A New Approach to Oppositions in Discourse

Abstract

This article proposes a flexible typology of syntactic frames (e.g., “X not Y,” “either X or Y”) which can act independently or cooperatively as triggers for noncanonical oppositions that co-occur in those frames. The provisional categories are based on Jones’s quantitative study of syntactic frames for housing co-occurring canonical (conventional) antonyms (e.g., “war” / “peace”) but are substantially revised to account for a broadening of the definition of opposition. These revisions also take into consideration how syntactic frames overlap and interact in context, making qualitative analyses necessary. Data are taken from U.K. newspaper reports of two major protest marches in London. Here, the use of common syntactic frames that house co-occurring canonical oppositions trigger a range of noncanonical examples, which contribute to a binary representation of people, groups, events, ideas, and processes. The typology proposed in this article can open up new ground for locating and investigating the ideological influence of oppositions in discourse in disciplines such as critical discourse analysis.

Keywords

opposition antonymy syntactic trigger difference equivalence critical discourse analysis ideology

This article explores how the syntactic structures that frame co-occurring “canonical” systemic oppositions (often called “antonyms”) within and between sentences in discourse can also trigger instances that rely on the context of their production and consumption for their oppositional status. Jones’s (2002) pioneering corpus study developed a typology of discoursal syntactic frames for antonyms, such as “X not Y,” “either X or Y,” and “X more than Y,” where X and Y are the antonym pair. The research presented here reconsiders, from a qualitative perspective, ways of categorizing “noncanonical” oppositions. In doing so, this research analyzes these oppositions’ contribution to the construction or reinforcement of polarized positions in texts, often with ideological implications.

For instance, during the 2010 U.K. general election campaign, an editorial column in the U.K. newspaper The Sun (7 April 2010:6) informed readers that on the subject of employment “we must decide between Labour’s tax on jobs or Tory growth,” creating an artificial mutually exclusive (nongradable) opposition between “tax [on jobs]” and “growth.” This is triggered by a combination of an “X or Y” frame and an ancillary canonical opposition between, traditionally, the two main political parties in the United Kingdom (Labour and Tory).¹ In this instance, the construction of such an opposition encourages readers to infer that “tax” and “growth” are incompatible, with the aim of encouraging Sun readers to vote for the Tory (Conservative) Party, which the newspaper openly supports.

This study seeks to refine, expand, and reevaluate the function of the syntactic structures that frame oppositions, as originally proposed by Jones, by considering their function as triggers for noncanonical oppositions. The revised classification is provisional and flexible, taking into account that qualitative analyses reveal how the frames overlap and interact with each other and that analysis of different genres and discourses, beyond the scope of this article, is likely to broaden and strengthen the proposed typology. This study ultimately aims to provide analytical tools for disciplines such as critical discourse analysis where, for instance, research often centers on how individuals, groups, processes, events, and ideas are represented in binary terms for ideological purposes.

The data for this study are taken from U.K. national news reports of two major protest marches in London in 2002 and 2003. Examples from the data are used to expand and adapt Jones’s typology of antonym frames based on a broader definition of oppositional relations, including the likelihood that oppositional concepts can be embraced by whole phrases and clauses as well as just individual lexical items such as “war” / “peace.” Therefore, although many standard studies of lexical semantic relations label these types of oppositional pairs as “antonyms” (sometimes narrowed down to refer to gradable opposites), in this article I use the term opposition, which encapsulates a broader sense of this type of relation. I also use the terms canonical and noncanonical—adapted from Murphy’s (2003) pragmatic approach to lexical semantic relations—to refer to oppositions that have a more or less stable basis in the linguistic system in which they participate. This is to avoid judging oppositional pairs as “good” or “bad” examples, for, as I argue, if an unusual oppositional pair (e.g., “cream” / “spleen”) resides in one of the frames common to conventional pairs, then in that instance it is not a bad opposition, just a context-bound one. It is also important to note that the terms canonical and noncanonical are not intended to treat oppositions as two mutually exclusive categories. The canonical status of oppositions ranges in a gradable cline from canonical to noncanonical, so the terms are necessarily imprecise. At the same time, this demonstrates the difficulties in avoiding representing ideas and concepts in anything other than a binary fashion, even in the realm of academic discourse.²

The Categorization of Oppositional Relations

Context-Free Categorization

The systematic relationship between words and word meanings has been one of the major preoccupations of lexical semanticists. Studies consistently foreground oppositeness as the most significant and pervasive of these relations, over and above other key relations such as synonymy, hyponymy, and meronymy (see, e.g., Lyons 1977; Cruse 1986, 2004; Justeson & Katz 1991; Mettinger 1994; Fellbaum 1995; Jones 2002; Murphy 2003; Jones et al. 2007; Izutsu 2008; Jeffries 2010b). Murphy (2003:169) calls binary opposition “the archetypal lexical semantic relation.” There is a general consensus that opposites tend to involve binary relations and that the relationship between them consists of maximal semantic similarity, differing usually on only one plane of difference (Murphy 2003; Davies 2008). For example, one of the most common and oft-quoted examples of an opposition, “hot” / “cold,” relies on the concept of temperature as the common ground between them. I call this the “plane of equivalence,” while degrees of temperature is the “plane of difference.”³ The opposition is generated via the interaction between the two conceptual planes.

Traditional studies tend to categorize oppositional types according to the type of context-free relationship that exists between the oppositional pair. Typically they differentiate between gradable oppositions (e.g., “hot” / “cold”), nongradable complementaries (“dead” / “alive”), converses (“husband” / “wife”), and directionals (“up” / “down”). Lyons’s (1977) categories and subcategories of opposition are a typical typology, expressed in diagrammatic form in Figure 1.⁴

Figure 1.

Diagrammatical representation of Lyons’s (1977) categories of opposition

Crucially, with a few exceptions, prior studies have mainly drawn on pairs of opposites in a syntax-free environment, or at best sentences containing co-occurring opposites invented for the purpose of exemplification and illustration (i.e., not taken from actual instances of discourse). This implies that the qualities specified in the categories are somehow intrinsic to the oppositional pairs, independent of their use in discourse, and moreover that there is a limited and stable set of pairs with independent existence in a language system (the langue in Saussurian terms). As a result, some studies make overt value judgments about the validity of some oppositional pairings. Justeson and Katz (1992:176) claim, for instance, that “there is something awkward about the conceptually opposite pair large-little lacking as they do the ‘clang’ association of clear antonyms,” arguing that “small” pairs up with “large” with a much more satisfying “clang” that it does with “big.” Cruse (1986:262) concludes his third and last chapter on opposites with a subheading “What makes a good opposition?” and attempts to answer this question.

These approaches limit what can be counted as an oppositional relation, often as a result of assuming that opposites are a relation between individual words codified for instance in lexical authorities such as thesauruses, which usually include antonyms as well as synonyms. This assumption sidesteps the complex relationship between words and concepts and the fact that often an individual concept can be expressed in phrasal or clausal form, and concepts in opposition to each other may be expressed using words that conventionally are not seen as opposites.⁵ Jones et al. (2007:131) argue that canonicity in antonym pairs should be judged by the relations developed by “convention” as well as their “semantic relatedness.” Some other pairs may occur as more peripheral examples of oppositions (e.g., “freezing” / “hot”). To use Jones et al.’s example, “fat” has much more credence than “overweight” as a canonical antonym of “thin” (based on statistics elicited in a corpus study, see Jones et al. 2007:136), even though the two pairs share a similar conceptual relation of opposition. However, in a sentence such as “I’d rather be overweight than thin,” the context in which “overweight” and “thin” occur (in an “X rather than Y” frame) makes them just as legitimate as oppositions as the same sentence in which “overweight” is replaced by “fat.” We could recast this declaration as “I’d rather have a bit of meat on me than look like a stick insect.” In this case similar concepts are being expressed and treated as oppositions, without any of the words in bold on either side of the formulation having any canonical lexical oppositional relation to each other. Frequency of use, therefore, might be an indicator of canonicity of antonyms but not of the viability of an opposition if placed in the appropriate syntactic frame. Importantly, in this example, the concepts of fat / thin are being treated as if they were mutually exclusive complementaries and not gradable on a scale of weight, owing to the syntactic structure (“rather X than Y”) in which they reside.

Context-Dependent Categorization

Recent approaches to categorizing oppositions according to their co-occurrence in common syntactic frames (such as “X not Y”) in stretches of real discourse have opened up the field for analyzing noncanonical examples such as the one above. Although syntactic frames are mentioned in passing in Fellbaum (1995), the most pioneering work has been conducted by Mettinger (1994) and more extensively by Jones (2002) in their corpus studies of oppositional relations (see also Jones et al. 2007; Davies 2007, 2008; Jeffries 2010a, 2010b).

