions were included. Chapter 6 presents the results of the studies of
non-linguistic factors: the first part compares British, American and Australian
English, and the second part deals with the spoken and the written material.
The investigative process had the following format. After extracting samples
of 200 tokens per variant (half British, half American), all the sample tokens
were imported into a database program (FileMaker Pro). Here, all the tokens
were classified in terms of various linguistic aspects (e.g. syntactic function in
the clause, the presence or absence of focus markers). For each variable, the distribution of syntactic variants in one category was compared with the distribution in the other category or categories. For instance, if we look at a certain factor (e.g. the presence of X) and find – in the two 100-tokens samples – that the
distribution between all and all of is 40-20 in one category (presence of X) and
60-80 in the other (absence of X), it can be assumed that there is a positive correlation between all and the first category and between all of and the second category (see further, Section 7.1). According to the null hypothesis (the expected
frequencies provided there is no correlation at all between categories and
variants), in each category the number of tokens would be the same for each syntactic variant86. In other words, judging by the data, this particular factor (presence of X) influences the choice of variant87.
Now consider the following – constructed but hopefully clarifying – example.
In a stratified sample of 400 tokens from the variable [all /of/ + PLURAL NOUN],
200 tokens are of the of-less all form, whereas 200 are of the all of form. In the
total ”population” (10,000 tokens), there are 90% all (9,000 tokens) and 10% all
of (1,000 tokens). The sample tokens are then analysed for the presence of focus
markers (such as nearly preceding the quantifier) in the NP or its near co-text.
The analysis yields the result presented in Table 4.2.
86
When there were fewer than 100 tokens of a particular variant in a particular variable, the frequencies were normalised to 100 on the basis of their internal distribution into different categories. For instance, if one variant comprises only 50 tokens and these are distributed over two
categories (14 in category A and 36 in category B), the normalised frequencies will be 28 and 72
respectively. This measure makes it easier to compare the low-frequency variants with the 100token samples of higher-frequency variants. We must, however, be aware that normalised frequencies may give a skewed picture of reality. For the chi-square testing, only absolute figures
were used.
87
Observe that, as mentioned in Section 2.1.2, only the correlations observed in both American and
British English are presented in tables and graphs, while correlations found in just one variety are
merely mentioned in the running text.
91
91
Table 4.2. Outcome of an analysis of the correlation between the presence
of a focus marker and all (constructed example)
all
all of
Total
N
%
N
%
N
%
+FM
50
25%
150
75%
200
100%
150
75%
50
25%
200
100%
−FM
+FM = the presence of a focus marker, −FM = the absence of a focus marker
At first sight, the figures suggest that all of is more frequent than simple all
under the +FM condition (when a focus marker precedes the quantifier). This
would, however, be an erroneous conclusion. If it is assumed that the distribution of the sample reflects the distribution of the total “population”, simple all
would still be more frequent in the total “population” (25% of 9,000 is 2,250)
than all of (75% of 1,000 is 750) under the +FM condition.
What the figures do show is that the +FM condition promotes the use of all
of. Under the +FM condition, all of is used in 75% of the cases, while according
to the null-hypothesis it would be used in 50% of the cases. Similarly all is used
in 75% of the cases under the −FM condition. In other words, it could be argued
that there is a positive correlation between +FM (presence of FM) and all of and
a negative correlation between +FM and simple all. This correlation may not
have been so easily spotted if absolute figures had been used, since simple all is
so much more frequent than all of. In Chapter 7 the same procedure will be used
with real figures.
The results of the analyses of linguistic and non-linguistic factors were tested
by means of the chi-square significance test88. This test checks the likelihood of
a correlation between a variant and a factor category occurring by chance (in
which case the null-hypothesis is not falsified) or not. If the test yields a small pvalue (e.g. 0.01) there is less than a one-in-100 chance that a correlation has
occurred by chance. Thus a relationship between factor and variant may exist.
The lower the value, the smaller the likelihood that the difference is a matter of
chance, but the significance borderline is often placed at 0.05 (one chance in 20).
One problem of using significance tests with very large figures is that even
small differences may seem to be highly significant. On the other hand, this
could also be regarded as something positive, since with a very large material
one can detect differences that would have gone unnoticed had the material been
smaller. It is important to keep in mind, however, that the only thing that statistical significance really tells us is that it is unlikely that a difference in frequency
is random; it says nothing about whether this difference is important or about
how it could be explained.
Besides yielding information on which correlations between variants and factors were statistically significant (and therefore relevant to discuss), the chisquare test can be used to compare the relative strength of different correlations
(see Chapter 8). The basis for the claims about relative correlation strength was
88
92
As mentioned in Footnote 87, only correlations occurring in both IND95 and NYT95 were included in the presentation of linguistic factors. The figures from the British and the American
material were conflated in the significance test, and the respective frequencies were sometimes
quite different even though the same tendency was observed in both corpora.
92
the phi coefficient, which is the standard instrument for measuring the strength
of correlations with this type of data. A table in Appendix E provides the exact
phi coefficient for all the statistically significant correlations.
For the type of analysis used in this study, where the isolated effect of one
factor category at a time was investigated, the chi-square test was best performed
on four-cell tables. In many cases in the material, there were only two variants
and two factor categories, which makes the significance testing straightforward;
if one variant correlates with one factor category, the other variant automatically
correlates with the other, as illustrated in Table 4.3.
Table 4.3. Four-cell table used for testing the correlation between two variants and a two-category factor
Variant X
Variant Y
Factor category A
+
−
Factor category B
+
−
An example of this is the correlation between the all + PLURAL NOUN variable
(two variants: all and all of) and the “presence of a focus marker” factor (two
factor categories: +FM and −FM), where all of correlates with +FM and all correlates with −FM.
In many cases, the picture was more complex, either because there were three
or more variants (four with all/whole + SINGULAR COUNT NOUN) or because there
were three or more factor categories. In order to be able to use four-cell tables in
the test, variants or factor categories (and in some cases both) had to be conflated.
When there were more than two variants and a two-category factor and I
wanted to measure the correlation between Variant X and one of the factor categories, the remaining variants were conflated, This is illustrated in Table 4.4.
Table 4.4. Four-cell table used for testing the correlation between Variant
X and Factor category A with a two-category factor
Variant X
Variant Y+Z
Factor category A
+
−
Factor category B
+
−
Using this methodology we can say something about the significance and
strength of the correlation between Variant X and Factor category A. An example is all/whole + MASS NOUN (three variants: all, all of and whole) in relation
to the presence of a focus marker (two factor categories: +FM and −FM), where
+FM correlated with all of.
When there were only two variants but a multi-category factor, I conflated
those factor categories that were not in current focus and could then say something about the correlation between Variant X and Factor category A (see Table
3.5).
93
93
Table 4.5. Four-cell table used for testing the correlation between Variant
X and Factor category A with a multi-category factor (only two
variants)
Variant X
Variant Y
Factor category A
+
−
Factor category B+C etc.
+
−
An example is the correlation between variants in the variable all + PLURAL
(two variants: all and all of) and syntactic function, where the corpus
material included tokens in six categories; however, only two of them (subject
function and prepositional complement function) correlated positively with variants. So when, for instance, testing the correlation between all and the subject
function, all other functions (object, prepositional complement etc.) were conflated.
Finally, the most complicated case occurs when there are both more than two
variants and more than two factor categories, as in Table 4.6. Here, both variants
and categories have to be conflated.
NOUN
Table 4.6. Four-cell table used for testing the correlation between Variant
X and Factor category A with a multi-category factor (more than
two variants)
Variant X
Variant Y+Z
Factor category A
+
−
Factor category B+C etc.
+
−
An example is the correlation between the all/whole + GEOGRAPHICAL NOUN
variable (three variants: all, all of and the whole of) and syntactic function.
When testing the correlation between the all variant and the subject function, all
other functions were conflated, and so were the all of and the whole of variants.
A few more observations should be made. First, as mentioned above, in most
cases where there were more than two variants, each variant in turn was tested in
relation to a particular factor category. In a few cases, however, two variants exhibited very similar correlation patterns and no separation was made. The reason
is that clear correlations could otherwise appear as less significant and also relatively weaker (compared with other correlations). An example is the both + PLURAL NOUN variable, where both the and both of the were conflated in the significance testing of a correlation with the modifier factor, since they had very
similar correlation patterns. Had both the and both of the been tested in isolation,
simple both and both of the would have been conflated in the first case (both the)
and simple both and both the in the second (both of the); consequently this
would have weakened the significance and strength of the correlation.
Second, when analysing correlations with one variant and/or factor category
at a time, the same data were used several times in cases where there were more
than two variants and/or factor categories. If the <0.05 alpha level is used in the
chi-square test, we accept that there is a one-in-20 risk that the correlation has
occurred by chance. If that should be the case, and the same data is used more
than once, this happens again and again. To minimise the risk of repeatedly re-
94
94
porting on correlations which have occurred by chance, the <0.01 level was applied as a standard. Instances where only the <0.05 level was reached were included in the presentation, but marked as less statistically significant.
Third, it should be kept in mind that the figures used for the chi-square test
are total figures in the case of variation according to region and medium. Conversely, in most of the cases of linguistic factors they are 200-token samples for
each variant (100 per corpus). In those cases where the material comprised fewer
than 200 tokens of a particular variant, normalised frequencies were used in the
bar charts and tables in Chapters 7 and Appendix D (see Section 7.1). However,
only absolute figures were used when testing correlations by means of the chisquare test. One reason for this is that it must always be certified that the expected frequency in a cell is more than. Normalised frequencies can give a
skewed picture.
4.4
Methodological problems
One very general problem that any corpus linguist interested in language variation will encounter is how to find comparable corpora of, for instance, different
regional varieties of a language. Since it would be too expensive and too timeconsuming for each researcher to compile a special corpus for every research
question, one will often have to make do with already existing corpora, compiled
according to different principles and sometimes perhaps not very comparable at
all (cf. Lindquist & Levin 2000, Kilgarriff 2001). The problem has to be taken
into account in the interpretation of results.
Investigating spoken language by means of corpora is another problematic
area. Corpora of spoken material are much less frequent. In most cases, they are
smaller and not so easily available as written corpora, owing to extensive transcription costs. Furthermore, transcription is not always entirely trustworthy, as a
certain amount of hearing and spelling mistakes can be included89. It can further
be discussed how “natural” corpora of “natural conversation” really are, when
recordings have been made non-surreptitiously, for ethical reasons, as is the
standard procedure today. To what extent do people change their linguistic behaviour when they know that they are being recorded? This fact ought not to be
particularly important in a study of variation concerning items like those
analysed here. One would expect that the “tape-recorder factor” would be more
influential when it comes to participants’ use of certain vocabulary items, such
as swearwords or stigmatised non-standard grammatical features (such as double
negations).
A problem regarding the written text material is the fact that many newspapers and publishing companies use style guides (particular to each “house
style”), thereby standardising their authors’ language. This too has to be remembered in variation research. On the other hand, if one finds variation with respect
to the particular variable at issue, one must conclude that house style editing has
not taken place or has not been effective. The newspapers used as material were
contacted and asked about their use of style guides. Only The New York Times
89
Misprints and spelling mistakes may of course occur in written sources as well.
95
95
answered (in spite of repeated queries to The Independent and The Sydney Morning Herald). Their style guide did not include anything that concerned the use
of quantifiers investigated in this thesis.
Another drawback of using newspaper corpora to look at regional variation
(related to the previous problem) is that most newspapers, as pointed out by Bell
(1991:12), use material not produced by the papers’ own journalists, but acquired from news agencies or other newspapers. The material may come from
other parts of the English-speaking world, skewing the quantitative data on regional variation. It is also the case that local writers rewrite material from other
papers. Furthermore, part of the text material in a newspaper corpus consists of
spoken dialogue written down (in interviews). Judging from the material used
here, there are few differences in the use of quantifiers between spoken and
written English (see further Section 6.2). One could of course speculate as to
whether there would have been more differences had another type of written
language been analysed in the study.
The use of newspaper text on CD-ROM also involves certain technical problems brought up by Minugh (1997:79). The search software of these computer
corpora was not made for linguistic research but for information retrieval, a fact
that makes searches for some words and structures complicated. The CD-ROM
search programs coming with some newspapers do not allow searches for very
frequent function words, which are treated as “noise words” or “stop words”.
This means that, when searching for all, both and half in these newspapers, other
measures had to be taken (see Appendix C).
With the newspaper corpora (especially The New York Times), searches often
yield duplicate tokens. These were excluded from the material to as large an extent as possible. If there are few tokens to a search, the deletion is seldom problematic. This is because one can easily spot redundant examples if they are
sorted in alphabetical order on the word to the left or right of the keyword. In
searches yielding thousands of lines of very similar structures, the problem is
less easily solved, since duplicate lines may turn up very far away from each
other. The method used here was to sort the concordance lines on the fifth words
to the left and right of the keyword. This revealed at least most duplicate lines
that were initially undetected.
Finally, a problem occurs with the linguistic factors. For some of these, the
analysis is straightforward and objective, as in the case of the presence or
absence of an adjacent of. In other cases, the analysis is a matter of more or less
subjective interpretation. Divisibility (see Sections 3.3.4.4. and 7.3.3) and
animacy (3.3.4.4 and 7.3.4) are examples of such cases. I have tried to at least
explain and motivate my choices, thereby allowing the reader a fair chance of
understanding the reasoning behind them.
4.5
Summary
This chapter has served as a background for the empirical analyses that will be
presented in the following three chapters (5–7). We started out by concluding
that computerised corpora provide a convenient source of material for the research into syntactic variation in natural language. There are, however, different
96
96
views about whether corpus linguistics is a whole theory or just a methodological tool and about how corpora should be used.
Section 4.2 gave a brief presentation of the corpora used in the study: three
newspaper archives – British, American and Australian – and two corpora of
spoken English, one from Great Britain and one from the U.S. This was followed by a more extensive section on the procedure used for the extraction of
data from the corpora, the sampling of material for the analyses of linguistic factors and the analysis itself. Complete lists of variables included and excluded are
available in the appendices. The chapter ended with a discussion of certain problems concerning the material and the use of it. One such problem is the disadvantage of comparing data from different corpora.
97
97
98
5. Overall frequency
distribution of variants
This chapter presents the results of the study of overall frequency distribution of
NPs with all, whole, both and half. The main purpose was to find out whether
some variants were more frequent than others overall. As discussed in section
2.1.3 and 2.1.4, fixed expressions and the result of knock-out effects were excluded from the study. However, the decision about which constructions to include and which to leave out is to some extent subjective. Therefore, Section 5.2
compares figures for a few of the variables with figures where the fixed expressions were included.
5.1
Results
This section presents the results in the form of tables showing the absolute
number of tokens in all the corpora taken together, the number of tokens per ten
million words and the relation between the variants expressed in percentages.
Each table is followed by authentic corpus examples illustrating the different
syntactic variants, in order to make the interpretation of the tables easier90. The
variables will be presented in the following order: NPs with all/whole and common nouns, NPs with both and common nouns, NPs with half and common
nouns, NPs with all/both/half and demonstrative pronouns, NPs with all/both
and personal pronouns, NPs with half, the indefinite article and a singular noun/
numeral, NPs with all/whole and geographical names, and finally the two semantically restricted subgroups: (i) NPs with all/whole and temporal nouns and (ii)
NPs with both and nouns denoting body parts or kinship.
5.1.1
NPs with common nouns
A general observation concerning all the variables with common nouns is that
the variants with of were less frequent than the of-less variants. These results
corroborate indications in reference grammars, where of constructions are generally introduced as possible alternatives (see, e.g., Quirk et al 1984:258). Similarly OED (s. v. all adj. 6 and both adj. 6) considers all of and both of to be rare.
A common knock-out effect occurred in cases where the NP was preceded by a
90
In this chapter, there will be no discussion of individual examples; they were merely included for
illustration of the table contents.
99
99
quantifier + or (as in some or all…). Here the of variant is the only valid alternative (see Appendix B), and such tokens were excluded from the statistics.
5.1.1.1
NPs with all/whole and common nouns
Tables 5.1 to 5.3 show the frequencies of NPs with all/whole and a determiner in
combination with singular count nouns, mass nouns and plural nouns (in that
order). The reason for separating these three groups is that they exhibit different
variation patterns. Singular count nouns (and, to some extent, mass nouns) are
used with whole whereas plural NPs generally are not (unless with a different
meaning). Moreover, whole is predominant with singular count nouns, while all
predominates with mass nouns, so not distinguishing between singular count
nouns and mass nouns would have yielded misleading figures91.
Let us first look at all in combination with a singular count noun in Table
5.192. In this variable, one particular phrase, all the way, was extremely frequent.
It could not be excluded as a fixed phrase, however, since there were also examples of all of the way, the whole way etc. Instead, figures where noun phrases
with way (all the way, all of the way, the whole way) have been excluded are
given in brackets.
Table 5.1. NPs with all/whole, DETERMINER and SINGULAR COUNT NOUN93
Number of Per cent
tokens
all + DETERMINER + SINGULAR COUNT NOUN94
3903
26%
(1699)
(13%)
all of + DETERMINER + SINGULAR COUNT NOUN
325
2%
(324)
(3%)
95
DETERMINER + whole + SINGULAR COUNT NOUN
10557
70%
(10513)
(81%)
the whole of + DETERMINER + SINGULAR COUNT NOUN
392
3%
(392)
(3%)
15177 100%96
Total
(12928) (100%)
91
92
93
94
95
96
100
Where a sentence was ambiguous between a singular count noun and a mass noun reading (e.g.
some NPs with the noun discussion), the token was included in the group with the most natural
quantifier, i.e. mass nouns if it was a token with all and singular count nouns if it was a token
with whole. This results in circular reasoning, but these cases were very few.
It should be observed that NPs with temporal nouns (e.g. all the day, the whole night) are not
included in this frequency count but presented separately below (5.1.6.1).
Throughout the study the figures also include cases where one or more modifying adjectives precede the noun, as in all the nice children.
A number of tokens of various types, e.g. all the/this while and for all the world as if, were excluded as fixed expressions (see Appendix B).
A number of tokens of various types, e.g. the whole gamut and the whole truth, were excluded as
fixed expressions (see Appendix B).
Here and throughout, the total sum is given as 100% also in those case where the percentages for
each variant add up to 99 or 101 because they have been rounded off (to avoid decimals in the
table). The rounding process also means that very low absolute figures sometimes result in 0%,
as in Table 5.2.
100
The variants are illustrated in (5:1) to (5:4).
(5:1) And you go to the, the dole office and they have and they have all the
alphabet out there, and you, you go for the letter which is the first letter of
your last name. (BNC)
(5:2) Throughout all of his career, Dwyer never lost his sense of humour and,
oddly enough for one so successful in the coaching caper, he was a particularly human sort of bloke. (SMH95)
(5:3) She spends her whole life looking out the window watching who’s going
up and down. (BNC)
(5:4) The police say that the numbers are between 50 and 150 for the whole of
the borough. (IND95)
In this variable, DETERMINER + whole was predominant (82% with all the way
excluded), which is in line with the claim, made by Quirk et al (1984:259) and
Putseys (1984:388f), that whole is more natural than all in combination with a
singular count noun. The corpora also contained almost four thousand instances
of all + DETERMINER, a variant which Quirk et al (1985:259, 381) suggest is formal and only used with divisible nouns. However, as mentioned above, in cases
where the determiner was the definite article, the all the way phrase constituted
the majority of the cases. There were also a large number of tokens of the all
my/his etc. life construction. This variable will be further discussed in terms of
divisibility in Section 7.3.3.
In Table 5.2 we find the frequencies for all/whole in combination with a mass
noun.
Table 5.2. NPs with all/whole, DETERMINER and MASS NOUN
Number of Per cent
tokens
all + DETERMINER + MASS NOUN97
18402
93%
all of + DETERMINER + MASS NOUN
982
5%
DETERMINER + whole + MASS NOUN
400
2%
the whole of + DETERMINER + MASS NOUN
12
0%
19796
100%
Total
Examples (5:5) to (5:8) illustrate the four variants.
(5:5) I mean that’s where all the action is. (LSAC)
(5:6) Let me stuff all of this stuff in here first. (LSAC)
97
A number of tokens of various types, e.g. all the rage and and all that jazz, were excluded as
fixed expressions (Appendix B).
101
101
(5:7) The Shoreline East was begun by executive order, not legislative action,
‘so there’s this whole uneasiness over whether efforts by the legislature
can save the service,’ Mr. Aniskovitch said. (NYT95)
(5:8) ‘Imagine grabbing the whole of the water in your hand and forearm’, says
David […] (IND95)
In the case of mass nouns, the form with the of-less all variant predominated as
expected (93%), slightly more than the whole variant did in NPs with a singular
noun. Interestingly, there were also 395 instances of whole combined with mass
nouns, a variant which Quirk et al (1985:260) regard as “unacceptable”. However, this is again a case of a recurrent phrase, the whole time, which was a frequent NP type, although in this case the other alternatives (all the time and all of
the time) were frequent as well. The all of variant was about as frequent proportionally as in the variable with singular count nouns. The form with the whole of
was very rare with mass nouns (only 13 instances in all). In this variant, there
were no tokens of NPs where the determiner was a demonstrative or possessive
determiner, only NPs with the definite article.
Table 5.3 shows how the two variants of the variable with all in combination
with a plural noun were distributed in the corpus material.
Table 5.3. NPs with all, DETERMINER and PLURAL NOUN
Number of Per cent
tokens
all + DETERMINER + PLURAL NOUN98
33637
90%
all of + DETERMINER + PLURAL NOUN
3556
10%
37193
100%
Total
In (5:9) and (5:10) we see examples of the variants.
(5:9) Bihac fulfils all those Balkan prejudices, its history a tale of betrayal and
unfortunate geography, of local warlords, political intrigue and greed.
(SMH95)
(5:10) We don’t want to miss all of the people falling down. (NYT95)
This variable has only two variants, since whole in combination with a plural has
a unitary rather than distributive meaning99. As with the other variables described, the of variant is the less frequent alternative. The use of all vs. all of in
plural NPs can further be compared with NPs with mass nouns (disregarding
whole variants which were quite rare). Finally, note that the of form was slightly
more frequent with plurals than with mass nouns. This variable will be discussed
again in Section 6.1.1 in relation to regional variation.
98
99
102
Tokens of /and/ all these/those kind, sort of/ + things were excluded as fixed expressions.
In the expression all (of) our/your/their lives, all has unitary rather than distributive meaning as
well (see 3.3.4.4), competing with our whole lives. Those tokens were excluded. It is possible
that all is used with this meaning with other nouns as well, but it would have been too timeconsuming to analyse all the 37,193 examples in the material.
102
5.1.1.2
NPs with both and common nouns
The next two tables give the figures for NPs with both and the definite article or
a demonstrative/possessive determiner. A distinction was made between NPs
with the definite article and NPs with a demonstrative or possessive determiner.
The reason for this separation is that the simple both variant (as in both children)
is usually presented in the literature as an alternative to the former only (e.g.
Seppänen & Seppänen 1986:169f; Berry 1997:138). A particular subgroup contains simple both and both with a possessive determiner in combination with a
noun denoting body parts or kinship (see Section 5.1.6.2). These tokens were excluded from the frequencies here. Table 5.4 shows the variable with both /THE
DEFINITE ARTICLE/ and its alternative variants.
Table 5.4. NPs with both, /THE DEFINITE ARTICLE/ and PLURAL NOUN
Number of Per cent
tokens
both + PLURAL NOUN100
10782
97%
both the + PLURAL NOUN
201
2%
both of the + PLURAL NOUN
118
1%
11101
100%
Total
Examples of the three variants are given in (5:11) to (5:13).
(5:11) The drifters came in to both piers and on a Saturday night, the village
was a busy place. (BNC)
(5:12) Pierre van Hooijdonk scored both the goals for Celtic. (IND95)
(5:13) Both of the authors contribute to Horticulture magazine, and Mr. Winterrowd is the author of ‘Annual for Connoisseurs,’ yet I can’t find seams
in this text; it is written in a single voice. (NYT95)
There was a clear predominance for the zero form (97%), and the variants with
both the and both of the were infrequent (2% and 1% respectively). This result
can be contrasted to the description in Leech (1989:66) that both the boys “can
be replaced” (my emphasis) by both boys; similarly, many other grammars present the three variants as if they were equally common. This is clearly misleading since both is the totally predominant variant. Remember also the knock-out
effect described in 3.3.4.2, i.e. that the both the and both of the forms are not
used if the NP has restricted reference, as in both halves (also see Section
5.1.6.2).
The other variable including both and a plural noun is presented in Table 5.5.
100
A number of tokens of various kinds, e.g. in both cases and the best of both worlds, were excluded as fixed expressions (see Appendix B).
103
103
Table 5.5. NPs with both, DEMONSTRATIVE/POSSESSIVE DETERMINER and
PLURAL NOUN
both these, my etc. + PLURAL NOUN
both of these, my etc. + PLURAL NOUN
Total
Number of Per cent
tokens
567
66%
295
34%
862
100%
Examples (5:14) and (5:15) illustrate the two variants.
(5:14) There is good reason for both these attitudes. (SMH95)
(5:15) […] well I’ll call up both of those folks. (LSAC)
With a possessive or demonstrative determiner following both, the of variant was
quite frequent, used in 34% of the cases, (also see 5.1.6.2). This variable exhibited clear regional variation, which will be discussed in 6.1.1 below.
5.1.1.3
NPs with half and common nouns
Let us now turn to NPs with half combined with determiners and common
nouns. In consistency with the presentation of all/whole and a common noun, the
tokens were separated according to number and countability, even though there
are no restrictions with respect to variation here. Table 5.6 shows half in combination with a singular count noun.
Table 5.6. NPs with half, DETERMINER and SINGULAR COUNT NOUN
Number of Per cent
tokens
half + DETERMINER + SINGULAR COUNT NOUN101
2827
80%
half of + DETERMINER + SINGULAR COUNT NOUN
722
20%
3549
100%
Total
The two variants are exemplified by (5:16) and (5:17).
(5:16) A few months ago the construction of a rail link caused the closure of
half the airport when a tunnel collapsed. (IND95)
(5:17) See, the thing is, we only pay half of the budget from the actual faculty,
[…] (LSAC)
The half of variant was used in 20% of the cases in NPs with singular count
nouns.
The frequency distribution for the variable with half and a mass noun is presented in Table 5.7.
101
104
Tokens of half the battle were excluded as fixed expressions.
104
Table 5.7. NPs with half, DETERMINER and MASS NOUN
Number of
tokens
half + DETERMINER + MASS NOUN
769
half of + DETERMINER + MASS NOUN
179
948
Total
Per cent
81%
19%
100%
Example sentences are given in (5:18) and (5:19).
(5:18) Drizzle with remaining olive oil and sprinkle with half the brown sugar
and salt. (SMH95)
(5:19) In some programs, more than half of the money the agency spends goes
to contractors. (NYT95)
The frequency distribution between half and half of was virtually the same as
with singular count nouns.
Table 5.8 finally shows the distribution in the variable with half and a plural
noun.
Table 5.8. NPs with half, DETERMINER and PLURAL NOUN
Number of Per cent
tokens
half + DETERMINER + PLURAL NOUN
1910
61%
half of + DETERMINER + PLURAL NOUN
1197
39%
3107
100%
Total
Examples (5:20) and (5:21) illustrate the variants.
(5:20) I don’t even know half the models by heart. (LSAC)
(5:21) Mr. Murphy vigorously disputes the contention that he is a racist, saying
that he parted on good terms with Mr. Griffith’s mother, and that half of
his clients are black. (NYT95)
Interestingly, the of form was about twice as frequent in plural noun phrases as
in the NPs with singular count nouns and mass nouns. It should be noted, however, that there were fairly great discrepancies between the regional varieties
(see Section 6.1.1) and also between different types of central determiners (see
7.2). We can finally conclude that the of form was more frequent in the half variable than in the all and both variables. One possible explanation is that all of and
both of are comparatively recent forms, whereas half of has existed much longer
(cf. 3.3.1 above).
5.1.2
NPs with demonstrative pronouns
The next presentation of frequency distribution concerns NPs with all, both and
half where the head is a demonstrative pronoun rather than a noun. We will first
105
105
look at all three of them, starting with all in Table 5.9, and then discuss them
after the tables.
Table 5.9. NPs with all and DEMONSTRATIVE PRONOUN102
Number of Per cent
tokens
all this/that/these/those
7521
85%
all of this/that/these/those
1325
15%
8846
100%
Total
Examples (5:22) and (5:23) illustrate the variants with and without of.
(5:22) It would be easy to dismiss all this as a backwater of policy in one of the
lesser departments of state. (IND95)
(5:23) You mean all of these are for me? (LSAC)
Table 5.10 shows the frequencies for the two variants with both and a demonstrative pronoun.
Table 5.10. NPs with both and DEMONSTRATIVE PRONOUN
Number of Per cent
tokens
both these/those
37
34%
both of these/those
73
66%
110
100%
Total
Here are two examples of both in combination with a demonstrative pronoun.
(5:24) In both these he had notable success. (IND95)
(5:25) You got a hundred on both of those? (BNC)
The frequency distribution for half in combination with a demonstrative pronoun
is presented in Table 5.11.
Table 5.11. NPs with half and DEMONSTRATIVE PRONOUN
Number of Per cent
tokens
half this/that/these/those
157
40%
half of this/that/these/those
235
60%
392
100%
Total
Examples of the variants are given in (5:26) and (5:27).
102
106
A third variant, the whole of this/that, as in If you sold the whole of this that’s the earning and
you’ve got to do this excuse me, in three weeks, occurred twice in the spoken British corpus
106
(5:26) Yeah, they’re having to sell some of it because they’re going from like a
five thousand square foot house to almost half that and um, […] (LSAC)
(5:27) I don’t even think I’ve done half of these yet. (BNC)
The variant with of was more frequent when the head of the NP was a demonstrative pronoun rather than a common noun (cf. with Tables 5.1 to 5.4 above).
This can be compared with Quirk et al’s (1985:373) somewhat vague description
that of is “often preferred” with demonstrative pronouns. The statement seems
more valid as far as NPs with both and half are concerned (compared to NPs
with all), since in the material there was a fairly great difference between the
quantifiers as regards the frequency of the of constructions (15% with all, 66%
with both and 60% with half). As for the tokens with both, however, the number
of tokens was quite low: only 107 tokens in all for both these/those and both of
these/those. We will return to these variables in Section 6.1.2, since they (especially NPs with both) exhibited regional variation.
5.1.3
NPs with personal pronouns
From NPs with demonstrative pronoun as heads, let us now move on to all and
both in combination with personal pronouns, one variant of which is referred to
as the floating quantifier when occurring in the subject function (see 3.3.4.3).
Only plural noun phrases were included in the study, even though the variable
also occurs in the third person singular (it all and all of it). The reason for this is
that in several cases it was very difficult to decide whether all belonged to the
NP as a quantifier or whether it was an adverbial modifying the verb phrase (as
in It all depends how you define major, IND95). As mentioned above, all tokens
of you all and all of you were excluded owing to you all sometimes being used
as a simple plural form of you, without having any quantifying function at all.
Frequencies for NPs with all are presented in Table 5.12.
Table 5.12. NPs with all and PERSONAL PRONOUN 103
we/us, they/them all
all of us, them
Total
Number of Per cent
tokens
16937
89%
2174
11%
19111
100%
Examples (5:28) and (5:29) illustrate the two variants.
(5:28) They think they can all be helped by others. (NYT95)
(5:29) Yeah all of us had to audition. (BNC)
In table 5.13, we see the frequencies for the two variants with both.
103
There is also a third alternative in subject function, a combination of the two other constructions,
as in They are all of them happy (see Section 3.3.4.3 above) and a non-standard alternative, the
both of them (see Section 6.2).
107
107
Table 5.13. NPs with both and PERSONAL PRONOUN 104
we/us, you, they/them both
both of us, you, them
Total
Number of Per cent
tokens
3257
81%
779
19%
4036
100%
The two variants are illustrated in (5:30) and (5:31).
(5:30) And that’s what I need to ask you both. (LSAC)
(5:31) So I’m gonna ask both of you for a word specially in a moment to
describe what it was like for that girl in the jungle. (BNC)
The tables show that the we/us (...) all/both variant was far more frequent (88%
with all and 81% with both) than the variant with of. Remembering (cf. Section
3.3.4.3) that in the literature, the floating quantifier has generally been considered the marked variant (or at least that is the conclusion that can be drawn
from the labelling of the phenomenon), this is quite interesting, as markedness
and frequency of usage are often considered to be related. It should be remembered, however, that, in the literature, floating quantifiers have mainly been
brought up in relation to NPs with a noun as the head of the NP rather than a
pronoun as is the case here. The former is not analysed in this study, so we do
not know the frequency of the floating construction (e.g. the children were all…)
compared to the “unmarked” alternative (e.g. all /of/ the children were…). The
variable with personal pronouns also exhibited some regional variation, which
will be discussed in Section 6.1.3.
It should further be kept in mind that the variables with personal pronouns
were affected by a knock-out effect. The we/us (…) all/both variant was never
used in minor clauses, such as short answers and appositions (see Appendix B.
5.1.4
NPs with half, the indefinite article and a singular
noun or numeral
NPs with half, the indefinite article and a singular noun/numeral usually have
two variants (see Table 5.14). The following sentences illustrate the two variants.
(5:32) Let’s make it half an hour, okay? (BNC)
(5:33) A half-hour before kickoff, he was in the training room being rubbed
down by a physical therapist. (NYT95)
104
108
The same form as described in Footnote 103 occurs with both as well, They are both of them
seaworthy (see section 3.3.4.3 above).
108
Table 5.14. NPs with half, THE INDEFINITE ARTICLE and SINGULAR NOUN/
NUMERAL
105
Number of Per cent
tokens
6089
78%
1685
22%
7774
100%
half a/an
a half 106
Total
The more recent half a/an variant (cf. 3.3.1.3) was far more frequent (80%) than
the a half form (20%). There were 1506 instances of a half in the material, so the
description in Quirk et al (1984:388) that a half “occurs occasionally” (my emphasis) seems to be an understatement. This is one of the cases where there was
clear regional variation in usage, which is discussed in the next Chapter (6.1.4).
Furthermore, a knock-out effect influenced some of the variation. In cases where
the quantified NP directly preceded another noun, as in a half-hour journey, and
the indefinite article determines the second rather than the first noun, only the
form with the indefinite article first is possible (*half an hour journey).
