4. Covert operations: the meanings of "war" and "violence"
There is a time for everything, and a season for every activity under the heavens: ... a
time to kill and a time to heal, a time to tear down and a time to build ... a time to love
and a time to hate, a time for war and a time for peace.
Ecclesiastes 3
Saussure once wrote that you could dedicate a whole book to the meaning of one
word and still not fully capture its significance. Malinowski made a similar point, writing that
"the elucidation of the 'meaning' of a word does not happen in a flash, but is the result of a
lengthy process" (Malinowski 1935:
44). Two lexical items are the focus for this chapter:
"war" and "violence". These words do not immediately call one's attention: the word "war"
would not be considered to inscribe a particular ideological perspective in accounts of
evaluative lexis (as in e.g. Martin and White, 2005, #711). Indeed, in my study into the
Australian Broadcasting Corporation's coverage of the invasion of Iraq, it took me some time
to see the complexity and significance of these two lexical items to the legitimation of the
Coalition's invasion of Iraq. In hindsight, this is no surprise: from what ideology tells us
about the nature of language, it is clear that significant domains and levels of meaning
operate beneath the surface of our conscious awareness. The elucidating their meanings
requires a number of methodological steps in a process, and at the same time, we need an
understanding of how words mean, such as has been provided to us by Saussure (see chapter
2). The intricacies of the words "war" and "violence" are those of any open system lexical
item: like any other content word, these two words display the intra-sign dependencies
Saussure theorized. Their meanings are a function of their relations to other words, relations
such as synonymy, antonymy, hyponymy, meronymy, or general lexical association i.e.
collocation. It is through these relations that any word - a concept defined in Halliday as "the
product of a large number of classificatory dimensions" (Halliday 2002c: 159) - is enabled to
have referential potential.
I will begin the exploration of these lexical items by reference to standard
lexicographical resources, drawing on the Oxford English Dictionary as a starting point. The
relative sizes of the entries for these two words is itself very revealing. I will also consider
the etymology of these words, and their location within the conceptual architecture of both
the OED thesaurus and Roget's thesaurus. These standard investigations will entail some
attention to the syntagmatic potential of these lexical items. As Saussure notes, “words
© Annabelle Lukin Please do not quote without permission - page 91 -
acquire relations based on the linear nature of language because they are chained together”
(Saussure 1974: 123). Words are only meaningful because they form structures, and lingusts
since at least Saussure's time have recognised that the separation of lexis from morphology or
syntax is untenable. Alongside the concept of "collocation", Firth coined the term
"colligation" to mean the meanings associated with the typical grammatical habits of a
particular lexical item (Firth 1957b: 13). Grammatical meanings are part and parcel of how
lexical items function. It is central to the meaning of "war" that it has long ago left behind its
verbal potential - that is, its capacity to construe an unfolding process. Instead, "war" is a
thing, a nominal element, subject to the kinds of grammatical environments and relations
afforded this particular grammatical category. It is, for example, central to the meaning of
"violence" that its earlier verb forms, it was transitive.
From the examination of these terms in the dictionary and thesaurus, the next step will
be to consider the "behaviour" of these words in large corpora. As noted in Chapter 1, these
two words function as key lexical nodes in the primary corpus of ABC TV news reports
(from 20/03/03-03/04/03). In Firth's terms, they are "key-words, pivotal words, leading
words" (Firth 1968a: 106-107) (see Chapter 2). In my primary corpus, these words behave
in the manner of two like poles on magnetics: they repel each other, and cannot be brought
into proximity. "Violence" in this corpus is not relevant to reporting "the Iraq war". The
dominant environment of use for "violence" is, instead, reports on protests against "the Iraq
war". The ABC journalists reporting on "the war" observed a strong boundary between the
concept of "war" and the concept of "violence". But this practice was not particular to the
ABC journalists reporting this particular instance of "war", as I show below, by examining
larger data sets which include further media reports and official press briefings on the "Iraq
war". Indeed, the discrimination between "war" and "violence" is not particular to the
reporting of this particular war. With many large corpora now available, such as the British
National Corpus, the Google Books Corpus, the corpus of the British Hansard, the Corpus of
Contemporary American English, it is possible to examine patterns of usage of such key
terms across many genres, and across time spans. Large scale corpus analyses confirm the
strong insultation between the categories of "war" and "violence". Reference to two early
modern English corpora (the Early English Books Online Corpus, and the Lampeter Corpus)
will enable me to explore earlier patterns of collocation around these two words.
The final section of this chapter will focus on the colligational patterning of these
words to bring out their typical structural environments. While corpus techniques allow
automated searching of patterns based on part-of-speech terms, deeper grammatical patterns
© Annabelle Lukin Please do not quote without permission - page 92 -
are also central to the development of a "configurative rapport (see Chapter 2). The analysis
of such patterns require the introduction of manual analysis based on larger linguistic units.
In addition, the nominal character of the word such as "war" means that, from the perspective
of a clause, it can function as Subject, Complement or Adjunct. Considering the clause from
an ergative perspective, the term "war" might function as Agent, Medium or Range. These
grammatical environments afford and constrain the way "war" can be construed. In narrower
structural terms, the concept may be located at the centre of a nominal group structure (e.g.
"the Iraq war"), or elsewhere as a pre- or post-modifying element (e.g. "war plans", "the plans
for post-war Iraq"). Again, this kind of analysis requires manual labour, and it will be based
largely on the primary corpus (2 weeks of TV news reports by the Australian Broadcasting
Corporation, from 20/03/03-02/03/03).