Mettinger and Jones draw from genre-specific corpora to log the common syntactic environments in which co-occurring opposites appear. Mettinger (1994:2) bases his analysis around a collection of 350 oppositional pairs chosen from the 1972 edition of Roget’s Thesaurus and another 350 pairs chosen from what he describes as his “opposites in context corpus,” which consisted of forty-three mainly crime fiction novels. Mettinger places his selection of examples into one of a number of syntactic frames and assigns a common function to each one. So, for instance, the “(either/whether) X or Y” frame functions to present “Choice (Exclusive).”⁶

Jones’s (2002) study of the syntactic environments for opposites is the most comprehensive. Using three thousand sentences selected from a corpus of 280 million words taken from the Independent newspaper (1988–1996), he preselects fifty-six canonical oppositional pairs (such as “peace” / “war”) and categorizes them according to syntactic frames in which they co-occur. The result is a detailed classification based on a combination of both the form and the function of these frames. His categories include “coordinated antonymy” (e.g., “X and Y,” “X or Y”) “transitional antonymy” (“from X to Y”), and “negated antonymy” (“X not Y”). However, a by-product of his findings is that many unconventional oppositions are generated as a result of what he calls “ancillary antonymy.” This is where a second pair of co-occurring lexical items becomes opposites because they feature in the same syntactic structure as one of Jones’s conventional pairs, such as in the following example:

(1) Robin Cook, Labour’s Health spokesman, demanded: “How can it be right to limit the hours worked by lorry drivers and airline pilots, but wrong to limit the hours of junior hospital doctors undertaking complex medical treatment?” (The Independent; Jones 2002:45)⁷

Here, the conventional distinction between “right” and “wrong,” according to Jones, helps create a contrast between “lorry drivers and airline pilots” and “junior hospital doctors undertaking complex medical treatment.” Jones’s aim is mainly to describe and categorize the frames. He provides a useful starting point by noticing the unconventionality of some “contrasts.” My aim is to investigate how noncanonical oppositions occur not only as “ancillaries” but also in all of the syntactic frames logged by Jones. In the process I have also reconsidered Jones’s typology and problematized the approach that draws conclusions based on quantifying frequency of occurrence. For instance in example (1) above, the connective “but” also arguably plays a role in framing (and triggering) the oppositions, and not just the canonical pair “right” / “wrong.” It is also significant that the ancillary “B-pairs” (Jones 2002:47) consist of extended noun phrases rather than just the conventional one-word lexical item. Furthermore, the contrasts implicit in these phrases are not necessarily always lexicalized. For instance, presumably Robin Cook is implying, by choosing these particular jobs to contrast with that of junior doctors, that the work of lorry drivers and airline pilots is not as complex, so that a simple / complex opposition is triggered at the conceptual level, even though the word simple is not employed.

A by-product of Jones’s detailed quantitative approach is that it opens up exciting possibilities for investigating noncanonical oppositions and the ideological repercussions of their consistent use in discourse. Murphy (2003:204) points out that Jones’s “taxonomy of antonym functions and collateral syntactic frames provides a means to identify situational or context-bound antonymy.” She claims,

Awareness of these frames and their functions gives us a means for recognizing context-dependent cases of antonymy. So, while Jones has used canonical antonyms to demonstrate the existence of these functions, other pairs, including non-canonical opposites and words that are not antonymous in neutral conditions, function as antonyms when in these frames. (Murphy 2003:204).

Before Jones’s study, Jeffries (1998:105-106) had already referred to the potential of syntactic structures—in this case “parallelism”—to create context-based oppositions. She describes how the U.K. Conservative party’s 1979 general election slogan “Labour say he’s black, We say he’s British” leads us to interpret “black” and “British” as opposites (see also Jeffries 2010b:2-3 for a more detailed examination of this example).⁸

One other more recent study of co-occurring oppositions in discourse is that of Izutsu (2008:673), who attempts to provide “linguistic evidence for the classification of opposition relations into three semantic categories (contrast, concessive, and corrective.)” She does not attempt to assign opposites into specific syntactic frames, but, using a model taken from cognitive grammar, she provides some helpful semantic distinctions that contribute to the way the provisional typology below is characterized, specifically the section on “concessives.”

Syntactic Frames and Triggers

This section provides an overview of the ways that key syntactic frames for canonical opposites—many identified by Jones (2002)—can, at the conceptual level, trigger oppositions between words, phrases, and clauses not conventionally recognized as having this relation. If oppositions are as omnipresent as many studies suggest, then it seems reasonable to assume that the common frames in which canonical oppositions co-occur will also structure the way we process noncanonical variations. I use the term trigger in much the same way that Levinson (1983:181-185), proposes that presuppositions can be triggered by certain syntactic formulae. More research is needed on how and to what extent noncanonical oppositions can be conceptually generated by appearing in these frames. Nevertheless, the qualitative analysis of the individual examples below suggests that these frames do play an important role in creating binary structures.

My approach is informed by the following principles: oppositional pairs have both lexical and conceptual foundations; these pairs exist as part of the system of lexical semantic relations (langue) and are updated through use (parole); some oppositional pairs can be viewed as more canonical examples than others owing to frequency of use and their entrenchment in a linguistic culture, while still allowing for the possibility that any pair of words/phrases/clauses can be classed as acting in an oppositional manner; the way we understand noncanonical oppositions may be based on processing a series of overlapping or blended canonical oppositions at a higher superordinate conceptual level.⁹

The following discussion and the revised typology resulting from it owe a huge debt to the groundwork undertaken by Jones (2002). The revisions are based on differing approaches and desired outcomes.

Jones’s typology consists of the following main categories of opposition in order of frequency (see Jones 2002:180; percentage indicated): “ancillary antonymy” (38.7 percent; a range of frames); “coordinated antonymy” (38.4 percent; e.g., “X and Y,” “[either] X or Y”); “comparative antonymy” (6.8 percent; e.g., “more X than Y,” “X rather than Y”); “distinguished antonymy” (5.4 percent; e.g., “the difference between X and Y”); “transitional antonymy” (3.0 percent; e.g., “turning X into Y”); and “negated antonymy” (2.1 percent; e.g., “X not Y,” “X instead of Y,” “X as opposed to Y”). More recently Jones, in collaboration with Murphy et al. (2009:2161-2162), constructed another category, “interrogative antonymy” (e.g., “X or Y”). Jones’s main categories and some textual examples from his 2002 study are summarized in order of frequency (most frequent first) in Table 1.

Table 1.

Summary of Jones’s (2002) Most Common Syntactic Frames, with Examples from His Data (Frame Type Frequency Indicated in Brackets in First Column)

Jones’s category	Jones’s descriptions, functions	Jones’s frames	Jones’s textual examples
Ancillary antonymy (38.7 percent)	One antonymous pair responsible for generating another new one	No specific frame associated with this category	Now it denotes high butter mountains and a low boredom threshold.
Coordinated antonymy (38.4 percent)	Inclusiveness and exhaustiveness of scale	X and Y	He took success and failure in his stride.
		X or Y	Most Ugandans
		X but Y	they don’t encourage it
Comparative antonymy (6.8 percent)	Places words in a comparative context or measures one antonym against another	More X than Y	Sometimes I feel more masculine than feminine
		X is more (adj) than Y	small monk tails are cheaper than large ones
		X rather than Y	people become active rather than passive adjuncts
Distinguished antonymy (5.4 percent)	Alludes to the inherent semantic dissimilarity (metalinguistic distinction)	The difference/distinction [etc.] between X and Y	The blurred distinction between fact and fiction
		Separating X and Y	separating out our masculine and feminine instincts
Transitional antonymy (3.0 percent)	Expresses a movement or change from one location or state to another	From X to Y	crossing the boundary from straight to gay people
		Turning X to Y	Even hard currency has turned soft
Negated antonymy (2.1 percent)	Negates one antonym as a device to augment the other	X not Y	pessimism not optimism
		X instead of Y	met him alive instead of dead
		X as opposed to Y	being young and keen as opposed to being old and keen

My own provisional typology includes the following revisions to Jones’s typology: (1) treating the frames as triggers for noncanonical oppositions where they occur in these frames; (2) using the term opposition instead of antonymy; (3) removal of the “ancillary” and “coordinated” categories; (4) the creation of new categories—“replacive opposition” (taking in “X rather than Y,” “X instead of Y”), “concessive opposition” (e.g., “X but Y,” “despite/while/although X, Y”), “binarized option” (e.g., “[either] X or Y”), and “parallelism” (a range of structures); (5) treating whole phrases and clauses as examples of oppositions; and, as the above list suggests, (6) placing some of Jones’s examples in different categories. These are not just cosmetic changes, as is discussed throughout the article. The revised typology with canonical and noncanonical examples is set out in Table 2.

Table 2.