5.1.5
NPs with all/whole and geographical names
The frequencies for the variants within the variable with geographical names are
given in Table 5.15.
Table 5.15. NPs with all/whole, /THE DEFINITE ARTICLE107/ and
GEOGRAPHICAL NAME
all + GEOGRAPHICAL NAME108
all of + GEOGRAPHICAL NAME
the whole of + GEOGRAPHICAL NAME
Total
Number of Per cent
tokens
125
16%
405
52%
250
32%
780
100%
Examples of the variants are given in (5:34) to (5:36).
(5:34) If the United States recognises all Jerusalem as the capital of Israel, that
will end the peace process. (IND95)
(5:35) So is there no AC in all of Albuquerque? (LSAC)
105
There is also a non-standard variant, a half a (see further Section 6.2).
A number of tokens of various kinds, e.g. a half-back and a half-smile were excluded as fixed
expressions (see Appendix B).
107
Cases where the geographical name was preceded by the definite article (as in all of the United
States) were excluded. “The definite article” in the table heading refers to cases of the whole of.
108
In the material from NYT95, the normal search procedures could not be used, a fact that might
have affected the results. See further Appendix C.
106
109
109
(5:36) The fall of the whole of western Bosnia would also destroy peace proposals for the republic offered by Western powers because it would leave
the Serbs with less than the 49 percent of the country the plan would
allow them. (NYT95)
The most frequent form overall was all of, which was used in half of the cases.
The second most frequent variant is the whole of (33%), whereas the form with
simple all is the least frequent of the three (16%). This is also the variant which
Quirk et al (1984:260) consider to be more formal than the other two. In the light
of this claim we will return to the construction in Section 6.2 to see if there are
differences between speech and writing. This is also one of the cases where there
was regional variation in the corpus material (see Section 6.1.5).
5.1.6
Subgroups
Finally, we will look at semantically restricted subgroups of a few of the larger
groups (Sections 5.1.1.1; 5.1.1.2), i.e. all/whole with singular count nouns and
both with plural nouns. The first subgroup comprises all/whole with temporal
nouns, and the second comprises both with nouns for body parts or kinship.
5.1.6.1
NPs with all/whole and temporal nouns
First, Table 5.16 presents the frequencies for the variable with temporal nouns.
Table 5.16. NPs with all/whole, /THE DEFINITE ARTICLE/ and TEMPORAL
109
NOUN
all + TEMPORAL NOUN110
all the + TEMPORAL NOUN
all of the + TEMPORAL NOUN
the whole + TEMPORAL NOUN
the whole of the + TEMPORAL NOUN
Total
Number of Per cent
tokens
5062
92%
43
1%
3
0%
402
7%
15
0%
5525
100%
The five alternatives are illustrated in (5:37) to (5:41).
(5:37) The debate dragged on all day before Mr. Dole interrupted it. (NYT95)
(5:38) […] I said goodnight Charlie, he say (sic) goodnight and that’s the only
thing he said to me all the weekend. (BNC)
(5:39)
A shoulder injury caused him to miss virtually all of the next season
[…]. (IND95)
109
The study comprises temporal nouns for parts of the day (e.g. morning, night), days of the week
(e.g. Monday), seasons (e.g. summer) and such nouns as week, month, season and term. These
nouns may also be preceded by a modifier, as in all next season.
110
Tokens of all day, night etc. long were excluded as fixed expressions.
110
110
(5:40) I think she would be available the whole day. (LSAC)
(5:41) ‘The trauma team spent almost the whole of the night trying to find a
suitable bed and were simply unable to do so,’ he said. (IND95)
This is the variable in the study that the largest number of possible variants. In
the literature, all, all the and the whole are generally the only alternatives presented, and indeed there were very few instances of all of the and the whole of
the (only 3 tokens of the former and 15 of the latter) in the material. All the was
also infrequent (43 tokens, 1%). Moreover, simple all (without of and the definite article) was the clearly predominant variant (92%) with the whole in second
place (7%). The data thus support Putsey’s (1984:389) statement that all is preferred to whole in NPs with temporal nouns. This variable also showed regional
variation, which will be discussed in Section 6.1.6.1. Again, some tokens were
excluded due to a knock-out effect, viz. quantified temporal noun phrases occurring in newspaper advertisements, where only the simple all form was used (see
Appendix B).
5.1.6.2
NPs with both and nouns for body parts or kinship
The second subgroup is related to the two variables with both and a plural noun,
since the variants in this group are (i) simple both (from the variable with the
definite article) and (ii) both in combination with a possessive determiner (with
or without of). They are combined with nouns that refer to body parts (like eyes
or arms) or kinship (like parents). There are also other nouns that may be used
either with a possessive noun or the definite article, depending on the context.
These were, however, included in the both (/of/the) + PLURAL NOUN variable
above. The results of the frequency study are presented in Table 5.17.
Table 5.17. NPs with both, /POSSESSIVE DETERMINER/ and NOUN FOR BODY
PARTS OR KINSHIP
both + NOUN FOR BODY PARTS/KINSHIP
both + POSSESSIVE DETERMINER + NOUN FOR BODY
PARTS/KINSHIP
both of + POSSESSIVE DETERMINER + NOUN FOR BODY
PARTS/KINSHIP
Total
Number of Per cent
tokens
766
77%
185
19%
39
4%
990
100%
Examples (5:42) to (5:44) illustrate the three variants.
(5:42) And I certainly love the way that both eyes are catching those lovely
highlights! (BNC)
(5:43) In the bathroom, Powles smashed a wine glass and cut both his wrists
with the broken glass. (SMH)
111
111
(5:44) I never heard of it either and he has it tattooed on both of his arms.
(LSAC)
If these figures are compared with those in Table 5.3 (both – both the – both of
the), simple both is less predominant here (70% compared to 97% above). On
the other hand, the simple both form is not always a natural choice (see Section
3.3.4.2). Finally, there was regional variation, described in Section 6.1.6.2.
5.2
Fixed expressions
The fact that the definition of fixed expressions is problematic was discussed in
Section 2.1.4. In this study, fixed expressions were excluded from the statistics
since competition between variants is eliminated in those cases, according to dictionaries and the corpus material. In order to examine what consequences the exclusion of fixed expressions had on the results, this section presents a few tables
giving figures both with and without fixed expressions included.
The first example concerns NPs with all and whole in combination with a singular count noun, where phrases such as for all the world as if and the whole
gamut were excluded (see Appendix B for a complete list of the constructions
excluded). Table 5.18 illustrates the difference between figures where fixed expressions were excluded and where they were included.
Table 5.18. Figures for all/whole, DETERMINER and SINGULAR COUNT NOUN
with fixed expressions excluded and included
Fixed expression
excluded
Number of
Per
tokens
cent
all + DETERMINER + SING. COUNT NOUN
3903
26%
all of + DETERMINER + SING. COUNT NOUN
325
2%
DETERMINER + whole + SING. COUNT NOUN
10557
70%
the whole of + DETERMINER + SING. COUNT NOUN
392
3%
15177 100%
Total
all + DETERMINER + SING. COUNT NOUN
all of + DETERMINER + SING. COUNT NOUN
DETERMINER + whole + SING. COUNT NOUN
the whole of + DETERMINER + SING. COUNT NOUN
Total
Fixed expression
included
Number of Per cent
tokens
4492
25%
325
2%
12501
71%
392
2%
17710
100%
What happens when fixed expressions are included is that the whole strengthens
its position as the predominant variant by one per cent, from 70% to 71%,
whereas two of the other variants recede by one per cent each (from 26% to 25%
112
112
for all and 3% to 2% for the whole of). The percentage for all of did not change
at all.
The second example regards all in combination with mass nouns, such as all
the fashion (see Appendix B for a complete list). In Table 5.19 we can see that
the inclusion of previously excluded fixed expressions does not change the
relative figures although it strengthens the predominant variant in absolute
figures with 207 tokens.
Table 5.19. Figures for all/whole, THE DEFINITE ARTICLE and MASS NOUN
with fixed expressions excluded and included
Fixed expression
excluded
Number of
Per cent
tokens
all + DETERMINER + MASS NOUN
18402
93%
all of + DETERMINER + MASS NOUN
982
5%
DETERMINER + whole + MASS NOUN
400
2%
the whole of + DETERMINER + MASS NOUN
12
0%
19796
100%
Total
all + DETERMINER + MASS NOUN
all of + DETERMINER + MASS NOUN
DETERMINER + whole + MASS NOUN
the whole of + DETERMINER + MASS NOUN
Total
Fixed expression
included
Number of
Per cent
tokens
18609
93%
942
5%
395
2%
13
0%
19959
100%
The main reason for choosing these particular groups of fixed expressions in this
part was that they included a sufficiently large number of tokens to be potentially relevant (although they, in fact, did not affect the original figures very
much).
The fixed expressions were excluded as a safety precaution, but the results
show that they could have been included without any great effects on the results.
Why then does the exclusion or inclusion of fixed expressions affect the results
to such a small extent? In the case of all/whole with singular nouns, the excluded
fixed all and whole expressions divided neatly in proportions similar to the free
expressions. In other cases, the fixed expressions are of the predominant variant,
e.g. all in NPs with mass nouns and simple both in the [both /of the/ + PLURAL
NOUN] variable. The reason as to why the exclusion of fixed expressions does
not affect the results in these cases is that the most frequent variant is so totally
dominating.
113
113
5.3
Summary
This chapter has presented the first empirical part of the study, giving the overall
frequency distribution of variants in each of the variables. In the majority of
cases, one variant was predominant, and in a few cases the other variants were
only marginally used, such as the both /of/ the variants, which were only used in
a few per cent of the cases. It is important to keep in mind, however, that the fact
that one variant predominates does not immediately entail that this variant is the
most appropriate one in all contexts. It might very well be that there are contexts
where a rarely used variant is the only (or at least the best) alternative. There
were also some cases where there was more balance in the use of two or more
syntactic variants. Some claims in the literature were corroborated, e.g. that the
of variant is less frequent than the one without of in NPs with common nouns,
and that of is more frequent when the head of the NP is a demonstrative pronoun
rather than a common noun. From this presentation of overall frequency distribution, we will move on to the investigation of two non-linguistic factors: region
and medium.
114
114
6. Non-linguistic factors
We will now move on to the analysis of factors influencing the choice of variant,
beginning with region and medium. The chapter starts by looking at how the
variants were distributed according to region (6.1), and then moves on to
distribution in spoken vs. written English (6.2).
6.1
Distribution according to region
This section deals with regional variation in three newspaper corpora: The New
York Times (1995) for American English, The Sydney Morning Herald (1995)
for Australian English and The Independent (1995) for British English111.
Generally, the regional differences found in the written material were also found
in the spoken material. I provide comments for those variables where this was
not the case. The differences between the varieties were significant at the 0.01
level (according to the chi-square test) unless otherwise indicated. For each of
the significant cases, a phi coefficient expressing the strength of the correlation
(0 = no correlation, 1 = perfect correlation) is given in Appendix E, Table E1
(see further Section 4.3.4).
It has been claimed that variants with of (as in all of the children) are more
frequent in AmE than in BrE (e.g. Swan 1994:35; Berry 1997:81) and that a half
(as in a half hour) is more frequent in AmE than in BrE (Benson et al 1996:21;
Berry 1997:70f). Are these and other suggestions in the literature corroborated
by the corpus results? Are there other, hitherto undetected, differences in frequency between regional varieties? We will begin with NPs with common
nouns.
6.1.1
NPs with common nouns
Table 6.1 shows the frequencies for all, both and half in NPs with plural nouns,
where the same tendency occurred with all three words. As noted in Section
5.1.1.1, NPs with all in combination with a common noun were separated into
singular count, mass and plural NPs, owing to their different variation patterns. I
have only included the plural nouns here (and similarly with half), in order to
avoid presenting too many tables,. This is also the noun type where the greatest
difference between the varieties was observed. NPs with both were, like in
Chapter 5, divided into two variables, one with simple both and alternatives with
the definite article and another with demonstrative/possessive determiners. In111
Remember that no spoken AusE material was used in the study.
115
115
teresting figures commented on in the text will be marked with bold type in the
tables here and throughout the rest of the study.
Table 6.1. NPs with all/both/half, DETERMINER and PLURAL NOUN in
different regional dialects
The New
The Sydney
The
York Times
Morning
Independent
Herald
N
%
N
%
N
%
12732
85% 7529
92% 10789
97%
all + DETERMINER +
PLURAL NOUN
112
all of + DETERMINER +
2254
665
8%
384
3%
14986
100% 8190
100%
11173
100%
5531
19
60
5610
99% 2351
0%
52
1%
18
100% 2421
97%
2%
1%
100%
2607
97
28
2732
95%
4%
1%
100%
15%
PLURAL NOUN
Total
both + PLURAL NOUN
both the + PLURAL NOUN
both of the + PLURAL NOUN
Total
both these, my etc. +
150
50%
136
75%
236
86%
150
50%
45
25%
40
14%
300
100%
181
100%
276
100%
845
54%
442
66%
525
71%
726
46%
230
34%
213
29%
1571
100%
672
100%
738
100%
PLURAL NOUN
both of these, my etc. +
PLURAL NOUN
Total
half + DETERMINER +
PLURAL NOUN
113
half of + DETERMINER +
PLURAL NOUN
Total
The table shows that the of variant was most frequent in AmE and least frequent
in BrE with all three quantifiers (all, both and half), thus corroborating the statements in the literature referred to above. Judging from the results of the corpora,
Swan’s (1994:35) claim that “American English usually has all of ” (my emphasis), is far too strong, however; simple all is clearly predominant also in most
of the AmE material.
In the both + PLURAL NOUN variable, the simple both form was most
predominant in the American material and both the was slightly more frequent in
112
The same kind of correlation (all – BrE, all of – AmE) recurred in the all/whole + MASS NOUN
variable of). Another regional difference worth reporting on is that the whole of was almost
exclusively used in the British corpora (NYT95: 8 tokens vs. IND95: 162 tokens and similarly in
the spoken corpora).
113
The correlation between the of-less form and BrE and the of-form and AmE recurred in the half +
SINGULAR COUNT NOUN variable.
116
116
BrE than in AmE and AusE. Both of the was very infrequent, used in only 1% of
the cases, regardless of regional variety. Swan (1994:109) writes that “in American English, both of is usual”, without any further comments. Since both of the
is so infrequent it is not likely that he is thinking of constructions like these, but
rather NPs with other types of determiners (e.g. both of these pronouns), or perhaps with demonstrative pronoun as heads (e.g. both of these) (see further in the
next section).
6.1.2
NPs with demonstrative pronouns
The next table (6.2) shows the regional variation in NPs with all, both and half
in combination with demonstrative pronominal heads.
Table 6.2. NPs with all/both/half and DEMONSTRATIVE PRONOUN in different
regional varieties
The New
The Sydney
The
York Times
Morning
Independent
Herald
all this/that/these/those
1678
78% 1276
86% 2292
90%
all of this/that/these/those
478
22% 203
14% 260
10%
2156 100% 1479 100% 2552 100%
Total
both these/those
both of these/those
Total
half this/that these/those
half of this/that/these/those
Total
6
13
19
32%
68%
100%
7
12
19
37%
63%
100%
17
10
27
63%
37%
100%
62
89
151
41%
59%
100%
61
65
126
48%
52%
100%
29
33
62
47%
53%
100%
In Section 5.1.2 we noted that of was more frequent in NPs with a demonstrative
pronoun as head than in NPs with a determiner and a noun. This difference recurred in all three newspaper corpora (cf. Table 6.1). As in the variables with
common nouns as heads, the of variant was most frequent in the American material and least frequent in the British material. The greatest difference occurred
in the both variable, but the tokens were very few. Furthermore, the correlation
between both and BrE and between both of and AmE+AusE was only significant
at the <0.05 level. The correlation in the variable with half was not statistically
significant.
6.1.3
NPs with personal pronouns
There was also some regional variation in the variable with all and personal pronouns, as in they … all vs. all of them. The results for NPs with all and both are
presented in Table 6.3.
117
117
Table 6.3. NPs with all/both and PERSONAL PRONOUN in different regional
dialects
The New
The Sydney
The
York Times
Morning
Independent
Herald
N
%
N
%
N
%
we/us, they/them all
4078
83% 2357
86% 5022
91%
all of us, them
863
17% 374
14% 520
9%
4941 100% 2731 100% 5542 100%
Total
we/us, you, they/them both
both of us, you, them
Total
865
255
1120
77%
23%
100%
368
93
461
80% 1033
20% 173
100% 1206
86%
14%
100%
The variant with of was most frequent in AmE and least frequent in BrE. To my
knowledge, this variation is not discussed in the literature on floating quantifiers,
and as can be seen, the difference was not so great. Perhaps this is just another
case of of variants being more frequent in AmE.
6.1.4
NPs with half, the indefinite article and a singular
noun or numeral
The most conspicuous case of regional variation concerns the distribution of half
a/an and a half in the different regional varieties, as presented in Table 6.4.
Table 6.4. NPs with half, THE INDEFINITE ARTICLE and SINGULAR NOUN or
NUMERAL in different regional dialects
The New
The Sydney
The
York Times
Morning
Independent
Herald
N
%
N
%
N
%
half a/an
1841
59% 1176
87% 2136
94%
a half
1254
41% 175
13% 127
6%
3095 100% 1351 100% 2263 100%
Total
According to Benson & et al (1986:21) and Berry (1997:70f), a half is more frequent in AmE than in BrE. This was corroborated by the corpus data: the a half
variant was far more frequent in AmE (40%) than in the BrE (6%) and AusE
(13%).
As described in Section 3.3.1.3, the a half variant was first attested in 835 and
half a not until 1377. We do not know how long it took for the latter to be firmly
established, but let us hypothesise that a half lived on as the most frequent form
for several hundred years after the first attestation of half a. It may thus have
been the most frequent form when the English language was exported to the
other side of the Atlantic, and may thereafter have lived on in American English.
118
This is a process observed for other linguistic features of English, such as the
word fall and the phrase I guess (Crystal 1995:93).
Moreover, Berry’s (1997:70f) suggestion that there is a difference in meaning
between the two syntactic variants (see 3.3.4.3) is quite interesting in the light of
the present findings: Do Americans refer to established units more often than
people speaking British English? It is conceivable that some speakers regard the
two as having different meanings, but that a process of dialectal conventionalisation, i.e. “when certain possibilities become conventionally established” in a
dialect, has taken place (Langacker (1988:38), see Section 2.1.2).
6.1.5
NPs with all/whole and geographical names
Another case of regional variation which I have not seen described in the literature was detected in the variable with geographical names (Table 6.5).
Table 6.5. NPs with all/whole, /THE DEFINITE ARTICLE/ and GEOGRAPHICAL
NAME in different regional dialects
The New
The Sydney
The
York Times
Morning
Independent
Herald
N
%
N
%
N
%
all + GEOGRAPHICAL NAME114
55
16%
18
12%
45
18%
all of + GEOGRAPHICAL NAME
258
73
59
23%
77%
50%
the whole of + GEOGR. NAME
21
6%
55
38% 148
59%
334 100% 146 100% 252 100%
Total
All of was the predominant form in the AmE material (78%), whereas the whole
of predominated in the BrE material (59%). The AusE material adhered to the
AmE preference for all of (50%), even though the whole of was frequent here as
well (38%). Simple all was the least frequent variant in all three corpora.
6.1.6
Subgroups
In this section, we will look at three subgroups, two of them previously discussed in Chapter 5 (NPs with all/whole and a temporal noun and NPs with both
and a noun for body parts or kinship), and one more (NPs with all/whole and a
collective noun).
6.1.6.1
NPs with all/whole and temporal nouns
Table 6.6 shows the results of the comparison of different variants with temporal
nouns in the regional varieties.
114
In the material from NYT95, the normal search procedures could not be used with the simple all
variant. This might have affected the results to some extent. See further Appendix C.
119
119
Table 6.6. NPs with all/whole, /THE DEFINITE ARTICLE/ and TEMPORAL NOUN
in different regional dialects
The New
The Sydney
The
York Times
Morning
Independent
Herald
N
%
N
%
N
%
all + TEMPORAL NOUN
1812
95% 752
92% 1276
90%
0%
1%
1%
all the + TEMPORAL NOUN
2
9
18
all of the + TEMPORAL NOUN
0
0%
1
0%
1
0%
the whole + TEMPORAL NOUN
103
5%
53
7% 110
8%
the whole of the + TEMP. NOUN
0
0%
0
0%
9
1%
1917 100% 815 100% 1414 100%
Total
There were small differences generally between the varieties, the simple all
variant being a little more frequent in AmE than in the other varieties. Benson et
al (1986:21) write that all the is a Briticism. Since this variant is so infrequent
overall, it is of course difficult to judge whether there are any real differences
between the varieties. However, when comparing absolute figures it can at least
be concluded that all the was most frequent in BrE (18 instances, 4 per 10 M
words), less frequent in AusE (9 instances, 2 per 10 M words) and least frequent
in AmE (2 instances, 0,3 per 10 M words). There was also a difference in the use
of the whole between the spoken American material and the other corpora. We
will return to this in Section 6.2.
6.1.6.2
NPs with both and nouns for body parts or kinship
The second subgroup also exhibited regional variation, as illustrated in Table
6.7.
Table 6.7. NPs with both, /POSSESSIVE DETERMINER/ and NOUN FOR BODY
PARTS OR KINSHIP in different regional dialects
The New
The Sydney
The
York Times
Morning
Independent
Herald
N
%
N
%
N
%
322
78% 179
75%
both + NOUN FOR BODY
86% 224
PARTS/KINSHIP
61
15%
26
13%
73
both my etc. + NOUN FOR BODY
24%
PARTS/KINSHIP
30
3
1%
1
0%
both of my etc. + NOUN FOR
7%
BODY PARTS/KINSHIP
413 100% 208 100% 298 100%
Total
The form with of was more frequent in AmE than in the other variants, whereas
both my was particularly frequent in BrE and simple both in AusE.
120
6.1.6.3
NPs with all/whole and collective nouns
One more subgroup (of the all/whole + SINGULAR COUNT NOUN variable), not
discussed in Chapter 5, should be mentioned, since it shows a particularly interesting case of regional variation. The tokens in this group were not separated
from the larger group, since the variants are the same. It is just treated here as a
special case of all/whole in combination with singular count nouns.
Levin (2001:124) found that all with collective nouns, like family and government, is more frequent in BrE than in AmE, while whole, on the other hand, is
more frequent in AmE than in BrE. This study, which used the same corpus material but took a larger number of different collective nouns (in fact all found in
the material) into account, corroborated these results, as illustrated in Table 6.8.
Table 6.8. NPs with all/whole, DETERMINER and COLLECTIVE NOUN in
different regional dialects
The New
The Sydney
The
York Times
Morning
Independent
Herald
N
%
N
%
N
%
22
62
81
all + DETERMINER + COLLEC7%
31%
24%
TIVE NOUN
all of + DETERMINER + COLLEC-
21
7%
15
8%
10
3%
280
87%
120
61%
246
73%
323
100%
197
100%
337
100%
TIVE NOUN
DETERMINER + whole
COLLECTIVE NOUN
Total
+
All was indeed more frequent with collective nouns in BrE (24%) and
(especially) AusE (31%) than in AmE (6%), whereas whole was more frequent
in AmE (88%) than in BrE (73%) and AusE (61%). A plausible explanation of
the more frequent use of all in BrE and AusE is that all is mainly used to indicate individual members of a group. When referring to individuals, BrE speakers use plural agreement (in verbs and pronouns) with collective nouns more
often than Americans do. It seems likely that BrE speakers should thus use all
more often as well. This argumentation is in line with Huddleston & Pullum
(2002:375), who claim that all foregrounds the individual members of a group,
while whole focuses more on the entity.
6.1.7
General observations
It was mentioned in Section 2.2.1 that differences between AmE and BrE are
often presented in a categorical way in the literature, suggesting that differences
are absolute, i.e. that one variant only exists in BrE and the other only exists in
AmE. As for the use of quantifiers, some authors are categorical, such as Benson
et al (1986:21) who state that a half is an Americanism, and that all the day is a
Briticism. Others are more balanced, such as Berry who writes that the a half
variant is “more common in American English than British English” (Berry
1997:81). We can see from the results that there were no absolute differences between regional varieties in the corpora. Furthermore, the same syntactic variant
121
121
generally predominated in all three varieties, such as simple both (as in both
children). The only exceptions were NPs with geographical names and NPs with
both and a demonstrative pronoun, where different variants predominated in the
different varieties. In the latter of these, the figures were too low to be reliable.
In some cases an alternative form was far more frequent in one variety than in
another, for example a half and the variants with of in American English, even
though the same form (half a/an in this case) predominated in both.
Finally, a general observation regarding Australian English can be made. In
Section 2.2.1 it was remarked that AusE shares features of both BrE and AmE,
owing to its British origin and later to its American influence in the form of
media etc. (cf. Peters & Fee 1999:135f). The results of the analysis of quantifiers
corroborate this, since in most cases the figures for AusE lie between those of
the BrE and the AmE material (cf. Estling Vannestål 2001a). Quite often the
AusE figures were closer to those of one of the varieties, generally BrE. So,
judging from the present material, Australians resemble BrE speakers more than
AmE speakers in their use of quantifiers.
6.2
Distribution in spoken and written
English
Biber et al (1999:277) point to an interesting difference in distribution between
all and both across different registers. In their corpus research they found that all
was far more frequent in conversation and fiction than in academic discourse,
whereas the opposite was true of both. The authors explain this fact with reference to the need for precision in academic discourse (both being more precise
than all), while conversation and fiction “have a tendency to opt for more categorical expressions (especially all)”.
Chapter 3 reported on some more specific statements concerning the formality
of particular syntactic variants (all + SINGULAR COUNT NOUN, both of, a half and
all + GEOGRAPHICAL NOUN). We will begin by comparing these with the corpus
data. Since no spoken AusE was included in the material, the tables presented
only show figures from AmE and BrE. The differences were significant at the
0.01 level unless otherwise indicated (see Appendix E, Table E2, for phi coefficients).
First, Quirk et al (1985:381) claim that all in combination with a singular
count noun is more formal than a construction with whole. Since NPs with the
definite article and NPs with other determiners proved to exhibit rather great differences in the spoken corpora, they will be separated in the presentation. Table
6.9 shows the frequencies for NPs with the definite article.
122
122
Table 6.9. NPs with all/whole, THE DEFINITE ARTICLE and SINGULAR COUNT
115
NOUN in the written and spoken material
The New
The Independent
York Times
N
%
N
%
all the + SINGULAR COUNT NOUN
736
626
16%
22%
(87)
(97)
(3%)
(3%)
all of the + SINGULAR COUNT NOUN
90
3%
28
1%
(88)
(3%)
(28)
(1%)
the whole + SINGULAR COUNT NOUN
2466
75%
3173
80%
(2465)
(93%)
(3159)
(92%)
11
0%
162
4%
the whole of + SINGULAR COUNT
(11)
(0%)
(162)
(5%)
NOUN
3303
100%
3989
100%
Total
(2651) (100%)
(3446) (100%)
LSAC
BNC
(spoken AmE)
(spoken BrE)
all the + SINGULAR COUNT NOUN
477
159
18%
43%
(56)
(56)
(7%)
(8%)
all of the + SINGULAR COUNT NOUN
11
1%
7
1%
(11)
(2%)
(11)
(1%)
the whole + SINGULAR COUNT NOUN
625
56%
661
73%
(609)
(90%
(658)
(82%)
0
0%
78
9%
the whole of the + SINGULAR COUNT
(0)
(0%)
(78)
(10%)
NOUN
1113
100%
905
100%
Total
(676) (100%)
(803) (100%)
Nothing in this corpus material corroborates Quirk et al’s statement. In the three
newspaper corpora and the spoken British corpus, the relative frequency of all
the, compared to the other syntactic variants, is quite similar (18% for the spoken British corpus, 22% and 16% for the newspapers). Interestingly, in the American spoken corpus, the use of all with singular count nouns is much more frequent than elsewhere (43%). The reason for this is the even greater predominance of the all the way phrase in the spoken American corpus compared to the
other corpora. When NPs with way are excluded, there is very little difference
between the corpora as regards the use of all the.
In Table 6.10, we see the frequency distribution for the variants in the variable with all/whole, a demonstrative or possessive determiner and a singular
count noun.
115
Tokens with all the way excluded are given in brackets.
123
123
Table 6.10. NPs with all/whole, DEMONSTRATIVE/POSSESSIVE DETERMINER
and SINGULAR COUNT NOUN in the written and spoken material
The New
The Independent
York Times
N
%
N
%
371
306
all this, my etc. + SINGULAR COUNT
35%
42%
NOUN
all of this, my etc. + SINGULAR
64
6%
21
3%
616
58%
351
48%
3
0%
48
7%
COUNT NOUN
my, this whole + SINGULAR COUNT
NOUN
the whole of this/my + SINGULAR
COUNT NOUN
Total
all this, my etc. + SINGULAR COUNT
1054
100%
LSAC
111
19%
726
100%
BNC
354
73%
NOUN
all of this, my etc. + SINGULAR
17
3%
15
3%
452
78%
98
20%
0
0%
15
3%
580
100%
482
100%
COUNT NOUN
my, this whole + SINGULAR COUNT
NOUN
the whole of this/my + SINGULAR
COUNT NOUN
Total
Here, however, all was more frequent in the newspaper corpora than in the spoken American material (35–42% in the newspapers and 19% in LSAC). Meanwhile, all was in fact predominant (73%) in the spoken British material (BNC).
Three of the writers of A Comprehensive Grammar of the English Language
(Quirk et al 1985) are British (and the fourth is from Sweden, where BrE has traditionally been more influential than AmE in the teaching of English). Therefore
it is quite strange that the results of the comparison of the American material
should be in line with their suggestion that all is more formal than whole, whereas the results of the comparison of the British material point in the other direction (i.e. all was more frequent in speech than in writing). A large number of the
examples in the BNC consisted of two particular phrase types (all my/his etc. life
and all this/that lot). With these tokens disregarded, the differences are a little
less conspicuous. Nevertheless, all this/my etc. was still much more frequent in
the spoken British material than in the other three corpora.
Next, the OED (s.v. both adj. 6) considered both of in combination with a
determiner to be “colloquial”. One could therefore expect this variant to be more
frequent in speech than in writing. Both of with the definite article was very
infrequent in the spoken material, occurring only seven times in LSAC and five
times in BNC. The form was slightly more frequent compared to the other
syntactic variants (simple both and both the) in the spoken American material
than in the written material, but since the figures were so low and the differences
124
124
not statistically significant, I refrain from concluding that the results corroborate
OED’s claim. It should also be remembered that the information in OED is not
very up-to-date (although it is not known when this particular entry was written)
and over time, some previously “colloquial” forms also enter the written
medium. The OED claim also concerned demonstrative determiners. Apart from
the variant with of being more frequent in the American than in the British
material in both media, of was indeed more frequent in speech. Again, however,
the figures were low, and the differences were not statistically significant.
Putseys (1984:379) writes that a half is more formal than half a, while Longman (2003) refers to a half as a form typical of spoken American English. The
syntactic variants are presented in Table 6.11.
Table 6.11. NPs with half, THE INDEFINITE ARTICLE and SINGULAR NOUN or
NUMERAL in the written and spoken material
The New
The Independent
York Times
N
%
N
%
half a/an
1841
59%
2136
94%
a half
1254
127
6%
41%
3095
100%
2263
100%
Total
half a/an
a half
Total
LSAC
N
%
191
68%
90
32%
281
100%
BNC
N
%
745
95%
39
5%
784
100%
In the British corpus material, there was no difference between speech and writing, a half occurring in 5% of BrE speech, and in 6% of BrE newspaper text. In
AmE, a half was more frequent in the newspaper text (40%) than in the spoken
material (32%), corroborating Putseys’ claim.
Simple all with a geographical name (as in All Paris welcomed the general) is
further considered formal by Quirk et al (1984:260). In the British material,
simple all was more frequent in writing than in speech, which might indicate that
the form is felt to be more formal. There was, however, an opposite tendency in
the American material. Finally, the figures were very low (too low to be submitted to the significance test), so it would be unwise to jump to conclusions.
Let us now move on to the subgroups. First, there was a small difference between speech and writing in the American corpora concerning NPs with temporal nouns (Table 6.12). As we saw in Table 6.6, there were virtually no differences in distribution between the newspapers. In the spoken corpora, however, the whole was more frequent in the spoken AmE material than elsewhere.
125
125
Table 6.12. NPs with all/whole, /THE DEFINITE ARTICLE/ and TEMPORAL
NOUN in the spoken material
The New
The Independent
York Times
all + TEMPORAL NOUN
1812
95%
1276
90%
all the + TEMPORAL NOUN
2
0%
18
1%
all of the + TEMPORAL NOUN
0
0%
1
0%
the whole + TEMPORAL NOUN
103
110
8%
5%
the whole of the + TEMPORAL NOUN
0
0%
9
1%
1917
100%
1414
100%
Total
all + TEMPORAL NOUN
all the + TEMPORAL NOUN
all of the + TEMPORAL NOUN
the whole + TEMPORAL NOUN
the whole of the + TEMPORAL NOUN
Total
LSAC
472
84%
6
1%
0
0%
81
14%
0
0%
559
100%
BNC
750
91%
11
1%
1
0%
55
7%
6
1%
823
100%
There was also a difference according to medium in the variable with both and a
noun denoting body parts or kinship. Table 6.13 shows the correlation.
Table 6.13. NPs with both, /POSSESSIVE DETERMINER/ and NOUN FOR BODY
PARTS OR KINSHIP in the written and spoken material
The New
The Independent
York Times
N
%
N
%
322
224
both + NOUN FOR BODY
78%
75%
PARTS/KINSHIP
61
15%
73
24%
both my + NOUN FOR BODY PARTS/
KINSHIP
both of my + NOUN FOR BODY
PARTS/KINSHIP
Total
both + NOUN FOR BODY
PARTS/KINSHIP
both my + NOUN FOR BODY
PARTS/KINSHIP
both of my + NOUN FOR BODY
PARTS/KINSHIP
Total
126
30
7%
1
0%
413
100%
298
100%
LSAC
N
%
25
66%
BNC
N
%
16
48%
8
21%
17
52%
5
13%
0
0%
38
100%
33
100%
126
The table indicates that the simple both form is more frequent in written than in
spoken English, since it correlated positively with both the American and the
British newspaper material. Also note that both my was proportionally more
frequent in the spoken British material, but on the other hand there were few
tokens in the spoken corpora.