(a) "War" in the dictionary and thesaurus
The Oxford English Dictionary (or "OED") traces the history of the modern word
"war" to a Late Old English word (wyrre/werre, c. 1050), coming into Old English from
North Eastern Old French (werre), a word with its own origins in Old High German, meaning
"confusion, discord, strife", and going back to a verbal form in the Old High German,
meaning "to bring into confusion or discord". The OED states in its etymology that although
the origins of the modern word go back to High German, "no Germanic nation in early
historic times had in living use any word properly meaning 'war'", that is, meaning "war" in
our modern sense of the term. Both noun and verb forms are listed in the entry for "war". In
its entry for the verbal form, the earliest citation is to the Anglo-Saxon Chronicle (dated at
1154). The OED describes this verb as transitive, and notes that it is now obselete17. Thus, in
its earlier life, "war" could could take a direct object. The meaning of the Old English verb
form is given, for the modern English reader, as "to make war upon". The modern definition
for this transitive verb requires the use of a preposition (upon), indicating that the modern
concept of war cannot now construe this process in transitive terms. This is a key point about
the meaning of "war" I return to later in this chapter.
The OED defines the modern English noun "war" as:
Hostile contention by means of armed forces, carried on between nations, states, or
rulers, or between parties in the same nation or state; the employment of armed forces
against a foreign power, or against an opposing party in the state.
17. There is a second verbal meaning listed for "war", also transitive, meaning to "To ‘worst’,
defeat in a contest or competition; to surpass, excel". According to the OED, the Google
Books Ngram corpus gives this word a frequency of 0.01-0.1 per million words.
© Annabelle Lukin Please do not quote without permission - page 93 -
Based on the OED citations, this meaning is not quite 900 years old. The earliest use
of the word in its current sense is an entry in the Anglo-Saxon Chronicle, dated to 1122.
a1122 Anglo-Saxon Chron. (Laud) ann. 1116 Se cyng Henri fylste his nefan..þe þa
wyrre hæfde togeanes his hlaforde þam cynge of France.
[Translation: The aforementioned king Henry assisted his nephew who then had war
against his lord the king of France]18
The OED lists the modern word as noun only. A search of the Google Books Corpus
for the word "war" as noun and verb returns no hits for the word as a verb (see Figure 4; note
that the Google Books Corpus requires a minimum of 40 tokens to display an ngram). As a
noun, "war" is both a countable noun and a mass noun. As a countable noun, "war" can have
a "particularized sense" where it means "A contest between armed forces carried on in a
campaign or series of campaigns". Note that the use of the word to mean "Actual fighting,
battle; a battle, engagement" is now obselete. So while the word can and is frequently used
with reference to "particularized" events, these events must be of a certain scale to attract the
designation of "war". As the OED notes, it is frequently used in this particularized sense with
the definite article "to designate a particular war" and "with (an) identifying word or phrase,
as in the Trojan war, the Punic Wars, the Wars of the Roses". As can be seen, instances of the
category of "war" attract names, i.e. are designated through proper nouns. "War" can also be
used metaphorically or rhetorically for "any kind of active hostility or contention between
living beings, or of conflict between opposing forces or principles". The dictionary also notes
that one can "declare", "wage" or "levy" war, and can "make war".
18. Thanks to A/Professor Rosemary Huisman, of the University of Sydney, for this
translation.
© Annabelle Lukin Please do not quote without permission - page 94 -
Figure 4: Google Books corpus showing absence of "war" as a verbal form since
1800.
The modern definition of "war" entails the notion of "nation" or "state". Thus, the
antagonists to "war" are political entities, either representing a political body as a whole, or as
parties internal to a political body. "War" is also defined as action "between" political parties:
entailed in its definition is a notion of action that is bidirectional. This feature of the
definition is echoed an analogy Clausewitz drew on to get at the meaning of war:
we shall do best by supposing to ourselves two wrestlers. Each strives by physical force
to compel the other to submit to his will: each endeavours to throw his adversary, and
thus render him incapable of further resistance (Clausewitz 1873: Chapter 1, Section
1).
Further, both the general and particularized sense of the word make reference to
"armed forces", which, according to the OED are a body of armed men constituting "the
fighting strength of a kingdom or a commander in a field". "War" involves these armed forces
engaging in "hostile contention" on behalf of a political entity. "Hostile contention" is the
defining action of "war". But it itself requires definition. To contend means:
To strive in opposition; to engage in conflict or strife; to fight. Const. with, against (an
opponent), for, about (an object).
The definition of "to contend" reiterates the bidirectional aspect of "war" already
noted. It adds a key part of the definition: that "war" is a striving "for" or "about" something.
Thus, "war" is collective action that is goal-directed, and purposeful. It is fighting "for"
something. The dictionary also reminds us that English allows as to construe "war" as a
locative state. It is possible to be "at war" or "in war". "War" is also construed as location in
© Annabelle Lukin Please do not quote without permission - page 95 -
the phrase "to go to war".
The OED dictionary entry includes a host of compound words in which "war" is a key
component. There are so many that the OED attempts some categorisation of these
compounds. The first category is defined as "simple attributive use", with sense of "of or
belonging to war" or "used or occurring in war", "suited or adapted for war". This category is
further sub-categorised as "general use" (40 compounds), words denoting "arms,
accoutrement, implements" (23 compounds), words denoting a "commander, officer, army"
(7 compounds), words denoting "cries, songs, musical instruments (11 compounds), words
referring to finance (13 compounds), and words denoting "literary or artistic works" (17
compounds). The second category, described by the OED as "objective, etc" (meaning a
compound word in which "war" functions as a kind of object), has 18 compounds. The third
category includes words which construe war as the instrument of some process, and this
category lists 27 compound forms. The final category, described as "special combination"
might perhaps also be described as "not elsewhere listed": it lists a diverse array of 87 further
compound words.