Summary of This Study’s Syntactic Frame Categories and Their Functions as Triggers

Category	Common syntactic frames, triggers	Canonical textual example	Noncanonical textual example	Function(s)
Negated opposition	X not Y; not X, Y	The government was elected to create unity in this country and not create division.	Clotted cream not ruptured spleen.	Emphasizes already inherent mutual exclusivity in canonical examples, and constructs mutual exclusivity in noncanonical ones. Often expresses preference for one state over another. Often combined with the contrastive “but” (e.g., not X but Y).
Transitional opposition	X turns into Y; X becomes Y	. . . turn the many decent, honourable and law-abiding people . . . into criminals.	Villages are turning into weekend rest centres or dormitories for commuting TV executives and merchant bankers.	Transformation from one state to a canonical or noncanonical opposite.
Comparative opposition	more X than Y; X is more A than Y	Dr Higgs was a lot more right than wrong . . .	. . . the marchers seemed more bemused than offended . . .	Measures X against Y by comparing them either directly (using the same relations of equivalence and difference) or indirectly, judged against another scale of equivalence and difference.
		. . . reward is more effective than punishment.	. . . more important than the fate of Labour is the fate of mankind.
Replacive opposition	X rather than Y; X instead of Y; X in place of Y	Wanting to be happy rather than sad, I accepted . . . (Jones 2002:79)	There has been speculation that Mr Michael may propose introducing a licensing system rather than an outright ban . . .	Expresses an alternative preferred or nonpreferred option to that which it is opposed.
		. . . the whole concept of managing by punishment instead of reward . . . (Jones 2002: 90)	In place of a charismatic leader, they have the belief that politicians are lying.
Concessive opposition	X but Y; despite X, Y; while X, Y; although X, Y; X, yet Y	[Unlikely to conjoin individual canonical lexical items unless expressing simultaneously contradictory states, e.g., “I was happy but also sad”.]	There was plenty of passion but the marchers remained good-natured.	Creates contrast between two conjoined phrases or clauses, often relating the unexpectedness of what is said in one clause in view of the circumstances described in the other clause.
			Despite the numbers, the march was peaceful.
			While it was true that militants . . . were out in force, the heart and mind of the protest was ordinary people.
Explicit opposition	X contrasted with Y; X opposed to Y; the distinction/ division/difference/ between X and Y; X against Y	Being young and keen as opposed to being old and keen . . . (Jones 2002:90)	The professionally-produced placards . . . contrasted with cobbled-together banners.	Where a linguistic item within the syntactic frame makes an explicit metalinguistic reference to either a presupposed or a constructed contrast between X and Y.
		. . . the division between gay and straight . . . (Jones 2002:81)	The Liberty and Livelihood March began ostensibly divided into two camps . . .
		This blurred distinction between fact and fiction . . . (Jones 2002:81)	“House music against war”
Parallelism	[No specific frames. Relies on repetitive structures]	You are as young as your faith, as old as your doubts (Jones 2002:56)	. . . they can walk over our lands but they can’t walk over us.	Repetition of a range of syntactic structures within which specific lexical items are foregrounded, inviting the addressee to relate them as oppositions. Often combined with other syntactic triggers such as “but” or other more canonical oppositional items. Often contain canonical oppositions which in Jones’s (2002) terms makes the second X/Y pair examples of “ancillary antonyms.”
			Leave us to our cow s*** and we will leave you to your city bull***t.
Binarized option	whether X or Y; either X or Y	Is she a good mommy or a bad mommy (Murphy et al. 2009:2162)	whether Mr Blair still treats those hundreds of thousands of people as an irrelevant minority, or accepts that this time, the countryside really has spoken	Offers and/or creates a choice between two mutually exclusive options.

Data

Most of the examples I use are from U.K. national news reports of the two biggest protest marches ever to take place on British soil (both in London): the Countryside Alliance protest against government legislation perceived to be eroding rural lifestyles, especially a ban on fox hunting (22 September 2002) and the Stop the War Coalition demonstration against the proposed invasion of Iraq by U.K. and U.S. armies (15 February 2003). A small corpus of reports on the protests was assembled, each published the day after each protest took place. For the Countryside Alliance protest, forty-one articles totaling 31,112 words were retrieved, from the Daily Mail, Daily Mirror, Daily Telegraph, Financial Times, The Guardian, The Independent, The Sun, and The Times. For the Stop the War Coalition protest, thirty-eight articles totaling 30,976 words were retrieved from the Independent on Sunday, Mail on Sunday, The Observer, News of the World, Sunday Mirror, Sunday Telegraph, and Sunday Times.

The approach taken in searching for the frames and analyzing the examples was a qualitative one. The decision to adopt these specific data was based on the discovery of a number of constructed oppositions in one specific news report on the Stop the War Coalition protest march in the Sunday Mirror (see Davies 2007). The article used a number of syntactic frames to artificially divide the protestors into “good” (i.e., peaceful and nonpolitical) and “bad” (i.e., troublemakers with a hidden political agenda). Subsequently a close reading of the other reports disclosed similar tendencies whereby a number of the frames categorized by Jones were used to create noncanonical binaries; other types of frames, not categorized by Jones, were also revealed.

Theoretically, any textual data can be used to search for oppositions. However, these texts have been especially useful, as the historic nature of both protests stimulated a variety of emotive responses, often expressing highly rhetorical polarized opinions, including the reported speech of politicians and protestors, slogans written and chanted by protestors, and the often partisan views of the journalists themselves. The data I analyze below include oppositions from all of these perspectives, but I do not quantify the frames in any way (as a computer-orientated corpus approach might). The contribution of constructed oppositions toward disseminating a particular ideological viewpoint is explored in the analysis and opens up fertile ground for further exploration.

A Proposed Typology of Syntactic Triggers for Oppositions

Negated Opposition

The “X not Y” structure is one of the archetypal frames for oppositions. Jones (2002:88) describes what he calls “negated antonymy” as potentially the “purest” form of oppositional relation with “its primary function being to generate a sharper contrast between the two words by making explicit their inherent antonymy.” By referring to their “inherent antonymy” Jones is assuming the existence of only canonical opposites in the X/Y slots. Nevertheless, the generation of a “sharper contrast” is an important rhetorical function of co-occurring canonical opposites in this particular frame, as (2) shows:

(2) The Government was elected to create unity in this country and not create division. (The Sun, 23 September 2002:4)

On a purely logical level, “and not create division” in the Y slot is redundant, as “unity” is semantically equivalent to the negation of its canonical opposite. In this case, however, the speaker—Labour MP Kate Hoey (rebelling against her own government’s policy on banning fox hunting)—is emphasizing her desire for unity between city and country by simultaneously appealing to the positive intentions of her own government, while warning against the negative consequences of ignoring the protestors’ campaign.

Other examples demonstrate how noncanonical oppositions might be constructed. For instance, in the following quote by Richard Burge, the chief executive of the Countryside Alliance, he employs a version of the “X not Y” frame (“not X, Y”):

(3) “We are not a colony, we are an equal and valued part of this nation.” (Daily Telegraph, 23 September 2002:2)

The colonizing force Burge refers to is the government, and those “colonized” are people who live in rural areas who will be most affected by the new legislation. The opposition generated here between “colony” and “an equal and valued part of this nation” is clearly noncanonical, including the fact that the Y of the pair is an extended noun phrase (not just one lexical item like “colony”).¹⁰ If it were possible to conceive of the word “colony” as having any kind of decontextualized canonical opposite, it may be something like “free or independent state.” The connection is not overly obscure, as the qualities of being equal and valued (Y) are not those normally associated with colonial rule, and so being X means being unequal and undervalued. We might then argue that this triggered opposition is conceptually absorbed into the meaning of “colony” and gives its negative value extra force. The implication is that the city-based government representatives are riding roughshod over rural areas as if they were some kind of foreign invader. However, if the speaker had simply declared “We are not a colony,” it would be difficult to infer a positive value, owing to the lack of a canonical opposite to “colony.”

Antiwar slogans used on placards and banners in the Stop the War Coalition march exploit the rhetorical potential of the “X not Y” frame to produce variations on “peace not war,” such as “make love not war” and “make tea not war.” The employment of one of a canonical oppositional pair (“war”) in the Y-slot of the frame is likely to trigger the concept of peace in the X slot. The correlation between “love” and “peace” is self-evident, and the partaking of “tea” is a stereotypically British pastime and its consumption associated with a peaceful environment.

A more obscure example is the contrasting of “clotted cream” with “ruptured spleen” in the following example:

(4) The family took ages to get out of the station at Waterloo, walking behind a huge, stately puppet of George Bush and a placard belonging to a group called Cornish Ravers that said: “Clotted cream not ruptured spleen.” (Independent on Sunday, 16 February 2003:2)

I do not have space here for a detailed analysis of how and why “clotted cream” is analogous with peace and “ruptured spleen” with war; ¹¹ suffice it to say that triggers additional to “not” include rhyme (“cream” and “spleen”), rhythmic and syntactic parallelism, the semantic relationship between “clotted” ( held together ) and “ruptured” ( broken ), and the culturally specific reference to a form of cream commonly associated with the consumption of scones and tea in the southwest of England. Again, these two-word noun phrases fail to conform to the traditional notion of single-word “antonyms,” creating as they do a more complex web of oppositional concepts than the kind of examples Jones uses in his corpora.