As far as non-standard variants are concerned, there are two variants that are
sometimes accounted for in the literature as mainly belonging to spoken language. Both these can be regarded as containing redundant material. The first
one is the both of us etc. as an alternative to we … both and both of us (cf.
Huddleston & Pullum 2002:377), which was found in 17 cases in the corpus material (the spoken corpora and in interviews in the newspapers). The second variant is a half a, observed by Jespersen (1909–49:361) and Svartvik & Sager
(1996:275). The latter point out that this form is “very colloquial” (my translation). In my material, it was used in as many as 156 cases, mainly in the AmE
corpora (73 in the spoken AmE corpus and 46 in NYT95). Still, it should be
noted that a corpus of spoken language necessarily contains false starts and misprints, and a half a could be an example of such features in some of the cases.
Examples (6:1) and (6:2) illustrate the non-standard variants.
(6:1) Doctor said he’ll send for the both of us, right? (BNC)
(6:2) Well, you’re pretty far out. A half an hour out of the city. (LSAC)
The conclusion drawn from the study of medium (conducted on BrE and
AmE material only) is that there were very little difference between the spoken
and the written material. There were some tendencies, but in most cases, either
the figures were low or a difference occurred in only one of the varieties. One
interpretation of the results is that there is indeed little difference between speech
and writing in the use of quantifiers. Another explanation could be that the registers used were too similar to reveal differences, for instance because newspaper
text includes so much dialogue. Perhaps the greatest difference between spoken
and written language with respect to quantifiers is that noted by Biber, referred
to above, viz. that all is more frequent in speech and both in writing.
6.3
Summary
This chapter dealt with two non-linguistic factors influencing the choice between
two or more syntactic variants: region and medium. Section 6.1 reported on differences between British, American and Australian English. Here it was concluded that two statements about regional differences in the use of quantifiers
were corroborated by the corpus results. First, of forms are generally more frequent in AmE than in BrE, and, second, the a half variant is used more often in
AmE than in BrE (where half a/an predominates). Moreover, a few previously
undetected regional differences were reported on. The idea that AusE, which has
been influenced by both BrE and AmE, takes an intermediate position between
the two varieties, was confirmed. From the results, it could, however, be concluded that Australians resemble British speakers more than Americans in their
127
127
use of quantified NPs. In the study of spoken and written material (conducted on
BrE and AmE material only) no great differences were detected, a fact which
may have different reasons.
128
128
7. Linguistic factors
The third aim of this study was to test a number of hypotheses about how linguistic factors in the noun phrase and its surrounding co-text influence the
choice of variant in the quantified NP types. The underlying assumption was that
there are indeed such correlations. We will move gradually from factors internal
to the NP (determiners, head and modifiers) to factors concerning elements outside the NP and then to the NP in its relation to the clause in which it occurs.
Some of the factors call for explanations, which will be given in each of the sections. Some of the factors were investigated in all of the variables, while others
were only relevant to one or a few of these (see Table 7.1 below). The following
factors will be analysed (and presented in the same order):
1. Type of central determiner in the NP
2. Factors relating to the NP head
a. noun vs. demonstrative pronoun as head
b. number and countability
c. divisibility in NPs with singular count nouns
d. animacy
e. natural vs. arbitrary time division in NPs with temporal nouns
f. type of “head” in NPs with half, the indefinite article and a
singular noun or numeral
3. The presence of certain elements in the NP or its near co-text
a. modifiers
b. an adjacent of
c. focus markers
4. Syntactic function of the NP
7.1
Introduction
The term “linguistic factor” is used to refer to the different linguistic characteristics of the NP and its co-text that might influence the choice of variant. “Category” is used for the groups into which a particular factor is divided. For instance, in the factor “the presence of an adjacent of ”, the categories are “no of ”,
“of within the NP” and “of preceding the NP)” (Figure 7.1).
129
129
FACTOR:
the presence of an
adjacent of
CATEGORY:
no of
CATEGORY:
of within the
CATEGORY:
of preceding the NP
Figure 7.1. Example of factors and categories in the study
So as to examine as many linguistic factors as possible, I selected certain variables from those presented in Chapter 5 (see Table 7.1). All three variables with
all and a common noun were included, plus the subgroup with temporal nouns,
since these variation patterns look so different. As for NPs with both, the both
(/of/the) + PLURAL NOUN variable and the subgroup with nouns denoting body
parts or kinship were analysed. As for NPs with half, only the group with plural
nouns was included, since the variation is the same for NPs with singular, plural
and mass nouns. In NPs with demonstrative pronouns as head as in all (of) these,
only those with all (not NPs with both and half) were included. The same applies
to NPs with personal pronouns, as in all of us/we all. In both these variables, it
was hoped that all would be a good representative for the other quantifiers, since
they occur in the same kind of constructions and had fairly similar variation
patterns. The variables with half, the indefinite article and a singular noun or
numeral and with all/whole and a geographical name were all included in the
analyses.
It should also be observed that not all of the linguistic factors accounted for
above are relevant to each of the variables. In a few cases, a factor is only relevant to one single variable (see the list at the beginning of this chapter). Number
and countability is a special case, since it compares variables with each other
rather than factor categories within a variable. Table 7.1 shows the factors where
several variables were investigated, each one marked with a plus sign. Note that
the variables are condensed in the table, so that, for instance, all/whole + SINGULAR COUNT NOUN has four variants: all + DETERMINER, all of + DETERMINER,
DETERMINER + whole and the whole of + DETERMINER. Similarly, all + PERSONAL
PRONOUNS has two variants: all of us and we (…) all.
In the variable with all/whole and singular count nouns, tokens with way were
excluded from the samples, since the phrase all the way was so dominating in
the material, and also because there are very few instances of all of the way and
the whole way. Another very frequent phrase was all my/his etc. life, but since
there were also quite a few cases of all of my/his etc. life and my/his etc. whole
life, the tokens were included. The same applies to the all/whole and mass noun
variable, where the whole time was predominant in the whole variant. Here too,
there were quite a few instances of alternative variants, all the time and all of the
time, so these tokens were not excluded.
130
130
Table 7.1. Linguistic factors relevant to more than one variable
Type of
AniModi- An adja- Focus
Synt.
det.
macy
fiers
cent of markers function
+
+
+
+
+
+
all/whole +
SING. COUNT
+
+
+
+
+
all/whole +
−
MASS NOUN
all + PLURAL
NOUN
+
+
+
+
+
+
both + PLURAL
−
+
+
+
−
+
+
+
+
+
+
+
−
−
−
+
+
+
all + PERSONAL
PRONOUN
−
−
−
+
+
+
half + THE
INDEF. ART.
all/whole +
GEOGR. NAME
all/whole +
TEMP. NOUN
both + NOUN F.
BODY PARTS etc
−
−
−
−
+
+
−
+
−
+
+
+
−
−
+
−
+
+
−
+
−
+
−
+
NOUN
half + PLURAL
NOUN
all + DEM.
PRONOUN
For this part of the study, only material from The New York Times and The
Independent was used. When the number of tokens of a particular variant exceeded 100 in either of the newspaper archives, 100 tokens per corpus were sampled according to the procedure accounted for in Section 4.3.3. When the frequency of a variant in a corpus was lower than 100, all these tokens were
used116. Three of the factors constitute a special case: type of central determiner,
noun vs. demonstrative pronoun as head and number and countability. For these,
all tokens occurring in the corpus were included since they had been separated in
the corpus queries, where particular words (e.g. all + the and all + this) had to be
searched for.
For each variable, the distribution of variants in one category was compared
with the distribution in the other category or categories (see Section 4.3.3). Such
comparison in a constructed example is illustrated in Table 7.2 and 7.3, the former exemplifying a case of correlation and the latter of non-correlation.
116
In these cases, the frequencies were normalised to 100 on the basis of their internal distribution
into different categories. Thus, for instance, if one variant comprises only 50 tokens and these are
distributed over two categories (14 in category A and 36 in category B), the normalised frequencies will be 28 and 72 respectively. This measure makes it easier to compare the low-frequency
variants with the 100-token samples of higher-frequency variants. Of course, we must be aware
that normalised frequencies may give a skewed picture of reality. For the chi-square testing, only
absolute figures were used.
131
131
Table 7.2. Constructed example of frequencies with correlation between the
presence of X and Variant 1
Presence of X
Absence of X
Total
Variant 1 Variant 2 Total
40
20
60
60
80
140
100
100
200
So how should this (and the following) table be interpreted? If a certain factor
(e.g. the presence of X) is analysed and the distribution of the two variants is 40–
20 in the presence of X and 60–80 in the absence of X, it can then be assumed
that there is a positive correlation between the presence of X and Variant 1, and
consequently also between the absence of X and Variant 2 (Table 7.2).
According to the null hypothesis, i.e. the expected frequencies when there is
no correlation at all between categories and variants (Table 7.3), the number of
tokens would be the same for each variant: for instance, 40–40 and 60–60.
Table 7.3. Constructed example of frequencies with no correlation
Presence of X
Absence of X
Total
Variant 1 Variant 2 Total
40
40
80
60
60
120
100
100
200
In other words, it can be concluded that, judging by the data in Table 7.2, this
particular linguistic factor (the presence of X) influences the choice of variant to
some extent. It should be remembered that a correlation found between such a
linguistic factor and a variant only implies that the likelihood for a certain choice
is greater than otherwise, not that a certain form is required. The chi-square test
was used to test the significance of all correlations. In the presentation, the correlations were significant at the <0.01 level unless otherwise indicated. The relative strength of the correlations as they appear in terms of phi coefficients (see
Section 4.3.4) will be discussed in Chapter 8. The phi coefficients are given in
Appendix E, Table E3.
When different factors are explored, it sometimes happens on closer inspection that two seemingly independent factors turn out to be related to each other.
This means that the correlation between one variant and a particular factor is
caused by the correlation between this variant and another linguistic factor.
There are some examples of such dependency in the material. Two different factors may also cancel each other out, so that a factor that should have exhibited a
correlation does not do so because another linguistic factor works in the opposite
direction.
The following model will be used for the presentation of the results. In cases
where there was a correlation in both the British and the American samples with
respect to a particular linguistic factor, frequencies demonstrating this tendency
are shown in a bar chart. Observe that the frequencies from the two corpora were
conflated in the graphs and also when tested for significance, but the separate figures (British/American) can be found in tables in Appendix D. In these tables,
132
132
correlations are presented in bold type and normalised frequencies in italics. If
no correlation was found, or if tendencies pointed in different directions in the
two corpus samples, this will only be mentioned briefly in the text. It should be
taken into account that there is a small risk of missing relevant information by
paying little attention to tendencies that were only spotted in one of the corpora.
As the aim was to find correlations that occur regardless of regional variety, I
decided that this was the best method (also see Section 2.1.2). Finally, it should
be remembered that a correlation may always be a matter of chance, even when
evidenced in both corpora and even if statistically significant.
We will now go through all the various linguistic factors in turn. The fist factor of study is the type of central determiner in NPs with common nouns.
7.2
Type of central determiner in the NP
In NPs with all/whole, both and half combined with a common noun, four different types of central determiners can follow the quantifier: (i) the definite article,
(ii) a possessive, (iii) a demonstrative and (iv) a specifying genitive construction.
In the case of both, there are three variables with different variation patterns
(both vs. both the vs. both of the; both these/my vs. both of these/my; both vs.
both my vs. both of my, the latter regarding NPs with nouns denoting body parts
or kinship, see Section 3.3.4.2). These variables cannot really be compared with
each other. The focus in this section is thus on comparable variables: NPs with
all/whole and NPs with half.
The presentation of type of central determiner differs from those where other
linguistic factors are involved, in that the whole material, not samples, was used
here, as described above. The graphs, which also look different from those in the
rest of the chapter (columns and rows reversed), illustrate the relative difference
(in percentages) between the determiners in the distribution of variants117. In
NPs with all and mass nouns, there were no statistically significant differences
between the determiners. As far as all and plural nouns is concerned, there was a
significant correlation between all of and a possessive determiner and a genitive
phrase in the American material, which did not recur with statistical significance
in BrE.
The first bar chart (Figure 7.2) shows the distribution of the four variants in
NPs with all/whole and single count nouns. Since the figures in the charts are
percentages (based on the average of the frequencies in the British and American
material) and the corpora contained very few genitives, the latter were excluded
from the graph.
117
For methodological reasons, in genitive constructions the statistics only comprise tokens where
the genitive is preceded by the definite article, a possessive determiner or a demonstrative determiner (not those where the genitive is a proper noun). Had these been included, the frequencies
would have been larger. However, this does not necessarily mean that the relations between the
variants would have been affected.
133
133
92
100
80
70
60
43
40
20
all
all of
whole
whole of
49
25
3 2
3
5
3
2
3
0
def. art.
poss. det.
dem. det.
Figure 7.2. Type of central determiner in NPs with all/whole, DETERMINER
and SINGULAR COUNT NOUN
There were clear differences between the different types of determiners. The
whole variant predominated most in NPs with the definite article, as in (7:1), and
was least frequent in NPs with a possessive determiner.
(7:1) ‘The whole question this season was going to come down to me and Kevin
being healthy,’ he said. (NYT95)
The all variant was much more frequent in NPs with a demonstrative determiner,
as in (7:2), and (especially) a possessive determiner, as in (7:3), compared to
NPs with the definite article.
(7:2) Hidden within these mags live Marie Stopes, Eleanor Marx, Olive
Schreiner - all that half-forgotten generation of feminists and idealists
who had faith in women, […] (IND95)
(7:3) ‘I’ve worked so hard all my life, said Mr. Nakane […] (NYT95)
A majority of NPs with all and a possessive included the noun life, as in (7:3).
The all of and the whole of variants were marginal with all three determiners.
Tokens where the determiner was a genitive construction were few but interesting, since there are two different ways in which the genitive determiner can
function. In NPs with whole, the quantifier generally quantifies the noun in the
genitive rather than the second noun, as in (7:4). The whole quantified NP functions as a determiner.
(7:4) I became fluent in Afrikaans, much to my whole family’s amazement.
(IND95).
Here, the genitive could not be replaced by a possessive determiner (*their
whole amazement), as it usually can when the second noun is quantified, as in
(7:5).
134
134
(7:5) […] he was restricted in his moves because he did not control all of the
county’s budget. (NYT95) (Æ all of its budget).
We will return to genitives in relation to syntactic function118 in Section 7.5.
Figure 7.3 illustrates the distribution of half and half of and plural nouns.
90
80
70
60
50
40
30
20
10
0
85
64
61
52
39
48
36
half
half of
15
def. art.
poss. det.
dem. det.
genitive
Figure 7.3. Type of central determiner in NPs with half, DETERMINER and
PLURAL NOUN
The half variant predominated in NPs with the definite article, a possessive
determiner and (less strongly) a genitive construction, as in (7:6) to (7:8). In contrast, half of was much more frequent in NPs with a demonstrative determiner, as
in (7:9).
(7:6) After a seven-month strike, half the internal programmers were moved to
other areas. (IND95)
(7:7) The two Buckinghamshire secondary schools in Burnham take half their
pupils from Slough. (IND95)
(7:8) There was little change in the first nine months of last year, according to
preliminary data from about half the state’s 234 hospitals […] (NYT95)
(7:9) ‘More than half of these children live in families with working parents,’
she said, […] (NYT95)
It should be observed that there were few tokens that included a demonstrative
determiner.
118
Also note that the genitive itself is not always a determiner, but is sometimes used as a descriptive/classifying modifier (see Section 7.4.1).
135
135
7.3
Factors relating to the NP head
This section will bring up various aspects, both grammatical and semantic, relating to the head of the noun phrase. The starting-point in this section is that in
NPs with a noun, it is this noun which is the headword. We will return to
whether this is an appropriate analysis in noun phrases with of (as in all of the
children) in Section 7.3.3. The presentation first deals with a grammatical factor:
noun vs. demonstrative pronoun as the head of the NP. It moves on to an aspect
involving both grammar and semantics: number and countability in NPs with
common nouns. The section ends with four semantic aspects: (i) divisibility in
NPs with singular count nouns, (ii) animacy in NPs with common nouns and
geographical names, (iii) natural vs. arbitrary time division in NPs with temporal
nouns and finally (iv) the semantic type of “head” in NPs with half, the indefinite article and a singular noun or numeral.
7.3.1
Noun vs. demonstrative pronoun as head
As described in Section 5.1.2, variants with of were more frequent when the
head of the noun phrase was a demonstrative pronoun, as in (7:10), than when
the internal structure of the NP was [DETERMINER + NOUN], as in (7:11)119.
(7:10) New norms, some protocols, some guidelines – all of these will develop
the way they develop in any part of society as it undergoes change.
(NYT95)
(7:11) He left his mark on all of these congregations, […]. (IND95)
This tendency recurred with all three quantifiers (all, both and half), and was
particularly strong with both of, as in (7:12), and half of 120, as in (7:13).
(7:12) Both of these were eclipsed by the devastating 7.1 in 1886 in Charleston,
S.C. (NYT95)
(7:13) More than half attended college, and half of those graduated. (NYT95)
There were fairly few instances of both of and a demonstrative pronoun, however. Quirk et al (1985: 258, 373) remark that the of form is often “preferred”
with demonstrative pronouns, but they do not provide any explanation of this
fact. One simple but possible reason is that it is a question of rhythm, i.e. that
speakers avoid two adjoining stressed syllables by the insertion of of. As noted
119
An additional observation was that in the singular count noun group the of-less all variant was
much more frequent with a possessive determiner (all my etc.) than with the definite article or a
demonstrative determiner (all the/this etc.). The reason was that this group included a great many
instances of all my/his etc. life.
120
As mentioned above, these two variables will not be part of the rest of the analysis of linguistic
factors. However, they were included here since the division had already been made in the
corpus queries.
136
136
in Section 4.2, many of the examples from the newspaper corpora are writtendown interviews, i.e. spoken (although edited) material.
7.3.2
Number and countability in NPs with common nouns
We saw in Section 5.1.1.1 that there are clear differences in variation pattern
with respect to number and countability as far as variants with all and whole are
concerned. This means that the material had to be organised into different
variables (all/whole + SINGULAR COUNT NOUN, all/whole + MASS NOUN and all +
PLURAL NOUN). In NPs with a singular count noun, the whole variant, as in
(7:14), was the preferred alternative, while the all variant predominated in NPs
with mass nouns, as in (7:15).
(7:14) They are not signing up to the whole Charter 88 agenda. (IND95)
(7:15) Leaving all the hard work to someone else is also an option. (IND95)
The variant with the whole of was the least frequent alternative in the all/whole +
variable, and almost non-existent in NPs with mass
nouns. Plural NPs with whole were rare and have a different meaning from the
majority of those with all (see Section 3.3.4.4).
One particularly interesting issue is NPs with mass nouns and the quantifier
whole, since Quirk et al (1985:260) do not consider this combination of quantifier and noun type acceptable. As we saw in Section 5.1.1.1, the phrase the
whole time constituted a large part of these examples. The word time is a mass
noun when describing “something that is measured in minutes, hours, years etc.
using clocks” (Longman 2003), and the countable form of time means something
else (‘an occasion’). It seems that in some cases, the word time in the first sense
may be thought of as a countable unit, in which case the whole would be the
most natural quantifier. There could be a difference between all (of) the time and
the whole time, the latter being more likely than the former to refer to a more
specific time period, whole being used for individuating the reference of the
noun phrase. Cf. He was silent the whole time, which refers to a specific period
of time, with He was interrupting me all the time, where the NP seems to have
more generic reference, meaning something like ‘continually’ (cf. Labov
1984:53). This interpretation strengthens the idea of time in the whole time being
perceived as a count noun. On the other hand, the whole time can also be used
more generically (see Section 3.3.4.4).
The borderline between count and mass is often fuzzy. Depending on the context, many words can belong to either of the categories (cf. 3.3.4.4). The contextual interpretation was taken into account with tokens which dictionaries (e.g.
Longman 2003) denote as both count and mass. As mentioned in Section 5.1.1.1,
the very few really ambiguous cases were classified according to the quantifier
with which they occurred, i.e. as a mass noun with all and as a singular count
noun with whole.
The other mass nouns used with whole in the material were acreage, air, architecture, attention, baggage, behaviour, brouhaha, chaos, chemistry, conduct,
demeanour, exhilaration, hurly-burly, hypocrisy, hysteria, machinery, mayorSINGULAR COUNT NOUN
137
137
alty, mystique, panoply, politics, renovation, rigmarole, shame, sharing, significance, socialising, stability, stock 121, stuff, technology, trade, uneasiness and
vastness, all of which are invariably mass nouns in Longman (2003). Abstract
nouns, which account for the majority of these examples, are more prone to be
used as count nouns than, for instance, nouns expressing a concrete substance
(?the whole butter). Again, it can be hypothesised that whole is used for individuation in a particular situation. A question arising is whether the people who
used whole in combination with these words were conceptualising them as count
nouns or whether their idiolects simply make no distinction between singular
count nouns and mass noun in the choice between the two quantifiers.
As observed in Section 5.1.1.3, the tokens with half were, in conformity with
the all groups, also divided into singular count nouns, mass nouns and plurals. In
the British material, the half of variant was more frequent in NPs with mass and
plural nouns than in NPs with singular count nouns. There was no difference between the noun types in the American corpus.
The variable all + DEMONSTRATIVE PRONOUNS was analysed for number only.
The singular tokens (all (of) this/that) invariably had complex antecedents, such
as phrases, clauses or sentences. Therefore, no distinction into singular count
and mass was made. The material exhibited no correlation with variants consistent across the two corpus samples. It can be noted, however, that, regardless of
variant, there were overall far more instances of singular NPs, as in (7:16) and
(7:17), than plural ones, as in (7:18) and (7:19).
(7:16) All this is pure Highland magic: here nature and the mark of man […] are
in rhapsodic equilibrium. (IND95)
(7:17) All of that makes sense. (NYT95)
(7:18) Beneath all these is an underworld of gnomes, doing the usual gnomish
things with wheelbarrows, watering cans and fishing rods. (IND95)
(7:19) Midwood has all of those, plus access to Brooklyn College across the
street, […]. (IND95)
The next section discusses the divisibility of the head in NPs with singular count
nouns, which is an issue clearly related to that of number and countability.
7.3.3
Divisibility
Most examples of NPs with singular count nouns and mass nouns as heads behaved in line with the expectations, derived from descriptions in the literature.
NPs with singular count nouns mainly take whole rather than all (of), whereas
NPs with mass nouns mainly take all (of) the rather than whole. However, as reported on in Sections 5.1.1.1 and 7.3.2, the material also included cases where
121
138
This is a word that can be both count and mass. In the context in which it occurs, it means ‘the
total value of a company’s shares’ (cf. Longman 2003), in which case Longman classifies it as a
mass noun.
138
whole is used with mass nouns, and similarly there are cases of all in NPs with
singular count nouns. I decided to look into divisibility in the NPs with singular
count nouns since Quirk et al (1985:259f) suggest that the of-less all variant is
only used with singular count nouns if the noun is divisible, and that all of and
whole are the most natural alternatives. The hypothesis was that the heads in all
tokens of all and a singular noun (as in all the/my/this book) in the material
would be divisible, or that there would at least be a higher proportion of indivisible nouns among those NPs with all of and whole (of) than among those with ofless all.
Divisibility is not always a straightforward concept. For instance, as mentioned in Section 3.3.4.4, there is often a possibility of regarding not clearly divisible nouns as divisible under certain circumstances. The main principle here
has been that in order to be categorised as divisible, the noun must be divisible
into parts that are similar in kind if not in size, just as is the case with mass
nouns. For instance, a word like forest was seen as divisible since it comprises a
number of trees. Although there is subjectivity involved in my analysis, I will
ensure that all cases of all in combination with indivisible nouns are accounted
for. All collective nouns, such as crew, family and government are clearly divisible, since they are as it were made up of different individuals and thus have a
kind of “plural” function, sometimes reflected in the choice of plural agreement
(see Section 3.3.4.4). Words denoting geographical areas, such as city, country
and world were also classified as divisible on the grounds that they can quite
easily be divided into smaller areas (e.g. city into streets or blocks, country into
regions and world into countries). Similarly, temporal nouns122 are divisible into
smaller time units (e.g. week into days). There were also a number of other
nouns which were seen as divisible, such as career (into positions), history (into
events), life (into years) and song (into notes). As mentioned in Section 3.3.4.4,
Quirk et al (1985:260) claim that all is mainly used with abstract divisible
nouns. Indeed, this was true of all tokens except two: forest and catalogue.
Figure 7.4 shows how the categories correlated with the variants. The hypothesis that of-less all is never used with indivisible nouns was not confirmed.
Still, indivisible nouns were in the minority in the all group (22 tokens out of
200), and indivisibility was far more frequent with the whole variants123. Also, in
accordance with Quirk et al (1985:260), the correlation pattern in NPs with all of
differs from that in NPs with all, instead resembling the variants with whole (of).
122
The occurrences of temporal nouns here are from NPs with demonstrative determiners, as in all
this night. Cases with the definite article are included in the special subgroup with simple all as
an alternative (cf. Section 5.1.6.1).
123
In the case of the the whole of + DETERMINER variant, the majority of BrE tokens were
indivisible (60 out of 100). The American material comprised only 10 tokens, 8 of which were
divisible and 2 indivisible.
139
139
200
180
160
140
120
100
80
60
40
20
0
178
132
113
120
87
80
68
divisible
indivisible
22
all
all of
whole
the whole of
Figure 7.4. Divisibility in NPs with all/whole, DETERMINER and SINGULAR
NOUN
The noun types in NPs with the of-less all variant classified as indivisible were
area, body, brain, budget, debt, fall 124, heart, investment, period, plan, rise,
trophy, valley, and way 125 (some of them occurring more than once). Here are a
few example sentences:
(7:20) We know all this area is a flood zone. (NYT95)
(7:21) I don’t think the Taliban came all this way to stop at the gates […]
(IND95)
The referents of these nouns are, in my view, at least not clearly divisible into
parts of an equal kind. Admittedly, it could be argued that some of them are divisible, even if those cases are not as obvious as those accounted for above. For
instance, nouns that can express ad hoc division, such as area and period, could
be divided at a lower level into sub-areas and sub-periods. Furthermore, some of
the words can be perceived as mass rather than count nouns, even though they
are countable in Longman (2003). In such cases, all would be a natural choice.
This reasoning applies, for instance, to a word like debt, which is used as a mass
noun in an expression like in debt.
To conclude, since the majority of tokens of all in combination with a singular count noun could quite easily be interpreted as having a divisible referent,
Quirk et al (1985:259, 381) are obviously right in their statement that divisibility
is a crucial factor. All and all of having different correlation patterns may be seen
124
Fall is used in this case with the meaning “a reduction in the amount, level, price etc. of something”, classified as a count noun by Longman (2003).
125
Note that the tokens with way here are used in NPs with demonstrative determiners, as in all this
way. Cases with the definite article (all the way) were excluded in this chapter, as accounted for
in Section 7.1. Quirk et al (1985:259) regard this noun as divisible.
140
140
as supportive of the point of view that NPs with these two forms should be
analysed as two different syntactic structures. Quirk et al (1985:258) and
Huddleston & Pullum (2002:333, 376), among others, suggest that all in a
phrase like all of the children is the head of the noun phrase (cf. Section 3.3.4.1).
If all is the head, of the children should be a postmodifier, which may explain
why the divisibility of the noun is less important than in NPs where all directly
precedes the [DETERMINER + NOUN] structure.
7.3.4
Animacy
Animacy in language, an important and sometimes complicated concept, was
discussed in Section 3.3.4.4 in relation to quantified noun phrases with all and
geographical names and with all/whole and collective nouns. There are different
ways of analysing collective nouns and geographical entities. Yamamoto
(1999:16ff, 139) classified virtually all cases as inanimate, although observing
that they can sometimes be regarded as having “inferred animate” reference.
Levin (2001:126), on the other hand, considered all collective nouns in his study
that refer to groups of human beings animate. I used a model where animacy was
divided into “real animate” and “inferred animate”, even though these two were
grouped together as animate in the graphs (see Appendix D for the separate
figures). Some collective nouns were invariably categorised as “real animate”,
e.g. family and staff. These are seldom looked upon as abstract entities, which is
reflected in the fact that they are frequently used with plural agreement in verbs
and pronouns (cf. Levin 2001:131)126. “Inferred animate” was used about words
which can be thought of either as groups of individuals or as (administrative/
sports/geographical etc.) entities, e.g. government, school, team and geographical names. Tokens of this type were included in the group if, in the co-text,
there was some kind of signal of animacy, underlined in the following examples:
(7:22) Nevertheless, the pair were wed in 1673 at a ceremony boycotted by
almost all the court except the loyal Sir Edward Carteret, […]. (IND95)
(7:23) All the world loves a winning Ferrari. (IND95)
(7:24) ‘Both of the companies are saying we believe in the world of interactivity
but we’re bringing this world into broadcast,’ Mr Gates said. (IND95)
All tokens classified as animate are of the “inferred animate” kind, as in (7:25).
(7:25) When the right-hander was named to the American League All-Star team
last week, all of Nicaragua rejoiced. (NYT95)
Section 7.3.4.3 will compare Quirk et al’s (1985:260) idea about geographical
names with the corpus material, but first we will look at common nouns.
126
In NPs with all, both and half and plural nouns, there were of course also a number of unambiguously animate cases, such as all the children or both doctors.
141
141
7.3.4.1
NPs with all/whole and singular count nouns
In Section 6.1.6.3 we saw that in NPs with collective nouns all was more frequent in BrE than in AmE. This fact was discussed in terms of all foregrounding
individuals, whereas whole foregrounds the entity (Huddleston & Pullum
2002:375). In line with this, one could expect a correlation in BrE (i) between all
and animate reference, and (ii) between whole and inanimate reference in the
variable with all/whole and singular count nouns, provided the material contains
a sufficiently large number of collective nouns. The hypothesis was corroborated, as illustrated by Figure 7.5, which only shows the British material since no
correlation occurred in AmE127.
100
71
67
50
90
85
33
inanimate
animate
29
15
10
0
all
all of
whole
the whole
of
Figure 7.5. Animacy in NPs with all, /DETERMINER/ and SINGULAR COUNT
NOUN
Animate reference (real and inferred) correlated positively with all and all of, as
in (7:26) and (7:27), while inanimate reference correlated with whole and the
whole of, as in (7:28) and (7:29).
(7:26) Among those who will not benefit are part-timers who joined TSB after
1976 but left before December 1990, when the bank admitted all its staff
into the scheme. (IND95)
(7:27) But Mark Jackson is going to make all of their team better. (NYT95)
(7:28) Indeed, it seems that the whole success story was sparked off by a
quarrel. (IND95)
(7:29) The Cornwall Coast Path runs 268 miles around the whole of the county,
[…] (NYT95)
127
142
Henceforth, correlations that only occurred in one of the corpora are only mentioned in the text
and not illustrated in the form of graphs and tables in the appendix. This is a special case, however, since the result corroborated the hypothesis about a region-specific grammatical tendency.
142
Moreover, in the simple all variant there were more tokens with animate reference proper than with inferred animacy (such as government and school). In
the other three variants it was the other way round.
7.3.4.2
NPs with plural nouns
The variables with all, both and half in NPs with plural nouns were all included
here. No statistically significant correlations occurred in NPs with all and half,
but there was a positive correlation in the variable with both (Figure 7.6).
160
140
120
100
80
60
40
20
0
144
105
95
97 102
inanimate
animate
56
both
both the
both of the
Figure 7.6. Animacy in NPs with both, /THE DEFINITE ARTICLE/ and PLURAL
NOUN
Here, inanimate reference correlated with simple both, as in (7:30). There was
also a correlation between animate reference and both the and (slightly more
saliently) both of the, as in (7:31) and (7:32).
(7:30) Both permissions were given, on appeal, by the Government in the 1960s
after refusals by the county council. (IND95)
(7:31) Both the newcomers were backed by a political group with ties to the
state’s largest street gang, Gangster Disciples. (NYT95)
(7:32) In Jackling’s eyes both of the Hewlett women were immensely desirable,
with long luscious lines like luxury lawnmowers. (IND95)
In all cases except simple both in NYT95, there were more tokens of animate
reference proper than of inferred animate reference. The largest proportion of
tokens of inferred animacy occurred in the both of the variant (in both corpora),
but note that in NYT95 there were very few tokens of this kind.
7.3.4.3
NPs with all/whole and geographical names
Quirk et al (1985:260) remark that there is a difference in the use of totalising
quantifiers depending on whether the noun phrase refers to a geographically or
143
143
politically defined area or to the people inhabiting it128. In the former case, the
the whole of variant, as in (7:33) is claimed to be preferred, while the simple all
variant, as in (7:34), is said to predominate in the latter.
(7:33) Ours is the cleanest restaurant in the whole of India. (NYT95)
(7:34) But Beijing is also the home of Tiananmen Square, the place, according
to large signs posted around its perimeter, that all China longs for.
(IND95)
The third alternative, all of, is not discussed by Quirk et al in terms of animacy.
Three interesting results could be obtained from the analysis (see Figure 7.7).
180
160
140
120
100
80
60
40
20
0
157
137
105
95
inanimate
63
animate
43
all
all of
the whole of
Figure 7.7. Animacy in NPs with all/whole, /THE DEFINITE ARTICLE/ and
GEOGRAPHICAL NAME
First, there was indeed a positive correlation between the whole of and inanimate
reference on the one hand and simple all and (inferred) animate reference on the
other, as in examples (7:33) and (7:34) above. The distinction was not absolute,
since the material comprised tokens both of the whole of used with animate reference, as in (7:35), and of simple all used with inanimate reference, as in (7:36).
(7:35) Meanwhile, the whole of Europe is praying loudly that your party loses
the next election. (IND95)
(7:36) This was when Dr. Weeks realized he had entered, as he said, ‘the largest
tomb in the Valley of the Kings and maybe the largest ever found in all
Egypt.’ (NYT95)
Second, even though this positive correlation strengthens the claim made by
Quirk et al that there is a difference between the variants, simple all was in fact
more frequent with inanimate reference than with animate (105 tokens of inani128
144
Note that in this variable, there can only be inferred animate reference.
mate vs. 95 tokens of animate reference). Third, all of stands somewhere between simple all and the whole of as regards frequency distribution. We will return to the concept of animacy in relation to syntactic function (7.5.5), since
these two factors seem to be related to some extent.