"War", therefore, acts as classifier to an enormous variety of things including family
life ("war-bride", "war wedding"; "war baby"; "war widow"; "war-orphaned"), artistic forms
and types of artist ("war photograph"; "war poet"), instruments of war, forms of attitude
towards war or mental strain/illness ("war-minded"; "war strain"; "war psychosis"; "warweary"; "war fever", "war-loving"; "war hysteria"; "war-guilt"), social institutions pertaining
to war ("War Department", "war hospital", "war cemetery", "war college"; "war memorial"),
the economy or finance associated with war ("war economy", "war chest"; "war expenditure";
"war debt"; "war bond"; "war surplus"), the effort and achievement particular to war ("war
production", "war footing", "war effort", "war service"; "war record", "war purpose"; "war
hero"; "war heroine"), space ("war zone"), time ("war years"; "war period"; "war
generation"), and war-related governance ("war measure"; "war-machine"' "war ration"). In
these compounds, the premodifier "war" creates a hyponym. In other words, each compound
is a more delicate instance of a general cateogory. Thus, "war hysteria" is a more delicate
form of the category "hysteria"; a "war bride" is a particular type of bride. We also find
compounds which construe a reduced form of a transitive relation. For example, in
compounds such as "war-breeder", "war-chronicler", "war-maker", "war-winner" "war" is
medium in a transitive clause. We also find compounds which construe war as agentive. The
forms described by the OED as "instrumental" are an example of this relation. In the
compound adjectives, such as "war-battered", "war-ravaged", "war-scarred", "war-torn",
© Annabelle Lukin Please do not quote without permission - page 96 -
"war-wearied", "war-wracked", "war" is the agent of the processes such as "ravaging",
"scaring", "wearying", "wracking".
From the proliferation of compound forms, it is clear the "war" has considerable
linguistic infrastructure. This degree of infrastructure has evolved through usage, i.e. parole,
and reveals the kind of communal pressure that has been exerted on this lexical item. The
proliferation of terms tells us "war" has considerable reach in the life of English-speaking
communities. That it is an event requiring organisation is formalised in various ways, such as
the names for social institutions which pertain to war, and in the titles that are given for those
with degrees of authority in managing the prosecution of war. That we name particular types
of art forms associated with war is also telling. All this lexical technology indicates that
"war" is not only a routine and naturalised part of English-speaking cultures; it is recorded
and memorialised in various art forms. It is, therefore, deeply institutionalised.
One further resource in developing our picture of the meaning of "war" is the
thesaurus. A thesaurus offers us a semantic classification of a word, in relation to a map of the
semantic fields of a language. In a thesaurus, a word is given a location by virtue of its
proximity to other words. The OED includes a thesaurus, in which it makes an initial
separation between "the external world", "the mind", and "society". "War" is located within
the domain of "society", which includes the following elements: "the community",
"inhabiting or dwelling", "armed hostility", "authority", "morality", "education", "religion",
"communication", "travel", "occupation", "leisure". "War", not surprisingly, falls into the
subcategory of "armed hostility"; but what is noteworthy is that "armed hostility" forms part
and parcel of society. It is of the same order as "morality", "religion", "inhabiting or
dwelling", "education", etc. See Figure 5.
© Annabelle Lukin Please do not quote without permission - page 97 -
Figure 5: "War" in the OED historical thesaurus
(b) "Violence" in the dictionary and thesaurus
The dictionary entry for the word "violence" is considerably shorter than that for the
word "war". The OED defines violence as:
The deliberate exercise of physical force against a person, property, etc.; physically violent behaviour or treatment; (Law) the unlawful exercise of physical force, intimidation
by the exhibition of such force.
The English word derives from an Anglo-Norman and Middle French term. The
earliest citation in the OED is attributed to St. Thomas Beckett in 1300. While "war" is a use
of force, "violence" is the "unlawful exercise of physical force". The word "violence" has
produced a standard set of morphological variants such as "violent", and "violently" (as well
as a couple of obscure other forms, e.g. "violency", "violented"). Like "war" it has some
history in a verbal form. Both "violent" and "violence" are listed as verbal forms, probably
derived from the Middle French form "violenter". Both are transitive constructions, meaning
to "compel, coerce, or constrain", or "to subject (a person) to violence; to attack, assault".
Both are now considered obselete or rare. Unlike the proliferation of compound forms
involving "war", according to the OED there are no compound forms based on "violence".
While the OED suggests the collocations of "domestic", "gang" and "mob" violence, it does
not formally lists these as compound forms. Why does "violence" lack the linguistic
© Annabelle Lukin Please do not quote without permission - page 98 -
infrastructure of "war"? One part of the explanation is likely the fact that "violence" is
inherently illegitimate. This could explain why it does not get used as a reference point for
various forms of human endeavours. Though nominal like "war", it does not have the
capacity as "war" does to be countable, and this naturally removes its potential to be granted
a status of a thing taking a proper noun; it cannot therefore stand is a label for a set of actions
over a particular time, in a particular location, between particular entities.
With respect to the thesaurus, "violence" is located within the general category of "the
external world", and is therefore outside "the social world". The "external world" category is
further divided into the following subcategories: "the universe", "the earth", "the living
world", "sensation", "matter", "abstract properties", "relative properties", and "the
supernatural". The category "abstract properties", contains the following further
subcategories: existence, creation, cause, present events, action or operation, time, space or
extent, motion or movement. Of these, the category of "action or operation", is subdivided
into various types, one of which is "behaviour or conduct". Once again, this category is
further subdivided, and includes the category of "bad behaviour". It is in this category that we
finally locate "violent behaviour". Figure 6 illustrates the location of "violence" within this
section of the OED historical. As we can see from its thesaurus location, violence is therefore
not a thing in itself, but is a type of thing: that is, it is defined as a kind of behaviour or
conduct inherently "bad". To label particular actions as "violence" inherently and
unequivocally delegitimises those actions.
Figure 6: "Violence" in the OED historical thesaurus
(c) "War" and "violence": what do collocational/colligational
© Annabelle Lukin Please do not quote without permission - page 99 -
patterns reveal?