Replacive Opposition

Jones includes the frames “X instead of Y” and “X opposed to Y” under his category of “negated antonymy.” In addition, he includes “X rather than Y” as an example of “comparative antonymy.” My view is that, taking the semantic content of these frames into account, “X rather than Y” and “X instead of Y” (and “X in place of Y”) have more in common with each other than the frames and categories in which Jones places them. I have included Jones’s frame “X opposed to Y” in yet a different category (“explicit opposition”; see below).

The category I call “replacive opposition” sits functionally somewhere in between negation and comparison. According to Quirk et al. (1972:671-672), a replacive “expresses an alternative to what has preceded [it]” and conjuncts such as rather indicate that “the proposed alternative is preferable.” They also note that instead might also be treated as a replacive “but more strongly implies a contrast” (Quirk et al. 1972:672).

There are subtle but important distinctions between negated and replacive oppositions that make the latter category useful. First, not does not always necessarily act to conjoin two entities or processes. It can act to negate one of these and leave the reader to infer what the implied alternative is. For instance, “Your dress is not nice” may trigger an implicature that “the dress is horrible” without this actually being articulated.¹² On the other hand, instead [of] and rather [than] necessarily have to conjoin two phrases representing particular concepts. Rather [than] often functions to express a preference for one option over another in a way which treats them as binary oppositions, that is, with only two choices. This is illustrated in example (5) (one of Jones’s canonical examples) and the noncanonical examples (6)–(8) from my data:

(5) Mr Shevardnadze stressed that work for peace rather than war should prevail. (The Independent; Jones 2002:76)

(6) That they had made tea—and coffee—rather than war was borne out last night by Scotland Yard which praised the marchers’ behaviour. (Sunday Times, 23 September 2003:3)

(7) There has been speculation that Mr Michael may propose introducing a licensing system rather than an outright ban in certain areas. (The Guardian, 23 September 2002:4)

(8) He predicted his plans would be published “in weeks rather than months” and that the divisive issue could finally be settled within a year. (The Independent, 23 September 2002:4)

In each case if we substitute not for rather than, there is much less emphasis on the actors involved making a conscious preference for X over Y, whether or not Y has already been previously adopted or chosen.

In examples (9)–(11), the use of instead of and in place of implies that what follows the preposition of is (or has been) already in place or preferred in the past.

(9) Instead of thinking short term, it was time to start thinking long term. (The Independent; Jones 2002:90)

(10) If Bush and Blair throw away the UN rulebook then we are left with the doctrine of might is right instead of the rule of law. (News of the World, 16 February 2003:8)

(11) In place of a charismatic leader, they have the belief that politicians are lying. (The Observer, 16 February 2003:4)

In example (10) the Scottish National Party leader Alex Salmond contrasts “might is right” with “the rule of law,” whose oppositional status relies on overlapping conceptual oppositions such as immorality / morality and illegality / legality , warning against replacing an existing rule of law with morally dubious action. In example (11), the opinion column writer is describing how the public’s faith in politicians has been replaced with cynicism, drawing on canonical oppositions such as honesty / dishonesty and charismatic / bland . All of the examples in this section have to a greater or lesser extent a replacive function in that they express how one of a pair of opposites either is or could be replaced by the other, whether it be by choice or presented as a fait accompli.

Comparative Opposition

Jones’s (2002:76) own definition of what he calls “comparative antonymy” is a helpful one. He defines it as “the co-occurrence of an antonymous pair within a framework that places those words in a comparative context or measures one antonym against the other.” He cites typical frames as “more X than Y,” “X is more [adj] than Y,” and “X rather than Y.” In this section I provide evidence that the first two of these frames can also act to trigger noncanonical oppositions and that the “X rather than Y” frame is best treated as a “replacive” rather than a comparative.

Where comparative frames differ from the negated and replacive ones is the presumption that the qualities expressed in the X/Y pair, or those expressed in the scale against which they are measured, are gradable, or at least treated as such in the specific discourse context. It is difficult to assign comparative qualities using “more” or “less” to words or phrases conventionally assumed to be complementaries, although any words or phrases, regardless of their gradability, can be treated as complementaries or gradable in specific contexts.

Jones (2002:77) draws attention to a useful distinction between “direct” and “indirect” comparison. Direct comparison utilizes the “more X than Y” frame, which is “used to identify the point on a semantic scale which most fittingly characterises that which is being described.” However, the frame “X is more [adj] than Y” is “indirect comparison” because “the sentences above compare antonyms against a separate, specified scale” (Jones, 2002:78).

The following two examples demonstrate direct comparison. In (12), the gradable scale on which the canonical pair “right” / “wrong” is compared is based on the “plane of equivalence” between the two, that is, judgments, and the “plane of difference” degrees of correctness .¹³

(12) And it is possible to accept both that Dr Higgs was a lot more right than wrong in her diagnoses, but that it is now impossible for her to return. (The Independent; Jones 2002:77)

(13) Perhaps a hundred anti-hunting protestors had gathered in Parliament Square, yelling abuse and banging drums, but there was no trouble, and the marchers seemed more bemused than offended by the occasional shouts of “Go home, scum.” (Daily Telegraph, 23 September 2003:1)

In example (13), the same frame is used to compare “bemused” and “offended,” not typical oppositions; hence ascertaining the semantic scale on which the comparison sits is not as straightforward. In this context, what unites the adjectives is their role in describing the reactions of the protestors to the insults hurled at them. The comparison is on a scale which registers the intensity of the protestors’ reactions. The journalist describes a scenario in which antihunting protestors are “yelling abuse and banging drums” and, using another oppositional trigger but, notes that the opposite of what might be expected to happen occurs: “There was no trouble.” The writer uses the reaction of bemusement to emphasize that in other circumstances protestors would take offence and hence trouble may ensue. So the opposition the reader is being asked to consider is that between a mild and strong reaction judged against what might be expected in that specific situation.

Examples (14) and (15) utilize indirect comparatives whereby the semantic scales against which the oppositional pair are compared do not belong specifically to the same dimension connecting the pair as opposites. So, in example (14), there is a scale, effectiveness , which is different from that of the actual canonical oppositional pair “reward” and “punishment,” which are being compared against this scale.

(14) Training would be based upon rewarding good behaviour, because behaviourists, Skinner argues, had found that reward is more effective than punishment. (The Independent; Jones 2002:78)

(15) But more important than the fate of Labour is the fate of mankind. (News of the World, 16 February 2003:8)

Example (15) is taken from an opinion column written by the leader of the Scottish National Party Alex Salmond, a vocal opponent of military intervention in Iraq. Although example (13) presents an interpretation of protestors’ reactions, which are gradable, here it is the evaluative graded quality of importance against which a stark choice between “Labour” and “mankind” is being presented, treated as if they are mutually exclusive complementaries. So even though theoretically importance can be measured subjectively as if it were a gradable quality, when there are only two things being measured, in effect there are no intermediate levels. The rhetoric (reinforced by the parallelism of “the fate of X/Y”) attempts to persuade people to abandon their support for Labour, and in Scotland at least, support the Scottish National Party, who they compete against to be the biggest party.

Both Jones (2002) and Mettinger (1994) label the frame “X rather than Y” as “comparative.” Jones creates a subcategory called a “preferential comparative.” Commenting on whether or not this frame should be classed as “negated antonymy,” he claims,

X rather than Y still reflects some sort of comparison, especially when considered in its literal sense. It also features than, the most reliable lexical signal of Comparative Antonymy. Arguably this makes it more analogous with sentences belonging to the class of Negated Antonymy. (Jones 2002:79)

It is clear that rather tends to express preference (as the section on “replacives” shows). However than specifically expresses comparison only if conjoined with more or less. Comparatives by their nature have to involve some kind of gradability whether directly or indirectly, whereas the “X rather than Y” frame does not.

The following noncanonical example helps bear this out:

(16) To my mind, Channel 4’s draconian action owes more to the tender feelings of media luvvies rather than a stand on a point of principle. (Daily Telegraph, 23 September 2002:11)

Here, a comparative meaning is triggered by more, whereas rather is mainly redundant. However, if more is omitted and the sentence adjusted to, for instance, “Channel 4’s draconian action is a result of the tender feelings of media luvvies rather than a stand on a point of principle,” then rather than implies that the X/Y pair are being treated as complementaries (having a replacive function) and no comparative/gradable quality is being triggered.

Concessive Opposition

The role of the coordinating conjunction but in framing and also generating oppositions would not on the surface seem to be a controversial one. In example (17), taken from Jones’s (2002) corpus, but clearly contributes to contrasting the canonical oppositions “easy” / “difficult” as well as “fall into” / “get out of”:

(17) The First Division of the Endsleigh League is like a well—easy to fall into but difficult to get out of. (The Independent; Jones 2002:57)

Indeed, but is typically labeled as a having “contrastive” or “adversative” function as a coordinating conjunction in many works on grammar. The importance of connectives such as but is highlighted in Izutso’s (2008) distinction among “contrasts,” “concessives,” and “correctives”—the three main categories she claims encapsulate opposition relations in context—whereby but is described as having all three functions depending on its syntactic context.