7.3.4.4
NPs with both and nouns for body parts or kinship
In the subgroup variable with both, there were no instances of inferred animacy,
since all nouns either denote body parts or people in the form of kinship expressions. Figure 7.8 illustrates the variation.
200
150
175
139
135
inanimate
animate
100
65
61
50
25
0
both
both my
both of my
Figure 7.8. Animacy in NPs with both, /POSSESSIVE DETERMINER/ and NOUN
FOR BODY PARTS OR KINSHIP
Here we find the same correlation as in the both (/of/the) variable: (i) between
simple both and inanimate reference (i.e. words for body parts), as in (7:37), and
(ii) between both (of) my and animate reference (i.e. words for kinship), as in
(7:38) and (7:39).
(7:37) I saw a man with a gun, holding it with both hands. (IND95)
(7:38) Both her daughters had dropped out of James Monroe High School in
the Bronx after hallway attacks, […] (NYT95)
(7:39) ‘They wanted the court to provide them with what their child most
needs, which is a legally recognized relationship with both of her
parents.’ Ms. Dohrn said […] (NYT95)
7.3.5
Natural vs. arbitrary time division
As described in Section 3.3.4.4, J. Hudson (1996:119) suggests that in NPs with
temporal nouns, there is a difference in the choice of variant according to
whether the noun belongs to a category of “natural” time division, or whether it
145
145
is arbitrary and only observable by means of a man-made measuring instrument.
Whereas the whole occurs with both kinds of temporal nouns (and thus is “unmarked”), simple all is mainly used with words from the natural categories129. J.
Hudson’s categorisation (Table 7.4, somewhat modified and extended) was used
as the basis for an analysis of the NPs with temporal nouns to check whether her
ideas would tally with my corpus material. Not all of the nouns in this table
occurred in the corpus samples, but all were represented in the total material.
Table 7.4. Natural and arbitrary words for time
Natural time division
afternoon
autumn
day
evening
fall
month
morning
night
season
spring
summer
winter
year
Arbitrary time
division
century
decade
hour
minute
second
term
week
weekend
J. Hudson’s argumentation in borderline cases is accounted for in Section
3.3.4.4. The difference between her analysis and mine is that she regarded week
as a natural category, since it is based on other natural categories (days and
nights). I disagree with this because with such a categorisation, century and
decade (based on years) would also have to be regarded as natural. The existence of words that denote ‘seven days’, ‘ten years’ and ‘hundred years’ (rather
than some other numbers of days and years) seems to be purely arbitrary.
The result is presented in Figure 7.9. There was indeed a correlation
(significant at the <0.05 level only), in line with J. Hudson’s suggestion, (i)
between natural temporal nouns and simple all, as in (7:40), and (ii) between
arbitrary nouns and the whole, as in (7:41).
(7:40) Staring at a screen all day has given me a headache. (IND95)
(7:41) ‘I wonder if anyone has had the experience of getting caught up in a
book for the whole weekend?’ he asked brightly, […]. (NYT95)
129
146
J. Hudson (1998:119) observes one exception, term, a man-made concept, which also occurs
with simple all.
146
200
180
160
140
120
100
80
60
40
20
0
176
162
natural
arbitrary
38
24
all
the whole
Figure 7.9. Natural and arbitrary time division in NPs with all/whole, /THE
DEFINITE ARTICLE/ and TEMPORAL NOUN
There were also cases of all used with words which were categorised as “arbitrary”. Of these tokens, the great majority include the word week, which J. Hudson classifies as belonging to the “natural” category130. Furthermore, a few instances of all weekend were found in the material.
As for the other words in this category (century, decade, term), there was one
single instance of all term 131 in the entire corpus material (from the spoken British corpus. There were no instances of NPs with all in combination with century
and decade. All other cases were used with the whole rather than all. This could
of course mean that they cannot be used with this quantifier, which strengthens
J. Hudson’s hypothesis about a relation between the natural–arbitrary division
and the use of quantifiers. There were very few instances of these words overall,
two of the whole century and one of the whole decade in the entire corpus material. A complementary search of the World Wide Web revealed that the whole is
more frequent than all with century and decade.
Perhaps the natural–arbitrary distinction is not the only important factor regarding the choice between all and the whole. We will see later on (Section
7.5.6.1) that the syntactic function of the NP is crucial to this choice (simple all
occurring almost exclusively in the adverbial function). Further, some of the
words belonging to the “arbitrary” category would be unnatural to use with
all/whole in adverbial function (second, minute, hour). So, the “syntactic function” factor could be the main reason for those words not being used with all
rather than their being examples of arbitrary time division.
130
Had Hudson’s categorisation of week as a natural category been used, there would have been a
slightly stronger correlation (significant at the <0.01 level) between “natural time division” and
all and between “arbitrary time division” and whole.
131
There were very few instances of term overall – only three occurrences of the whole term in
IND95 and none in NYT95.
147
147
7.3.6
Type of “head” in NPs with half, the indefinite article
and a singular noun or numeral
As described in Section 3.3.4.3, Berry (1997:70f) claims that there is a meaning
difference between the variants in the half + THE INDEFINITE ARTICLE + SINGULAR NOUN/NUMERAL variable. According to this idea, half a/an refers to a
part-whole relationship (half a bottle out of a whole bottle), whereas a half denotes an established unit (e.g. a half-bottle). This postulated difference in meaning was problematic, since the context was often not informative enough to
allow such conclusions based on the corpus material. Instead, I decided to semantically examine different types of NP “heads”132 to see whether these different NP heads would influence the choice of variant. Additionally, it was hoped
that such a study would reveal something relevant to Berry’s claim. The “heads”
were categorised in the following way:
(a)
(b)
(c)
(d)
nouns expressing time or space (e.g. hour, year, inch, mile)
nouns for partitive relations133 (e.g. bottle, cup, pint, pound)
numerals and nouns to do with figures (e.g. dozen, million, percent)
other nouns not regularly associated with measurement (e.g. game, share,
victory)
The main outcome of the analysis was that in the American material, there was
very little difference in distribution of variants between the different “head”
types. The differences in the British material were also quite small. Nevertheless,
they were at least a little more conspicuous than in AmE.
Let us now have a look at a correlation that was consistent across both corpora, although it was very weak in the American material and only significant at
the <0.05 level (see Figure 7.10). The correlation only involved two of the categories, so only these are included in the figure.
Half a/an correlated positively with a numeral or noun to do with figures, as
in (7:42), and a half with a partitive construction, as in (7:43).
(7:42) I saw about 30 horses and bought half a dozen, and one of them was
Antonin. (IND95)
(7:43) The menu does not vary much, either: hot chili and vegetables, a piece of
bread, an orange and a half-pint of milk. (NYT95)
132
What is referred to as the “head” in this context is the word that is quantified by half. From a
strictly syntactic point of view, this may be erroneous, since in an NP such as half a dozen boys,
the real head would be boys rather than dozen. However, the item of relevance to the present analysis is the quantified word. For practical reasons this factor was included in the umbrella section concerning various aspects of the (real) head of quantified NPs (also cf. Section 7.3.3).
133
In a few cases in the material, the partitive relation was understood rather than overtly expressed.
148
148
90
80
70
60
50
40
30
20
10
0
80
50
25
partitives
numerals
13
half a/an
a half
Figure 7.10. Type of “head” in NPs with half, THE INDEFINITE ARTICLE and
SINGULAR NOUN or NUMERAL
The correlation (i) between half a and numerals and (ii) between a half and partitives could possibly be related to Berry’s suggestion about the semantic difference between half a/an (part of whole) and a half (established unit). Interestingly, the differences between the variants were smaller in AmE than in the
British material. This may be explained by the fact that a half was much more
frequent in AmE (41%) than in BrE (6%) (see Section 6.1.4). With a half being
so frequent in AmE, people may be less inclined to use it for semantic distinction
(cf. Langacker on conventionalisation of variants in Section 2.1.2). If, however,
a variant is used sparingly, as in BrE, speakers may be more inclined to uphold
this distinction, instead reserving a half for cases of established units. On the
other hand the “established unit” concept probably applies mainly to certain
words, such as bottle, as in (7:44). This is the word Berry uses for illustration.
(7:44) He would buy a dozen beers, a half-bottle of whisky or gin and a few tonics […] (IND95)
It seems more unlikely that NPs like half an hour/a half hour or half a dozen /a
half dozen could be distinguished into a part-whole meaning and an establishedunit meaning.
7.4
The presence of certain elements in
the NP or its near co-text
7.4.1
Modifiers
This section deals with the syntactic complexity of the noun phrase in terms of
modification. Two simplifications were made as a starting-point for the classification (cf. Section 3.1.1.3). First, the distinction between modification and com-
149
149
plementation is problematic, so I decided that all elements in the noun phrases
that were neither heads nor determiners would be categorised as modifiers, regardless of their semantic function. Second, every token where two nouns occurred immediately after each other was categorised as a premodified noun,
leaving the fuzzy distinction between compounds and premodified nouns aside.
Apart from two nouns written together as one word, the only cases where two
juxtaposed nouns were regarded as a compound were (i) when there was a hyphen between the two nouns and (ii) in geographical names like the Channel
Islands, both of which were very rare in the material.
The analysis of modification was only performed on variables with common
nouns and the all/whole + temporal noun variable (cf. Sections 5.1.1 and 6.1.6.1)
since modification is not used in the other variables134. Four categories were
identified: NPs with (i) no modifier, (ii) premodifier(s), (iii) postmodifier(s) and
(iv) both pre- and postmodifier(s). I decided at a fairly early stage that no more
fine-grained division (into different types of pre- and postmodifiers) would be
made, since the number of tokens in the samples is only 200 per variant.
The premodifier category consisted of adjectives (e.g. all the right noises),
participles (e.g. all the required defences) and nouns (e.g. all the two-digit
numbers)135. As far as postmodification is concerned, the material comprised relative clauses (e.g. all the changes that have been made), non-finite clauses (e.g.
all the people depicted in it), prepositional phrases (e.g. all the changes in health
care) and also some adjectives (all the guests present) and adverbs (e.g. all the
losers tonight). There were also a number of tokens that contained both pre- and
postmodification. The charts in this section present the frequencies boiled down
to two categories: modification and no modification. Tables D6:1 to D6:4 in Appendix D supply the frequencies for the different modifier categories.
No significant correlations were spotted in NPs with half and plural nouns or
NPs with all/whole and temporal nouns. Let us now look at those variables
where correlations occurred, starting with singular count nouns (Figure 7.11).
There was a positive correlation between the two of-less variants and the “no
modifier” category, as in (7:45) and (7:46). A positive correlation also occurred
between all of/the whole of and the “modifier” category, as in (7:47) and (7:48).
(7:45) He lived here all his life […]. (NYT95)
(7:46) Genetic tests of risk will in fact render private insurance schemes
ineffective as a way of caring for the whole population. (IND95)
(7:47) The military has run the country of 45 million people for virtually all of
its modern history. (NYT95)
134
One exception is NPs with personal pronouns, where postmodification is quite common in the all
of variant. Since a floating quantifier is never used with postmodification, all tokens of postmodified NPs with all of us/them were excluded from the study.
135
A genitive sometimes has modifier rather than determiner function. A few tokens of such “classifying” or “descriptive” genitives (cf. Quirk et al 1985:327) occurred in the corpus material, e.g.
all of the children’s programs (NYT95) and all the womens’ aid centres (IND95). None, however, appeared in the samples analysed here.
150
150
(7:48) Tears streamed down Sampras’s face for practically the whole of the
final set, […] (IND95)
180
160
140
120
100
80
60
40
20
0
167
144
119
100100
no modifier
modifier
81
56
33
all
all of
whole
the whole
of
Figure 7.11. The presence of modifiers in NPs with all/whole, DETERMINER
and SINGULAR COUNT NOUN
Furthermore, the all, all of and the whole of variants were more often premodified than postmodified. Contrastingly, the whole variant was more frequently
postmodified, usually by means of an of phrase, as in (7:49).
(7:49) The whole world of our imagination breeds in childhood […] (NYT95)
Figure 7.12 shows the variation in the all/whole + MASS NOUN variable.
180
160
140
120
100
80
60
40
20
0
161
104
116
96
85
no modifier
modifier
38
all
all of
whole
Figure 7.12. The presence of modifiers in NPs with all/whole, DETERMINER
and MASS NOUN
151
151
In this variable whole correlated positively with the “no modifier” category, as in
(7:50), and all/all of correlated with the “modifier” category, as in (7:51) to
(7:52). Note that in NPs with singular nouns, the all variant behaved differently,
correlating with the “no modifier” category instead.
(7:50) He would have waited at Manger Square the whole time, […]. (NYT95)
(7:51) Despite all its artful stage management, this week’s Labour Party conference has been full of the unexpected. (IND95)
(7:52) In February 1994, he imposed a quarter point cut in base rates against all
of the advice from both Treasury and Bank officials. (IND95)
The correlation between whole and the “no modifier” category can, to a great extent, be explained by the fact that tokens of the whole time constitute a majority
of the cases, and most of these lack modification, as in (7:50) above, even
though some of them are postmodified, mainly by relative clauses, as in (7:53).
(7:53) The whole time I was there I never had a head coach who liked my style
of play. (NYT95)
As in the variable with singular count nouns, the head of an NP with whole was
more often postmodified than premodified. Like the the whole time tokens, other
postmodified NPs with whole were often postmodified by of phrases (7:54).
(7:54) The whole stability of the London concert scene is now under threat.
(IND95)
There was also a correlation in NPs with all and plural nouns (Figure 7.13).
150
124
105
100
95
76
no modifier
modifier
50
0
all
all of
Figure 7.13. The presence of modifiers in NPs with all, DETERMINER and
PLURAL NOUN
152
152
All correlated positively with the presence of a modifier, as in (7:55), while there
was a correlation between all of and the absence of a modifier, as in (7:56). The
all variant again behaved differently than in NPs with singular count nouns.
(7:55) But given all the warnings about air pollution, aren’t cyclists just
making things worse for themselves […] (IND95)
(7:56) All of her teams have gone to post-season tournaments […]. (NYT95)
In both variants, premodification was more frequent than postmodification.
The strongest correlation occurred in NPs with both (see Figure 7.14).
200
172
139
150
113
no modifier
87
100
61
50
modifier
28
0
both
both the
both of the
Figure 7.14. The presence of modifiers in NPs with both, /THE DEFINITE
ARTICLE/ and PLURAL NOUN
Simple both correlated with the absence of a modifier, as in (7:57), and both (of)
the with the presence of a modifier, as in (7:58) to (7:59).
(7:57) Both men are retired from British Gas and are not paid a salary. (IND95)
(7:58) Both the final teams will be named today […] (IND95)
(7:59) He said he had been close to both of the boys who were killed […]
(NYT95)
There was a particularly strong correlation between both of the and the presence
of a premodifier, as in (7:60).
(7:60) Both of the arrested teen-agers live within a mile of the crime scene […]
(NYT95)
It can be concluded that in the case of both /of the/ + PLURAL NOUN, the presence
of one or more modifiers is an important conditioning factor for the choice of
153
153
variant. The simple both form seems to have become almost a kind of “fixed
frame” with an open slot for unmodified (often fairly short) nouns. In contrast,
the two other variants (both the and both of the) are the preferred alternatives for
more elaborate variants with modification of various types.
7.4.2
An adjacent of
It was hypothesised that, for the avoidance of repetition, there is a stylistic barrier to using all of, whole of, both of and half of if there is an of in the near co-text
of the quantifier. This of either introduces a postmodifier, as in (7:61), or directly
precedes the NP, as in (7:62).
(7:61) The Walnut Creek CA-based building products company Fibreboard
Corp agreed on Jun 20, 1995 to sell nearly all the assets of its wood
products business […]. (NYT95)
(7:62) Under the program, officials said, the government will guarantee interest
rates of about half the current levels […]. (NYT95)
All variables except half + THE INDEFINITE ARTICLE + SINGULAR NOUN or
NUMERAL (where there is no of form) and all/whole + TEMPORAL NOUN (where
the of forms very infrequent) were included. In three variables, (i) NPs with personal pronouns, (ii) NPs with demonstrative pronouns as heads, and (iii) NPs
with geographical names, an adjacent of only occurs before the noun phrase.
Therefore, the only cases where the presence of an adjacent of can be studied are
when the NP has prepositional complement function, as in (7:62) above.
The analysis of the corpus material corroborated the hypothesis in most of the
variables. The both + plural noun variable showed no correlation, but on the
other hand, the tokens of both of the were very few. In the variable with half and
a plural noun, there was a (not statistically significant) correlation between the
of-less variant and the presence of an adjacent of in the American material. As
for all with personal pronouns (all of us – we all etc.), there were very few instances overall of variants where an adjacent of occurs. Strangely, all those five
tokens (one in NYT95 and four in IND95) were of the all of us form, as in
(7:63). This is contrary to the idea that of-less variants would be preferred when
there is an adjacent of in the near co-text of the quantifier.
(7:63) “I know it’s the way the legal system works, but the way we let this guy
carry on and make buffoons out of all of us,” she said. (NYT95)
This surprising correlation between all of and the presence of an adjacent of may
be explained by the syntactic function of the NP (Section 7.5.3). The analysis
showed a positive correlation between all of us and the prepositional complement category. In fact, all the five tokens of all of us had this function.
154
154
In the other variables, there was a statistically significant correlation between
the of-less variant and the presence of an adjacent of. Figures 7.15 to 7.17 illustrate the variation pattern in NPs with all/whole + SINGULAR COUNT NOUN,
all/whole + MASS NOUN and all + PLURAL NOUN respectively.
250
200
196
189
178
149
150
no of
an adjacent of
100
51
50
11
22
4
0
all
all of
whole
the whole
of
Figure 7.15. The presence of an adjacent of in NPs with all/whole,
DETERMINER and SINGULAR COUNT NOUN
250
200
195
183
171
150
no of
100
an adjacent of
50
29
5
18
0
all
all of
whole
Figure 7.16. The presence of an adjacent of in NPs with all/whole,
DETERMINER and MASS NOUN
155
153
155
200
188
164
150
no of
an adjacent of
100
36
50
12
0
all
all of
Figure 7.17. The presence of an adjacent of in NPs with all, DETERMINER
and PLURAL NOUN
The correlation between the presence of an adjacent of and the of-less variant
was particularly clear in NPs with demonstrative pronouns as heads. Only two
instances of all of in combination with an adjacent of occurred (see Figure 7.18).
250
200
198
183
150
no of
an adjacent of
100
50
17
2
0
all
all of
Figure 7.18. The presence of an adjacent of in NPs with all and
DEMONSTRATIVE PRONOUN
As regards NPs with geographical names as heads, it is difficult to decide
whether there was a positive correlation between the two variants with of (all of
and the whole of) and the absence of an adjacent of (see Figure 7.19). In the
British corpus, there was a weak correlation between all of and the whole of and
the absence of an adjacent of. In the American material only the correlation
between all of and the “no of ” category was confirmed, not the one between the
whole of and the “no-of ” category. On the other hand, the number of tokens of
the whole of in this corpus sample is very low – only 21 in all (see Table D7:5) –
so normalised frequencies may give a less reliable picture of the relations.
156
156
250
193
179
200
186
150
no of
an adjacent of
100
50
21
14
7
0
all
all of
the whole of
Figure 7.19. The presence of an adjacent of in NPs with all/whole, /THE
DEFINITE ARTICLE/ and GEOGRAPHICAL NAME
Finally, there was a correlation (significant at the <0.05 level only) in the variable with both and nouns for body parts or kinship (Figure 7.20).
250
200
190
200
187
150
no of
an adjacent of
100
50
10
13
0
0
both
both my
both of my
Figure 7.20. The presence of an adjacent of in NPs with both, / POSSESSIVE
DETERMINER/ and NOUN FOR BODY PARTS OR KINSHIP
The two forms lacking of, simple both and both my, correlated with the presence
of an adjacent of. In contrast, both of my correlated with the absence of an adjacent of in the American material. There were no examples whatsoever of an adjacent of in NPs with the both of my variant (but the British corpus contained only
one single token of the both of my variant).
7.4.3
Focus markers
In Section 3.3.3.2 the possibility of there being (at least) two
was discussed: one where totality is in focus (corresponding
primitive ALL, see Wierzbicka 1996:47), and another where all
large quantity. Certain types of elements in the NP or its near
meanings of all
to the semantic
rather expresses
co-text, e.g. ap-
157
157
proximators and other quantifiers136, were expected to put focus on the totality
meaning137. Accordingly, since it was hypothesised that all of would be more
frequent with totality meaning, it could be expected that the all of variant would
correlate with the presence of a focus marker. Most of the focus markers belonged to one of four groups:
(a)
(b)
(c)
(d)
quantifiers: many, cardinal numbers etc.
approximators138: about, almost, nearly, substantially, virtually etc.
exceptions: except X, bar X etc.
negators (where the quantifier is clearly within its scope139): not, never
etc.
There were also some other elements which were considered to function as focus
markers: literally, fully, throughout etc.
Focus markers are also used in NPs with half. In addition to approximators,
quantifiers and negators, a few more types occur in these NPs, viz. at least, close
to, fully, (no) less than, (no) more than, not even and only. In the case of half, it
is not totality that is in focus, but rather the measuring function of half.
The variables included are (i) NPs with all/whole and common nouns (ii) NPs
with all and demonstrative pronoun heads (iii) NPs with half and plural nouns
and (iv) NPs with half and the indefinite article + a numeral or singular noun.
7.4.3.1
NPs with all
Let us now look at the results as far as NPs with all are concerned. Approximators (such as almost and nearly) and quantifiers (in the form of numerals)
were the most frequent types. Indeed, my hypothesis was confirmed in virtually
all of the variables investigated (all + common noun, demonstrative pronoun and
personal pronoun). Figure 7.21 shows the correlation in NPs with all/whole and
singular count nouns.
Interestingly, in addition to the fairly strong correlation between all of and the
presence of a focus marker, as in (7:64), there was also a less salient correlation
between the whole of and a focus marker, as in (7:65). This could indicate that
whole also has both a totality and a large-quantity meaning, as suggested in
passing in Section 3.3.3.2. Alternatively, it is just an indication that there is more
focus on the totality meaning with of constructions
136
Quirk et al (1985:258) claim that an of variant is preferred (esp. in AmE) in an [all +
DETERMINER + NOUN] variant if the noun phrase includes another quantifier, such as many: all of
the many children. This claim, however, is not discussed in terms of totality vs. large-quantity
meaning.
137
Other elements can emphasise large-quantity meaning instead, such as for in For all the
headlines about cuts in the high street banks, jobs in the financial sector have gone up by
200,000 […] (IND95), which could be paraphrased as ‘…in spite of the many cuts…’ .
138
As mentioned in Section 3.1.1, de Haan (1989: 32) notes that in grammatical descriptions, approximators and other elements preceding the determiner(s) in an NP have often been considered
not to belong to the noun phrase, but rather to be modifiers of the whole clause.
139
The following sentence is an example of a token that was not regarded as including a focus
marker, since the quantified NP is not within the scope of the quantifier: She had never all her
life cared very much for women (IND95).
158
158
250
198
184
200
171
150
no focus marker
116
84
100
50
focus marker
29
16
2
0
all the
all of the
the whole
the whole
of the
Figure 7.21. The presence of a focus marker in NPs with all/whole,
DETERMINER and SINGULAR COUNT NOUN .
(7:64) He has spent virtually all of his professional career within the City
University system, serving as president of Kingsborough since 1971.
(NYT95)
(7:65) The guitarist spends almost the whole of this period standing on one leg,
his face screwed up, […] (IND95)
In NPs with mass nouns, there was also a fairly clear correlation between all of
and the presence of a focus marker (see Figure 7.22).
250
200
199
194
163
150
no focus marker
focus marker
100
37
50
6
1
0
all
all of
whole
Figure 7.22. The presence of a focus marker in NPs with all/whole,
DETERMINER and MASS NOUN
Here is an example of the correlation:
159
157
159
(7:66) He works between 20 and 25 days per month; not all of his work is as
lucrative as recording: […] (NYT95)
The correlation with the presence of a focus marker recurs with all of in combination with plural nouns (Figure 7.23), as in (7:67).
(7:67) Today all of the family except Franz share one thing in common: they
now vote for the Greens […] (IND95)
184
200
166
150
No focus marker
100
Focus marker
34
50
16
0
all
all of
Figure 7.23. The presence of focus markers in NPs with all, DETERMINER
and PLURAL NOUN
Moving to NPs with personal pronouns, we find a correlation between the all of
us variant and the presence of a focus marker, as in (7:68) (see Figure 7.24).
(7:68) Virtually all of them are built with imported lumber, because Japan’s
trees are smaller and younger than those found elsewhere. (NYT95)
250
198
178
200
150
no focus marker
100
focus marker
50
22
2
0
us all
all of us
Figure 7.24. The presence of focus markers in NPs with all and PERSONAL
PRONOUN
160
160
There are two possible explanations for this correlation. The first one (along the
lines discussed in Section 3.3.3.4), is that there might in fact be two meanings of
all in combination with personal pronouns: one focusing on totality, and the
other just expressing large quantity. This could also be related to the possibility
that the we/us (...) all variant is more frequent in evaluative statements, as in
Men are all the same. The second explanation is a stylistic one: the use of a
focus marker in combination with the we…all variant could sometimes result in
a clumsy or ambiguous variant (e.g. ?nearly we all). There was only one single
token in the 200-token sample of a focus marker (a negator) combined with the
we/us (...) all variant, as in (7:69).
(7:69) Will we not all end up, like the inmates, with no rights but plenty of
drugs? (NYT95)
As for the NPs with a demonstrative pronoun as head, only a few tokens that
included a focus marker. Most of these were of the all of type, in line with the
findings above. There were also few instances of focus markers in the variable
with all/whole + GEOGRAPHICAL NAME. Here, no correlations were found.
7.4.3.2
NPs with half
Focus markers are also used in NPs with half, both in (i) NPs with a definite determiner and a common noun (as in nearly half the children) and in (ii) NPs with
the indefinite article and a singular noun or numeral (as in almost half a mile).
Just as in the case of all, it was hypothesised that the presence of a focus marker
in noun phrases with half would be correlated to the form with of. The focus
marker actually occurred far more often in noun phrases with half than in noun
phrases with all.
The most frequently used focus markers in NPs with half + PLURAL NOUN
were approximators, quantifiers in the form of numerals and comparing phrases
such as more than and less than. The correlations are presented in Figure 7.25.
200
155
150
100
128
72
45
50
no focus marker
focus marker
0
half
half of
Figure 7.25. The presence of a focus marker in NPs with half, DETERMINER
and PLURAL NOUN
161
161
In accordance with the expectations, there was a positive correlation between the
of variant and the presence of a focus marker, as in (7:70). In this case, however,
focus markers frequently occurred in the of-less tokens, as in (7:71), as well.
(7:70) Fewer than half of the duffers are white-collar professionals, it says, with
40 percent blue-collar or clerical workers. (NYT95)
(7:71) While some companies such as North East Gas and Severn Trent gave
callers a full answer every time, North West Water’s employees proved
unable to come up with a proper answer in nearly half the cases.
(IND95)
There seems to be more focus on the measuring function of half if it is followed
by of, as in half of the duffers. The focus marker modifies the quantifier (fewer
than half), rather than the whole noun phrase (fewer than half of the duffers).
This may become clearer if the phrase is “divided” by means of an of (also cf.
Section 7.3.4).
Figure 7.26 illustrates the variation in the half + THE INDEFINITE ARTICLE +
SINGULAR NOUN or NUMERAL variable. Here I did not have a pre-existing idea
about which of the variants might correlate with the presence of a focus marker.
In the British material, tokens including a half century used as a more or less
fixed cricket term140 were excluded, since the distribution between different categories was more similar to the distribution in the American material then. The
correlation occurred whether the tokens were excluded or not, however. Here the
focus markers mainly consisted of approximators and comparisons.
There was a positive correlation between the presence of a focus marker and
the half a/an variant, as in (7:72).
(7:72) After years of being a world class performer, Edwards is now out on his
own, with a best this year of 18.43m, which is more than half a metre
further than any of his nearest rivals have managed. (IND95)
The correlation may have something to do with the postulated semantic difference between the two variants (see Section 7.3.6). If we reason in terms of what
the focus marker modifies, as in NPs with plural nouns, it may be more natural
for focus markers to be used in partitive relationships than with established units.
That is, in the half a/an variant, a focus marker clearly modifies half (nearly half
a bottle), whereas in the a half variant, the modification seems to concern the
whole NP (nearly a half-bottle).
140
162
A half century should perhaps have been excluded from the whole study as a fixed expression,
since no instances of half a century with this meaning occurred in the corpus material. However,
in dictionaries I only found the term century (not combined with half). A search of the World
Wide Web showed that both half a century and a half century are used in this context. This suggests that a half century is not an entirely fixed phrase even though the corpus material pointed
in that direction.
162
200
150
177
140
no focus marker
focus marker
100
60
50
23
0
half a/an
a half
Figure 7.26. The presence of a focus marker in NPs with half, THE
INDEFINITE ARTICLE and SINGULAR NOUN or NUMERAL
7.5
Syntactic function of the NP
This final section deals with a linguistic factor which involves the NP in its relation to the whole clause in which it occurs. The noun phrase is typically used in
one of the clausal functions of subject, object or predicative, but it can also be
used in some functions as part of other clausal elements. The following list provides examples of noun phrases with all/whole, both and half in the different
functions described by Huddleston & Pullum (2002:327), previously accounted
for in Section 3.1.2141.
•
•
•
•
•
•
subject: Both prices exclude tax. (IND95)
object: I’ve not expended all of my energy in topical poems […].
(NYT95)
predicative: I mean, that’s the whole basis of our justice system.
(NYT95)
complement in a prepositional phrase142: And that is true for all of us as
well. (NYT95)
subject-determiner in a noun phrase: […] his whole life’s work.
(IND95)
adjunct in a clause: Was the course closed all winter […]. (NYT95)
141
Huddleston & Pullum (ibid, see Section 2.1.2) account for two more syntactic functions that
noun phrases can take on: NP modifiers (as in the opera Carmen) and vocatives. These two did
not occur in the material, but one could imagine vocatives such as All of you, come here! and perhaps even All (of) the boys, come here!
142
Observe that some words that are usually used as prepositions (e.g. with) can also function as
subordinating conjunctions. Consequently, in a phrase such as with both cheeks burning, the
quantified noun phrase is not a prepositional complement but the subject of a non-finite clause
(Quirk et al 1985:1003).
163
163
•
•
•
•
modifier in an adjective phrase: […] [more than half a percentage point
higher] […]. (IND95)
modifier in an adverb phrase: [A half mile later], they were to stand at
the shore […]. (NYT95)
modifier in a prepositional phrase: [Half an inch to the right] and the
ball would have hit the flag […] (IND95)
supplement: Opic does business with 144 countries, all of them – […]
(NYT95)
Since there are so many different categories, bar charts would not be useful and
will therefore not be used in this section. Instead tables will be given in the text.
None of the variables appeared in all the syntactic functions, apart from the variable with half, the indefinite article and a singular noun or numeral. This will become evident from the tables (7.5–7.13) below.
Before going into the presentation of findings, I offer a number of clarifications. The “subject” category comprises subjects in both active and passive
clauses, and in both finite and non-finite clauses. Similarly, direct and indirect
objects (the latter very infrequent) were grouped together in the statistics. What
Huddleston & Pullum (ibid) refer to as “complement in a prepositional phrase”
will, for simplicity’s sake, be called “prepositional complement”.
Huddleston & Pullum (ibid) specify the NP-determiner function as “subjectdeterminer in NP”, explaining it in terms of the relation between the genitive NP
and the following NP, which bears “significant resemblance to that between a
subject NP and the verb in clause structure” (as in Sue’s analysis of the passive
construction Æ Sue analysed the passive construction). This relation does not
apply to all the quantified NPs occurring in the present study, so the function
will just be referred to as “NP determiner”. The function is mainly used with
NPs with half, the indefinite article and a singular noun/numeral (see 7.5.4).
There were also some instances in the all/whole + SINGULAR COUNT NOUN variable (but only one in the 200-token sample). In the all + PLURAL NOUN variable
there can sometimes be ambiguity between an interpretation as an NP determiner
and another syntactic function, as in […] all our friends’ children attend it […]
(IND95). This noun phrase could be interpreted either as ‘all the children of our
friends’ (NP determiner) or ‘the children of all our friends’ (subject function).
No such cases occurred in the sample, however. In the both + PLURAL NOUN variable, the majority of NPs with a genitive construction were used in the determiner function, as in both bands’ success (IND95).
Henceforth, the term “adverbial” will be used instead of “adjunct”, which
Huddleston & Pullum (ibid) prefer, since the latter (to a greater extent than the
former) has been used with different meanings in different linguistic theories. Interestingly, Huddleston & Pullum do not include the adverbial function in their
account of clausal functions that noun phrases can take on, but defer it to their
list of “other functions”, such as determiners and various types of modifiers. The
reason might be that noun phrases are not so frequent in this function, but it still
seems somewhat strange to group it with complements, modifiers and determiners, none of which can stand on their own as elements of a clause.
What Huddleston & Pullum (ibid) refer to as “supplement” is usually called
“apposition” or “appositive noun phrase” in the linguistic literature (cf. e.g.
164
164
Quirk et al 1985:1300ff; Biber et al 1999: 638ff). Henceforth, a new “fragment”
category will be used to comprise appositions, independent prepositional phrases
and short answers. An example of a fragment is given in (7:73).
(7:73) Which of the following office products is ergonomically correct? […]
Answer: all of the above – if you believe what the manufacturers say.
(NYT95)
7.5.1
NPs with common nouns
The first table (7.5) shows in which different syntactic functions NPs with all/
whole and a singular count noun occurred in the material.
Table 7.5. Syntactic functions of NPs with all/whole, DETERMINER and
SINGULAR COUNT NOUN
all
NYT95
all of whole
Subject143
Object
Predicative
Prep compl.
NP determ.
Adverbial
Fragment
Total
10
17
2
16
0
54
1
100
18
39
0
30
0
13
0
100
45
21
3
31
0
0
0
100
the
whole
of
30
30
0
40
0
0
0
100
Total pop.