The dictionary definitions illuminate some continuity between "war" and "violence" both are about the use of force - but in every other respect they are distinct. What picture
emerges when we examine these two terms from the perspective of their typical collocations
as well as from their relative frequencies? I turn now to the British National Corpus, as the
most representative corpus of English available. The corpus is multi-generic, including a 90,
000, 000-word balanced written corpus (i.e. with equal samples of a range of written
registers) and a c. 10, 000, 000-word corpus of orthographically transcribed spoken
discourse. The written registers include academic prose, fiction, newspaper discourse,
unpublished letters, school and university essays. The spoken discourse section includes
"context-governed" speech, i.e. samples of dialogic speech from four general domains:
"educational and informative", "business", "public or institutional" and "leisure"19. It also
includes "demographically sampled" speech, i.e. speech from natural conversation by
speakers sampled from different social classes and geographic locations, and with due regard
for differences in gender and age. The BNC can reasonably claim to be the most
representative corpus of a modern language currently available, though it is British English,
and is synchronic (that is, sampled from roughly the same time period, the late 20th century.).
The concept of "collocation" was discussed in Chapter 2, in particular as introduced
by Firth. Collocation, Firth argues, is not "mere juxtaposition", but is rather "an order of
mutual expectancy". Words are mutually expectant and mutually prehended (Firth 1957b:
12); they indicate a "multiplicity of systems derived from carefully contextualized structures"
(Firth 1968c: 18). Firth's conception is widely cited in corpus linguistics20. The advent of
sizeable, searchable corpora has opened up considerable opportunities to advance both
linguistic theory and method in relation to habitual patterning in lexicogrammar, as well as
the pressure to operationalize the notion of collocation for empirical research (Bartsch &
19. For further details of the corpus composition, see natcorp.ox.ac.uk.
20. Bartch and Evert argue in a recent paper that there have been no studies based on a
Firthian view: "there has not, to our knowledge, been any systematic large-scale study resting
on a Firthian notion of collocation. Studies typically take as their vantage point specific types
of multi-word expressions (such as support verb constructions or verb-particle constructions,
e.g. Baldwin 2008), or rely completely on intuitions of annotators (e.g. lexicographers’
judgements). This study tries to avoid such initial assumptions and aims to evaluate
association measures solely based on a Firthian notion of collocation, i.e. it works on the
basis of co-occurrence in context without phraseological or lexicographical assumptions
guiding the experiment." (Bartsch & Evert 2014: 50)
© Annabelle Lukin Please do not quote without permission - page 100 -
Evert 2014: 48-49). Following Firth, Bartsch and Evert define collocation as "the habitual
and recurrent juxtaposition of words with particular other words" (Bartsch & Evert 2014:
48). Baker et al. (Baker et al. 2008: 278) include the concept of a "span" in their definition
of collocation, which is "the above-chance frequent co-occurrence of two words within a predetermined span". Baker et al. note that collocates do not need to be adjacent to the node to
influence the meaning of node word; five words to left and right of the node word is a typical
"collocation window" for the investigation collocational patterns (Baker et al. 2008: 278).
Such a span gives a way of bringing out the "more general associations" of a particular item
under investigation (Brezina, McEnery, & Wattam 2015).
The methods for exploring, measuring and interpreting collocations are the subject of
considerable debate amongst scholars (see e.g. (Bartsch & Evert 2014; Brezina et al. 2015)).
Three basic criteria relevant to the study of collocation are 1. distance from the node word, 2.
frequency of the collocational pattern, and 3. exclusivity, in other words, the degree to which
the collocation relation is exclusive such that these two words more typically go together than
with other words (Brezina et al. 2015). Brezina et al. 2015 note as an example the expression
"in love": while looked at from the perspective of "love" this combination is dominant, when
consider from perspective of "in", the phrase is one of very many possible collocations. In
other words, the relationship between "in" and "love" "is not exclusive" (Brezina et al. 2015:
140). Other criteria discussed (e.g. Gries 2013) include directionality (attraction strength is
rarely symmetrical between two words), dispersion (the spread of the node and its
collocations across a given corpus), and type-token distribution (the "level of competition"
for a slot around a node word from other collocate types). Brezina et al. 2015 add a seventh
feature of collocational patterns, which they describe as the "connectivity between individual
collocates" Brezina et al. 2015: 141. They note that collocates of words "do not occur in
isolation, but are part of a complex network of semantic relationships which ultimately
reveals their meaning and the semantic structure of a text or corpus" (Brezina et al. 2015:
141)21.
Various statistical measures of collocational strength have been developed in corpus
linguistics. Establishing an association measure requires a measure of the "random cooccurence baseline" (named the "expected frequency", where "expected" here means
expected by a computer and not a human), compared with the observed collocational
21. How collocational graphs demonstrate "semantic structure" is not made entirely clear by
(Brezina et al. 2015).
© Annabelle Lukin Please do not quote without permission - page 101 -
frequencies. The various association measure "can be understood as different ways of
comparing the observed and expected values, putting different weight on different aspects of
the collocation relationship" (Brezina et al. 2015: 145). The collocational relationship is
complex, so that "no single association measure can capture all of its aspects" (Brezina et al.
2015: 144). All are concerned with testing the hypothesis that the co-occurence - collocation
- is non-random. These measures include mutual information (MI, as well as MI2, and MI3),
log likelihood, dice co-efficient, T-score and Z-score. The British National Corpus makes a
set of these calculations available, including Mutual Information - a measure of "mutual
attraction" - which I will use to retrieve collocations. In particular, I will use the MI3 value,
since, as Daille (Daille 1995) has argued, it provides a balance between exclusion and
inclusion of lower frequency collocations. Thus, MI3 gives more weight to frequency, and, as
Brezina et al. argue, "no matter how suggestive, an association which is not repeated enough
will be less influential than an assocation that is more firmly established in the discourse"
(Brezina et al. 2015: 160). MI is a calculation based on the relation between observed
collocate frequency, and "expected" collocate frequency (where "expected" is determined
only by the relative frequencies of each of the two items under consideration). The higher the
value, the more frequent the association between the node word and its collocate. In addition
I will calibrate the selection collocations with reference to their Log Likelihood scores
(Dunning 1993)22. LL scores give a measure of how much evidence one has to reject the null
hypothesis (that is, that the observed collocational frequency of two terms is simply due to
chance). An LL value of 3.83 converts to a p value of less than 0.05; 6.63 to less than 0.01;
10.83 to less than 0.001; 15.13 to less than 0.0001.