Jones (2002:57), however, does not include but in any specific category (including “coordinated antonymy”), even though elsewhere he claims that “the word but acts unambiguously as a signal that what comes next should be contrasted with what went previously.” When it features in his ancillary antonymy examples, he claims that “traditionally, the adversative conjunction was thought to be a powerful signal of contrast (Nesfield 1898:79), but corpus evidence suggests that this is perhaps the most dispensable contrast-generating device” (Jones 2002:60, emphasis added). The claim that but seems to play only a marginal role in generating oppositions seems to contradict Izutsu’s classification system as well as much of the evidence in my own examples. This is likely a result of differing methodologies of data collection, classification, and analysis, as I will show.

I synthesize both Izutsu and Jones’s approach to this kind of frame/trigger in favor of creating a new category, the “concessive” opposition, which consists of but and also subordinators such as while, (al)though, despite, and yet, whose roles are to subvert expectations set up in one of the clauses. Some of the confusion around the roles of these conjunctions can, I believe, be attributed to three factors. First, there is the functional difference in purely syntactic terms between but as a coordinator and the others as subordinating conjunctions. Second, they play a much more prominent role in conjoining multiword phrases and clauses than the other frames mentioned so far, a phenomenon that is generally bypassed in studies of co-occurring oppositions. Third, in some cases the opposition generated by concessives is mainly between the expected and the unexpected , concepts that are not inherently part of the meaning of the words themselves participating in the opposition but absorb that conceptual opposition as part of their meaning in the specific context in which they co-occur.

Leech (2006:24) defines a concessive as “an adverbial clause or other adverbial which expresses a contrast of meaning or implication of ‘unexpectedness’ in its relation to the matrix clauses of the sentence of which it is part.” Note here that Leech attributes the qualities of both contrastiveness and unexpectedness as part of the concessive function.

Izutsu (2008:661) concurs on the general function of the concessive that “involves some background assumption or expectation.” One example she uses is “Bill studied hard but he failed the exam” (2008:649), whereby one would expect Bill to pass his exam given the assertion in the first clause that he studied hard.¹⁴ The unexpectedness of his failure in the second clause is signaled by the conjunction. Similarly, in (18) the assertion of a passionate Countryside Alliance demonstration is subverted by its good-natured character in the second clause, creating a noncanonical opposition between “passion” and “good-natured.” The implication here is that “passion” is equivalent to bad-natured (i.e., potentially troublesome) by being counterposed with “good-natured.”

(18) There was plenty of passion but the marchers remained good-natured. (Financial Times, 23 September 2002:3)

(19) Two anti-hunt demonstrators were arrested for public order offences, but a 150-strong anti-hunt counter-protest in Parliament Square remained peaceful. (The Guardian, 22 September 2009:8)

Izutsu (2008:649), however, believes that but plays qualitatively different roles, belonging not only in the category of concessive but also as a “contrastive” and even, in a minor way, as a “corrective” (e.g., “John is not American but British”)—three categories that “are semantically distinct from one another” based on “formal (lexical or syntactic) differences.” She argues that contrast involves comparing two distinct entities, whereas concession involves a subverted expectation about the same entity. So presumably, to adapt the concessive example above, the proposition “Bill studied hard but Phil failed his exams” would be an example of a contrast and not a concessive because of the differing referents (Bill and Phil). It might be tempting therefore to create two separate oppositional categories—“contrastive” and “concessive.” But it is problematic to separate the two so categorically.

To illustrate it is useful to compare examples (18) and (19). In (18), the referent is clearly the same in both clauses: the marchers, whose possible and actual emotions are being contrasted. In (19), two sets of demonstrators (anti- and prohunt) are being counterposed both in terms of their views and their behavior. However, this does not seem to fundamentally change the role of but, which in (19) also seems to be acting as a concessive. The assertion of the first clause, that there were arrests for public order offenses on the Countryside Alliance demonstration, sets up the expectation that the march could have been marred by outbreaks of violence. This expectation is subverted by the information in the second clause. We can check this if we substitute and for but in (19). We can still conceptualize an opposition here, without the necessity of an adversative trigger, because “public order offences” and “peaceful” do not require a huge leap of faith to understand the contrast. However, the “denial of expectation” meaning has disappeared, with both clauses having equal status. We can swap them and conjoin them with and and the meaning is not fundamentally altered. This is not the case with but, which suggests an element of surprise. If we swap the clauses so that the “arrests” clause appears after but, it is the disruption that is unexpected, not the peace. Izutsu (2008:657-658) claims that contrast “does not undergo a noticeable change in meaning even after the order of clauses is reversed and also . . . it is not strikingly affected by replacing the connective [but] with and or by removing it completely.” This seems to assume that the two conjunctions have a purely syntactic role and are empty of semantic content, and it begs the question of why language users would bother choosing one over the other in the first place.

Using the term “contrastive” as a category of opposition relations is also not as preferable as “concessive” because often “contrast” is synonymous with “opposite” (or opposition), unless it refers to a nonbinary contrast (see Figure 1). The term “contrastive” also implies that the other categories are in some way noncontrastive. All the concessive examples that only have one referent in the examples here still include contrast (opposition) of some kind, even if just between the concepts of expectedness / unexpectedness . In example (20), which has only one referent (the couple and their child), the writer implies a contrast between the reader’s expectations that a family born and bred in London would have no sympathies with protestors from the countryside marching through their city, and the actuality that the family joins the protest in support. The opposition is therefore also between city and country on a conceptual level.

(20) Susie Plant, 27, Georgie Denham, 22, and Arthur Godsal, five today, were all London born and bred, but felt compelled to join the protest. (The Times, 23 September 2002:4)

In seeking to downplay the role of but, Jones (2002:57) states that the majority of the sentences he classifies as ancillary antonymy “do not use a contrastive conjunction; antonymy alone is usually considered sufficient to signal clausal opposition. Indeed, when a conjunction is used to link clauses, it is more likely to be additive than adversative.” Jones gives the example in (21):

(21) On Saturday night, as news of Claudio’s death spread, the police presence in Vaulx was heavy, and the violence relatively light. (Jones 2002:58)

It is true that because of the canonical status of “heavy” / “light” as opposites, no other conjunction with a contrastive role is necessary to trigger the opposition; and has a neutral role in conjoining the two clauses, so they could be swapped around without changing the meaning. What Jones does not take into account is that whichever conjunction is used by the writer—and or but—is likely to be a conscious choice between two very different meanings. If the writer had chosen but,” then the inference is likely to be that one would initially expect the police presence to be strongly associated with violence (which didn’t happen), while with and, one infers that the violence was kept to a minimum because of the heavy police presence. It seems therefore problematic to claim that conjunctions are more likely to be additive than adversative when there are other signals of opposition; it depends entirely on what message the writer wants to transmit.

Other conjunctions that perform a concessive (and therefore also an inherently contrastive) role include yet, despite, although, and while. In terms of purely syntactic roles, apart from yet, they differ from but by being subordinators, which means the clauses they head can occur first or second in the sentence. However, the semantic roles they play are very similar to but. If we compare examples (22)–(24) to examples (18) and (19), which use but, it is apparent that these examples imply that given the size of the march, the anger (“passion”) of the protestors, the counter demonstration, and the “bone-chilling three-hour wait,” one would expect some breaches of the peace. That this did not happen is surprising and signaled by the concessive conjunctions (but in (18) and (19), yet and despite in (22)–(24)).

(22) Despite the presence of 150 anti-hunt protesters in Parliament Square, the march was peaceful. (Daily Mail, 23 September 2002:1)

(23) Despite the numbers, the march was peaceful. (The Financial Times, 23 September 2002:1)

(24) From the time the first marchers began arriving to the moment when the columns of people filling the streets finally started to move off towards Hyde Park, it was a bone-chilling three-hour wait. Yet it was always good-humoured. (Sunday Mirror, 16 February 2003:6)

(25) Although the demonstrators included rich landowners and rosy-cheeked farmers in green wellies and waxed jackets, they were joined by an amazing array of like-minded folk such as poverty-stricken labourers and humble craftsmen. (The Sun, 23 September 2002:4)

(26) While it was true that militants, anarchists, anti-capitalists and anti-Americans—what one weary PC called “the great unwashed”—were out in force, the heart and mind of the protest was ordinary people. (Sunday Mirror, 16 February 2003:2)

Jones (2002:57) notes the role of “while” in some of his ancillary antonymy sentences, but only goes as far as saying that “one effect of using subordinators in this way is to signal that the first clause contains ‘given’ information, while that contained in the second clause is ‘new.’” He does not assign the “while X, Y” frame to any separate category.

Examples (25) and (26) also show the problem in trying to distinguish, as Izutso does, between concessives and contrastives on the basis of the number of referents involved. In both of these examples, two sets of referents are being contrasted: rich / poor and extreme / ordinary protestors. Yet this contrast is triggered by subordinators typically performing the role of concessives, so that the second clause is treated as surprising given the circumstances expressed in the first.