458
152
3081
14
all
IND95
all of whole
15
22
0
18
0
40
5
100
22
49
2
14
0
10
2
100
36
21
0
40
1
0
2
100
the
whole
of
22
34
1
43
0
0
0
100
403
49
3510
210
There was a clear correlation between the all variant and the adverbial function,
as in (7:74).
(7:74) He has voted Tory all his life but thinks he will support Tony Blair’s
New Labour next time. (IND95)
The great majority of these tokens included the noun life, as in (7:74), whereas
the rest of the NPs included a temporal noun and a demonstrative determiner, as
in (7:75).
(7:75) All this week the Daily Poem will feature the theme of insects. (IND95)
The other variants (all of, whole and the whole of) were never or seldom used as
adverbials. Instead, they occurred almost exclusively in the subject, object and
prepositional complement functions, while predicatives and fragments were rare
(and so they were in the other variables). There was a correlation (i) between the
143
The order of categories is the same as in Huddleston & Pullum’s (1985:327) presentation.
165
165
whole and the subject category, as in (7:76), and (ii) between all of and the object function, as in (7:77).
(7:76) The whole sales organization picks very carefully what we choose to
make. (NYT95)
(7:77) The Rams are getting all the revenue, and taxpayers are picking up all of
the cost. (NYT95)
To conclude, in this variable, the variation (especially that between all and the
other variants) can to a fairly great extent be explained by the syntactic function
of the noun phrase.
Moving to all/whole+ MASS NOUN, the picture is quite complicated, because
of the many categories and several correlations, as illustrated in Table 7.6.
Table 7.6. Syntactic function of NPs with all/whole, DETERMINER and MASS
NOUN
Subject
Object
Predicative
Prepositional complement
Adverbial
Fragment
Total
all
20
26
2
44
8
0
100
Total population
5856
NYT95
all of whole
10
30
15
28
1
3
34
1
2
69
5
1
100
100
515
73
all
12
31
3
40
13
1
100
5242
IND95
all of whole
9
24
20
37
3
0
18
10
16
61
3
1
100
100
79
92
The predominant phrase in the whole variant, the whole time, is mostly used as
an adverbial, and there was thus a strong positive correlation between whole and
this category, as in (7:78).
(7:78) It took 10 years and an Israeli Supreme Court order lifting the veils of
secrecy for the public to learn that Professor Klingberg had been in
Ashkelon Prison the whole time as a convicted Soviet spy. (NYT95)
The other variants, all and all of, were not very frequent in the adverbial function, but were mainly used in the subject, object and prepositional complement
functions. All of correlated with the subject function, as in (7:79).
(7:79) All of this food is treated in essentially the same way. (IND95)
There was further a positive correlation between both the all and the all of variants and the object function, as in (7:80) and (7:81).
(7:80) The 61-year-old manager knows all the jargon and uses it. (NYT95)
166
166
(7:81) We do all of the finish work by hand. (NYT95)
Finally, the all variant correlated with the prepositional complement function, as
in (7:82)
(7:82) After all the magnificence, it was interesting to discover what the
children had absorbed. (IND95)
As for all + PLURAL NOUN, the correlation pattern was a little less complex than
in the variables accounted for above (see Table 7.7).
Table 7.7. Syntactic functions of NPs with all, DETERMINER and PLURAL
NOUN
all
Subject
Object
Predicative
Prepositional complement
Adverbial
Fragment
Total
Total population
NYT95
all of
28
43
35
35
1
0
22
32
0
0
4
0
100
100
12732
2254
IND95
all
all of
26
25
0
47
2
0
100
43
29
0
27
0
1
100
10789
384
Just as in NPs with mass nouns, there was a positive correlation between all of
and the subject category, as in (7:83). A positive correlation was also found
between the prepositional complement category and all, as in (7:84) (see Table
7.7).
(7:83) Almost all of the main courses come with heaping side dishes like earthy
rice and red beans […] (NYT95)
(7:84) This monster, compared with all the other quarks, is like a big cowbird’s
egg in a nest of little sparrow eggs. (NYT95)
Now, could the correlation pattern (all – prepositional complement function and
all of – subject function) observed in both NPs with mass noun and NPs with
plural nouns be explained? There are at least two possible explanations where
this factor is related to other linguistic factors. One explanation of all being favoured (as opposed to all of) in the prepositional complement category is that in
about one third of the tokens in the variable with plural nouns the preposition
preceding the quantified noun phrase is of 144, as in (7:85)145.
144
There were not so many instances of a preceding of among the mass nouns in prepositional
complement function: only 12 out of 83.
145
The Oxford Advanced Learner’s Dictionary (2000) brings up Of all the… as an informal phrase
expressing anger, but there does not seem to be any such negatively evaluative connotation in the
tokens occurring in the present material.
167
167
(7:85) Of all the ‘high islands’ of Polynesia, Rarotonga is the most gentle and
serene, and also the least well-known. (IND95)
The presence of an adjacent of may (see Section 7.4.2) affect the choice between
a quantifier with of and one without of in the direction of the latter. Another, perhaps even more plausible possibility, which applies to both NPs with plural and
mass nouns, is that the correlation can be explained in terms of totality vs. largequantity meaning (see Sections 3.3.3.2 and 7.4.3). When analysed in more detail,
it emerged that the great majority of the tokens of NPs with of-less all used as
prepositional complements seem to have large-quantity meaning, as in (7:86)
and (7:87).
(7:86) But the chance to enjoy a pint at home without all the hassle of actually
having to go out and buy it is not a new idea. (IND95)
(7:87) That means that for all the turnovers by the defence, the offense is
negating them almost one for one. (NYT95)
One fact strengthening the analysis is that none of these tokens had a focus
marker. It could also be the case that the correlation between the all of form and
subject function is related to the same meaning distinction, since it seems that a
large number of the tokens occurring in the subject function had totality meaning, as in (7:88) and (7:89). The analysis is strengthened by the fact that there
were fairly many cases of focus markers (underlined) being used in NPs with
subject function.
(7:88) But not all of the evidence pointed in the same direction. (NYT95)
(7:89) Virtually all of the old rocks come from lakes, caves or riverbeds, […]
(IND95)
There was no statistically significant correlation in the half + PLURAL NOUN variable. In the both + PLURAL NOUN variable there was a statistically significant
correlation (i) between both and the subject function and (ii) between both
the/both of the and the object function in the British corpus. There was, however,
no such correlation in the American material.
7.5.2
NPs with demonstrative pronouns
Table 7.8 shows the five syntactic functions which NPs with demonstrative
pronouns assumed in the corpus samples.
168
168
Table 7.8. Syntactic function of NPs with all and DEMONSTRATIVE PRONOUN
NYT95
IND95
all
all of
all
all of
Subject
39
37
54
61
Object
23
18
15
12
Predicative
1
1
4
1
Prepositional complement
23
20
31
38
Fragment
6
4
6
6
100
100
100
100
Total
Total population
1678
478
2292
260
The correlation pattern found with all/whole + MASS NOUN and all + PLURAL
NOUN recurs in this variable. There was thus a positive correlation in both corpora (i) between the all of variant and the subject function, as in (7:90), and (ii)
between the all variant and the prepositional complement function, as in (7:91).
The latter was only significant at the <0.05 level.
(7:90) So all of these were a success. (IND95)
(7:91) On top of all that, the heat wave hit. (NYT95)
Probably, the same explanation (totality vs. large-quantity meaning) could be applied here as well. Furthermore, as regards prepositional complements, the all of
variant is perhaps avoided when the preposition preceding the quantified noun
phrase is of. In about half of the cases of the prepositional complement category,
of was the preceding preposition. Another finding was that the fragment category, although rare in this variable as well, contained more tokens (in both variants) than in the other variables. Here are two examples:
(7:92) It is a relief to turn off the video-machine, to order room service at the
Hotel St George, […], all this during Ramadan. (IND95)
(7:93) Something is touched or awakened or threatened – all of that – by its
very intensity. (NYT95)
7.5.3
NPs with personal pronouns
Quantified NPs with personal pronouns were only used in those three different
syntactic functions which were the most frequent ones in the majority of the
other variables as well: i.e. subject, object and prepositional complement functions (see Table 7.9).
169
169
Table 7.9. Syntactic function of NPs with all and PERSONAL PRONOUN
NYT95
IND95
we all
all of us
we all
all of us
Subject
53
38
74
72
Object
21
22
18
19
Prepositional complement
5
10
25
43
100
100
100
100
Total
Total population
4078
863
5022
520
There was a positive correlation (i) between the we/us (...) all variant and the
subject function, as in (7:94), and (ii) between the all of variant and the prepositional complement function, as in (7:95).
(7:94) We all know Mike Fratello is a good coach, making the best of a
decimated squad. (NYT95)
(7:95) Death is eventually, if partially, outfaced in all of them save for that
involving the young boys where Macdonald does not manage to avoid a
certain sentimentality. (IND95)
The correlation between the subject category and the we…all variant can perhaps
be explained by the all of variant including the object form of the personal
pronoun (us and them). Some people might be disinclined to use this variant in
the subject function and instead prefer the we…all variant. The correlation
between the prepositional complement category and the all of variant is somewhat more surprising: it is contrary to the results of three other variables (all +
MASS NOUN, all + PLURAL NOUN, all + DEMONSTRATIVE PRONOUNS), where the
prepositional complement category correlated with the of-less variant. However,
in those other variables, the preposition preceding the NP was in many cases of
(which probably led some speakers to avoid a quantifier variant with of). In
contrast, in the group with personal pronouns, of was quite infrequent as the
preposition introducing the prepositional phrase.
Another thing worth mentioning is that there were a fairly large number of instances of sentence fragments, such as appositions and short answers, in the
corpus material. This is a case of a knock-out effect, since the we…all variant is
never used in these fragments. Accordingly all the tokens of the all of type, as in
(7:96), were excluded from the study (see Appendix B).
(7:96) All New York vitamin stores have permanent discounts of around 20 per
cent – all of them. (IND95) (* … of around 20 per cent – they all)
7.5.4
NPs with half, the indefinite article and a singular
noun or numeral
This variable is the only one which appears in all the different syntactic functions: subject, object, predicative, prepositional complement, modifier in an ad-
170
170
jective or adverb phrase, modifier in a prepositional phrase, NP determiner and
apposition (which is part of the fragment category). The term “NP determiner”
calls for some extra clarification in this variable. It was used for the three different cases exemplified in (7:97) to (7:99), where the quantified NP determines the
head of the matrix NP (indicated by square brackets).
(7:97) He sprayed at least [half a dozen shots] at two electrical workers
standing on the sidewalk […] (NYT95)
(7:98) The Years of Challenge (1897-1904) is the fruit of [half a century’s
fascination] with the revolutionary Austrian. (IND95)
(7:99) Its enterprising policy has enabled [half a dozen of its tournament
winners] to make the trip to Vegas this year. (IND95)
A possible alternative syntactic analysis of half a dozen in (7:99) is as the head
of the matrix NP with of its tournament winners functioning as a postmodifying
prepositional phrase. An argument for such an analysis is that half a dozen can
stand on its own: …has enabled half a dozen to make the trip ... This analysis
was discarded, however, since of here is clearly partitive rather than having a
conventional prepositional function (cf. Sections 3.3.4.1 and 7.3.4).
Three of the categories in the syntactic-function factor were unique to this
variable in the sample material146: the adjective phrase modifier, as in (7:100),
the adverb phrase modifier, as in (7:101), and the prepositional phrase modifier,
as in (7:102).
(7:100) […] a reasonable steak, to be properly cooked, should be at least a halfinch thick and therefore weigh at least 6oz […] (IND95)
(7:101) That was nearly a half-century ago. (NYT95)
(7:102) Half a century after WWII, Japan’s Parliament is considering whether
to apologize for killing millions of people during the war. (NYT95)
As for the results, there were few positive correlations that were consistent
across the two corpora (see Table 7.10). The NPs functioned most frequently as
prepositional complements and NP determiners.
146
The last two of these, adverb phrase modifier and prepositional phrase modifier, both occurred
with the phrase all the way in the all/whole + SINGULAR COUNT NOUN variable. These tokens
were, as we have seen, not included in the samples for the analysis of linguistic factors.
171
171
Table 7.10. Syntactic function of NPs with half, THE INDEFINITE ARTICLE
and singular NOUN or NUMERAL
NYT95
IND95
half a/an
a half
half a/an
a half 147
Subject
0
2
1
3
(4)
Object
10
5
12
(20)
32
Predicative
1
1
2
1
(1)
Prepositional complement
13
20
(18)
29
30
NP determiner
35
35
36
(49)
50
Adverbial
4
2
0
0
(0)
AdjP/AdvP modifier
11
16
6
3
(4)
PP modifier
7
9
6
2
(1)
Fragment
3
2
8
3
(4)
100
100
100 100 (100)
Total
Total population
1841
1254
2136
127
(84)
There was one correlation between the half a/an variant and the prepositional
complement category, as in (7:103).
(7:103) Garnish each with half a lobster tail and 1 claw and serve immediately.
(NYT95)
In the British material, there was a correlation between a half and the object
category. The main reason for this correlation was the great frequency of a half
century used as a cricket term, as in (7:104), in combination with verbs such as
complete, make, reach and score.
(7:104) Lest we get too carried away, Gough yesterday became the first
Englishman to score a half century […] since Chris Lewis against the
West Indies at Edgbaston in 1991. (IND95).
In the American material, there was a positive correlation between a half and the
noun phrase determiner category, as in (7:105). With all the tokens of a half century excluded, this correlation was confirmed in the British material.
(7:105) A half dozen sites were studied with such methods in the 1970’s and
1980’s, with tantalizing results. (NYT95)
It is possible that the correlation (i) between half a/an and the prepositional complement and (ii) between a half and the NP determiner function could be related
to the “part-whole” vs. “established-unit” distinction discussed in Sections 7.3.6
and 7.4.3.
147
172
The figures in brackets indicate number of tokens when a half century used as a cricket term was
excluded (see Section 7.3.6).
172
7.5.5
NPs with geographical names
Table 7.11 illustrates the syntactic functions in which quantified NPs with a geographical name were used in the corpus samples.
Table 7.11. Syntactic function of NPs with all/whole, /THE DEFINITE
ARTICLE/ and GEOGRAPHICAL NAME
NYT95
IND95
all
all of
the
all
all of
the
whole
whole
of
of
Subject
23
29
36
20
38
42
Object
10
13
24
24
26
28
Predicative
0
1
0
0
2
1
Prepositional complement
44
48
41
44
37
50
Fragment
8
2
6
0
2
1
100
100
100
100
100
100
Total
Total population
55
258
21
45
59
148
There was a positive correlation (i) between the simple all variant and the subject function, as in (7:106), and (ii) between all of and the whole of and the object category, as in (7:107) and (7:108).
(7:106) And the broader question that all America will be pondering is whether
the march will serve merely as a springboard for Mr Farrakhan’s
political career, […]. (IND95)
(7:107) We understand that they probably intend to remain for a few weeks, but
they must eventually leave all of northern Iraq. (NYT95)
(7:108) Vladimir Zhirinovsky, himself born in Alma-Ata, wants the whole of
Kazakhstan back. (IND95)
As described in Section 3.3.4.4 and evidenced in Section 7.3.4.3, the simple all
variant is often preferred in cases of (inferred) animacy, where the speaker refers
to the inhabitants of a country, town etc. rather than to the country, town etc. as a
geographically or politically defined area. This fact may explain the correlation
regarding the present linguistic factor, since animate NP heads tend to be more
active than inanimate ones. They, therefore, more often function as the subject of
clauses (or Agents in a two-participant clause, as, for instance, Hopper &
Thompson (1980:252) put it), whereas inanimate NP heads more typically function as objects. Also supporting the idea that simple all is more “personified”
than the whole of is the fact that when the members of the object category are
sub-classified into direct and indirect objects, indirect objects proved only to
occur with the simple all variant. Again, there seems to be a case of two interrelated factors.
173
173
7.5.6
Subgroups
7.5.6.1
NPs with all/whole and temporal nouns
The last variable but one – NPs with temporal nouns – is the one where the most
conspicuous correlation between variant and category was found. There proved
to be a clear difference in syntactic function between variants with all and
variants with whole. This piece of information is notably absent in the school
and reference grammars consulted. Table 7.12 shows this difference.
Table 7.12. Syntactic functions of NPs with all/whole, /THE DEFINITE
ARTICLE/ and TEMPORAL NOUN
NYT95
IND95
all
the whole
all
the whole
Subject
0
0
10
14
Object
2
4
16
35
Prepositional complement
0
0
33
40
Adverbial
41
11
98
96
100
100
100
100
Total
Total population
1812
103
1276
110
NPs with simple all were almost exclusively used in adverbial function (98 and
96 tokens respectively), as in (7:109):
(7:109) Monica seemed relaxed all week but it wasn’t like before. (IND95)
Note that in the British material, the correlation was not far from absolute: 96
tokens of all in adverbial function and 4 tokens in other functions vs. 11 tokens
of whole in adverbial function and 89 tokens in other functions. Overall, there
were only six examples in the 200-token sample where the simple all variant was
used in another function, viz. as a direct object combined with spend or take, as
in (7:110). In these cases, the whole spend all day is very similar in meaning to a
phrase like be somewhere/do something all day, where all day has adverbial
function.
(7:110) It’s like living next door to a gun nut who spends all day and half the
night shooting at beer bottles. (NYT95)
The other frequent variant used in NPs with temporal nouns was the whole.
Simple all was mainly used as an adverbial, while the whole correlated with
three other categories: subject, as in (7:111), object as in (7:112), and prepositional complement, as in (7:113). In the American material, however, the adverbial category predominated also with the whole, as in (7:114).
(7:111) It was a fantastic feeling but the whole week was unbelievable,
something I’ll never forget. (IND95)
174
(7:112) Couples in new relationships tend to streamline their plans by spending
the whole weekend in bed. (NYT95)
(7:113) But the best bargains of the whole year are just before Christmas, […].
(IND95)
(7:114) It stayed open the whole night. (NYT95)
The correlation with syntactic function may be related to the discussion in 7.3.2,
where it was suggested that whole can be used for specificity and individuation
of singular count noun referents, while all sometimes has a less specifying (more
generic) function. It is not implausible that this applies to NPs with temporal
nouns as well. Therefore, it could be part of the explanation as to why whole is
more often used in subject, object and prepositional function, whilst all is more
often used in adverbial function.
Quirk et al (1985:259) write that the simple all form is the only possible variant in negative expressions such as I haven’t seen him all day. They fail, however, to mention that the crucial factor here is the adverbial function of the NP
rather than the negation, since the whole indeed occurs in negated clauses, as in
(7:115).
(7:115) The whole weekend is not going to be good. (NYT95)
7.2.6.2
NPs with both and nouns for body parts or kinship
Finally, we will take a look at the syntactic functions of NPs with nouns for
body parts or kinship, as illustrated in Table 7.13.
Table 7.13. Syntactic functions of NPs with both, /POSSESSIVE DETERMINER/
and NOUN FOR BODY PARTS OR KINSHIP
NYT95
IND95
both
both my both of
both
both my both of
my
my
subject
24
18
0
58
57
45
object
22
21
29
26
37
100
prep. comp.
21
14
18
0
53
55
fragment
1
0
0
1
0
0
100
100
100
100
100
100
Total
Total pop.
322
61
30
224
73
1
There were no correlations in the variable with both in combination with a
semantically non-specified noun, as accounted for in Section 7.5.1. In this subgroup, however, there was a correlation between the simple both form and the
prepositional complement function, as in (7:116). The two variants including a
possessive determiner (both /of/ my) correlated with the subject function, as in
(7:117) and (7:118).
175
175
(7:116) Back in the West End nightclub, a twentysomething boy slinks by in an
Adidas jacket zipped to his chin, with a backpack on both shoulders.
(IND95)
(7:117) Both my parents were very brave. (IND95)
(7:118) Both of her lungs had collapsed, and she also had liver damage.
(NYT95)
As we saw in Section 7.3.4.2, both correlated with inanimate nouns (i.e. body
parts). The majority of the prepositional complements are of this type, such as in
both hands, with both feet etc., which tend to become almost like fixed frames
with open prepositional and nominal slots. Virtually all instances of the both (of)
+ POSSESSIVE DETERMINER variant in prepositional complement function include
a noun for kinship, as in (7:119), rather than for body parts, as in (7:120).
(7:119) […] Lot, the Sodomite made good, was later seduced by both his
daughters in turn (IND95)
(7:120) Dare had arthroscopic surgery on both of his knees […]. (NYT95)
7.6
Summary
This chapter explored linguistic factors which influence the choice of variant in
quantified NPs with all/whole, both and half. The underlying assumption was
that there are such correlations, and a number of different hypotheses were tested
on the corpus material. In some cases, I had a pre-existing hypothesis about how
a particular factor could correlate with a particular variant. These hypotheses
were based upon claims in the literature or upon my own intuition, and this was
tested on the corpus material. In other cases, a linguistic factor was investigated
without there being a pre-existing hypothesis of correlation.
We will not go into all of the analyses again, but merely point to a few interesting results. Firstly, my hypothesis about all of being particularly frequent
when the quantifier has totality meaning was corroborated indirectly by the correlation between this variant and the presence of a focus marker, such as almost
and many. Focus markers also played an important role in variables with half.
Secondly, there was a very strong correlation in NPs with all/whole and a
temporal noun between the simple all variant and the adverbial function. Animacy is a fascinating topic, and, thirdly, we saw that the hypothesis about NPs
with geographical nouns was corroborated: simple all was particularly frequent
in cases of (inferred) animate reference, whereas the whole of correlated with inanimate reference. A final result worth mentioning is the strong correlation in
the both + PLURAL NOUN variable concerning the presence of a modifier. Both
was particularly frequent when the NP was unmodified, and both the and both of
the correlated with modification. One important general conclusion to be drawn,
besides those for the particular variables, is that none of the linguistic factors
176
176
studied in this chapter can explain all of the variation. There were also a few
cases of interrelated factors. There may, of course, be others, not discovered in
the present study.
In the final chapter I will summarise the results from another perspective.
Here, I go through the different variables in turn to see what particular factors
were relevant in each case. My aim is to provide a more coherent and holistic
picture of the four quantifiers and the NP types in which they occur.
177
177
178
8. Conclusion
The principal aim of this study was to investigate variation in noun phrases including all/whole, both and half. More specific aims concerned overall frequency distribution of variants, non-linguistic factors (frequency distribution across
region and medium) and finally linguistic factors.
The first section (8.1) of this chapter summarises and visualises the results
presented in Chapters 5 to 7. In 8.2 I make some general observations regarding
the study and its relation to this particular research area, providing some
examples of how the results could be used in reference grammars. Lastly, I
conclude by pointing at a few possible areas of future research.
8.1
Summary and visualisation of
results
In Chapters 6 and 7 we investigated a number of factors to find out how they
correlated with different variants. The findings were presented for each of the
factors in turn. We shall here take the other perspective and look at each of the
variables and visualise them in the form of figures including all statistically significant correlations between variants and factors.
The type of figure varies depending on how many variants occurred in each
particular variable. Thus, a variable with two variants will be diagrammatically
illustrated by an arrow, a variable with three variants by a triangle (exemplified
in Figure 8.1) and a variable with four variants (all/whole + SINGULAR COUNT
NOUN) in the form of a square. The most frequent variant overall is indicated by
bold type and the relative frequencies between the variants (presented in Chapter
5) are indicated by percentages (from the total material) given in brackets. For
each variant, an example sentence is provided, illustrating several factors.
It should be noted that only statistically significant correlations are included
in the figures. All of them are significant at the <0.01 level, except those in
brackets, which are significant at the <0.05 level only. It is also very important
to remember that the presentation, in the form of figures and example sentences,
is probabilistic rather than absolute. That is, it should not be read as if, for instance, with certain factors present, the same variant is always used. The correct
interpretation is instead that “in the presence of factor X and Y, this particular
variant is more likely than otherwise”.
The closer a particular factor category is located to a certain variant, the
stronger the correlation between them (see Section 4.3.4 for a description of how
the relative strength of correlations has been calculated). Observe that the location of the correlations in the figure only gives their relative positions in terms of
179
179
strength. Put in other words, just because the distance between one factor and a
certain variant is twice as long as that of another factor, the correlation with the
latter is not twice as strong as with the former. Exact phi coefficient figures for
the factors (forming the basis of the ranking of correlations and obtained via the
chi square test) are given in Appendix E. Knock-out factors are indicated by
square brackets outside the figure, but close to the variant to which they are related. These will not be commented on here but are described in Appendix B.
Some correlations are characterised by a plus/minus relationship, as in the
presence or absence of a modifier. In the figures, such correlations are presented
as “modifier”/“no modifier” etc. In other cases, the particular category correlating with the variant is given, such as “animate reference/inanimate reference”
or “subject function”/”object function”/”adverbial function”.
Here is a constructed example of an illustrating figure (Figure 8.1).
Variant X (79%)
modifier
animate ref.
BrE
animate ref.
no modifier
Variant Y (12%)
(AmE)
inanimate ref.
no modifier
Variant Z (9%)
Figure 8.1. Figure of syntactic variation in a variable with three variants
and three factors involved in correlations (constructed example)
The figure shows that this variable comprises three variants (X, Y and Z), of
which Variant X was the most frequent one overall. Three factors (“the presence
of a modifier”, “animacy” and “region”) were involved in statistically significant
correlations (mainly at the <0.01 level) with particular variants. The “modifier”
category exhibited a positive correlation with Variant X (as did the “no modifier” category with Variants Y and Z). “Animate reference” correlated positively
with Variants X and Y (and “inanimate reference” with Variant Z). As regards
the regional factor, one of its categories (BrE) correlated positively with Variant
X, whereas another category (AmE) correlated positively with Variant Z (only at
the <0.05 level, as indicated by the brackets). Finally, “the presence (or absence)
of a modifier” exhibited the strongest correlation with all three variants, and is
thus located closest to the variants, while the weakest one concerned “region”.
How much stronger the “modifier” correlation was is not shown in the figure.
180
180
Illustrations of the same kind will be presented for each of the variables in an
attempt at visualising the complexity of the syntactic variation in general as well
as clarifying the particular patterns relevant for each variable.
8.1.1
NPs with common nouns
8.1.1.1
NPs with all/whole and common nouns
Figure 8.2 illustrates the variation pattern in the all/whole +
NOUN variable.
all (13%148)
adverbial function
poss./dem. det.
no focus marker
divisible noun
no modifier
animate ref.
(an adjacent of)
inanimate ref.
subject function
no modifier
indivisible noun
no focus marker
the definite article
SINGULAR COUNT
all of (2%)
a focus marker
a modifier
animate ref.
object function
(no adjacent of)
inanimate ref.
a modifier
no focus marker
whole (82%)
the whole of (3%)
poss./dem. det. = possessive/demonstrative determiner
Figure 8.2. Significant correlations in NPs with all/whole, DETERMINER and
singular COUNT NOUN
The following four variants occurred:
(8:1) He has lived all his life openly gay in Dinnington, […] (NYT95)
(8:2) […] he was planning to lay off virtually all of his remaining support staff,
(NYT95)
(8:3) ‘The whole picture is so drastic,’ said Rosalie Cream, […](NYT95)
148
With all the way excluded: all 4%, all of 2%, whole 91%, the whole of 3%.
181
181
(8:4) My object here, however, is not to analyse the whole of the report which
was published just under a week ago. (IND95)
In this variable whole predominated clearly. In the all variant, the strongest correlation was with the adverbial function and possessive/demonstrative determiner, because of the frequent occurrence of phrases like all my life and all this
week 149. Although phrases with inanimate nouns like life and year dominated,
there were also quite a few instances of collective nouns, which explains why all
correlated with animate reference. All of correlated most strongly with the presence of a focus marker and a modifier. Meanwhile, the whole variant showed its
strongest correlation with the definite article and the absence of a focus marker.
The variant including the whole of, finally, correlated most strongly with the
absence of a focus marker and the presence of a modifier. Note that the presence/absence of an adjacent of correlation was only significant at the <0.05
level.
The all/whole + MASS NOUN variable is visualised in Figure 8.3.
all (93%)
nfm
prep. comp.
modifier
an adjacent of
object function
BrE, AusE
AmE
object function
no adjacent of
[Knock-out:
subject function
some or…]
a modifier
a focus marker
all of (4%)
an adjacent of
no modifier
no focus marker
adverbial function
whole (3%)
nfm = no focus marker, prep. comp. = prepositional complement function
Figure 8.3. Significant correlations in NPs with all/whole, DETERMINER and
MASS NOUN
There are three variants in this variable:
(8:5) I do think those people who marry into my family find it increasingly
difficult to do so because of all the added pressure […] (IND95)
149
182
Another frequent adverbial phrase is all the way, which was excluded from the study of linguistic
factors because of its great predominance.
182
(8:6) Almost all of the really cool software arriving in this office in the last six
months […] came on laser-pitted CD-ROM disks. (NYT95)
(8:7) He had breathed unaided the whole time, […] (IND95)
Here we again see one variant (all in this case) predominating strongly. All
(which was more frequent in the British and Australian material) correlated most
strongly with the absence of a focus marker and the prepositional complement
function. In contrast, the strongest correlation in the all of variant (which was
more frequent in the American corpus) was the presence of a focus marker and a
modifier. Whole correlated most strongly with the adverbial function (because of
the many instances of the whole time) and with the absence of a focus marker.
The correlation pattern for all + PLURAL NOUN is presented in Figure 8.3.
all (92%)
an adjacent of
prep. complement function
BrE
a modifier
no focus marker
a focus marker
no modifier
AmE
subject function
no adjacent of
[Knock-out: some or…]
all of (8%)
Figure 8.4. Significant correlations in NPs with all, DETERMINER and
PLURAL NOUN
This variable comprises only two variants, all and all of, the former being more
than ten times more frequent than the latter:
(8:8) Little did I know that I would live in Malacanang Palace for 20 years and
go to all the major corridors of power in the world, […] (IND95)
183
183
(8:9) We were a commuter school, and almost all of our kids were local.
(NYT95)
There were slightly fewer correlations in this variable. In NPs with all, the
strongest correlations concerned the presence of an adjacent of and the prepositional complement function. All of correlated most strongly with the absence
of an adjacent of and the subject function. It was also more frequent in the
American material.
8.1.1.2
NPs with both and common nouns
Figure 8.5 shows the correlations occurring in the both + PLURAL NOUN variable.
both (97%)
[knock-out: restricted reference
(except nouns denoting body
parts or kinship), opening times
in advertisements etc.]
inanimate reference
AmE
BrE
animate reference
modifier
animate reference
modifier
both the (2%)
[knock-out:
one or…]
both of the (1%)
Figure 8.5. Significant correlations in NPs with both, /THE DEFINITE
ARTICLE/ and PLURAL NOUN
Examples (8:10) to (8:11) illustrate the three variants.
(8:10) Draws seem possible in both matches, […] (IND95)
(8:11) Other Conservative MPs pointed out that both the cancelled meetings
would have covered the same ground […] (IND95)
(8:12) The authorities said Mr. Green knew both of the slain postal workers,
(NYT95)
The both variable was perhaps the most interesting one with respect to overall
distribution. In grammar books the three variants (both – both the – both of the)
are generally presented as if they were equally common, but the corpus study
showed that simple both is extremely predominant (97% overall). Simple both
(most frequent in AmE) correlated with the absence of a modifier and inanimate
184
184
reference, while both the (most frequent in BrE) and both of the correlated with
the presence of a modifier and with animate reference.
8.1.1.3
NPs with half and common nouns
The correlation pattern of half + PLURAL NOUN is presented in Figure 8.6.
half (63%)
no focus marker
BrE, AusE
poss. det./def. art./gen.
dem. determiner
AmE
a focus marker
[knock-out: some or…]
half of (37%)
poss. det. = possessive determiner, def. art. = the definite article, gen. = genitive, dem.
determiner = demonstrative determiner
Figure 8.6. Significant correlations in NPs with half, DETERMINER and
PLURAL NOUN
The two variants are illustrated in (8:13) to (8:14).
(8:13) In the club itself, half the members are women. (SMH95)
(8:14) He figures the 30 Dow stocks operate in some 70 industries and get
about half of their sales from abroad […]. (NYT95)
Half without of was the predominant variant, and half of was more frequent in
AmE than in BrE and AusE. In addition to the correlations with regional
varieties, half correlated with the absence of a focus marker and also with the
presence of a possessive determiner, the definite article or a genitive. Half of
correlated with the presence of a focus marker and a demonstrative determiner.
185
185
8.1.2
NPs with all and demonstrative pronouns
Figure 8.7 shows the pattern for all + DEMONSTRATIVE PRONOUN.
all this/that/these/those (89%)
an adjacent of
BrE
prep. comp. function
AmE
no adjacent of
subject function
all of this/that/these/those (11%)
prep. comp. = prepositional complement
Figure 8.7. Significant correlations in NPs with all and DEMONSTRATIVE
PRONOUN
These are the two variants:
(8:15) But a question arises out of all this: […] (IND95)
(8:16) All of this is lucidly and soberly recounted by Mr. Stille, […] (NYT95)
In this variable, the variant without of predominated again. As regards the all
this etc. variant, only two correlations were statistically significant at the <0.01
level: the presence of an adjacent of and British English. The correlation with the
prepositional complement function was only significant at the 0.05 level. All of
this etc. (most frequent in AmE) correlated most strongly with the subject
function and the absence of an adjacent of. Also remember that of variants were
more frequent when the head of the noun phrase was a demonstrative pronoun,
as in this variable, rather than a noun, as in all (of) the book/children.
186
186
8.1.3
NPs with all and personal pronouns
The correlation pattern for all + PERSONAL PRONOUN is illustrated in Figure 8.8.
we/us, they/them all (88%)
subject function
no focus marker
BrE
AmE
a focus marker
prep. comp. function
[Knockout: fragment]
all of us/them (12%)
prep. comp. = prepositional complement
Figure 8.8. Significant correlations in NPs with all and PERSONAL PRONOUN
In this variable, there are also two variants, as illustrated by (8:17) and (8:18).
(8:17) Why are we all so depressed? (IND95)
(8:18) […] foreign investors are now flocking back to all of them. (NYT95)
The we/us (...) all variant predominated strongly over the variant with all of
followed by a personal pronoun. The former (most frequent in BrE) correlated
significantly with the subject function and the absence of a focus marker,
whereas the latter (most frequent in AmE) correlated with the prepositional
complement function and the presence of a focus marker.