In line with the recommendation in Brezina et al. 2015 that a standard notation be
developed for making explicit the parameters used in extracting collocations for a given node
word from a given corpus, the "collocation parameters notation" (CPN) for the collocations
of the node words "war" and "violence" using the whole BNC is set out in Table 4.
Statistic Statistic Statistic
ID
name
cut-off
value
L and R
span
Minimum
Minimum
collocate
collocation
frequency (C) frequency
Filter
22. Dunning 1993 demonstrates the value of LL for the analysis of co-occurence in data in
which one does not assume a normal distribution.
© Annabelle Lukin Please do not quote without permission - page 102 -
-
MI3
9
5L-5R
5
N/A
collocates of punctuation
removed (including enclitic
's); some function words
(conjunctions, deictics)
removed
Table 4: CPN for "war" and "violence" search in the BNC.
Before proceeding to some discussion of the collocations of the node words, let us
first consider the frequency and distribution of the node words "war" and "violence" in the
BNC. "War" turns up 27, 217 times, across 2, 204 texts, giving an average usage per text (i.e.
per text in which the lexical item appears) of 121. "Violence" turns up 5, 507, in 1, 087 texts,
giving an average usage per text of 5. The overall relative frequency of "war" to "violence" in
the BNC is nearly 5 to 1. The hit per text ratio of "war" to "violence" is 24 to 1. With the
collocation frequency set at 5, and an MI3 value of 9, the BNC returned 1, 202 collocates for
"war", and 453 for "violence". "War" has 2.65 times the number of collocates of "violence",
providing further evidence for my suggestion that "war" is a concept with wide reach in the
lives of English speakers. Tables 5 and 6 set out the figures just discussed, including the
number of collocates based on LogLiklihood, which gives a slightly higher number.
"war"
No. of hits
No. of texts
Average hits No. of collocates (collocate freper text
quency min. 5)
27217
2204
121
1, 202 [with MI3=9]
up to 1251 LL>3.84 (p<0.05)
up to 1057 LL>6.63 (p<0.01)
"violence"
5507
1087
5
453 [with MI3=9]
up to 567 LL>3.84 (p<0.05)
up to 479 LL>6.63 (p<0.01)
Table 5: Frequency and distribution of "war" and "violence" in the BNC
Ratio of no. of hits
in corpus
"war"/violence
4.94/1
Ratio of number of Average hits per No. of collocates
texts
text
(collocate frequency
min. 5)
2/1
24/1
2.6/1
Table 6: Ratio of frequency and distribution of "war" to "violence" in the BNC
Collocates, as Saussure and Firth argue, help us understand the meanings associated
with a word. With so many collocations for these both words, it is not feasible to explore
them all. Appendix 1 sets out the top 100 collocates for the word "war" in the BNC. The cut© Annabelle Lukin Please do not quote without permission - page 103 -
off of 100 is arbitrary, and is simply a means to acquire a working list. Even at 100, the MI3
score is high (>17 for "war", and > 12 for "violence"), suggesting there are many further
significant collocational relations with the word "war". Collocates consisting of punctuation,
and function words were manually removed (although some of these will be discussed
below). The BNC also returns enclitics (e.g. 's) as distinct collocates so these were also
removed.
To begin with, the two node words prehend themselves. For "war", the explanation is
in the point made earlier: the word has so many lexical environments. The concept is
elaborate, and so when it becomes topical, there are many structural forms into which it can
enter. There is by contrast a singular explanation for why "violence" prehends "violence": it is
the oft heard expression that "violence begets violence" - see Figure 7. While our collective
wisdom tells us that violence leads to more violence (this collocation is repeated in other
large corpora, as I show below), the "Great War" was thought, by some at least, to be the war
that would end wars Wells, 1914, #67384. That "war" could attract this belief signals a
potential, that further analysis below brings out.
Figure 7: "Violence begets violence": BNC concordance lines for "violence" and
"begets"
The two node words also exhibit some collocational attraction, measured by an MI3
score of 11.4754. There are a total of 32 instances where these terms are collocates within
the parameters set for this study. These collocations are dispersed across 29 texts. With "war"
appearing in 2204 texts in the BNC, this means just over 1% (1.3%) of the texts in which
"war" appears the word "violence" is also present. The figure is 2.6% for the presence of
"war" in texts that feature the word "violence". See Table 7. Again, this confirms earlier
© Annabelle Lukin Please do not quote without permission - page 104 -
evidence that these words act like two north or two south poles of two magnets.
node word ranking
n
n expected
n
observed
n texts
Coll. value
war
239 in collocates for
"violence"
27217
11.523
32
29
MI3:11.4735
LL: 24.54
violence
596 in collocations for "war"
5507
11.509
32
29
MI3: 11.4754
LL: 24.43
Table 7: Collocation of "war" and "violence" in BNC
What do the collocations of these words reveal about their meanings? Corpus
linguists analyse collocations by seeking "semantic prosodies". McEnery 2006, drawing on
Stubbs 2002, uses the term "semantic prosody" to refer to the "meaning arising from the
interaction between a given node word and its collocates" (Stubbs 2002). This is a Saussurean
principle: meaning arises in the play of signs, and so syntagmatic relations are key to the
meaning of a word, both in an instantial and in a systemic sense. McEnery also argues that
semantic prosodies typically encode attitudes and evaluations, and he goes on to suggest that
the evaluations conveyed by such prosodies are typically negative (although he does not
explain or elaborate this claim) (McEnery 2006: 23). Semantic prosodies are more than
simply positive versus negative, as McEnery's discussions of the collocations and keywords
of his corpora illustrate.