Therefore, in my typology I include but, while, although, yet, and despite under the category of concessive triggers. The distinction between whether the attributes of one referent or two are being contrasted is a useful one but does not determine whether each opposition is to be treated either as a contrast or as a concessive (to use Izutso’s distinction). Also, especially in the case of the multiword oppositions in the examples above, each one has to be examined qualitatively in its context of production and consumption to bring out the full complexity of the oppositions being triggered. Indeed, arguably there can be several overlapping conceptual oppositions between both one and two referents. For instance, in (25) and (26), two types of protestors are being opposed, but also if the demonstration is being treated as one entity, then two sides of the same referent are being opposed to emphasize the diversity of the protest. The role of the subordinate concessives as conjoiners of whole clauses rather than simple single-word canonical oppositions is likely to be the reason why they have been overlooked as opposition frames in other studies, in which the triggering of noncanonical oppositions has little priority.

Transitional Opposition

When a Daily Mail journalist reporting on the Countryside Alliance demonstration claims the fox hunting bill will “turn the many decent honourable and law-abiding people . . . into criminals” (Daily Mail, 23 September, 2002:39), the contrast is exemplified in terms of the transformation of what the journalist deems legitimate into illegitimate behavior. Unlike the other categories of opposition, which assert, express a preference for, compare, or demonstrate the unexpectedness of one state or opinion over another, what Jones (2002:85-87) calls “transitional antonymy” and Mettinger (1994:54) “mutation” often involves either a past state that has since changed in the present or a present state that has the potential to turn into its opposite in the future. “Prototypical” frames, according to Jones, include “from X to Y,” “X gives way to Y,” and “X turning into Y” (or “turns X to Y”), an example of the latter framing the canonical pair “weakness” / “strength” in example (27):

(27) Just as the Princess has grown, turning weakness to strength, so, surely must this country’s economy continue to grow. (The Independent; Jones 2002:86)

Another difference between transitional oppositions and the other triggers outlined so far is the variety of forms this category can take, owing to more reliance on semantic content than the others. Many of the frames employ the prepositions to / from; however, synonyms of turn combined with other prepositions are many. Jones logs, for instance, “X falling into Y,” “slip from X to Y,” “lurched from X to Y,” “swung from X to Y,” and “crossing the boundary from X to Y,” among others.

The noncanonical examples in my data make specific rhetorical points about social or attitudinal change, using both phrasal and clausal oppositions in the examples below, in frames as varied as “X has supplanted Y” and “spurned X for Y.”

(28) In the minutes before the march begins, anyone will tell you why protest has supplanted politics. (The Observer, 16 February 2003:4)

(29) British marchers have spurned isolation for solidarity, and fear for fury. (The Observer, 16 February 2003:4)

In these specific cases there is speculation about how the protest marches have transformed the “public” from passive agents at the mercy of an authoritative government to a strong united mass force capable of challenging New Labour.

Explicit Opposition

Sometimes oppositions are framed and triggered using lexis that explicitly draws attention to existing canonical oppositions or signals the user’s intention to treat them as such if they are noncanonical. The clearest examples are those that use phrases such as “X contrast(s)(ed) with Y” or “X opposite/opposed to Y.” For instance, take (30) and (31):

(30) The country people came to London to join in a well-organised, well-behaved march through the streets. They contrasted dramatically with the crowds who sat down for CND, or marched against the Vietnam war, in the days when the amplified voices of Michael Foot and Tony Benn filled Trafalgar Square and sent Leftwing pulses racing. (Daily Mail, 23 September 2002:39)

(31) Dozens of causes were represented. The professionally-produced placards of the protest groups with their fierce messages—“Blair and Bush—Wanted for Murder”—contrasted with cobbled-together banners. “Notts County supporters say Make Love Not War,” said one. (Sunday Mirror, 16 February 2003:2)

In example (30) the columnist John Mortimer, writing for the Daily Mail and an overt supporter of the aims of the prohunting Countryside Alliance demonstrators, is trying to draw a clear distinction between the current demonstration and those the Daily Mail is unlikely to have supported in the past. It is possible that the paper, not noted for supporting mass protests against government policy, is making a determined effort to prove to its readership that this support is the exception rather than the norm. Some evidence for this can be seen in the previous paragraph, where Mortimer describes how to the farming community, up until this particular protest, “marching seemed for political hotheads interested in remote events in distant countries.” As the contrast is made explicit through the trigger, the reader is left to interpret in what ways the Countryside Alliance protest is not like others. If the former is “well-organised” and “well-behaved,” for instance, then we are to infer that the others were not. These others were those that marched against nuclear weapons Campaign for Nuclear Disarmament (CND) or against war, and they were led by former high profile left-wing Labour MPs (Foot and Benn), whose voices were “amplified.” So the Daily Mail reader is reassured that support for this particular march is legitimate, as it is as unlike other kinds of protests. Example (31) also draws unconventional oppositions between “placard” / “banner” and “professionally-produced” / “cobbled-together,” again using an explicit opposition trigger, “contrasted with.”¹⁵

The closest equivalent to the explicit opposition category in Jones’s (2002:81) system is what he calls “distinguished antonymy,” which involves an oppositional pair co-occurring “within a framework that alludes to the inherent semantic dissimilarity of those words.” These sentences are “metalinguistic in the sense that a distinction between antonyms is overtly referred to” (Jones 2002:82). Many of the examples Jones quotes utilize the frameworks “[noun phrase] between X and Y,” with the noun phrases including difference, division, discrepancies, distinction; or “[verb phrase] between X and Y,” such as distinguish and discriminate. More obscure examples of Jones’s “distinguished antonymy” include the gulf between, the gap between, a clear distance between, barriers between, and no man’s land between. When these noun phrases are used, according to Jones, the opposition is already presupposed, as in (32):

(32) But it made the point that the division between gay and straight is one of the many rifts in our society. (The Independent; Jones 2002:81)

Here the writer takes the “gay” / “straight” distinction for granted to make a broader point about social division. The noun phrases in Jones’s data are often followed by a predicate involving a form of the verb “to be” and then a complement, in which the writer is trying to describe this presupposed distinction in some way. It is also worth noting that in all of Jones’s own examples, these noun phrases are modified by a definite determiner like the or this, which act to reinforce the presupposed nature of the distinctions, and hence reaffirm their conventionality. As Jones has already preselected the pairs of opposites for his study, based on their canonical status, it is unsurprising that “the [noun] between X and Y” frameworks feature significantly in this way. In my data, however, there are other frameworks involving some kind of metalinguistic contrast between an X/Y pair: “X contrasts/ed with Y,” “divided into X and Y,” “X and Y polarized,” “different strategy between X and Y,” “X is different from Y,” and “X against Y.” In many of these cases, the context determines how the opposition can be made clear; the distinction would otherwise not be obvious.

The following is a good example of the point above. It is taken from the Daily Telegraph’s report on the Countryside Alliance march:

(33) The Liberty and Livelihood March began ostensibly divided into two camps—those whose primary concern was liberty and the hunting issue, and those whose livelihoods depended on farming and a healthy rural economy. (Daily Telegraph, 23 September 2002:4)

The writer deems it necessary to explain why the organizers of the protest felt compelled to split the march initially into two separate units under the banners of “Liberty” and “Livelihood.” The organizers themselves have constructed a noncanonical opposition, and the writer has drawn attention to that differentiation with the trigger “divided into.” The uninitiated reader needs the further information, in the form of the postmodification after the hyphen, to understand how it is that the two concepts can possibly be treated oppositionally. Even then, to fully process the opposition, the reader would have to understand how the concepts of principle and necessity are interwoven with freedom / restriction and possibly wealthy / poor .

Jones’s category of “distinguished antonymy” is an important one. However, I have subsumed his frames under a heading that more precisely emphasizes their metalinguistic functions (“explicit opposition”). This takes into account other frames, such as the examples above, which use different techniques to differentiate between pairs that cannot be taken for granted as oppositions out of context. It also makes sense to include what perhaps would be an archetypical explicit oppositional frame—“X as opposed to Y”—which Jones includes as an example of “negated antonymy.”

Parallelism

Parallelism involves the repetition of linguistic structures at various levels (phonological, syntactic, lexical, etc.) within which specific lexical items are foregrounded, inviting the addressee to relate them in some way. This usually involves relations of equivalence or difference between the foregrounded items. Parallel structures, according to Short (1996:14), “invite the reader to search for meaning connections between the parallel structures, in particular in terms of the parts which are varied.” Short (1996:67-68) also claims that many parallel structures “push readers towards perceiving semantic relations between words and phrases which do not exist as such in the language system as a whole,” including the creation of noncanonical oppositions.