187
187
8.1.4
NPs with half, the indefinite article and a singular
noun or numeral
The half + THE INDEFINITE ARTICLE + SINGULAR NOUN/NUMERAL variable and its
correlations are visualised in Figure 8.9.
half a (80%)
BrE
a focus marker
prep. comp. function
AusE
(numeral)
(partitive noun)
NP det. function
no focus marker
AmE
[knock-out: directly
preceding another noun]
a half (20%)
prep. comp. = prepositional complement, NP det. = NP determiner
Figure 8.9. Significant correlations in NPs with half, THE INDEFINITE
ARTICLE and SINGULAR NOUN or NUMERAL
The following variants occur:
(8:19) There was an Emmy award for a television special and gold-album
recordings, each selling more than half a million. (IND95)
(8:20) Onions were chopped and thrown into a pot to sweat in olive oil; […], a
half-cup of sugar and a can of tomato paste followed. (NYT95)
The half a/an variant predominated overall. In this case, however, there was
strong regional variation; in American English, the a half variant was much
more frequent than in British and Australian English. The strongest correlations
in the half a/an variant was (apart from BrE) the presence of a focus marker and
prepositional complement function. The correlation with a numeral was only sig-
188
188
nificant at the <0.05 level. The strongest correlations in the a half variant were
(apart from AmE) the absence of a focus marker and NP determiner function150.
The correlation with partitive noun was only significant at the <0.05 level.
8.1.5
NPs with all/whole and geographical names
Figure 8.10 illustrates the variable comprising all/whole + GEOGRAPHICAL NAME.
all (16%)
animate reference
subject function
(an adjacent of)
(no adjacent of)
object function
inanimate reference
AmE
object function
inanimate reference
BrE
all of (51%)
the whole of (43%)
Figure 8.10. Significant correlations in NPs with all/whole, /THE DEFINITE
ARTICLE/ and GEOGRAPHICAL NAME
Three variants are used:
(8:21) […], not just Bill Clinton but all America is the loser. (IND95)
(8:22) The first covers all of California, […] (NYT95)
(8:23) St Edmundsbury and Ipswich, which covers almost the whole of Suffolk,
is a relatively new diocese […] (IND95)
All of was the most frequent variant overall. This is another case of regional
variation, however. Simple all correlated most strongly with animate reference
and the subject function. All of correlated most strongly with American English
and inanimate reference, whereas the whole of correlated most strongly with
British English and inanimate reference. The correlation between all of and the
absence of an adjacent of was only significant at the <0.05 level.
150
It should be observed that in the case of syntactic function, the correlation was only significant
when all tokens including the cricket term a half-century had been excluded from the data.
189
189
8.1.6
Subgroups
8.1.6.1
NPs with all/whole and temporal nouns
This is the correlation pattern for all/whole + TEMPORAL NOUN:
all (91%)
[knock-out: opening times
in advertisements etc.]
adverbial function
(natural time)
AmE
(BrE, Aus)
(arbitrary time)
subject function
object function
prep. complement function
the whole (7%)
Figure 8.11. Significant correlations in NPs with all/whole, THE DEFINITE
ARTICLE and TEMPORAL NOUN
There are two main variants151:
(8:24) Busloads have been arriving all day, […] (NYT95)
(8:25) The life of Catherine Cobb spanned nearly the whole century […]
(IND95)
Again, we have a case of strong predominance for one variant: simple all. The
whole was the second most frequent variant, while the others (all the, all of the
and the whole of the) were very infrequent. Only the “syntactic function” factor
correlated in a statistically significant way with the variants. Simple all (most
frequent in AmE) showed a very strong correlation with the adverbial function
(the strongest of all correlations in the study), whereas the whole (most frequent
151
190
Three more variants, all the, all of the and the whole of the occurred in the material (see Section
5.1.6.1), but since they were very infrequent they were not included in the analysis of linguistic
factors. Hence the reason why the total percentage sum is only 98%.
190
in BrE and AusE) correlated with the prepositional complement, object and subject functions. The “time division” factor correlated at the <0.05 level only.
8.1.6.2
NPs with both and nouns for body parts or kinship
The final variable is both in combination with a noun referring to body parts or
kinship, as illustrated in Figure 8.12.
both (77%)
inan. ref.
prep. comp.
AusE
(an adjacent of)
(an adjacent of)
BrE
subject function
animate reference
both my (19%)
(no adjacent of)
AmE
subject function
animate reference
both of my (4%)
inan. ref. = inanimate reference, prep. comp. = prepositional complement function
Figure 8.12. Significant correlations in NPs with both, /POSSESSIVE
DETERMINER/ and NOUN FOR BODY PARTS/KINSHIP
The three variants occurring in this variable are illustrated in (8:26) to (8:28).
(8:26) Englewood Cliffs is very small and very well-off, with so few black
residents they could be counted on the fingers of both hands. (SMH95)
(8:27) Both my parents went to Oxford University – they both read maths there
[…] (IND95)
(8:28) Both of their children were born after the divorce. (NYT95)
Simple both (particularly frequent in Australian English) correlated most
strongly with inanimate reference and the prepositional complement function,
while both my and both of my (most frequent in the British and American material respectively) correlated with animate reference and the subject function. The
correlation concerning an adjacent of was only significant at the <0.05 level.
191
191
8.2
Concluding remarks
After going through the different variables, it has become clear that the picture
of syntactic variation presented in grammar books is a very simplified one. Some
of the variables exhibited more complex patterns, i.e. correlated positively with
more factor categories than others. There may also be other factors that are
equally or more important. We have looked at region and medium, and a number
of linguistic factors. Other non-linguistic factors may be involved (e.g. age, educational background) and it is also possible that there are other syntactic and semantic/pragmatic ones that could be examined (e.g. type of verbs interacting
with the quantified NPs). Furthermore, there could be lexical and textlinguistic
factors influencing the choice of variant.
Besides covering a grammatical area that has been little analysed before in
terms of variation, this investigation differs from many other studies of syntactic
variation, because it focuses on a very small part of the language. I have accordingly been able to take a large number of factors into account, giving a complex
picture of the various aspects that are at play in the choice of one variant over
another. Many other works dealing with syntactic variation are either mainly
theoretical, discussing, for instance, what should be considered the meaning constant in variation research, or case-studies of just one or a few factors. Studies
which have endeavoured to take both linguistic and non-linguistic factors into
account are particularly rare (for exceptions, see, for instance, Tottie 1991; Biber
et al 1998; Levin 2001).
Another important aspect is that the analysis is based on a large authentic material in the form of computerised text corpora. Working with authentic texts is
not always an easy and straightforward task, since real language seldom fits
neatly into linguistic categories. This has been discussed in the method and material chapter and reflected in the large number of footnotes commenting on the
results. Still, in my opinion, basing one’s claims on what real people have said
and written gives more validity than relying on intuition only, especially when
the researcher is not a native speaker of the language under investigation. The
study has, among other things, pointed out differences between the words all,
both and half, which are often treated in the grammatical literature as being very
similar in form and function, perhaps owing to a shortage of empirical
information on how the words are really used. Similarly, the study has provided
descriptions of usage that are not brought up in the grammatical literature, by
taking due consideration of (i) overall relative frequency of variants, (ii)
differences according to region152 and (iii) various linguistic factors influencing
the choice of variant. An additional contribution of the study has been to corroborate some previous statements in the literature. Besides all the specific new
insights obtained about linguistic and non-linguistic factors, it was observed that
152
192
No clear differences between speech and writing consistent across British and American English
were observed in the study. The reason is either that there are no such differences or that the
written and spoken corpus materials were too similar, since newspapers contain a great deal of
spoken language from interviews.
192
in most of the variables, one variant predominated greatly, sometimes almost to
the exclusion of other variants. This fact contrasts with how syntactic variation
with these NP types is usually presented in grammar books, i.e. as if different
syntactic variants were equally frequent. In many areas, the results are too
detailed or uncertain to be useful for pedagogical and reference grammars. In
some cases, however, the application of the results would clearly improve grammar books. In a reference grammar, such information could, for instance, take
the following shape:
• There are three variants including both and a plural noun: both books, both
the/these/my books or both of the/these/my books. The first variant is far more
frequent than the other two and is the variant that is generally used in cases
where there are only two possible referents, such as both halves and both
twins. The variants both the and both of the are often the preferred alternatives when the NP includes a modifier, as in both /of/ the accused men.
Sometimes, especially in NPs with nouns designating body parts and kinship,
a possessive determiner is a more natural alternative (both my parents, both
my hands etc.) than the definite article.
• There are (mainly) two variants to express totality in a noun phrase including
a temporal noun: all day and the whole day. All the day (esp. BrE), all of the
day and the whole of the day also occur, but are quite unusual. All is almost
exclusively used when the NP is used in adverbial function, as in I haven’t
seen him all day (esp. when the clause is negated). In contrast, the whole is
the preferred variant when the NP is used in another syntactic function, as in
The whole weekend has been marvellous.
• In cases where we wish to use a personal pronoun together with all, there are
two possibilities: we … all, as in Why can’t we all go back to school? or all
of us, as in Why can’t all of us go back to school? The former variant (sometimes referred to as “the floating quantifier”) is the more frequent one overall, and the latter is more frequent in American than in British English. The
we/us (...) all variant is particularly frequent when the NP has subject function, as in the example above, while the all of us variant predominates in prepositional complement function, as in There’s a Jekyll and Hyde in all of us.
• When an NP includes a geographical name and a word for totality, there are
three alternatives: all France, all of France and the whole of France.
Remember that of has to be used in the third variant. All of is the most
frequent variant in American English, whereas the whole of predominates in
British English. Geographical names can denote two different things, a geographical/political area or the people inhabiting it. This distinction is often
shown in the choice of quantifier, since the whole of is more frequent in the
former function and simple all in the second, while all of is about equally frequent in the two functions. The area–people distinction is also reflected in
simple all being more frequent in subject function (which is often used about
people), while the whole of is more frequent in object function (which is
often used for things).
193
193
• Half a kilo and a half kilo are used with the same sense by many people. The
latter variant is a particularly frequent alternative in American English. Some
people make a semantic distinction between the two, using a half as a more
established unit, as in a half-bottle, and half a/an as an ordinary partial expression (comparable with some etc.). This difference is reflected in the fact
that a half is the more frequent alternative when the NP is used as a determiner of the head of the NP, as in a half dozen cows. On the other hand, half
a/an is particularly frequent when there is an approximator like almost or
nearly in the noun phrase.
The distinction between totality and large-quantity meaning might also be given
some treatment in reference grammars, pointing to all of mainly being used
when the totality meaning (sometimes signalled by focus markers like nearly or
except one) is involved.
In Section 2.1.2, we saw that the whole idea of grammatical synonymy has
been questioned. My general conclusion with respect to syntactic variation is
that there are indeed cases where two or more grammatical structures seem to
compete with each other, even though a number of linguistic and non-linguistic
factors (sometimes many, sometimes few) influence the variation to a greater or
lesser extent. I thereby refute a very strict interpretation of the “two forms = two
meanings” hypothesis. The conclusion is strengthened by the fact that virtually
no absolute correlations occurred. In other words, it was not the case that with
one factor category present, the same variant was always chosen. Also, the material comprised examples of very similar phrases, as in (8:29) to (8:31).
(8:29) Both major parties have been slow to adopt Asian candidates. (SMH95)
(8:30) We have become the target of both the major parties. (IND95)
(8:31) […] and the looming Federal election provides them with a unique
opportunity to garner further concessions from both of the major parties.
(SMH95)
In the examples, both the statistically significant correlations observed with this
variable (animacy and the presence of a modifier) are out of play, since all three
examples are of the same type (animate reference and no modifier). The NPs in
the examples are different in syntactic function, but this factor did not show a
significant correlation in the material. We must of course always be aware that
there may be correlations that were not detected here. Also, Langacker’s ideas
about dialectal conventionalisation of certain variants (Section 2.1.2) and Sapir’s
comment that people do not uphold semantic distinctions (Section 3.3.4.1) are
important to keep in mind.
I can envisage a number of possible studies following up the present one. The
figures obtained here could, for instance, be used in more advanced factor analysis, (see Section 2.1.5). The area of totality vs. large-quantity meaning could be
more closely examined, for instance, by means of elicitation tests with native
speakers, or by looking more closely into the interplay with other factors. The
194
194
same goes for many other of the areas that have only been touched upon briefly,
such as divisibility and animacy in quantified noun phrases.
As far as the four quantifiers are concerned, other aspects could be investigated as well. Quantifiers could, for instance, be explored from a contrastive
point of view, using parallel corpora of English and Swedish or other languages.
Another field would be to compare how the language of learner corpora differs
from native speakers’ language in the use of quantified NPs. To conclude, syntactic variation is a very rich and interesting field, and there are many areas of
syntactic variation where the method used for the present study could be applied.
195
195
196
Appendices
A
Variables
Table A1. NPs with all/whole, both and half, DETERMINER and COMMON
NOUN
Variables153
all + DETERMINER + SINGULAR COUNT NOUN
all of + DETERMINER + SINGULAR COUNT NOUN
DETERMINER + whole + SINGULAR COUNT NOUN
the whole of the + SINGULAR COUNT NOUN
Examples
all the book
all of the book
the whole book
the whole of the book
all + DETERMINER + MASS NOUN
all of + DETERMINER + MASS NOUN
DETERMINER + whole + MASS NOUN
the whole of + DETERMINER + MASS NOUN
all the stuff
all of the stuff
the whole stuff
the whole of the stuff
all + DETERMINER + PLURAL NOUN
all of + DETERMINER + PLURAL NOUN
all the children
all of the children
both + PLURAL NOUN
both the + PLURAL NOUN
both of the + DETERMINER + PLURAL NOUN
both children
both the children
both of the children
half + DETERMINER + NOUN
half of + DETERMINER + NOUN
half the book
half of the book
153
The formulae also cover cases where one or more modifying adjective(s) precede(s) the noun.
This is true of all the variables with common nouns.
197
197
Table A2. NPs with all/whole, both and half and DEMONSTRATIVE PRONOUN
all this/these
all + DEMONSTRATIVE PRONOUN
all of + DEMONSTRATIVE PRONOUN
all of this/these
both + DEMONSTRATIVE PRONOUN
both of + DEMONSTRATIVE PRONOUN
both these
both of these
half + DEMONSTRATIVE PRONOUN
half of + DEMONSTRATIVE PRONOUN
half these
half of these
Table A3. NPs with all, both and PERSONAL PRONOUN
all of us
all of + PERSONAL PRONOUN
PERSONAL PRONOUN + all
we/us all
both of + PERSONAL PRONOUN
PERSONAL PRONOUN + both
both of us
we/us both
Table A4. NPs with half, THE INDEFINITE ARTICLE and SINGULAR NOUN OR
NUMERAL
half a/an + SINGULAR NOUN or NUMERAL
a half + SINGULAR NOUN or NUMERAL
half an hour/half a dozen
a half hour/dozen
Table A5. NPs with all/whole and GEOGRAPHICAL NAME
all India
all + GEOGRAPHICAL NAME
all of + GEOGRAPHICAL NAME
all of India
the whole of + GEOGRAPHICAL NAME
the whole of India
Table A6. NPs with all/whole, /THE DEFINITE ARTICLE/ and TEMPORAL NOUN
all day
all + TEMPORAL NOUN
all the + TEMPORAL NOUN
all the day
all of the + TEMPORAL NOUN
all of the day
the whole + TEMPORAL NOUN
the whole day
the whole of the + TEMPORAL NOUN
the whole of the day
Table A7. NPs with both, /POSSESSIVE DETERMINER/ and NOUN FOR BODY
PARTS OR KINSHIP
both + NOUN FOR BODY PARTS/KINSHIP
both eyes/parents
both his eyes/parents
both + POSSESSIVE DETERMINER + NOUN FOR BODY
PARTS/KINSHIP
both of + POSSESSIVE DETERMINER + NOUN FOR BODY both of his eyes/parents
PARTS/KINSHIP
198
B
Exclusions
B1
Exclusions
Knock-out effects
As described in Section 2.1.3, tokens where a factor “knocks out” the variation
were excluded from the statistics. There were four cases of knock-out effects.
The first one concerns NPs with both (/of/ the) and a plural noun, where simple
both is the only alternative if (i) the noun has restricted reference (see Section 3.
3.4.2) and (ii) the noun does not denote body parts or kinship:
(1)
Koy Detmer, the Colorado quarterback who had torn a knee ligament just
two weeks before, played in parts of both halves […] (NYT95)
Second, all cases of all of + PERSONAL PRONOUNS were excluded when used in
language fragments, such as short answers, appositions and independent prepositional phrases. The reason is that the syntactic alternative, the we/us (...) all
variant, is never used in these linguistic structures. Example (2) illustrates such a
case.
(2)
I feel that more information is needed. For all of us. (IND95)
A third knock-out effect occurred in NPs with all/whole and temporal nouns and
in NPs with simple both and the word days. Since newspaper material was used
as material for the investigation, a number of tokens came from advertisements
of opening times of restaurants, museums etc. and from information about events
etc. Tokens of all + TEMPORAL NOUN, as in (3), and of simple both + PLURAL
NOUN, as in (4), were excluded since no variation exists in this context.
(3)
Closed lunch Saturday and all Sunday. (NYT95)
(4)
The festival runs both days from 10 A.M. to 4:30 P.M.; admission is free.
(NYT95)
There is finally a more general knock-out effect occurring in all cases where
there is normally variation between a variant with of and one without (as in all of
the children vs. all the children). If the quantifier is preceded by another quantifier + or, as in (5), variation is impossible. Here some requires an of-construction, with the consequence that all without of cannot be used.
(5)
Fire experts say that if the door had been accessible or the sprinkler system
had worked, some or all of the victims might have escaped before being
overcome by smoke and carbon monoxide. (NYT95)
199
199
Fixed expressions
As mentioned in Section 2.1.4, a large number of tokens were excluded as fixed
expressions since they did not exhibit any variation. The basis for deciding what
constructions to exclude was twofold. First, these constructions were found as
whole phrases in one or more dictionaries154. Second, there was no variation in
the corpora. If variation was found in expressions which dictionaries considered
to be fixed, these were included in the statistics (see below). Definitions given
with some of the figurative expressions below were taken from Longman Dictionary of Contemporary English (2003). The following constructions were
excluded as fixed expressions:
(1) NPs with all/whole, DETERMINER and SINGULAR COUNT NOUN:
•
•
•
•
•
•
•
•
•
•
•
•
all the/this while
for all the world as if, for all the world like (‘exactly as if/like’)
make all the difference
the whole gamut (‘all the possibilities between two extremes’)
/go/ the whole hog (‘to do something as completely or as well as you
can, without any limits’)
the whole idea 155
the whole lot
the whole matter 156
the whole point /of/
the whole shebang (‘the whole thing’)
the whole shooting match (‘the whole of a situation, or an event that is
the best or most complete of its kind’)
the whole thing 157
(2) NPs with all/whole, DETERMINER and MASS NOUN:
•
•
•
•
•
154
155
156
157
158
200
all my love 158
all the best (‘used to express good wishes for the future’)
all the fashion
all the go (‘very fashionable’)
all the rage (‘very popular and fashionable’)
The dictionaries consulted were Cambridge International Dictionary of English (1995), Collins
Cobuild English Dictionary for Advanced Learners (2001), Longman Dictionary of Contemporary English (2003), Norstedts stora engelsk-svenska ordbok (1999) and Oxford Advanced
Learner’s Dictionary (2000).
The whole idea was not found in dictionaries, but since it is very similar in meaning to the whole
point and exhibited no variation in the corpus, it was excluded as a fixed expression.
The whole matter was not found in dictionaries, but since it is very similar in meaning to the
whole thing and exhibited no variation in the corpus it was excluded as a fixed expression.
One example of all the thing was found in the spoken American corpus, but this single case did
not justify the dismissal of the whole thing as a fixed expression.
Tokens were deleted when used as isolated phrases, as in ‘All my love, Gracie’ says the message
[…] (IND95), and otherwise included, as in A person with whom I’ve had many creative differences […] but who tonight has all my love and respect and affection (NYT95).
200
•
•
•
/and, or/ all that, this 159 + a common noun of a semantically rather
non-specific kind, very often with a negative connotation (baloney, bit,
blah blah blah, bollocks, bullshit, business, crap, crud, garbage, guff,
hoopla, junk, lot, nonsense, rot, rubbish, scum, shit, stuff, thing)160
and all that jazz (‘and things like that’)
the whole truth
(3) NPs with all/whole, DETERMINER and PLURAL NOUN
•
/and/ all these, those /kind, sort of/ + things 161
(4) NPs with both and PLURAL NOUN
•
•
•
•
•
in both cases
in both senses /of the word/
on both accounts/counts
the best of both worlds (‘a situation in which you have the advantages
of two different things without any of the disadvantages’)
the worst of both worlds (a jocular distortion of the best of both worlds)
(5) NPs with half, DETERMINER and SINGULAR COUNT NOUN
•
/be/ half the battle (‘you have done the most difficult part of something
and the rest is easy’)
(6) NPs with half, the indefinite article and singular or numeral
• a half-back (‘a player in football, rugby, hockey…’ or ‘a player in
American football …’)
159
This construction sometimes includes a premodifier, such as bloody or kind of (as in all this
bloody shit)
160
Tokens were deleted when linked by and (as in […] trying to act like white guys and all this
stuff, LSAC), or used as appositions ([…] he told me I don't wanna talk to you any more, blah,
blah, all that rubbish, BNC), otherwise included (as in Took us half the night to get all that stuff
in, BNC). Three examples of and all of this/that + [common noun] were found in the spoken
American corpus and one in The Sydney Morning Herald. However, these few cases did not justify the dismissal of /and/ all this/that + [common noun] as a fixed expression. In those cases
where expressions that are generally regarded as fixed were included (e.g. all the time, see
below), there were several tokens of an alternative construction (e.g. 53 tokens of the variant expression all of the time). As for cases of all this/that [+ common noun] which were not linked by
and or used as appositions, however, the material included a larger number of tokens of the of
variant, as in And we can share all of this stuff (LSAC). Accordingly, these cases were not considered fixed and instead included in the material. The same goes for cases where the head of the
noun phrase was a pronoun (and all this/that), rather than determined noun (e.g. and all that
shit). Here, a great many examples of all of (and all of this/that) were found. Accordingly, and
all that could not be excluded as a fixed expression.
161
Two examples of and all of these/those things were found in The Sydney Morning Herald, one in
The Independent and one in The New York Times, but these few cases did not justify the dismissal of /and/ all these/those things as a fixed expression.
201
201
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
a half-breed (‘a word which is now considered offensive, meaning
someone whose parents are of different races’)
a half-brother
a half-caste (same as half-breed)
a half-circle
a half-fare
a half-life (‘the half life of a radioactive substance is the length of time
it takes to lose half of its radioactivity’)
a half-light (‘the dull grey light you see when it is almost dark but not
completely dark’)
a half-marathon
a half-moon
a half-negro
a half-nelson (a term in wrestling)
a half-nod
a half-note
a half-peace
a half-price
a half-sister
a half-smile
a half-stroke
a half-swing
a half-tone
a half-truth
a half-volley (a term in tennis or cricket)
a half-wit (‘a stupid person or someone who has done something stupid’)
(7) NPs with all/whole /THE DEFINITE ARTICLE/ and TEMPORAL NOUN:
•
B2
all day, night etc. long
“Fixed” expressions included in the study
The following constructions, which are generally regarded as fixed expressions,
were included because they exhibited variation in the corpus material:
•
•
•
•
162
163
202
all /of/ the time, the whole time 162
and all /of/ that
for all /of/ (‘in spite of’)163
half /of/ the story (‘an explanation that is not complete’)
In Section 7.3.4 it was suggested that there is sometimes a meaning difference in terms of specific-generic reference between all the time and the whole time.
Jespersen (1933:185) points to the “obliteration of the original meaning of all” in the for all construction.
202
C
Search procedures used with the
different corpora
The five corpora used are all of different types and thus required somewhat
different search techniques, which will be accounted for here.
The British National Corpus (BNC)
The BNC has its own search program, called SARA. This was, however, not used
for the present study. Instead an interface created by a research team at the
English department of the University of Zurich (Sebastian Hoffmann, HansMartin Lehmann and Peter Schneider) was accessed via the Internet. After a
restriction to the particular material that was to be used (i.e. spoken texts in
dialogue settings), lexical searches (e.g. all the) were made, concordance lines
were sorted alphabetically, irrelevant examples were deleted manually (which
sometimes required an expansion of the co-text) and finally each file was
downloaded as a text file and saved on the computer hard disk. In some cases
(see Section 5.1) tokens were divided (manually) into groups (e.g. [all the +
SINGULAR COUNT NOUN], [all the + MASS NOUN] and [all the + PLURAL NOUN])
and put in separate text files. Tokens in subgroups (e.g. all/whole + TEMPORAL
NOUN) were extracted manually from the larger text files (all/whole + SINGULAR
COUNT NOUNS in this case). In searches for [all + GEOGRAPHICAL NAME] a
specific procedure had to be used, since such a query involves a large number of
different lexical items. I made a query for just all, and then went through the
concordance lines manually to find those cases where all was immediately
followed by a geographical name (by looking for capital letters).
The Independent
The Independent on CD-ROM comes with a search software called Freeway.
This program is easy to handle and allows for searches of all, whole, both and
half. Of, however, is a noise word, a fact which made it necessary to use another
search program in order to search for constructions like all of the children.
Searches were made in Freeway for all newspaper texts including all, whole,
both and half and these were downloaded as text files and saved on the computer
hard disk. Subsequently, the concordance programme Wordsmith was used for
lexical searches on these text files, after which the same procedures were carried
out as with the BNC.
The Longman Spoken American Corpus
This corpus does not have its own search program, so lexical searches were
carried out with the Wordsmith program directly on the text files on the corpus
203
203
CD-ROM. The same lexical search procedures were used as for the BNC and
The Independent.
The New York Times
This corpus provided the greatest search difficulties of all the corpora, owing to
the fact that all four quantifiers are considered noise words by its own search
software (Proquest). Fortunately, the CD-ROM can be used directly with the
Wordsmith program, so lexical searches were carried out this way instead, after
which the same procedures as those described above were used. A particular
problem concerned the search for the [all + GEOGRAPHICAL NAME] construction.
The procedure used with the other corpora could not be undertaken with The
New York Times, owing to the fact that (i) searches were carried out directly on
the CD-ROM text files, (ii) the text consisted of mainly one extremely large text
file (plus a few smaller ones, comprising, for instance, headlines) and (iii) Wordsmith does not allow more than 16,000 concordance lines for a search. In order
to get around the problem in the best way possible, searches were made with all
followed by particular geographical names instead, covering a large number of
countries (taken from the Longman dictionary of contemporary English), all
continents, all American states, all New York city areas, a number of large
American cities and a large number of world capitals and other important cities
(for instance, all those occurring in the other corpora). The consequence of this
could be that the figures for all + GEOGRAPHICAL NAME (from The New York
Times) in Chapters 5 and 6 are slightly lower than would have been the case had
the normal procedure been undertaken.
The Sydney Morning Herald
Finally, The Sydney Morning Herald has its own software (Fairfax). This program does not use noise words, so lexical searches could be made directly in
Fairfax. The texts which included the four quantifiers were then downloaded as
text documents, after which the same procedures were carried out by means of
Wordsmith as with the other corpora.
204
204
D
Frequency tables for the analysis of
linguistic factors
The following tables show the frequencies of all statistically significant correlations. They are presented factor by factor in the same order as in Chapter 7. In all
tables except the first one (type of central determiners), the frequencies are those
of the samples (100 tokens per variant and corpus). In the total corpus population there were sometimes great discrepancies in frequency between the different
syntactic variants. The total frequencies of tokens found in the corpora are
therefore given in the “total population” row of each table. With two of the linguistic features (the presence of modifiers, D6, and the presence of an adjacent
of, D7), subcategories are indicated by smaller type. Normalised frequencies
(when there were fewer than 100 tokens) are given in italics and significant correlations (at the <0.01 level unless otherwise indicated) by bold type.
D1.
Type of central determiner
Table D1:1. Type of central determiner in NPs with all/whole, DETERMINER
and SINGULAR COUNT NOUN164
NYT 95
definite
possessive
demonstrative
genitive
article
determiner
determiner
N
%
N
%
N
%
N
%
All
87
3% 308
62 22%
1
7%
41%
All of
88
3%
53
7%
3
1%
8
53%
Whole
2465
52%
214 77%
6
40%
93% 396
The whole of
11
0%
3
0%
0
0%
0
0%
2651 100% 757 100%
279 100%
15 100%
Total
IND95
definite
possessive
demonstrative
genitive
article
determiner
determiner
N
%
N
%
N
%
N
%
All
97
3% 249
55 31%
2
22%
46%
All of
28
1%
16
3%
4
2%
1
11%
Whole
3159
45%
106
59%
4
44%
92% 241
The whole of
162
5%
31
6%
15
8%
2
22%
3446 100% 537 100%
180 100%
9 100%
Total
164
Here and throughout, all tokens including way as the head of the NP (all the way, all of the way,
the whole way) were excluded from the statistics in this variable.
205
205
Table D1:2. Type of central determiner in NPs with half, DETERMINER and
PLURAL NOUN
Half
Half of
Total
definite
article
N
%
658
56%
516 443%
1174 100%
Half
Half of
Total
definite
article
N
%
394
71%
162
29%
556 100%
D2.
NYT95
possessive
demonstrative
determiner
determiner
N
%
N
%
118
4 11%
55%
97
45%
34 89%
215 100%
38 100%
IND95
possessive
demonstrative
determiner
determiner
N
%
N
%
95
5
24%
80%
24
20%
16 76%
119 100%
21 100%
genitive
N
65
79
144
%
45%
55%
100%
genitive
N
31
11
42
%
74%
26%
100%
Divisibility
Table D2:1. Divisibility in NPs with all/whole, DETERMINER and SINGULAR
COUNT NOUN
all
NYT95
all of whole
Divisible
Indivisible
Total
86
14
100
54
46
100
Tot pop
458
152
206
the
whole
of
43
80
20
57
100
100
3081
14
all
IND95
all of whole
92
8
100
59
41
100
403
49
the
whole
of
25
40
60
75
100
100
3510
210
206
D3.
Animacy
Table D3:1. Animacy in NPs with all/whole, DETERMINER and SINGULAR
COUNT NOUN
all
165
NYT95
all of whole
Inanimate
Animate
86
14
82
18
the
whole
of
74
100
26
0
Proper
Inferred
4
10
11
7
8
18
Total
100
100
Total pop.
458
152
IND95
all of whole
all
67
33
71
29
the
whole
of
85
90
15
10
0
0
22
11
4
25
4
11
1
9
100
100
100
100
100
100
3081
14
403
49
3510
210
Table D3:2. Animacy in NPs with both, /THE DEFINITE ARTICLE/ and
PLURAL NOUN
Inanimate
Animate
NYT95
both both of
the
the
56
43
76
24
44
56
IND95
both both of
the
the
49
54
68
32
51
46
Proper
Inferred
10
14
25
19
47
9
22
10
32
19
32
14
100
100
100
100
100
100
5531
19
60
2607
97
28
both
Total
Total population
both
Table D3:3. Animacy in NPs with all/whole, /THE DEFINITE ARTICLE/ and
GEOGRAPHICAL NAME
NYT95
all of
all
Inanimate
Inf. animate
Total
54
46
100
Total pop.
55
165
the
whole of
82
76
24
18
100
100
258
21
IND95
all of
all
51
49
100
45
the
whole of
61
75
39
25
100
100
59
148
This is the only case in the study where a correlation occurring in just one of the corpora is reported in a table rather than in the running text. The reason for the inclusion of the table is that
the figures corroborate a region specific correlation hypothesis.
207
207
Table D3:4. Animacy in NPs with both /POSSESSIVE DETERMINER/ and
NOUN FOR BODY PARTS AND KINSHIP
both
NYT95
both my
Inanimate
Animate proper
Total
62
38
100
29
71
100
both of
my
25
75
100
Total pop.
322
61
30
D4.
IND95
both my
both
73
27
100
32
68
100
both of
my
0
100
100
224
73
1
Natural and arbitrary time division
Table D4:1. Natural and arbitrary time division in NPs with all/whole, /THE
DEFINITE ARTICLE/ and TEMPORAL NOUN
NYT95
IND95
all
the whole
all
the whole
Natural
85
77
87166
89
Arbitrary
13
11
15
23
100
100
100
100
Total
Total population
D5.
812
103
1276
110
Type of “head”
Table D5:1. Type of head in NPs with half, THE INDEFINITE ARTICLE and
SINGULAR NOUN OR NUMERAL
Nouns for time/space
Partitive nouns
Numerals etc.
Others
Total
Total population
166
208
NYT95
half a/an
a half
34
39
4
10
38
44
18
13
100
100
1841
1254
IND95
half a/an
a half
32
18
9
15
12
36
23
55
100
100
2136
127
The correlation was only significant at the <0.05 level.
208
D6.
Modifiers
Table D6:1. The presence of modifiers in NPs with all/whole, DETERMINER
and SINGULAR COUNT NOUN
NYT95
IND95
all
all of whole
the
all all of whole
the
whole
whole
of
of
No modifier
62
60
57
40
91
70
76
74
Modifier
30
40
24
26
60
9
43
38
6
3
0
28
5
5
13
17
0
20
10
10
13
8
3
33
4
6
8
15
3
42
12
6
Total
100
100
100
100
100
100
100
100
Total pop.
458
152
3081
14
403
49
3510
210
(a) premodifier
(b) postmodifier
(c) both
Table D6:2. The presence of modifiers in NPs with all/whole, DETERMINER
and MASS NOUN
NYT95
IND95
all
all of whole
all
all of whole
No modifier
59
59
45
57
81
80
Modifier
19
19
44
41
41
55
(a) premodifier
(b) postmodifier
(c) both
Total
Total population
18
20
3
25
15
1
3
16
0
13
33
9
23
18
3
2
16
1
100
100
100
100
100
100
5856
515
73
5242
79
92
209
209
Table D6:3. The presence of modifiers in NPs with all, DETERMINER and
PLURAL NOUN
all
No modifier
Modifier
(a) premodifier
(b) postmodifier
(c) both
IND95
all
all of
35
65
55
45
33
20
6
32
14
4
31
24
10
22
10
13
100
100
100
100
12732
2254
10789
384
Total
Total population
NYT95
all of
41
50
50
59
Table D6:4. The presence of modifiers in NPs with both /THE DEFINITE
ARTICLE/ and PLURAL NOUN
NYT95
IND95
both both both of both
both both of
the
the
the
the
No modifier
50
32
37
29
81
91
Modifier
9
19
51
68
62
71
(a) premodifier
(b) postmodifier
(c) both
Total
Total population
D7.