Table 8 sets out the top 100 collocates of both "war" and "violence". The collocates
have been organized in the first instance by a grammatical principle: the first column of Table
8 distinguishes the lexical items into three grammatical categories: premodifiers of "war",
process items (both as verbal group elements, and where they are nominalised), and common
and proper nouns. Premodifiers are then split into Classifiers and Epithets; Classifiers are
further subcategorized through semantic categories: 'geographic', 'ordinal', 'temporal', 'stateinternal', 'means', 'extent', 'goal-orientation', 'gender', and 'legality'. Verbal group elements are
differentiated based on Halliday's process type schema: 'relational' (which are also potentially
auxialiary verbs), 'material and transitive', 'material and intransitive', 'material and both
transitive and intransitive', 'verbal and intransitive', and 'verbal and transitive'. Finally, I make
further simple distinctions in the category of 'common nouns' between thoses that refer to
humans and human collectives, and all other associated 'things'.
What do these associations show? Note first of all there are no positive associations
with the word “violence”. While some associated terms are neutral, many are either always
negative (sectarian, racist, unlawful, escalate, rape, flared, denounce, vandalism, robbery,
disorder, cruelty, hatred, looting), or they become so when coupled with “violence”, as in the
© Annabelle Lukin Please do not quote without permission - page 105 -
collocation of “domestic”, “sexual” or “physical” with “violence”. And not only does
“violence” collocate with “unlawful”, it also collocates with various acts deemed unlawful in
most if not all jurisdictions: vandalism, robbery, looting. The term “offences” means unlawful
acts. “Violence” also collocates with “aggression” (the MI3 score is 15.9203; there are 32
instances across 12 texts), a word with an overall much lower frequency than for either "war"
or "aggression" (the frequency of "war" to "aggression" in the BNC is nearly 22/1). With
respect to the process collocations, “violence” attracts transitive verbs, both material and
verbal. As explored earlier “violence” inself comes from a transitive verb. “Violence” is
something that has an impact. I note too that in this comparison, that the lexical verbs
associated with are greater in number than for “war” (though we need to keep in mind we are
dealing only with the top 100 collocations). These items also tend towards being lexis with a
reasonable degree of specificity. In terms of the range of classifiers, “violence” is classified
with respect to forms of state-internal violence, by the means adopted, and by its goal
orientation (“political”, “anti-foreigner”) or lack of same (“mindless”). “Violence” is also
gendered, taking the classifier “male”.
The prosodies around war are much more mixed. There are some negative
associations (the “horrors” of war, “criminals” and “crimes”); “war”, like “violence”, has
victims. But war is also associated with the word “hero”, with “peace”, with “independence”
and “revolution”. It also has many collocations which are non-evaluative, being concerned
with time and place. These collocates locate individual instances of the category “war”,
providing the schema for individuating wars and giving them proper names, which is a
crucial step in the process of them becoming the topic of historical narrative, and to their
being memoralised and commemorated. In this top 100 list we find a preoccupation still with
a war which is two and a half thousand years old, the Peloponnesian War. Modern British
discourse, across 100 million words and various quite distinct contexts, is still preoccupied
with the wars of Ancient Greece, though perhaps this is explained by Hanson's claim that
"The Greek and Roman writers who created the discipline of history defined it largely as the
study of wars" (Hanson 2010: 3). This historical orientation to the recounting of war not
doubt explains the collocations of "war" with “memorials” and “museums”. “War" is also
associated with "Office”, a sign of the institutionalized nature of war (Malešević 2010). The
capitalization of this lexical item is a sign of the official status: it is a reference to the British
War Office, an institution with its origins in the 17th century. Since 1964, the "War Office"
became the Ministry of "Defence"23, a noteworthy resemanticizing of this government
23. See https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/49053/
© Annabelle Lukin Please do not quote without permission - page 106 -
function. Despite this change, the "Office" remains a dominant collocation even into the
1990s.
Figure 8: Collocation of "war" and "Office" in the BNC
While "violence" is inherently unlawful, war is not. While "crime" can collocatewith
war, we also find the collocate "laws". War is governed by international law - there are "laws
of war". While we also have the concept of "war crimes", these are crimes against the laws of
war. In addition, one of the collocations in the top 100 collocates of "war" is the word "just".
"Just" has more than one function; but one of its collocational patterns with "war" is
reference to "Just War Theory", a notion with a long history in Western theology and
philosophy. Part of this heritage is the evolution of the "law of war" (see Chapter 1 of Stone,
1954, #61831). The unlawful use of force in the UN Charter is considered to be "aggression".
The relevant chapter of the Charter is Chapter 7, titled "Action with Respect to Threats to the
Peach, Breaches of the Peace and Acts of Aggression". While "war" is the use of force,
"aggression" is unlawful under the UN Charter. The UN Charter was adopted in 1945, the
history_of_mod.pdf
© Annabelle Lukin Please do not quote without permission - page 107 -
UN did not achieve consensus on the meaning of "aggression" until 1974, the culmination of
over 50 years of discussions through the tenure of both the League and the United Nations
Stone, 1977, #53986. According to Wilmshurst, 2008, #1223, the definition has "scarcely
ever been used for its primary purpose as a guide to the Security Council in determining
aggression by States". The BNC returns only a total of 6 collocations between "war" and
"aggression", construing, for instance "Japanese aggression", "Islamic aggression", and
"aggression" by Serbia - the context is the delegitimation of the use of force in each of these
cases.