Jones (2002:56) also describes parallelism as “an important contrast-generating device.” However, he includes the phenomenon as a “contrast-generating element” of ancillary antonymy, and in doing so limits the examples he finds to parallel structures containing at least two pairs of opposites—one pair being ancillary to the other canonical pair and also sometimes noncanonical. One of his examples is as follows:

(34) He leans forward and quotes from a piece of writing in French by Samuel Ullman, which roughly translates as: “You are as young as your faith, as old as your doubts.” (The Independent; Jones 2002:56)

Here the conventional “young” / “old” pairing placed in the parallel structure of “as X[adj] as your A[abstract noun]” creates a slightly less canonical, but still easily comprehensible, contrast between “faith” and “doubts.” However, parallelism can also create unusual contrasts in conjunction with common opposition triggers, such as not and but, without having to rely on ancillary pairings, as my data demonstrate.

It is significant that the four parallelism examples in (35)–(38) are all highly rhetorical, spoken or written by politicians, campaigners, or the journalists themselves.

(35) “ Leave us with our cow s*** and we will leave you to your city bulls***.” (The Sun, 23 September 2002:4)

(36) Jones, who lives in Hertfordshire and shoots, said: “The message we want to give to the Government is that they can walk over our lands but they can’t walk over us.” (The Times, 23 September; Jones 2002:4)

(37) It wasn’t a march, it was an invasion—central London taken over by a million or more peace-lovers. (Independent on Sunday, 16 February 2003:4)

(38) If Mrs Thatcher presided over the collapse of heavy industry, Tony Blair has watched the slow death of farming. (Daily Mail, 23 September 2002:39)

Example (35) is the slogan on one of the Countryside Alliance banners, which employs the oppositions between “us” / “you,” “our”/ “your” in the frame “leave [‘us’ / ‘you’] with [‘our’ / ‘your’] X/Y” to create an inventive distinction between “cowshit” and “bullshit” (asterisks original). These draw on a canonical opposition between “cow” and “bull” and so utilize lexis commonly associated with the countryside, though in this context the “bullshit” associated with the city is by no means intended to be associated with the excrement of the male of the bovine species. Instead, the implied opposition involves an understanding that in English idiom “bullshit” means “lies” or “nonsense.” This being the case, the inference has to be that the way of life of the farming community is the opposite of that of the city—honest and sensible. The parallelism here relies on and also triggers the overlapping conceptual contrasts between country / city, female / male , honesty / lies , and sense / nonsense .

In (36), the actor/footballer Vinnie Jones combines a negator and concessive (“X but not Y”) with the parallelism of “they can [not] walk over X/Y” to create an opposition between “us” and “lands” utilizing a contrast between the literal and metaphorical meanings of “to walk over” to express his resistance. The Independent on Sunday headline in (38) again contrasts individual lexical items “march” / “invasion” in the simple parallel structure “it was [not] a[n] X/Y” to emphasize the size of the antiwar protest.

In (38), the Daily Mail writer relies on the reader’s cultural background knowledge to process an opposition between Thatcher and Blair as former (Conservative) and current - in 2002 - Labour) prime ministers who, it is claimed, oversaw slumps in the economic fortunes of urban and rural areas respectively. The parallelism here relies on a combination of a repeated syntactic structure and some synonymy between the two halves. The second clause uses the same syntactic formula as the first with near-synonymic expressions (“presided over” / “watched,” “collapse” / “slow death”) which then draws attention to the simultaneous difference and equivalence between “heavy industry” and “farming,” which become indexical of an urban / rural opposition. The parallel structure can be expressed as shown in Table 3.

Table 3.

The Parallel Structures in Example (38)

Mrs Thatcher	presided over	the	collapse	of	heavy industry
U.K. prime minister	stative verb phrase	determiner	abstract noun phrase	preposition	noun phrase
Tony Blair	has watched	the	slow death	of	farming

The two clauses are symmetrical in that they begin and end with oppositional pairs while being held together by two pairs of synonyms in the middle.

The range of parallel structures that can trigger oppositions is likely to be vast, providing potential for further research in this area.

Coordination and Category Issues

Some of the examples in the previous section on parallelism use coordinators such as and and but to conjoin two parallel clauses. It has already been demonstrated in the section on concessives how but is an important opposition trigger. It is inadvisable therefore to label examples like (35) and (36) simply as instances of either a concessive or a parallelism, when it is evident that often an opposition is triggered by a combination of frames.

In the discussion above, while I have tried to stress that each type of frame outlined plays an important role in the triggering of oppositions, the categories themselves are not watertight, being merely a set of descriptive tools rather than fixed category classes inherent to the language involved. Many textual oppositions involve a web of triggers, sometimes two or three syntactic ones working together, plus the semantic elements that make up the words involved in the “X / Y” oppositional pairings themselves. The meanings of any given pair of oppositions will be governed by the context in which they appear. Despite occasionally alluding to the fuzzy boundaries between categories, Jones’s method is to treat them as closed sets. This becomes apparent in the very last sentence of Jones’s book when he presents “a new definition of antonymy,” based on the statistics he has gathered from his corpus data. His conclusion is,

Antonyms are pairs of words which contrast along a given semantic scale and frequently function in a coordinated and ancillary fashion such that they become lexically enshrined as “opposites.” (Jones 2002:179)

This conclusion is based on the fact that 38.4 percent of his corpus consists of those pairs he classes as examples of “coordinated antonymy” and 38.7 percent as “ancillary antonymy.” One obvious issue here is that a sizeable number of his examples utilize more than one syntactic frame. For instance, many of the examples of ancillary antonymy also rely on other triggers such as but, while, and and, and may also contain parallelism. The reason for assigning them to the category of ancillary antonymy rather than coordinated (or vice versa) seems arbitrary, as (40)–(42), taken from Jones’s section on ancillary antonymy, show:

(40) Since then, of course, they’ve all had knighthoods, usually when they’re too old to play Hamlet but too young to play butlers in Hollywood movies. (The Independent; Jones 2002:45)

(41) It is meeting public need, not private greed. (The Independent; Jones 2002:46)

(42) Kennedy dead is more interesting than Clinton alive. (The Independent; Jones 2002:49)

Presumably the examples above are included in the ancillary antonymy statistics, yet as the underlined words show, they could also be classed as examples of “coordinated antonymy” (“concessive” in my typology), “negated,” and “comparative antonymy,” respectively, according to Jones’s criteria.

Another problem with Jones’s conclusion lies in the fact that his category “coordinated antonymy” is unduly capacious. Under this heading he includes the frames “X and Y,” “X or Y,” “both X and Y,” “either X or Y,” “neither X nor Y,” “X and Y alike,” and “whether X or Y.” Such a range of frames provide hugely diverse functions. For instance, the headline of the Daily Mail’s (28 April 2010) editorial column commenting on the recent U.K. general election campaign uses the “X or Y” frame to offer a constructed choice between two alternatives: “Face up to our debt or end up like Greece.” However, an “X and Y” frame suggests that the X / Y pair are to be treated as equivalent, rather than showing a preference for one over the other. Also, in my protest data, where an oppositional pair occurs in an “X and Y” frame, it tends to be canonical because there is little in the coordinator and to trigger opposition, unlike or, which is a highly potent oppositional trigger. For instance, in (43) the Sunday Mirror attempts to show the diversity of protestors on the Stop the War Coalition march:

(43) They came from left, centre and right. They were old, middle-aged and young. Rich and poor. Lords, ladies, gentlemen. Students, housewives, bosses and workers from every sphere of British industry. (Sunday Mirror, 16 February 2003:4)

In this example, the pairs “left” / “right,” “old” / “young,” “rich” / “poor,” “bosses” / “workers” are all conjoined by and, relying mainly on their canonical status for their oppositional meaning. The coordinator does not act as an oppositional frame/trigger in the same way that or might. This is apparent in the first two uses of and in the list where a middle ground example (“centre” and “middle-aged”) is included in between the canonical X/Y pairings. The writer, in attempting to show the inclusive nature of the protest, is not relying on a binary construction but uses a list of three, which would not work in the case of an oppositional frame such as “X or Y.” If, for instance, the writer had described some of the protestors as “young and poor,” it is unlikely we would interpret “young” and “poor” as oppositional in this case.¹⁶ Putting the coordinating conjunctions and and or under one heading because of their grammatical function as coordinators ignores their semantic function. These conjunctions, one could argue, can behave as opposites themselves—and having an inclusive function and but an exclusive function. In my revised typology, I have found it unhelpful to have a “coordinated” category for the reasons explained below.

First, the coordinator and is unlikely to generate novel oppositions. It can only frame those that already have canonical oppositional status. Second, or can function both as an inclusive coordinator—therefore having more in common with and—and as offering a mutually exclusive choice between two options, often occurring in the structure “[either] X or Y.” Jones (2002:66) seems to treat the “[either] X or Y” frame as having mainly the first function, such as in the sentence “Most Ugandans, married or unmarried, had several lovers.” This inclusive function therefore does not trigger oppositions as such, but merely frames those we associate with canonical opposites. Jones’s (2002:61) criterion for including an opposite under the class of “coordinated antonymy” is that it must “signal inclusiveness or exhaustiveness of scale.” By “exhaustiveness” he means that what is being referred to takes in everything in between and including (where gradable) X and Y. Of the 533 examples of “X or Y” in his database, Jones does not give an example in which the frame is used to express a choice between the two opposites.