15
4
0
25
19
7
40
19
9
2
7
0
38
13
11
39
14
18
100
100
100
100
100
100
5531
19
60
2607
97
28
An adjacent of
Table D7:1. The presence of an adjacent of in NPs with all/whole,
DETERMINER and SINGULAR NOUN
NYT95
IND95
all
all of whole
the
all
all of whole
the
whole
whole
of
of
No of
97
80
92
69
90
98
98
88
An adj. of
2
10
2
12
3
20
8
31
(a) in postmod.
(b) preceding
2
1
1
1
12
8
10
0
1
7
0
2
18
13
9
3
Total
100
101
100
100
100
100
100
100
Total pop.
458
152
3081
14
403
49
3510
210
210
210
Table D7:2. The presence of an adjacent of in NPs with all/whole,
DETERMINER and MASS NOUN
NYT95
IND95
all
all of whole
all
all of whole
No of
89
93
82
90
96
99
An adjacent of
1
4
7
11
11
18
(a) in the postmodifier
(b) preceding
Total
Total population
7
4
1
0
7
0
12
6
3
1
10
1
100
100
100
100
100
100
5856
515
73
5242
79
92
Table D7:3. The presence of an adjacent of in NPs with all, DETERMINER
and PLURAL NOUN
NYT95
IND95
all
all of
all
all of
No of
84
80
94
94
An adjacent of
6
6
16
20
(a) in the postmodifier
(b) preceding
(c) both
Total
Total population
6
10
0
3
3
0
4
14
2
2
4
0
100
100
100
100
12732
2254
10789
384
Table D7:4. The presence of an adjacent of in NPs with all and
DEMONSTRATIVE PRONOUN
NYT95
all of
92
98
2
8
100
100
all
No of
An adjacent of (preceding)
Total
Total population
1678
478
IND95
all of
91
100
0
9
100
100
all
2292
260
211
211
Table D7:5. The presence of an adjacent of in NPs with all/whole, /THE
167
DEFINITE ARTICLE/ and GEOGRAPHICAL NAME
NYT95
IND95
all
all of
the
all
all of
the
whole of
whole of
No of
88
88
91
95
98
98
An adjacent of (prec.)
2
12
5
2
12
9
100
100
100
100
100
100
Total
55
Total population
258
21
45
59
148
Table D7:6. The presence of an adjacent of in NPs with both and NOUN FOR
168
BODY PARTS OR KINSHIP
NYT95
both
both of
my
my
96
92
100
0
8
4
both
No of
An adjacent of
(a) in the postmodifier
(b) preceding
(c) both
both
94
6
IND95
both
both of
my
my
95
100
0
5
0
4
0
0
8
0
0
0
0
1
5
0
0
5
0
0
0
0
Total
100
100
100
100
100
100
Total population
322
61
30
224
73
1
D8.
Focus markers
Table D8:1. The presence of focus markers in NPs with all/whole,
DETERMINER and SINGULAR COUNT NOUN
NYT95
IND95
all
all of whole
the
all
all of whole
the
whole
whole
of
of
63
53
No focus
80
92
100
92
98
91
marker
Focus
8
0
20
8
2
9
47
37
marker
100
100
100
100 100
100
100
100
Total
Total pop.
167
168
212
458
152
3081
14
403
49
3510
The correlation was only significant at the <0.05 level.
Again, we have a case of a correlation that was only significant at the <0.05 level.
210
Table D8:2. The presence of focus markers in NPs with all/whole,
DETERMINER and MASS NOUN
NYT95
IND95
all
all of whole
all
all of
No focus marker
83
80
100
95
99
Focus marker
5
0
1
20
17
100
100
100
100
100
Total
Total population
5856
515
73
5242
79
whole
99
1
100
92
Table D8:3. The presence of focus markers in NPs with all, DETERMINER
and PLURAL NOUN
NYT95
IND95
all
all of
all
all of
No focus marker
84
82
91
93
Focus marker
9
7
16
18
100
100
100
100
Total
Total population
12732
2254
10789
384
Table D8:4. The presence of focus markers in NPs with all and PERSONAL
PRONOUN
No focus marker
Focus marker
Total
Total population
NYT95
we all
all of us
87
99
1
13
100
100
4078
863
IND95
we all
all of us
91
99
1
9
100
100
5022
520
Table D8:5. The presence of focus markers in NPs with half, DETERMINER
and PLURAL NOUN
NYT95
IND95
half
half of
half
half of
No focus marker
24
25
40
54
Focus marker
60
46
76
75
100
100
100
100
Total
Total population
845
726
525
213
213
213
Table D8:6. The presence of focus markers in NPs with half, THE
INDEFINITE ARTICLE and SINGULAR NOUN OR NUMERAL
NYT95
IND95
half a/an
a half
half a/an
a half
No focus marker
65
75
88
94 (89)
Focus marker
12
6 (11)
35
25
100
100
100
100
Total
Total population
214
1841
1254
2136
127
214
E
Phi coefficients for the correlations
The following tables show the phi coefficients obtained in the chi-square test for
the correlations between variants and factors analysed in the study (see Section
4.3.4). These form the basis of the relative positioning of correlations with factor
categories within each of the figures in Chapter 8. The possible values range
from 1 (absolute correlation) to 0 (no correlation).
Table E1 presents the phi values for correlations between variants and regions. All correlations with the phi coefficient indicated were significant at the
<0.01 level. Some correlations were significant at the <0.05 level only. These
were also included to indicate that this factor category is potentially interesting
although more likely than the other correlations to have occurred by chance.
Table E1. Phi coefficients for correlations with regions
Variable
Significant
correlations
all + DETERMINER + MASS NOUN
BrE, AusE
all of + DETERMINER + MASS NOUN
AmE
Phi coeff.
.104
. 119
all + DETERMINER + PLURAL NOUN
all of + DETERMINER + PLURAL NOUN
BrE
AmE
.152
.166
both + PLURAL NOUN
both the + PLURAL NOUN
AmE
BrE
.075
.094
both these, my etc. + PLURAL NOUN
both of these, my etc. + PLURAL NOUN
BrE, AusE
AmE
.332
.332
half + DET. + SINGULAR COUNT NOUN
half of + DET. + SINGULAR COUNT NOUN
BrE, AusE
AmE
.111
.111
half + DETERMINER + PLURAL NOUN
half of + DETERMINER + PLURAL NOUN
BrE, AusE
AmE
.151
.151
all + DEMONSTRATIVE PRONOUN
all of + DEMONSTRATIVE PRONOUN
BrE
AmE
.117
.142
both + DEMONSTRATIVE PRONOUN
both of + DEMONSTRATIVE PRONOUN
BrE
AmE, AusE
we/us/they/them all
all of us/them
BrE
AmE
<0.05
<0.05
.144
.094
215
215
Table E1. Phi coefficients for correlations with regions (continued)
we/us/they/them both
BrE
both of us/them
AmE, AusE
.097
.086
half a
a half
BrE, AusE
AmE
.380
.380
all + TEMPORAL NOUN
the whole + TEMPORAL NOUN
AmE
BrE, AusE
0.068
<.0.05
all of + GEOGRAPHICAL NAME
the whole of + GEOGRAPHICAL NAME
AmE
BrE
.440
.442
both + NOUN FOR BODY PARTS/KINSHIP
both my + NOUN FOR BODY PARTS ETC.
both of my + NOUN FOR BODY PARTS ETC.
AusE
BrE
AmE
.095
.129
.171
all + DET. + COLLECTIVE NOUN
DET + whole + COLLECTIVE NOUN
BrE, AusE
AmE
.245
.204
Table E2 shows the few correlations found between media and different
variants. Most of these were only observed in one of the corpora. The reason for
including tables anyhow was (as described in Section 2.1.2) that they generally
comprised much larger frequencies than the 200-token samples used for the analysis of linguistic factors; consequently, they could be expected to be more reliable.
Table E2. Phi coefficients for correlations with media
Variable
Significant
correlations
all the + SINGULAR COUNT NOUN
spoken AmE
Phi coeff.
.209
all this/my + SINGULAR COUNT NOUN
all this/my + SINGULAR COUNT NOUN
spoken BrE
written AmE
.301
.156
both of these + PLURAL NOUN
spoken
.208
a half
written AmE
.048
the whole + TEMPORAL NOUN
spoken AmE
.158
both + NOUN FOR BODY PARTS/KINSHIP
written
.095
The values for correlations between variants and linguistic factors are presented
in table E3. In cases where there were more than one correlation, these are
ranked according to their phi coefficients. Correlations that were only significant
at the <0.05 level were placed at the bottom of the particular ranking list.
216
216
Table E3. Phi coefficients for correlations with linguistic factors
Variable
Significant
Phi coeff.
correlations
all + DET. + SINGULAR COUNT NOUN
adverbial function
.527
poss/dem. determiner
.481
no focus marker
.423
divisible noun
.405
no modifier
.264
animate ref. (in BrE)
.233
all of + DET. + SINGULAR COUNT NOUN
DET.
+ whole + SINGULAR COUNT NOUN
a focus marker
a modifier
animate ref. (in BrE)
object function
.423
.264
.233
.179
the definite article
no focus marker
indivisible noun
no modifier
subject function
inanimate ref. (in BrE)
an adjacent of
.441
.423
.301
.264
.250
.233
<0.05
the whole of + DET. + SING. COUNT NOUN
no focus marker
a modifier
inanimate ref. (in BrE)
no adjacent of
.423
.243
.233
<0.05
all + DETERMINER + MASS NOUN
no focus marker
prep. comp. function
a modifier
an adjacent of
object function
.296
.272
.241
.163
.129
all of + DETERMINER + MASS NOUN
a focus marker
a modifier
subject function
no adjacent of
object function
adverbial function
no focus marker
no modifier
an adjacent of
.296
.241
.176
.163
.129
.571
.296
.241
.163
DETERMINER +
whole + MASS NOUN
217
217
Table E3. Phi coefficients for the correlations with linguistic factors
(continued)
all + DETERMINER + PLURAL NOUN
an adjacent of
prep. comp. function
a modifier
no focus marker
.185
.161
.146
.136
all of + DETERMINER + PLURAL NOUN
no adjacent of
subject function
no modifier
a focus marker
.185
.168
.146
.136
both + PLURAL NOUN
no modifier
inanimate reference
.517
.236
both the + PLURAL NOUN
a modifier
animate reference
.517
.236
both of the + PLURAL NOUN
a modifier
animate reference
.517
.236
half + DETERMINER + PLURAL NOUN
no focus marker
possessive determiner
the definite article
genitive
.235
.145
.145
.145
half of + DETERMINER + PLURAL NOUN
a focus marker
demonstr. determiner
.235
.145
all+ DEMONSTRATIVE PRONOUN
an adjacent of
prep.compl. function
.176
<0.05
all of + DEMONSTRATIVE PRONOUN
subject function
no adjacent of
.195
.176
we/us/they/them all
subject function
no focus marker
.280
.211
all of us/them
prep.compl. function
a focus marker
.327
.211
218
218
Table E3. Phi coefficients for the correlations with linguistic factors
(continued)
a focus marker
.265 (.237)
half + INDEF. ART + SING. NOUN or NUM.
prep. comp. function
.170
numeral
<0.05
a half + INDEF. ART + SING. NOUN or NUM.
no focus marker
NP determ. function
partitive noun
all + GEOGRAPHICAL NAME
animate reference
subject function
an adjacent of
.235
.145
<0.05
all of + GEOGRAPHICAL NAME
inanimate reference
object function
no adjacent of
.235
.152
<0.05
the whole of + GEOGRAPH. NAME
inanimate reference
object function
.235
.152
all + TEMPORAL NOUN
adverbial function
natural time division
.732
<0.05
the whole + TEMPORAL NOUN
prep. comp. function
object function
subject function
arbitrary time division
.565
.325
.255
<0.05
both + NOUN FOR BODY PARTS/KINSHIP
inanimate reference
prep. comp. function
an adjacent of
.378
.239
<0.05
both my + NOUN FOR BODY PARTS/KINSHIP
animate reference
subject function
an adjacent of
.378
.320
<0.05
both of my + NOUN F. BODY PARTS/KINSHIP
animate reference
subject function
no adjacent of
.378
.320
<0.05
169
.265 (.237)
.146169
<0.05
In this linguistic factor, the significant correlation concerns tokens where a half century used as a
cricket were excluded. Otherwise, the correlation only occurred in the American material.
219
217
219
220
References
Primary sources
The British National Corpus (BNC)
The Independent on CD-ROM, 1995 (IND95)
The Longman Spoken American Corpus (LSAC)
The New York Times on CD-ROM, 1995 (NYT95)
The Sydney Morning Herald on CD-ROM, 1995 (SMH95)
Secondary sources
Aarts, Jan. 1976. The description of linguistic variation in English: From Firth till the
present day. English Studies, 57:1/6. (239–251)
Abney, Steven. 1987. The English noun phrase in its sentential aspect. Cambridge, Mass:
MIT.
Aijmer, Karin. 1985. What happens at the end of our utterances? The use of utterancefinal tags introduced by and and or. In Togeby, Ole (ed.). Papers from the Eighth
Scandinavian Conference of Linguistics. København: Institut for Nordisk Filologi.
(366–389)
Aijmer, Karin. 1996. Conversational routines in English. London: Longman.
Aldridge, Maurice. 1982. English quantifiers. A study of quantifying expressions in
linguistic science and modern English usage. Amersham: Avebury.
Algeo, John. 1988. British and American grammatical differences. International Journal
of Lexicography, Vol. 1, No. 1. (1−31)
Allan, Keith. 1999. The semantics of English quantifiers. In Collins, Peter & David Lee
(eds.). The clause in English: In honour of Rodney Huddleston. Amsterdam: John
Benjamins. (1–31)
Aston, Guy & Lou Burnard. 1998. The BNC handbook: Exploring the British National
Corpus with SARA. Edinburgh: Edinburgh University Press.
Bach, Emmon. 1989. Informal lectures on formal semantics. New York: State University
of New York Press.
Barrett, David. 1953. Totality and determination. Neuphilologischer Mitteilungen. Vol.
54. (134–144)
Barwise, Jon & Robin Cooper. 1981. Generalized quantifiers and natural language.
Linguistics and Philosophy 4. (129–219)
Bauer, Laurie. 1994. Watching English change. London: Longman.
Bell, Allan. 1991. The language of news media. Oxford: Blackwell.
Benson, Morton, Evelyn Benson & Robert Ilson. 1986. Lexicographic description of
English. Amsterdam: John Benjamins.
Berry, Roger. 1997. Determiners and quantifiers. Collins COBUILD English Guides,
volume 10. London: HarperCollins.
221
221
Berry, Roger. 1998. Determiners: a class apart. English Today 53, Vol. 14, No. 1. (27–34)
Biber, Douglas. 1987. A textual comparison of British and American writing. American
Speech, 62, 2. (99–119)
Biber, Douglas. 1988. Variation across speech and writing. Cambridge: Cambridge
University Press.
Biber, Douglas & Edward Finegan. 1991. On the exploitation of computerized corpora in
variation studies. In Aijmer, Karin & Bengt Altenberg (eds.). English corpus
linguistics: Studies in honour of Jan Svartvik. London: Longman. (204–220)
Biber, Douglas, Susan Conrad & Randi Reppen. 1998. Corpus linguistics. Investigating
language structure and use. Cambridge: Cambridge University Press.
Biber, Douglas, Stig Johansson, Geoffrey Leech, Susan Conrad & Edward Finegan. 1999.
Longman grammar of spoken and written English. London: Longman.
Blair, David & Peter Collins (eds.). 2001. English in Australia. Amsterdam: John
Benjamins.
Bod, Rens, Jennifer Hay & Stefanie Jannedy. 2003. Probabilistic linguistics. Cambridge,
Mass.: MIT Press.
Bolinger, Dwight. 1977. Meaning and form. London: Longman.
Börjars, Kersti. 1998. Feature distribution in Swedish noun phrases. Oxford: Publications
of the Philological Society.
Bowen, Rhonwen. 2003. Noun complementation in English: A corpus-based study of
structural types and patterns. Göteborg: Göteborg University.
Brown, Gillian & George Yule. 1983. Discourse analysis. Cambridge: Cambridge University Press.
Butler, Susan. 2001. Australian English – an identity crisis. In Collins, Peter & David
Blair (eds.). English in Australia. Amsterdam: John Benjamins. (151–161)
Cambridge international dictionary of English. 1995. Paul Procter (ed.). Cambridge:
Cambridge University Press.
Carden, Guy. 1976. English quantifiers. New York: Academic Press.
Chafe, Wallace. 1994. Discourse, consciousness and time: The flow and displacement of
conscious experience in speaking and writing. Chicago: University of Chicago Press.
Chesterman, Andrew. 1991. On definiteness. Cambridge: Cambridge University Press.
Chomsky, Noam. 1957. Syntactic structures. The Hague: Mouton.
Collins Cobuild English dictionary for advanced learners. 2001. Third edition. John
Sinclair (ed.). London: Collins.
Collins, Peter & David Blair. 2001. Language and identity in Australia. In Collins, Peter
& David Blair (eds.). English in Australia. Amsterdam: John Benjamins. (1– 13)
Comrie, Bernard. 1996. Markedness. In Verschueren, Jef, Jan-Ola Östman, Jan
Blommaert & Chris Bulcaen (eds.). Handbook of pragmatics. Amsterdam: John
Benjamins. (1–13)
Cornbleet, Sandra & Ronald Carter. 2001. The language of speech and writing. London:
Routledge.
Coveney, Aidan. 1996. Variability in spoken French. A sociolinguistic study of
interrogation and negation. Exeter: Elm Bank.
Croft, William. 1983. Quantifier scope ambiguity and definiteness. In Amy Dahlstrom,
Claudia Brugman & Monica Maccaulay (eds.). Proceedings of the Ninth Annual
Meeting of the Berkeley Linguistics Society, 12–13 February 1983. Berkeley: Berkeley
Ling. Soc. (25–36)
Crystal, David. 1995. The Cambridge encyclopedia of the English language. Cambridge:
Cambridge University Press.
Culicover, Peter. 1982. Syntax. Second edition. New York: Academic Press.
Dahl, Östen & Kari Fraurud. 1996. Animacy in grammar and discourse. In Fretheim,
Thorstein & Jeanette Gundel (eds.). Reference and referent accessibility. Amsterdam:
John Benjamins. (47–64)
222
222
van Eijck, Jan. Quantifiers. 1994. In Asher, R.E. & J.M.Y. Simpson (eds.). The encyclopedia of language and linguistics. Oxford: Pergamon Press. (3423–3429)
Estling Vannestål, Maria. 2000. All, the whole, both and half and their Swedish
equivalents in English and Swedish descriptive grammars. In Engdahl, Elisabet &
Kerstin Norén (eds.). Att använda SAG. 29 uppsatser om Svenska Akademiens
grammatik. Göteborgs universitet: Meddelanden från Institutionen för svenska språket
vid Göteborgs universitet. (111–124)
Estling Vannestål, Maria. 2001a. Prepositional variation in American and British English.
In Modiano, Marko (ed.). Studies in Mid-Atlantic English. Gävle: Gävle University
Press. (74–86)
Estling Vannestål, Maria 2001b. A/one half of → half of → half – a case of grammaticalization? Göteborg University. Unpublished paper.
Fillmore, Charles, J. 1992. “Corpus linguistics” or “Computer-aided armchair linguistics”. In Svartvik, Jan (ed.). Directions in corpus linguistics. Berlin: Mouton de
Gruyter. (35–60)
Francis, Gill. 1993. A corpus-driven approach to grammar. Principles, methods and
examples. In Baker, Mona, Gill Francis & Elena Tognini-Bonelli (eds.). Text and
technology. In honour of John Sinclair. Amsterdam: John Benjamins. (137–156)
Francis, Nelson. 1958. The structure of American English. New York: Ronald Press.
Fries, Charles C. 1967. The structure of English. London: Longman.
Givón, Talmy. 1993. English grammar: A function-based introduction. Amsterdam: John
Benjamins.
Gleason, Henry A. 1965. Linguistics and English grammar. New York: Holt, Rinehart
and Winston.
Goldberg, Adele. 1995. Constructions: A construction grammar approach to argument
structure. Chicago: University of Chicago Press.
Görlach, Manfred. 1991. Englishes: Studies in varieties of English 1984-1988.
Amsterdam: John Benjamins.
Grevisse, Maurice. 1993. Le bon usage. Grammaire française. Paris: Duculot.
Gries, Stefan. 2001. A multifactorial analysis of syntactic variation: Particle movement
revisited. Journal of Quantitative Linguistics. Vol. 8, No. 1. (33–50)
Gustafsson, Marita. 1983. Fronting of adverbials in four genres of English. In Jacobson,
Sven (ed.). Papers from the Second Scandinavian Symposium on Syntactic Variation,
Stockholm, May 15–16, 1982. Stockholm: Almqvist & Wiksell. (7–18)
de Haan, Pieter. 1989. Postmodifying clauses in the English noun phrase. A corpus-based
study. Amsterdam: Rodopi.
Haegeman, Liliane & Jacqueline Guéron. 1999. English grammar: a generative perspective. Malden: Blackwell.
Halliday, M. A. K. 1994. An introduction to functional grammar. Second edition. London: Edward Arnold.
Hargevik, Stieg. 1983. Various factors influencing the choice of auxiliary need in
Present-day English. Jacobson, Sven (ed.). Papers from the Second Scandinavian
Symposium on Syntactic Variation, Stockholm, May 15–16, 1982. Stockholm:
Almqvist & Wiksell. (19–30)
Haspelmath, Martin. 1995. Diachronic sources of ‘all’ and ‘every’. In Bach, Emmond,
Eloise Jelinek, Angelika Kratzer & Barbara H. Partee. Quantification in natural
languages. Dordrecht: Kluwer. (363–382)
Haspelmath, Martin. 1997. Indefinite pronouns. Oxford: Clarendon.
Hawkins, John. 1978. Definiteness and indefiniteness. A study in reference and
grammaticality prediction. London: Croom Helm.
Heine, Bernd. 1997. Cognitive foundations of grammar. Oxford: Oxford University Press.
223
223
Hellberg, Staffan. 1992. Bestämd form: du vet vilken jag menar. In Teleman, Ulf (ed.).
Grammatik på villovägar. Skrifter utgivna av Svenska språknämnden, 73. Solna:
Almqvist & Wiksell. (33–38)
von Heusinger, Klaus. 1997. Salience and definiteness. The Prague Bulletin of
Mathematical Linguistics, 67. (5– 23)
Hoeksema, Jacob. 1996. Floating quantifiers, partitives and distributivity. In Hoeksema,
Jacob (ed.). Partitives. Berlin: Mouton de Gruyter. (57–106)
Hopper, Paul J. 1991. On some principles of grammaticization. In Traugott, Elizabeth C.
& Bernd Heine (eds.). Approaches to grammaticalization, volume 1. Amsterdam: John
Benjamins. (17–35)
Hopper, Paul J. & Sandra A. Thompson. 1980. Transitivity in grammar and discourse.
Language 56. (251–299)
Hopper, Paul J. & Elizabeth C. Traugott. 2003. Grammaticalization. Second edition.
Cambridge: Cambridge University Press.
Huddleston, Rodney. 1984. Introduction to the grammar of English. Cambridge:
Cambridge University Press.
Huddleston, Rodney & Pullum, Geoffrey. 2002. The Cambridge grammar of the English
language. Cambridge: Cambridge University Press.
Hudson, Jean. 1998. Perspectives on fixedness: Applied and theoretical. Lund: Lund
University Press.
Hudson, Richard. 1990. English word grammar. Oxford: Blackwell.
Hundt, Marianne. 1998a. It is important that this (should) be based on the analysis of
parallel corpora. On the use of the mandative subjunctive in four major varieties of
English. In Lindquist, Hans, Staffan Klintborg, Magnus Levin & Maria Estling (eds).
The major varieties of English. Papers from MAVEN 97. (159–176)
Hundt, Marianne. 1998b. New Zealand English grammar, fact or fiction? A corpus-based
study in morphosyntactic variation. Amsterdam: John Benjamins.
Jackendoff, Ray. 1968. Quantifiers in English. Foundations of Language, 4 (1968). (422–
442)
Jacobson, Sven. 1975. Factors influencing the placement of English adverbs in relation to
auxiliaries. Stockholm: Almqvist & Wiksell.
Jacobson, Sven. 1980a. Issues in the study of syntactic variation. In Jacobson, Sven (ed.).
Papers from the Scandinavian Symposium on Syntactic Variation, Stockholm, May
18–19, 1979. Stockholm: Almqvist & Wiksell. (23–36)
Jacobson, Sven (ed.). 1980b. Papers from the Scandinavian Symposium on Syntactic
Variation. Stockholm: Almqvist & Wiksell.
Jacobson, Sven. 1982. Syntactic variation. Theory and practice. Umeå Papers in
Linguistics, no. 3.
Jacobson, Sven. 1983. Papers from the Second Scandinavian Symposium on Syntactic
Variation, Stockholm. Stockholm. Almqvist & Wiksell.
Jacobson, Sven. 1986a. Synonymy and hyponymy in syntactic variation. In Jacobson,
Sven (ed.). Papers from the Third Scandinavian Symposium on Syntactic Variation,
Stockholm, May 11–12, 1985. Stockholm: Almqvist & Wiksell. (7–17)
Jacobson, Sven (ed.). 1986b. Papers from the Third Scandinavian Symposium on
Syntactic Variation. Stockholm: Almqvist & Wiksell.
Jacobson, Sven. 1989. Some approaches to syntactic variation. In Fasold, Ralph &
Deborah Schiffrin. Language change and variation. Amsterdam: Benjamins. (381–
394)
Jespersen, Otto. 1909–49. A modern English grammar on historical principles. Part III:
Syntax, volume 2. Copenhagen: Eijnar Munksgaard.
Jespersen, Otto. 1924. The philosophy of grammar. London: Allen & Unwin.
Jespersen, Otto. 1933. Essentials of English grammar. London: Allen & Unwin.
224
224
Johansson, Stig. 1980. Corpus-based studies of British and American English. In
Jacobson, Sven (ed.). Papers from the Scandinavian Symposium on Syntactic
Variation, Stockholm, May 18–19, 1979. Stockholm: Almqvist & Wiksell. (85–100)
Johnson, Daniel & Tara Sanchez. 2002. Selected papers from NWAV 30. University of
Pennsylvania Working Papers in Linguistics, Vol. 8.3. Philadelphia: University of
Philadelphia Press.
Jörgensen, Nils & Jan Svensson. 1987. Nusvensk grammatik. Malmö: Liber.
Jucker, Andreas. 1992. Social stylistics: syntactic variation in British newspapers. Berlin:
Mouton de Gruyter
Kachru, Braj. 1985. Standards, codification and sociolinguistic realism: The English
language in the outer circle. In Quirk, Randolph & Henry Widdowson (eds.). English
in the world. Teaching and learning the language and literatures. Cambridge:
Cambridge University Press. (11−30)
Kachru, Braj. 1992a. Introduction: The other side of English and the 1990s. In Kachru,
Braj (ed.). The other tongue. English across cultures Second edition. Urbana:
University of Illinois Press. (1–15)
Kachru, Braj. 1992b. Models for non-native Englishes. In Kachru, Braj (ed.). The other
tongue. English across cultures Second edition. Urbana: University of Illinois Press.
(48–74)
Kennedy, Graeme. 1987. Quantification and the use of English: A case study of one
aspect of the learner’s task. Applied Linguistics, Vol. 8, No. 3. (264–286)
Kennedy, Graeme. 1992. Preferred ways of putting things with implications for language
teaching. In Svartvik, Jan (ed.). Directions in corpus linguistics. Berlin: Mouton de
Gruyter. (335–371)
Kennedy, Graeme. 1998. Introduction to corpus linguistics. London: Longman.
Kilgarriff, Adam. 2001. Comparing corpora. International Journal of Corpus Linguistics.
Vol. 6, No. 1. (97–133).
Klégr, Ales. 1987. Whole and determiners. Philologica Pragensia 30:1 (27–35)
Labov, William. 1978. Where does the sociolonguistic variable stop? A response to
Beatriz Lavendera. Sociolinguistic Working Papers 44. Austin: Southwest Educational
Development Laboratory. (43–70)
Labov, William. 1984. Intensity. Meaning, form, and use in context: Linguistic applications. Washington DC: Georgetown University Press (Georgetown University Round
Table on Languages and Linguistics). (43–72)
Langacker, Ronald. 1987. Foundations of cognitive grammar, volume I: Theoretical
prerequisites. Stanford, CA: Stanford University Press.
Langacker, Ronald. 1988. An overview of cognitive grammar. In Rudzka-Ostyn, Brygida.
Topics in cognitive linguistics. Amsterdam: John Benjamins. (3–48)
Langacker, Ronald. 1991a. Foundations of cognitive grammar, volume 2: Descriptive
application. Stanford, CA: Stanford University Press.
Langacker, Ronald. 1991b. Concept, image, and symbol. The cognitive basis of grammar.
Berlin and New York: Mouton de Gruyter.
Langacker, Ronald. 1999. Grammar and conceptualization. Berlin: Mouton de Gruyter.
Lavendera, Beatriz. 1978. Where does the sociolinguistic variable stop? Language in
Society, 7. (171-183)
Leech, Geoffrey. 1989. An A-Z of English grammar and usage. London: Edward Arnold.
Lehrer, Adrienne. 1987. English quantifiers from noun sources. In Steele, Ross & Terry
Threadgold. Language topics: Essays in honour of Michael Halliday. Amsterdam:
John Benjamins. (96–105)
Levin, Magnus. 2001. Agreement with collective nouns in English. Lund: Lund Studies in
English.
225
225
Lindquist, Hans. 2000. Livelier or more lively? Syntactic and contextual factors influencing the comparison of disyllabic adjectives. In Kirk, John (ed.). Corpora galore.
Analyses and techniques in describing English. Amsterdam: Rodopi. (125–132)
Lindquist, Hans & Magnus Levin. 2000. Apples and oranges. On comparing data from
different corpora. In Mair, Christian & Marianne Hundt (eds.). Corpus linguistics and
linguistic theory. Amsterdam: Rodopi. (201–214).
Longman dictionary of contemporary English. 2003. Fourth edition. Harlow: Longman
Dictionaries. [referred to as Longman 2003]
Lyons, John. 1977. Semantics, 2 volumes. Cambridge: Cambridge University Press.
Mair, Christian. 1990. Infinitival complement clauses in English: A study of syntax in
discourse. Cambridge: Cambridge University Press.
Mair, Christian. 1995. Changing patterns of complementation, and concomitant grammaticalisation, of the verb help in Present-day British English. In Aarts, Bas & Charles
Meyer (eds.). The verb in contemporary English: Theory and description. Cambridge:
Cambridge University Press. (258–272)
Mair, Christian.1998. Corpora and the study of the major varieties of English: Issues and
results. In Lindquist, Hans, Staffan Klintborg, Magnus Levin & Maria Estling (eds).
The major varieties of English. Papers from MAVEN 97. (139–157)
Melchers, Gunnel & Philip Shaw. 2003. World Englishes. London: Edward Arnold.
Mindt, Dieter. 1997. Corpora and the teaching of English in Germany. In Wichmann,
Anne, Steven Fliegelstone, Tony McEnery & Gerry Knowles (eds.). Teaching and
language corpora. London: Longman. (40–50)
Minugh, David. 1997. All the language that’s fit to print. In Wichmann, Anne, Steven
Fliegelstone, Tony McEnery & Gerry Knowles (eds.). Teaching and language
corpora. London: Longman. (67–82)
Minugh, David. 2000. You people use such weird expressions: The frequency of idioms
in newspaper CDs as corpora. In Kirk, John (ed.). Corpora galore. Analyses and
techniques in describing English. Amsterdam: Rodopi. (57–71)
Modiano, Marko. 1996. A Mid-Atlantic handbook. Lund: Studentlitteratur.
Moon, Rosamund. 1998. Fixed expressions and idioms in English. Oxford: Clarendon
Press.
Newbrook, Mark. 1992. Unrecognised grammatical and semantic features typical of
Australian English: A checklist with commentary. English World-wide 13. (1–32)
Newbrook, Mark. 2001. Syntactic features and norms in Australian English. In Collins,
Peter & David Blair (eds.). English in Australia. Amsterdam: John Benjamins. (111–
132)
Norstedts engelsk-svenska ordbok. 2000. Third edition. Stockholm: Norstedts.
O’Grady, William. 1982. The syntax and semantics of quantifier placement. Linguistics,
20 (519–539)
Övergaard, Gerd. 1995. The mandative subjunctive in American and British English in
the 20th century. Stockholm: Almqvist & Wiksell.
Oxford advanced learner’s dictionary of current English. 2000. Sixth edition. Hornby,
Albert, Sally Wehmeier & Michael Ashby (eds.). Oxford: Oxford University Press.
Oxford English dictionary (OED). 1933–1986. James A. H. Murray (ed.). Oxford: Oxford
University Press.
Paolillo, John. 2002. Analyzing linguistic variation. Statistical models and methods.
Stanford: Center for the Study of Language and Information.
Payne, John. 1994. Nouns and noun phrases. In Asher, R.E. & J.M.Y. Simpson (eds.).
The encyclopedia of language and linguistics. Oxford: Pergamon Press. (2848–2855)
Peters, Pam. 2001. Corpus evidence on Australian style and usage. In Collins, Peter &
David Blair (eds.). English in Australia. Amsterdam: John Benjamins. (163–178)
226
226
Peters, Pam & Margery Fee. 1989. New configurations: The balance of British and
American English features in Australian and Canadian English. Australian Journal of
Linguistics 9:1. (135–147)
Pettersson, John Sören. 1996. Grammatological studies. Writing and its relation to
speech. Uppsala: Reports from Uppsala University, Department of Linguistics.
Poutsma, Hendrik. 1914. A grammar of Late Modern English for the use of continental,
especially Dutch students. (Part I. The sentence). Nordhoff: Groningen.
Putseys, Yvan. 1985. A modular approach to the grammar of English predeterminers.