Premodifier
of Nominal
Group
Structure
Classifiers
"War"
"Violence"
Geographic: world,
Gulf, Boer, Vietnam,
Falklands, Korean,
Crimean, Iran-Iraq,
Franco-Prussian,
Spanish, Iraq,
Peloponnesian,
American, Imperial,
Russo-Japanese, ArabIsraeli, Pacific
Geographic: Natal
Ordinal:
Temporal: immediate
State-internal: township/s, sectarian,
ethic, loyalist, communal,
intercommunal, mob, PAV/VPN24, street
Ordinal: second, first,
II, last
Temporal: 1914-1918,
post-Cold, 1939-45,
1914, post-Second, preFirst, post-World, early, Means: physical, sexual
last
State-internal: civil
Means: Cold, nuclear
Extent: widespread
Extent: great, all-out
Goal-oriented (including absence of):
Goal-oriented
drug-related, political, anti-foreigner,
(including absence of): racial/racist, domestic, mindless,
gratutious
Gender:
Legality: just**
Gender: male
Legality: unlawful
Process
(including in
nominalised
form
Epithet
phoney
sporadic, serious, worst
Relational/auxiliary
be/is/was/were/been,
has/have/had
be/is/are/was/were/been
has/have/had
involving
24. 'VPN' and 'PAV' are the Slovakian and English acronyms for a political movement
established in Slovakia in 1989 - see https://en.wikipedia.org/wiki/Public_Against_Violence
© Annabelle Lukin Please do not quote without permission - page 108 -
Material and transitive brought
killed
won
begets
provoke/provoked
rape
marred
inciting
perpetrated
intimidation
harassment
Material and
intransitive
waged/waging,
broke (out),
end/ended/ending
came
began/beginning
outbreak
erupted
flared
end
continuing
outbreaks
upsurge
act
resorted (to),
use/used
Material and both
transitive and
intransitive:
fight/fought/fighting
escalate/escalating/escalation/escalated
Verbal and
intransitive:
declared/declaration,
Verbal and transitive:
Common
nouns
threat/threats/threatened/threatening
advocated
condone
abuse
Associated things
(various)
Peace, aftermath,
crimes, memorial, time,
effort, years, period,
attrition, horrors, laws,
Museum, Office,
wagon, trade,
revolution,
independence, boars
Associated humans/
human collectives
prisoner/s, veterans,
women, victims, police, IRA, spectator,
criminals, military,
Sind
guerrilla, victims, Hitler,
hero, army
Proper nouns Associated places
sex, crime/s, offences, vandalism,
robbery, disorder, bloodshed, fear,
cruelty, football, dishonesty, corruption,
hatred, looting, campaign, aggression
France, Germany,
Britain, Russia, zone,
Europe,
Table 8: Semantic prosodies: collocational patterns in the top 100 collocations of
"war" and "violence" [**'just' has various functions; it is noted here only in relation to the
expression "just war"]
(d) Colligational patterns in a news corpus
So far we have consider these terms - "war" and "violence" - from largely lexical
standpoints, through considering their dictionary meanings, their thesaurus locations, and
their collocations in a large multigeneric corpus. But lexical items are woven in intricate
wasy in text. As Saussure noted, signs are linear: they enter into syntagms. And from
Halliday we know that these syntagms are multifunctional. Linguistic structure is, therefore,
© Annabelle Lukin Please do not quote without permission - page 109 -
polyvalent, meaning in three distinct ways. Since this kind of analysis must be done
manually, I now return to my primary corpus, the ABC TV reports of the 2003 invasion of
Iraq. From this corpus of 123 news reports (c. 45, 000 words) from 20/03/03 to 03/04/03 I
have already examined the context in which the lexical item "violence" is selected. How is
the word "war" used? More specifically, how does the word "war" enter into structural
patterns, and what implications do these patterns have for understanding the patterning of
ideology? This lexical item is prominent in the corpus, being the most frequent open system
lexical item (310 instances, and a frequency of 0.72% in the corpus). The corpus also shows
some compounds in which "war" is an element, such as "anti-war", "post-war", "war-weary",
as well as the possessive form "war's" - the corpus has a total of 43 instances across these
categories. "War" can also be pronominalised as "it". There are approximately 29 instances in
which "it" refers to a nominal group with "war" as part of the structure. The figure is fuzzy,
because the texts at time display a textural looseness that mean the lexicalisation of a pronoun
can be open to interpretation (see Lukin, 2013, #33388 for some discussion of this issue).
This gives us a set of 384 (including 2 examples with Subject deletion) instances of the use
of "war" across the corpus, and the opportunity to examine how it is put into structure.
Studio Host 1:
Tonight the war against Iraq begins with Baghdad under attack
Studio Host 2:
President bush promises to disarm Saddam and free the Iraqi people
George W. Bush:
"This will not be a campaign of half measures, and we will accept no
outcome but victory"
Studio Host 2:
Good evening, welcome to a special edition of ABC news. The
Second Gulf War has begun.
Table 9: Opening of ABC TV 7pm bulletin announcing the invasion of Iraq in 2003
To set the parameters of the discussion, let us begin with the opening of the first
bulletin on ABC TV announcing the invasion of Iraq. This short extract is set out in Table 9.
In this short extract, the lexical item "war" appears twice. It is, as in all instances in this
corpus, and in line with modern usage, a nominal element, and more specifically, "war" is
located in these examples as Thing in the nominal group structure. Apparent also in this
example is the move from a common noun in the first usage - "the war against Iraq" - to a
proper noun - the Second Gulf War. The common noun has specific deixis (the use of the
definite article), indicating that this instance of the category of "war" is already treated as a
known quantity, despite it only having just begun. As a nominal element, it is able to be
recruited into a number of grammatical roles, interpersonally as Subject or Complement, or
as a nominal element within the Adjunct function, a non-nuclear element of clause structure.
In both examples above, "war" appears in a nominal group functioning as Subject of the two
© Annabelle Lukin Please do not quote without permission - page 110 -
clausees in which it appears. Looked at from an ergative point of view, a nominal element can
fulfill the grammatical role of Agent, Medium or Range. If Agent, "war" is construed as an
entity with the force to act over things of any kind, whether sentient (people, animals), or
non-sentient. Non-sentient entities may be material things, or can be abstract or complex
entities. If Medium, "war" is the entity which undergoes the process, but it has no role in
bringing the process about. If an entity is given the grammatical role of Range, it is an
extension of the process, the scope over which the process unfolds. In the two examples
above, "war" is Medium in clauses which are intransitive. This war simply "begins".
Example Speaker
Number
+condion Agent
al/
projectin
g clause
Process
Medium
1
Simon
Crean
It
will make Australia
2
Simon
Crean
The war
will create a
generation
of hate
3
Hosne
Mubarak
it
could
inflame
world
opinion
4
Hosne
Mubarak
it
could
spark
further
terrorist
attacks
5
School
girl
protestor
it
is killing
innocent
people
6
ABC
(Many*
journalist fear)
*Many
Syrians
the war
will
an already
destablize delicately
balanced
region
7
ABC
(Many
this war
journalist believe)
*Many
Jordania
ns
will
inflame
the
passions
of an
already
volatile
region
8
ABC
(They*
the war
journalist say)
*Many
Jordania
ns
will hurt
innocent
civilians
+Range
+Circums
tance
a terrorist
target
© Annabelle Lukin Please do not quote without permission - page 111 -
in
Australia's
region
9
ABC
(They*
the war
journalist say)
*Many
Jordania
ns
will
damage
Arab
economie
s
10
ABC
(if)
journalist
will
remove
Saddam
Hussein
it
11
ABC
the war in is [also] lives
journalist
the north claiming
Table 10: "War" as Agent in transitive, material process clauses = 11 instances in
total
Elsewhere in the corpus we find "war" given the grammatical role of Agent. Table 10
sets out all instances of the clause configuration in which "war" is grammatical Agent in the
environment of an material process - 11 clauses out of the set of 384 instances, or 3%, in
which "war" takes on a transitive, agentive role. In Example 5, "It is killing innocent people",
a school girl interviewed at an anti-war protest construes "war" as an Agent with considerable
power and consequences, as well as with actuality, through her choice of an unfolding present
tense. Australia's then Opposition Leader, Simon Crean, is also quoted stating "it will make
Australia a terrorist target". Here again, "war" is Agent, and its effect is very much negative.
This same view is presented in the words of then Egyptian president, Hosne Mubarak, who
states that the war could "inflame world opinion" and could "spark further terrorist attacks".
In their words, the destruction of war is a potential, rather than an actuality, though Mr Crean
is quite certain of its terrible consequences will come (notably, consequences for Australia,
rather than Iraqi civilians). There are six uses of the lexical item "war" by ABC journalists in
this corpus in which "war" is given the grammatical role of Agent in a material process. Four
of these examples are irrealis, by virtue of associated grammatical features (one appears in a
hypotactic conditional clause, and three in projected clause). Example 11 is barely transitive,
on the basis that the element analysed here as Medium is an instance of metonymy. While
"war" can certainly be construed as a powerful and destructive agent - and at the time of this
reporting, Iraqi civilians were being killed already in their thousands - this configuration is
minimal in this corpus.
So what work does this lexical item do in this environment? Given the 384 instances
mentioned above, can we see patterns that help us understand more about the meaning and
operation of this lexical item? Even with this small set of instances, the diversity of potential
combinations of structures is significant. Taking only instances of "war" a nominal group
element in Subject, Complement or Adjunct position, we now have a smaller set of 256
examples from our original set. Across this set, "war" is Subject in 36% of cases,
© Annabelle Lukin Please do not quote without permission - page 112 -
Complement 37% of cases, and Adjunct 27% cases. Taking the category of Adjunct first,
"war" is seen across many circumstantial environments: time, location, matter,
cause,
accompaniment, such as "to war", "as a result of the war", "at the start of the war", "on the
rights and wrongs of the war", "against this war", "in this war", "without the full force and
fury of the war", and so on. Notably, as Adjunct "war" is removed from the centre of the
clausal configuration. It becomes the circumstance around which other actions are given
focus. Take for example to expression "to go to war", where we find "war" as a location.
This construction creates a meaning in which "war" is a state of being in some time period
not clearly outlined. It gives a particular categorization to events that might be construed in
other ways, e.g. as "colonial aggression". Note that "aggression" is not a state which can be
locatively construed, as "war" can be; instead, it interprets action as a type (see earlier
discussion of the location of "violence" in the OED Thesaurus). A rhetorical implication of
"going to war" is that the condition of "war" exists already. One example from the corpus is
the following: "Juanita today Australia has gone to war". Australia's actions of deploying
troops and aircraft are construed not as part of war, but distinct from the already existing state
of war.
as Complement
as Subject
Material
30%
55%
Mental
14%
-
Relational
46%
43%
Verbal
10%
1%
100%
100%
Table 11: "War" as Complement and Subject, and by process type in ABC TV
corpus
Taking the remaining categories, "war" is equally Subject and Complement in this
news corpus. The distribution across these categories and by process type is set out in Table
11. This table shows the diversity of grammatical roles of this entity. "War" is diverse in its
structural presentation, a sign of it being a potential reference point for many semantic
domains, and further evidence for the claim earlier about the elaborate linguistic
infrastructure associated with "war". As Complement, its process type relation extends across
the major types, even if the typical dominance of material and relational remains. As Subject,
it is largely working in the environment of material and relational processes. The many
selections and configurations are too diverse to review here; let us take a further reduced
© Annabelle Lukin Please do not quote without permission - page 113 -
subset, those clauses in which "war" is the Head of the nominal group structure in a central
nuclear role in the clause. Out of the total set of origin instances (384), we now have a set of
85. Thus, of our original set, 22% are lexical usages which put "war" into the Head of a
nominal group structure and into a central clausal role. The majority of the clauses are
Material or Relational, with "war" as Subject (89%). Of this set, we have already seen those
instances in which "war" is Subject of a transitive material process (see Table 10). "War" is
about four times more likely, based on this corpus, to be Actor in an intransitive environment,
than to be construed as Agent of a transitive clause (41 to 11). If material, "war" is typically
Medium in an intransitive clause, as we saw in the opening extract above.
[EXTRACT ENDS]
© Annabelle Lukin Please do not quote without permission - page 114 -
© Copyright 2026 Paperzz