Former U.S. President George W. Bush famously declared after the attack on the World Trade Center, “Either you are with us or you are with the terrorists,”¹⁷ creating an artificial dichotomy between terrorists and the rest of the world, designed to make it difficult not to support U.S. policy without aligning with terrorism. Similarly in example (44) from the protest march data, the journalist offers Tony Blair a choice between only two options—treating the Countryside Alliance protestors as irrelevant or relevant—regardless of whether it is possible to adopt more than two positions:

(44) The only question now is whether Mr Blair still treats those hundreds of thousands of people as an irrelevant minority, or accepts that this time, the countryside really has spoken. (The Times, 23 September 2002:1)

It seems therefore that the frame “[either] X or Y,” when acting to proffer a choice between two alternatives, whether conventional or contextually constructed, needs its own category. In recent work (Jones & Murphy 2005; Jones 2006; Murphy et al. 2009), the need for a category that embraces the frame “[either] X or Y” has been recognized, and so “interrogative antonymy” has been created, which “involves the forcing of a choice between the two members of the antonym pair” (Murphy et al. 2009:2161). This is a sound definition, but not all instances of these forced or constructed choices involve being placed in interrogative structures such as “Is she a good mommy or a bad mommy?” (Murphy et al. 2009:2162), as my examples show. To encapsulate those choices that are presented as between only two options, whether in the form of an interrogative or not, I prefer the term binarized option.¹⁸

Conclusion

This article has aimed to produce a tentative typology of syntactic frames that are able to trigger noncanonical oppositions where they co-occur in discourse and to demonstrate the potential for using these frames to reveal their contribution to the ideological positions relying on constructed binaries prevalent in news texts. I have drawn from and adapted the frames proposed by Jones (2002), whose detailed corpus approach is as yet by far the most extensive.

It is important to note the differences between Jones’s approach and the one undertaken in this article. Jones is, in the main, treating his frames as environments in which to house co-occurring pairs rather than as triggering oppositions. This is because the pairs he chose are canonical and hence the semantic content itself is enough for an opposition to be recognized. The value of his quantitative approach is that he has confirmed or even revealed the existence of many of these frames. My approach has been to explore the triggering potential of the frames, which involves a much closer analysis of their functions—leading to some refinement of Jones’s categories—with a view to examining how the oppositions instantiated in the frames might be processed in their context.

Jones treats as significant the proportion of “coordinated” and “ancillary” oppositions he finds in his data, to the extent that he builds an entire definition of “antonymy” based around their prevalence. His decision to quantify his categories, producing statistics that place each example under only one heading, inevitably leads to unreliable conclusions, given that in my data at least I have found that many examples rely on the overlapping functions of several categories. Jones’s treatment of antonymy as a single-word phenomenon only also means that his results are unlikely to include many examples of those coordinators that tend to be used in longer stretches of discourse, such as but, while, although, and yet.

To give a full account of textual opposition, both quantitative and qualitative approaches are valuable. The corpus approach can be used to search for the oppositional frames to discover where creative examples of opposition have been generated and may be clustered in texts for a particular effect. However, the close examination of the text explored in my study has revealed other kinds of frames that can also trigger noncanonical oppositions and may not be obvious to the casual reader. The qualitative approach opens up ways to explore the effects of binary positions on text consumers. The typology of opposition triggers can be used, for instance, in studies that employ critical discourse analytical methodologies to expose the way ideologies are embedded in texts. Much of our perception of reality is filtered through news texts, and they regularly use constructed binaries as part of their rhetorical armory, resulting in hiding or ignoring alternative perspectives and representations. The more insidious uses of constructed oppositions include those that exacerbate or even create conflict by representing individuals, groups, races, and nations as if they were a homogenous group, to be either unequivocally promoted or maligned. My hope is that other researchers will take on board the opposition typology presented in this article and add it to their analytical tool kit.

Footnotes

Appendix

Typographical Conventions

X / Y	Indicates X and Y (representing words, phrases, clauses) are in an oppositional relationship
Bold	Indicates all co-occurring examples being treated as oppositional pairs both in and out of context, e.g., “The oppositional pair ‘hot’ / ‘cold’ are equivalent as examples of temperature” or “I like my coffee hot not cold”
Italic	Indicates syntactic triggers for oppositions, e.g., “the negator not acts as an oppositional trigger in the previous sentence” or “I like my coffee hot not cold”
Bold italic	Indicates words or phrases which are both syntactic triggers and a member of the X / Y of an oppositional pair, e.g., “ Yesterday I wanted coffee but today I wanted tea”
small capitals	Indicates concepts treated on their own, e.g., “‘Hot’ and ‘cold’ are both equivalent as examples of temperature”
italicized small capitals	Indicates concepts which act as the plane(s) of difference for an oppositional pair, e.g., “What separates hot and cold is their position on a scale temperature levels ”
bold small capitals	Indicates two concepts that are being placed in a position of opposition, e.g., “‘Fire’ and ‘ice’ stand in a position of opposition as representative of the canonical conceptual oppositions HOT / COLD ”

I have inserted these typological conventions into Jones and others examples for consistency, where they are not included in the originals.

The author(s) declared no potential conflicts of interest with respect to the authorship and/or publication of this article.

The author(s) received no financial support for the research and/or authorship of this article.

Notes

References

Cruse

D. Alan

. 1986. Lexical semantics. Cambridge, UK: Cambridge University Press.

Cruse

D. Alan

. 2004. Meaning in language: An introduction to semantics and pragmatics. 2nd edn. Oxford, UK: Oxford University Press.

Davies

Matt

. 2007. The attraction of opposites: The ideological function of conventional and created oppositions in the construction of in-groups and out-groups in news texts. In Lesley

Jeffries

Dan

McIntyre

Derek

Bousfield

(eds.), Stylistics and social cognition, 79-100. Amsterdam: Rodopi.

Davies

Matt

. 2008. Oppositions in news discourse: the ideological construction of “us” and “them” in the British press. Huddersfield, UK: University of Huddersfield dissertation.

Fellbaum

Christiane D

. 1995. Co-occurrence and antonymy. International Journal of Lexicography 8(4). 281-303.

Izutsu

Mitsuko Narita

. 2008. Contrast, concessive, and corrective: Toward a comprehensive study of opposition relations. Journal of Pragmatics 40(4). 646-675.

Jeffries

Lesley

. 1998. Meaning in English: An introduction to language study. New York: Palgrave.

Jeffries

Lesley

. 2010a. Critical stylistics: The power of English. Basingstoke, UK: Palgrave Macmillan.

Jeffries

Lesley

. 2010b. Opposition in discourse: The construction of oppositional meaning. London: Continuum.

10.

Jones

Steven

. 2002. Antonymy: A corpus-based perspective. London: Routledge.

11.

Jones

Steven

. 2006. A lexico-syntactic analysis of antonym co-occurrence in spoken English. Text and Talk 26(2). 191-216.

12.

Jones

Steven

Murphy

M. Lynne

. 2005. Using corpora to investigate antonym acquisition. International Journal of Corpus Linguistics 10(3). 401-422.

13.

Jones

Steven

Paradis

Carita

Murphy

M. Lynne

Willners

Caroline

. 2007. Googling for opposites—A Web-based study of antonym canonicity. Corpora 2(2). 129-155.

14.

Justeson

John S.

Katz

Slava M.

1991. Redefining antonymy: The textual structure of a semantic relation. Literary and Linguistic Computing 7(3). 176-184.

15.

Leech

Geoffrey

. 2006. A glossary of English grammar. Edinburgh, UK: Edinburgh University Press.

16.

Levinson

Steven C.

1983. Pragmatics. Cambridge, UK: Cambridge University Press.

17.

Lyons

John

. 1977. Semantics. Cambridge, UK: Cambridge University Press.

18.

Mettinger

Arthur

. 1994. Aspects of semantic opposition in English. Oxford, UK: Oxford University Press.

19.

Murphy

M. Lynne

. 2003. Semantic relations and the lexicon. Cambridge, UK: Cambridge University Press.

20.

Murphy

M. Lynne

Paradis

Carita

Willners

Caroline

Jones

Steven

. 2009. Discourse functions of antonymy: A cross-linguistic investigation of Swedish and English. Journal of Pragmatics 41. 2159-2184.

21.

Nahajec

Lisa

. 2009. Negation and the creation of implicit meaning in poetry. Language and Literature 18(2). 109-127.

22.

Nesfield

J. C.

1898. Manual of English grammar and composition. Toronto: Macmillan.

23.

Quirk

Randolph

Greenbaum

Sidney

Leech

Geoffrey

Svartik

Jan

. 1972. A grammar of contemporary English. London: Longman.

24.

Short

Mick

. 1996. Exploring the language of poems, plays and prose. London: Longman.