Saint Louis University Research Journal of the Graduate School of Arts and Sciences,
16:2 (369–392)
Quirk, Randolph. 1995. Grammatical and lexical variance. London: Longman.
Quirk Randolph, Sidney Greenbaum, Geoffrey Leech & Jan Svartvik. 1985. A comprehensive grammar of the English language. London: Longman.
Reed, Ann. 1996. Partitives, existentials, and partitive determiners. In Hoeksema, Jacob
(ed.). Partitives. Berlin: Mouton de Gruyter. (143–178)
Röhr, Heinz Markus. 1994. Writing. Its evolution and relation to speech. Bochum:
Brockmeyer.
Romaine, Susanne. 1984. On the problem of syntactic variation and pragmatic meaning in
sociolinguistic theory. Folia Linguistica, Tomus 18:3-4. The Hague: Mouton. (409–
437)
Rydén, Mats. 1979. On syntactic variability. Nordic Linguistic Bulletin. Vol. 3, No. 2. (4–
15)
Saeed, John. 1997. Semantics. Oxford: Blackwell.
Sanchez, Tara & Uri Horez. 2003. Selected papers from NWAV 31. University of Pennsylvania Working Papers in Linguistics, Vol. 9.2. Philadelphia: University of
Philadelphia Press.
Sankoff, David. 1978. Linguistic variation: Models and Methods. New York: Academic
Press.
Sankoff, David. 1988. Variable rules. In Dittmar, Norbert, Klaus Mattheier & Ulrich
Ammon (eds.). Sociolinguistics, an international handbook of the science of language
and society, volume. 2. Berlin: Mouton de Gruyter. (984–997)
Sankoff, Gillian. 1973. Above and beyond phonology in variable rules. In Bailey, James
N. & Roger W. Shuy (eds.). New Ways of Analyzing Variation in English. Washington
D.C.: Georgetown University Press. (44–61)
Sapir, Edward. 1930. Totality. (Language Monographs VI). Baltimore: The Linguistic
Society of America/Waverly Press.
Seppänen, Aimo & Ruth Seppänen. 1986. Notes on quantifiers and definiteness in
English. Studia Neophilologica 58:2 (169–187)
Sinclair, John, 1991. Corpus, concordance, collocation. Oxford: Oxford University Press.
Sinclair, John. 1997. Corpus evidence in language description. In Wichmann, Anne,
Steven Fliegelstone, Tony McEnery & Gerry Knowles (eds.). Teaching and language
corpora. London: Longman. (27–39)
Strevens, Peter. 1978. British and American English. London: Cassel.
Stubbs, Michael. 1996. Text and corpus analysis. Oxford: Blackwell.
Svartvik, Jan & Olof Sager. 1996. Engelsk universitetsgrammatik. Second edition.
Stockholm: Almqvist & Wiksell.
Svensson, Patrik. 1998. Number and countability in English nouns. An embodied model.
Uppsala: Swedish Science Press.
Swan, Michael. 1995. Practical English usage. Second edition. Oxford: Oxford University Press.
Sweet, Henry, 1898. A new English grammar, logical and historical. (Part 2: Syntax).
Oxford: Clarendon Press.
227
227
Teleman, Ulf, Staffan Hellberg & Erik Andersson (eds.). 1999. Svenska akademiens
grammatik. Stockholm: Norstedts Ordbok.
Thompson Sandra A. & Anthony Mulac, 1991. A quantitative perspective on the grammaticization of epistemic parentheticals in English. In Traugott, Elizabeth C. & Bernd
Heine (eds.). Approaches to grammaticalization, volume 2. Amsterdam: John Benjamins. (313–329)
Tognini-Bonelli. Elena. 2001. Corpus linguistics at work. Amsterdam: John Benjamins.
Tottie, Gunnel. 1991. Negation in English speech and writing. A study in variation. San
Diego: Academic Press.
Tottie, Gunnel. 2002. An introduction to American English. Malden: Blackwell
Publishers.
Trudgill, Peter & Jean Hannah. 2002. International English. Fourth edition. London:
Edward Arnold.
Wales, Katie. 1996. Personal pronouns in Present-day English. Cambridge: Cambridge
University Press.
Wierzbicka, Anna. 1996. Semantics. Primes and universals. Oxford: Oxford University
Press.
Winford, Donald. 1996. The problem of syntactic variation. In Sociolinguistic variation:
data, theory and analysis. Selected papers from NWAV 23 at Stanford. Arnold,
Jennifer, Renee Blake, Brad Davidson, Norma Mendoza-Denton, Scott Schwenter &
Julie Solomon (eds.). Stanford: CSLI Publications. (177–192)
Yamamoto, Mutsumi. 1999. Animacy and reference. Amsterdam: John Benjamins.
228
228
Index
Aarts, Jan, 18
Abney, Steven, 43
academic prose, 34, 85
active voice, 21, 164
adjacent of, 96, 129–130, 154–157, 168
adjective, 25, 31, 39, 41, 44, 47, 49, 50,
53, 55–56, 63, 150, 170–171
adjective phrase, 42, 164
adjunct, 42, 163–164
adverb, 31, 39, 47, 50, 58, 71, 73, 150
adverb phrase, 42, 164, 171
adverbial, 23, 31, 33, 41, 65, 73, 82,
107, 147, 164
affix, 23, 34
agreement, 29, 32, 77, 79–81, 121, 139
Aijmer, Karin, 25
Aldridge, Maurice, 47, 49, 61, 67, 68,
73, 76, 77
Algeo, John, 28
Allan, Keith, 68, 70
American English (AmE), 22, 23, 27–
30, 34, 48, 85–86, 91, 115–128
anaphora, 46
animacy, 32, 79–81, 96, 141–145, 173
any, 21, 47, 48, 59
apposition, 24, 67, 73, 108, 164–165,
170–171, 199
approximation, 39, 58, 64, 157–158,
161–162, 194
arbitrary time division, 81, 145–147
Aristotle’s syllogism, 47
armchair linguistics, 84
Aston, Guy , 86
Australian English (AusE), 27–30, 48,
85–86, 96, 115–128
auxiliary, 23, 29, 31, 33, 72, 75
Bach, Emmon, 48
bare NP, 40, 43
Barrett, David, 77
Barwise, Jon, 47–48
Bauer, Laurie, 27, 33
Bell, Alan, 85, 89–90, 95–96
Benson, Morton, 115, 118, 120, 121
Berry, Roger, 17n, 70–71, 72–73, 103,
115, 118–119, 121, 148–149
Biber, Douglas, 13, 14, 20, 21, 22, 29,
30–31, 37, 38, 39, 40, 42, 83, 85, 86,
122, 127, 165, 192
Blair, David, 27, 29
bleaching, 53
Bloomfield, Leonard, 30
Bod, Rens, 19, 24n, 26, 27
body part, 71, 78, 87, 111–112, 120,
126–127, 145, 175–176, 191
Bolinger, Dwight, 21, 34
Börjars, Kersti, 39, 43
Bowen, Rhonwen, 41
British English (BrE), 27–30, 34, 85–
86, 89, 115–128
British National Corpus, 54n, 85–86,
203
broken quantity, 76
Brown corpus, 34, 85n
Brown, Gillian, 14n
Burnard, Lou, 86
Butler, Susan, 29
Carden, Guy, 48
Carter, Ronald, 30
Cedergren, Henrietta, 19
central determiner, 40, 55–56, 131,
133–135
Chafe, Wallace, 30
Chesterman, Andrew, 44–46, 65
chi-square test, 27, 92–95, 115, 132,
215
Chomsky, Noam, 26, 83–84
classifying genitive, 40n
clitic, 34
Cobuild Direct Corpus, 54
cognitive grammar, 21–22, 48, 60, 69,
82
cognitive semantics, 22, 48
collective noun, 29, 32, 75, 78, 79–81,
121, 139, 141–142
Collins, Peter, 27, 29
collocation, 25
common case, 42
229
229
comparison , 31, 58, 162
complementation, 28, 31, 41–42, 149
compound noun, 41n, 42
Comrie, Bernard, 73
concordancing, 31
constant referent, 44
construction grammar, 22
continuous quantity, 76
conventionalisation, 22, 119, 149, 194
conversation, 30–31, 39, 86, 95, 122
conversational formula, 25
conversational routine, 25
Cooper, Robin, 47–48
Cornbleet, Sandra, 30
corpus linguistics, 17, 83–84, 95
corpus-based, 14, 37, 83–84
corpus-driven, 84
correlation, 22–23, 24, 27, 91–95, 131–
133, 179–180
co-text, 14
countability, 77–79, 137–138
Coveney, Aidan, 19, 20n, 33
Croft, William, 44, 48
Crystal, David, 28, 119
Culicover, Peter, 47
Dahl, Östen, 79–80
de Haan, Pieter, 37–39, 43, 158n
de Saussure, Ferdinand , 30
de-categorialisation, 53
deep structure, 21, 72
definite article, 19, 29, 33, 40, 45, 48,
51, 55, 56–57, 59, 61, 67, 70–71, 87,
88, 133–135
definite determiner, 52, 59, 62, 66, 70,
80n
definite pronoun, 59
definiteness, 43, 45–46, 48, 58–60
delayed pronoun, 73
demonstrative determiner, 55, 133–135
demonstrative pronoun , 45, 59, 105–
107, 117, 130–131, 136–137 138,
156, 161, 168–169, 186
descriptive, 17, 27, 37, 38
determined NP, 40
determiner, 33, 37, 38–39, 40–41, 42,
43, 45–46, 47, 50, 51–53, 54–55, 56,
57, 59–60, 62–63, 65, 66, 70, 71, 77,
133–135
dialect, 22, 26, 28–29, 85, 115–128
dictionary, 27, 50, 83, 168n, 200, 204
direct object, 72, 79, 174
discrete quantity, 76
distributive meaning, 45, 76–77, 102
230
disyllabic adjective, 31
divisibility, 77–78, 96, 138–140
DP hypothesis, 43
dual, 62–63, 66, 76n
each, 17, 21–22, 47, 68, 72–73, 77n
either, 17, 76n
elicitation test, 19, 48, 194
entire, 15n, 49n
Estling Vannestål, Maria, 29, 32, 33,
54, 59, 122
every, 21–22, 47, 77n
exception, 58, 64, 158
existential quantification, 47
expanding circle, 28
factor analysis, 27
factor category, 22, 94–95, 179–180
factor strength, 92–94, 132, 179–180
Fairfax, 204
fast genre, 34
Fee, Margery, 29, 122
figurative meaning, 25, 200
Filemaker Pro, 91
Fillmore, Charles, 84
Finegan, Edward, 14
fixed expression, 25–26, 65, 89, 112–
113, 200–202
floating quantifier, 57–58, 72–75, 89,
107–107, 118, 193
FLOB corpus, 34, 85
focus marker, 63, 157–163
formal semantics, 44, 45, 47–48
formality, 28, 56n, 122
fraction, 55, 60, 66, 77
frame, 26, 154, 176
Francis, Gill, 84
Francis, Nelson, 38n
Fraurud, Kari, 79–80
Freeway, 203
French, 17, 27, 45
frequency, 14, 19, 28, 54, 73, 83
Fries, Charles C., 38n
Frown corpus, 34, 85
function word, 15n, 34, 50, 96
functional grammar, 22
fused-head construction, 40, 66
generalized quantification, 47
generative grammar, 26, 48, 47, 67, 72
generic reference, 44, 70, 75, 88, 137,
202n
genitive case, 40, 41, 45, 67, 88, 133–
135, 150n, 164
230
geographical name, 55, 79–81, 109–
110, 119, 125, 143–145, 150, 173,
189, 193
Givón, Talmy, 22, 38n, 70
Gleason, Henry A., 39, 49
Goldberg, Adele, 22
Görlach, Manfred, 28
Government and Binding, 43
grammatical synonymy
, 13,
22, 68, 194
grammaticalisation, 34, 50, 52–54
Grevisse, Maurice, 45
Gries, Stefan, 27, 31
Grimm, Jacob, 30
grounding, 48
Guéron, Jacqueline, 73
Gustafsson, Marita, 31
Haegeman, Liliane, 73
half century, 161, 172, 189n, 219n
Halliday, Michael A. K., 30, 70
Hannah, Jean, 13, 28
Hargevik, Stieg, 31
Haspelmath, Martin, 53, 59, 75
Hawkins, John, 44
head of noun phrase, 37, 40, 41–42, 43,
45, 48, 49, 52, 53, 57, 58, 66–68, 71,
72, 74, 75, 88, 107–108, 117, 136–
149
headedness, 43, 66–68
Heine, Bernd, 71
Hellberg, Staffan, 46, 58
Helsinki Corpus, 52, 54
Hoeksema, Jacob, 72–73
Hopper, Paul J., 34, 53–54, 173
Horez, Uri, 19f
house style, 95
HPSG, 43
Huddleston, Rodney, 27, 39–47, 54–55,
57–62, 66, 69, 73, 76n, 77–79,
81,121, 127, 141–142, 163–164,
165n
Hudson, Jean, 14, 25, 50, 69, 75, 81,
145–147
Hudson, Richard, 43n, 61, 75
Hundt, Marianne, 27, 28, 29
inalienable possession, 71
indefinite article, 40, 52–53, 55, 56–57,
88
indefinite pronoun, 59
Independent, The, 23n, 85, 90, 96, 115,
201n, 203
indirect object, 79, 164, 173
individuation, 80n, 138, 175
inferencing, 47
inferred animacy, 79–80, 143–145
informational exposition, 31
inner circle, 28
intensity, 64–65
interactive style, 31
interplay of factors, 26–27
intonation, 48
intrinsic definiteness, 60
intuition, 14, 31, 83–84
involved style, 29, 31
Jackendoff, Ray, 67
Jacobson, Sven, 19, 20, 23, 24, 26, 31,
33
Jespersen, Otto, 53, 57, 59–60, 63, 66,
71, 73, 74n, 77, 127, 202n
Johansson, Stig, 28
Johnson, Daniel, 19
Jörgensen, Nils, 59
Jucker, Andreas, 28
Kachru, Braj, 27, 28
Kennedy, Graeme, 14, 47, 83, 84, 85
Kilgarriff, Adam, 95
kinship, 71, 87, 111–112, 120, 126–
127, 145, 175–176, 191
Klégr, Andres, 77, 78n
knock-out effect 23–24, 71, 75, 88, 99,
103, 108, 109, 111, 170, 180, 199
Labov, William, 19, 20, 26, 34, 64–65,
75, 137
Langacker, Ronald, 21–22, 48–49, 66,
68–69, 119, 149, 194
language change, 27, 33–34
large-quantity meaning, 62–65, 68, 158,
168–169, 194
Lavendera, Beatriz, 20
layering, 34, 53
learner corpus, 195
Leech, Geoffrey, 55, 70, 103
Lehrer, Adrienne, 47, 50, 104
Levin, Magnus, 29, 32, 79–81, 95, 121,
141, 192
lexical factor, 27, 31–32
lexical variation, 83
lexis, 26, 28, 29, 31–32
limiter, 39
Lindquist, Hans, 31, 95
linguistic factor, 17, 31–33, 129–177
linguistic patterning, 14
LOB corpus, 34, 85n
logic, 44, 47–48, 77
logical use, 46
231
231
Longman Spoken American Corpus
(LSAC), 54n, 85–86, 203–204
Lyons, John,43, 44–45, 46, 65n
Mair, Christian, 28, 29, 31, 33–34, 86
markedness, 50, 73, 74, 108
mass noun, 39, 40, 43n, 45, 51n, 56, 70,
72, 76, 77–78, 100n, 137–138
matrix NP, 67, 88, 171
meaning constant, 20
measurement, 148
medium, 13, 14, 30–31, 34, 85, 95,
122–127, 216
Melchers, Gunnel, 27–28
metonymy, 79
Middle English, 52
Mindt, Dieter, 83
Minimalist theory, 43
Minugh, David, 85n, 96
mode, 30n
Modiano, Marko, 28
modification, 28, 41–42, 149–154, 162
Moon, Rosamund, 25–26
Mulac, Anthony, 54
natural time division, 81, 145–47
negation, 23, 27, 30, 31, 32, 43, 47, 48,
64, 95, 175
negator, 158, 161
neither, 17
New York Times, The, 23n, 85, 90, 95–
96, 115, 201n, 204
New Zealand English, 27
Newbrook, Mark, 29–30, 48, 60n
newspapers, 31, 34, 85–86, 89–90, 95–
96, 111, 115, 199, 203–204
no, 27, 32, 60, 65n
nominal, 40, 43, 48
non-count, 76n
non-linguistic factor, 27–31, 115–128
non-referring expression, 44, 70
non-restrictive relative clause, 41n
non-standard English, 17, 48, 95, 107n,
109, 127
normalised frequencies, 91n, 95, 131n
NP complementation, 41–42, 149
NP dependent, 38
null hypothesis 91, 132
number, 130–131, 137–138
numeral, 40, 47, 53, 67, 69, 148–149,
158
NWAV(E), 19
O’Grady, William, 72–73
object, 39, 42, 43, 72, 74–75, 77, 82,
94, 163–176
232
of-dropping, 70
Old English, 50, 52
orthography, 28, 42
outer circle, 28
Övergaard, Gerd, 28, 29
Oxford English Dictionary (OED), 50–
54, 99, 124–125
Paolillo, John, 26
partitive, 54, 60, 66, 67, 68, 69, 73,
148–149, 162, 171
Partitive Constraint, 66
passive voice, 21, 29, 31, 164
Payne, John, 43
personal pronoun, 24, 52, 57, 74–75,
81, 89–90, 107–108, 117–118, 130,
169–170, 187, 193
Peters, Pam, 29, 122
Pettersson, John Sören, 30
phi coefficient, 93, 132, 180, 215–219
phonetics, 28, 30
phonology, 19, 20, 62
phrasal verb, 25, 27, 31
polysemy, 21n
population, 80–81, 89, 91–92
possessive determiner, 33, 45, 71, 87,
133–135
possessive pronoun, 29
postdeterminer, 40, 56, 58
postmodification, 39, 41, 66, 68, 70, 89,
141, 150–154
post-positioning, 73
Poutsma, Hendrik, 53, 67–68, 73n
pragmatic, 20, 21, 34, 61, 64–65, 66n
predeterminer, 40, 49, 55, 65
predicative, 25, 39, 42, 56, 58, 74, 75,
80n, 163
premodification, 23, 39, 41–42, 150–
154
preposition, 15n, 33, 66, 68, 163n, 167,
169, 170
prepositional complement, 74, 75, 154,
163–176, 193
prepositional variation, 29, 32–33
probability, 24, 26
pronominal head, 55
pronoun, 21, 24, 29, 31, 32, 33, 39, 40,
43, 45–46, 49, 50, 51, 52, 54, 55, 57,
58, 59, 60, 62, 67, 73, 74–75, 77, 80,
81
proper noun, 39, 88, 133n
proportional sample, 90
propositional content, 17, 20, 26, 34
Proquest, 204
232
prototype, 79
proverb, 25
pseudo totaliser, 61
Pullum, Geoffrey, 27, 39–47, 54–55,
57–62, 66, 69, 73, 76n, 77–79,
81,121, 127, 141–142, 163–164,
165n
Putseys, Yvan, 56, 57, 58, 63, 70, 76,
101, 125
quantification, 14, 44, 47–49, 75, 78
quantifier, 21, 47–49, 50–81
quantifier scope, 48
Quirk, Randolph, 14, 15n, 19, 27, 28,
37, 38–42, 46n, 55–56, 58–59, 62,
66, 70–73, 75, 77–78, 80–81, 84,
86–88,99, 101, 102, 107, 109–110,
122–125, 136–137, 139–141, 143–
144, 150n, 158n, 163n, 164, 175
reanalysis, 53
Received Pronunciation, 28
redundancy, 61–63, 68, 73, 127
Reed, Ann, 60, 66–68, 73
reference, 44–46, 48–49, 69, 70, 71, 75,
88, 137, 141–145, 199
reference grammar, 13, 14, 15, 28, 37,
38, 58, 63, 74, 80, 86, 99, 193–194
reference mass, 48–49, 69
referent, 41n, 44–46, 48–49, 59, 75n,
80–81, 140
referring expression, 37
reflexive pronoun, 59
region, 13, 14, 22, 27–30, 34, 48, 85,
95–96, 122–127, 215–216
register, 13, 14, 26, 20–31, 34, 39, 86,
127
relational dependency, 46
representativeness, 83–84
restricted reference, 71, 199
Röhr, Heinz Markus, 30
Romaine, Susanne, 20, 26
Rousseau, Pascal, 27
Rydén, Mats, 18, 19, 26
Saeed, John, 44
Sager, Olof , 28, 57, 127
salience, 46, 71n
sampling, 89–91
Sanchez, Tara, 19n
Sankoff, David, 19, 24, 26, 27
Sankoff, Gillian, 19
Sapir, Edward, 21, 30, 59, 60, 61, 66,
69, 77, 194
SARA, 203
scalar implicatures, 47
school grammar, 14, 15, 58, 63, 86
scope of negation, 30
search program, 90, 96, 203–204
semantic factor, 32–33
semantic primitive, 48, 157
Seppänen, Aimo, 67, 70, 103
Seppänen, Ruth, 67, 70, 103
Shaw, Philip, 27, 28
significance test, 92–94
simile, 25
Sinclair, John, 14, 15n, 31, 68, 83
slot-and-filler analysis, 38–39
sociolinguistics, 19–20, 27, 34
some, 29, 47–48, 60, 62–63
specific reference, 44–45, 70, 88
specifying genitive, 40, 45, 133
speech, 22, 27, 30–31, 59, 86, 122–127
standard English, 17, 27–29, 48, 87
stratified sample, 90–91
stress, 42, 62, 63n, 66n, 78, 136
Strevens, Peter, 28
Stubbs, Michael, 84
style, 26, 27, 31, 86, 95–96
subject, 33, 39, 42, 72, 75, 79, 82,
107n, 163–176
subject determiner, 42, 163–164
subjunctive , 28, 29, 45
supplement , 42, 164
Svartvik, Jan, 28, 57, 127
Svensson, Jan, 59
Svensson, Patrik, 50, 77
Swan, Michael, 17n, 70, 75, 77, 80, 84,
115–117
Swedish, 17, 22, 27, 59, 62, 63, 74,
80n, 195
Sweet, Henry, 30, 65n, 76
Sydney Morning Herald, The, 85, 96,
115, 201n, 204
synchronic variation, 34, 70
syntactic factor, 30
syntactic function, 42, 55n, 79, 82, 91,
94, 147, 154, 163–176
syntactic variation, 14, 15, 17, 19–35,
57, 80, 86, 192–195
teaching material, 27, 83
Teleman, Ulf, 59, 62–63, 74
temporal noun, 57, 70, 78, 81–82, 110–
111, 119–120, 139, 145–147, 174–
175, 190–191, 193, 199, 203
that-clauses, 31
theory-neutral, 17, 43
Thompson, Sandra A., 54, 173
Tognini-Bonelli, Elena, 84
233
233
totality meaning, 17n, 61–65, 66n, 68,
82, 158–163, 168, 194
totaliser, 59–61, 80
Tottie, Gunnel, 23, 27, 28, 29, 31, 32,
48
transformation, 21, 49, 50, 67, 70, 72
Traugott, Elizabeth C., 34, 53–54
Trudgill, Peter, 13, 19, 28
truth condition, 20, 21
unbounded noun, 76n
uncountable, 76n
uniqueness condition, 45
unitary meaning, 76–77
universal quantifier, 47, 65, 75
usage guide, 15, 17n, 84
van Eijck, Jan, 47–48
VARBRUL, 27
variable, 15, 19, 20
variable referent, 44
variable rule, 26, 27n
variant, 13, 14, 15
variation continuum, 25
verb complementation, 28, 31
vocative, 42, 163n
von Heusinger, Klaus, 45–46, 71n
Wales, Katie, 62
Wierzbicka, Anna, 22, 48, 157
Winford, Donald, 26
Word Grammar, 43n
Wordsmith, 87, 90, 203–204
World Wide Web, 147, 162n
writing, 29, 30–31, 85, 86, 122–127
Yamamoto, Mutsumi, 79–80, 141
you all, you-all, y’all, 62, 107
Yule, George, 14n
234
234
Acta Wexionensia
Denna bok ingår i en tredje serie Acta Wexionensia. Tidigare arbeten finns listade nedan.
Serie I
I History & Geography (ISSN 0349-0564)
1:1 Lars-Olof Larsson, 1979: Småländsk bebyggelsehistoria: I. Från Vikingatid
till Vasatid. 1:1 Kinnevalds härad. ISBN 91-7636-000-8
1:2 Lars-Olof Larsson, 1980: Småländsk bebyggelsehistoria: I. Från Vikingatid
till Vasatid. 1:2 Allbo härad. ISBN 91-7636-001-8
1:3 Lars-Olof Larsson, 1980: Småländsk bebyggelsehistoria: I. Från Vikingatid
till Vasatid. 1:3 Konga härad. ISBN 91-7636-001-6
1:4 Lars-Olof Larsson, 1981: Småländsk bebyggelsehistoria: I. Från Vikingatid
till Vasatid. 1:4 Norrvidinge och Uppvidinge härader. ISBN 91-7636-006-7
2. Stefan Åhman, 1983: Pottaskebränning i Sverige och Danmark under 1600talet. ISBN 91-7636-007-5
3. Lars-Olof Larsson, 1983: Bönder och gårdar i stormaktspolitikens skugga.
Studier kring hemmansklyvning, godsbildning och mantalssättning i Sverige
1625-1750. ISBN 91-7636-011-3
4. Tommy Book, 1984: Angerdorf och Exklav. Två företeelser i den berlinska
geografin. ISBN 91-7636-015-6
5. Tommy Book, 1987: Belgrad – Beograd – en stadsgeografisk studie, 1987.
ISBN 91-7636-041-5
6. K. Johansson, N. Johansson, L. Runesson & T. Söderblom, 1988: Växjö under det industriella genombrottets tidevarv. ISBN 91-7636-047-4
7. Tommy Book, 1991: Berlin – sönderbrutet och hopfogat. En politiskgeografisk studie. ISBN 91-7636-085-7
8. Lennart Johansson, 1992: Brännvin, postillor och röda fanor. Om folkrörelser,
politik och gammalkyrklighet i sekelskiftets Växjö. ISBN 91-7636-096-2
9. Sven Ola Swärd, 1992: Upplysning, uppbyggelse – bildning och nöje. Kultur i
Växjö från 1840-talet till 1970. ISBN 91-7636-097-0
II Economy & Politics (ISSN 0281-1588)
1. Jan Ekberg, 1983: Inkomsteffekter av invandring. ISBN 91-7636-010-5
2. Raimo Issal, 1984: Att överleva med bidrag. Konsten att leda ett företag när
den kommersiella marknaden inte räcker till. ISBN 91-7636-016-4
3. Jan Ekberg, 1991: Vad hände sedan? En studie av utrikes födda på arbetsmarknaden. ISBN 91-7636-084-9
4. Conny Johannesson, 1994: Etik i förhandlingsstaten. ISBN 91-7636-116-0
III Språk och litteratur (ISSN 0283-8583)
1. Bo Seltén, 1987: Svengelsk ordbok. ISBN 91-22-00844-6
Serie II
Samhällsvetenskap (ISSN 1401-629X)
1. Tommy Book, 1996: Balkanstaden – i ständig förvandling – stadsbyggnadsmässigt och etniskt. ISBN 91-7636-143-8
2. Stig-Arne Mattson, 1999: Effektivisering av materialflöden i supply chains.
ISBN 91-7636-205-1
Humaniora (ISSN 1401-6281)
1. H. Lindquist, S. Klintborg, M. Levin & M. Estling (red.), 1998: The major
varieties of English: papers from MAVEN 97. ISBN 91-7636-194-2
2. Peter Aronsson & Lennart Johansson (red.), 1999: Stationssamhällen. Nordiska perspektiv på landsbygdens modernisering. ISBN 91-7636-149-6
Serie III (ISSN 1404-4307)
1. Installation Växjö universitet 1999. Nytt universitet – nya professorer. 1999.
ISBN 91-7636-233-7
2. Tuija Virtanen & Ibolya Maricic, 2000: Perspectives on Discourse: Proceedings from the 1999 Discourse Symposia at Växjö University. ISBN 91-7636237-X
3. Tommy Book, 2000: Symbolskiften i det politiska landskapet – namn-heraldikmonument. ISBN 91-7636-234-5
4. E. Wåghäll Nivre, E. Johansson & B. Westphal (red.), 2000: Text im Kontext,
ISBN 91-7636-241-8
5. Göran Palm & Betty Rohdin, 2000: Att välja med Smålandsposten. Journalistik och valrörelser 1982-1998. ISBN 91-7636-249-3
6. Installation Växjö universitet 2000, De nya professorerna och deras föreläsningar. 2000. ISBN 91-7636-258-2
7. Thorbjörn Nilsson, 2001: Den lokalpolitiska karriären. En socialpsykologisk
studie av tjugo kommunalråd (doktorsavhandling). ISBN 91-7636-279-5
8. Henrik Petersson, 2001: Infinite dimensional holomorphy in the ring of formal
power series. Partial differential operators (doktorsavhandling). ISBN 91-7636282-5
9. Mats Hammarstedt, 2001: Making a living in a new country (doktorsavhandling). ISBN 91-7636-283-3
10. Elisabeth Wåghäll Nivre & Olle Larsson, 2001: Aspects of the European Reformation. Papers from Culture and Society in Reformation Europe, Växjö 26-27
November 1999. ISBN 91-7636-286-8
11. Olof Eriksson, 2001: Aspekter av litterär översättning. Föredrag från ett
svensk-franskt översättningssymposium vid Växjö universitet 11-12 maj 2000.
ISBN 91-7636-290-6.
12. Per-Olof Andersson, 2001: Den kalejdoskopiska offentligheten. Lokal press,
värdemönster och det offentliga samtalets villkor 1880-1910 (Doktorsavhandling). ISBN: 91-7636-303-1.
13. Daniel Hjorth, 2001: Rewriting Entrepreneurship. Enterprise discourse and
entrepreneurship in the case of re-organising ES (doktorsavhandling). ISBN: 917636-304-X.
14. Installation Växjö universitet 2001, De nya professorerna och deras föreläsningar, 2001. ISBN 91-7636-305-8.
15. Martin Stigmar, 2002. Metakognition och Internet. Om gymnasieelevers informationsanvändning vid arbete med Internet (doktorsavhandling). ISBN 917636-312-0.
16. Sune Håkansson, 2002. Räntefördelningen och dess påverkan på skogsbruket. ISBN 91-7636-316-3.
17. Magnus Forslund, 2002. Det omöjliggjorda entreprenörskapet. Om förnyelsekraft och företagsamhet på golvet (doktorsavhandling). ISBN 91-7636-320-1.
18. Peter Aronsson och Bengt Johannisson (red), 2002. Entreprenörskapets dynamik och lokala förankring. ISBN: 91-7636-323-6.
19. Olof Eriksson, 2002. Stil och översättning. ISBN: 91-7636-324-4
20. Ia Nyström, 2002. ELEVEN och LÄRANDEMILJÖN. En studie av barns lärande med fokus på läsning och skrivning (doktorsavhandling). ISBN: 91-7636351-1
21. Stefan Sellbjer, 2002. Real konstruktivism – ett försök till syntes av två dominerande perspektiv på undervisning och lärande (doktorsavhandling).
ISBN: 91-7636-352-X
22. Harald Säll, 2002. Spiral Grain in Norway Spruce (doktorsavhandling).
ISBN: 91-7636-356-2
23. Jean-Georges Plathner, 2003. La variabilité du pronom de la troisième
personne en complèment prépositionnel pour exprimer le rèfléchi (doktorsavhandling). ISBN: 91-7636-361-9
24. Torbjörn Bredenlöw, 2003. Gestaltning – Förändring – Effektivisering.
En teori om företagande och modellering. ISBN: 91-7636-364-3
25. Erik Wångmar, 2003. Från sockenkommun till storkommun. En analys av
storkommunreformens genomförande 1939-1952 i en nationell och lokal kontext
(doktorsavhandling). ISBN: 91-7636-370-8
26. Jan Ekberg (red), 2003, Invandring till Sverige – orsaker och effekter. Årsbok från forskningsprofilen AMER. ISBN: 91-7636-375-9
27. Eva Larsson Ringqvist (utg.), 2003, Ordföljd och informationsstruktur i
franska och svenska. ISBN: 91-7636-379-1
28. Gill Croona, 2003, ETIK och UTMANING. Om lärande av bemötande i professionsutbildning (doktorsavhandling). ISBN: 91-7636-380-5
29. Mikael Askander, 2003, Modernitet och intermedialitet i Erik Asklunds tidiga
romankonst (doktorsavhandling). ISBN: 91-7636-381-3
30. Christer Persson, 2003, Hemslöjd och folkökning. En studie av befolkningsutveckling, proto-industri och andra näringar ur ett regionalt perspektiv.
ISBN: 91-7636-390-2
31. Hans Dahlqvist, 2003, Fri att konkurrera, skyldig att producera. En ideologikritisk granskning av SAF 1902-1948 (doktorsavhandling). ISBN: 91-7636-393-7
32. Gunilla Carlsson, 2003, Det våldsamma mötets fenomenologi – om hot och
våld i psykiatrisk vård (doktorsavhandling). ISBN: 91-7636-400-3
33. Imad Alsyouf, 2004. Cost Effective Maintenance for Competitive Advantages
(doktorsavhandling). ISBN: 91-7636-401-1.
34. Lars Hansson, 2004. Slakt i takt. Klassformering vid de bondekooperativa
slakteriindustrierna i Skåne 1908-1946 (doktorsavhandling). ISBN: 91-7636402-X.
35. Olof Eriksson, 2004. Strindberg och det franska språket. ISBN: 91-7636-403-8.
36. Staffan Stranne, 2004. Produktion och arbet i den tredje industriella revolutionen. Tarkett i Ronneby 1970-2000 (doktorsavhandling). ISBN: 91-7636-404-6.
37. Reet Sjögren, 2004. Att vårda på uppdrag kräver visdom. En studie om vårdandet av män som sexuellt förgripit sig på barn (doktorsavhandling).
ISBN: 91-7636-405-4.
38. Maria Estling Vannestål, 2004. Syntactic variation in English quantified
noun phrases with all, whole, both and half (doktorsavhandling). ISBN: 91-7636406-2.
Växjö University Press
Växjö universitet
351 95 Växjö
www.vxu.se
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement