When Evaluative Adjectives Prevent Contradiction in a Debate

This paper argues that some words are so highly charged with meaning by a community that they may prevent a discussion during which each participant is on an equal footing. These words are indeed either unanimously accepted or rejected. The presence of these adjectival groups pushes the antagonist to find rhetorical strategies to circumvent them. The main idea we want to develop is that some propositions are not easily debatable in context because of some specific value-bearing words (VBWs), and one of the goals of this paper is to build a methodological tool for finding and classifying these VBWs (with a focus on evaluative adjectives). Our study echoes the importance of “cultural keywords” (as reported by Wierzbicka, Understanding cultures through their key words: English, Russian, Polish, German, and Japanese, 1997) in argument (as reported by Rigotti & Rocci, Argumentation in practice, 2005), but is rather based on a German approach developed by (as reported by Dieckmann, Sprache in der Politik: Einführung in die Pragmatik und Semantik der politischen, 1975), (as reported by Strauss and Zifonun, Der politische Wortschatz, 1986), and (as reported by Girnth, Sprache und Sprachverwendung in der Politik: Eine Einführung in die linguistische Analyse öffentlich-politischer Kommunikation, 2015) about “Miranda” and “Anti-Miranda” words that is expanded and refined here. In particular, our study tries to understand why some statements, fueled by appreciative (Tseronis, 2014) or evaluative adjectives, have such rhetorical effects on a pragmatic level in the particular context of a vote on the Swiss popular initiative called “for more affordable housing”. This context is fruitful since two parties offer reasons for two opposing policy claims: namely, to accept or to reject an initiative. When one party uses arguments containing such universally unassailable adjectival groups to defend a “yes” vote (in our example, pleading for more affordable housing rents), the opposing party cannot use a symmetrical antonym while pleading for the “no” vote. The methodological tool that is proposed here could shed light on the use of certain rhetorical and referential strategies in conflicting policy proposition contexts.


Introduction: About Undisputable Values and Their Effects
On the 7th of January 2019, a few weeks before a Swiss vote on the popular initiative called "more affordable housing", Jean-Luc Addor, a right-wing member of the National Council, wrote the following sentences on the Swiss People's party official website: "Launched by the left-wing party, the popular initiative 'more affordable housing' obviously seduces with its catchy name. Indeed, who in this country does not want to pay less for housing?" (Addor, 2020). 1 It is not the only example of a metadiscourse about the name of this initiative. 2 It seems to have worried groups opposed to the initiative and was often described as "catchy".
The main idea we want to develop is precisely that some propositions are not easily debatable in context because of some specific value-bearing words (VBWs), and therefore they constitute a powerful rhetorical strategy, pushing the opponent into silence or into an avoidance strategy. These specific VBWs will be flagged here as "Miranda words", a concept used by German-speaking discourse analysts, which may be an interesting tool to analyse argumentation and rhetorical moves in a debate. But their model is not completely clear on how to identify this type of word within a corpus. This is the main methodological question we want to address. The answer should probably imply focusing on VBWs rather than on a grammatical class (adjectives): Some effects that we are interested in could be triggered by nouns or adverbs and not only adjectives. For the clarity of the paper, we will consider nevertheless the prototypical case of evaluative and axiological adjectives. The reasons for this are threefold: They are sufficient to build the foundations of a methodological process which is one of the goals of this paper; they are identified and the most described in the aforementioned literature about evaluation in discourse; and they are more used in predicative expressions ("X is eval. adj"). Therefore, we focus on them in this study.
In our case, the name of this initiative "For a more affordable housing" is intrinsically evaluative, since it presupposes (Beaver et al. 2021) that housing rents are mainly not affordable enough and implies that house renting can be assessed against a standard, namely what can be counted as affordable. Subjective evaluative adjectives play a crucial role in argumentation, since they are often sufficient to determine an entire category of claims or standpoints: "The types of standpoints supported by When Evaluative Adjectives Prevent Contradiction in a Debate argumentation vary from descriptive standpoints ('The King of the Netherlands is inaugurated in Amsterdam') to evaluative standpoints ('The Mahler concert in the Concertgebouw was excellent') and prescriptive standpoints ('You should come with me to church this Sunday')" (van Eemeren et al. 2014: 7, italics are ours). This is echoing a familiar tripartition in many textbooks about debate and argument: propositions of fact, propositions of value, and propositions of policy.
Our approach to argumentation is embedded in linguistics and pragmatics, and the question of determining subjective or evaluative utterances is an important topic in these domains (Bednarek 2006(Bednarek , 2009Englebreston 2007;Jackiewicz 2014;Kerbrat-Orecchioni 1980;Legallois & Lenepveu 2014;Martin & White 2005;Tseronis 2014; Wiebe et al. 2005). In argumentation studies, with the notable exception of Tseronis (2014) about "appreciative modality" and whose aim is "to study the effect of including appreciation in the argumentative discourse" ( §14, we translate), these prolific studies are rarely quoted, to the best of our knowledge. Another exception is the domain of argumentation mining, whereas questions about linguistic marks on objectivity, certainty, and evaluation are important (Park & Cardie 2014;Rubin et al. 2006;Stede & Schneider 2019). Unlike Tseronis, who starts from a formal approach to argumentation, we will focus less on the different marks of evaluation in sentences than on their social reception in context. In this sense, our study will be essentially framed into rhetorical effects of these forms rather than into linguistic descriptions.
The methodological tool we want to build is just a means towards this end: the study of the rhetorical effects of VBWs in an argumentative situation, especially a deliberative one. Our general background is based on rhetorical-pragmatic discourse analysis crossing argumentation theories (Oswald & Herman, 2016), and by rhetorical effects, we mean the positioning revealed by the speaker's discourse in a social space (πόλις-a City in the Greek sense of community of values) as well as the epistemic effects created by this positioning, that is, the potential impact that the discourse can have on the audience's beliefs (strengthening, weakening or changing beliefs). For this case, we are interested in this question about a rhetorical effect: How may certain evaluative adjectives have an impact on actual political deliberation, especially on a balance between a protagonist and an antagonist? It should be highlighted that our case study illustrates our theoretical proposal; however, our results should be considered as a rationale that may explain the pragmatic effects we are interested in, but these effects could be or should be tested by further experiential studies.
We will first present some theoretical backgrounds that help build our research on qualifying adjectives (Sect. 2) and a practical background for the case that we will study, namely the Swiss political voting system and the case of an initiative put to the vote of Swiss citizens in 2020 (Sect. 3). 3 Then, we will explain how we classify adjectives and specifically the evaluative ones in our corpus (Sect. 4). In Sect. 5, we will present and use a not widely known German approach to political discourse analysis using, among others, the concept of Miranda words. This category seemed promising to us, for mainly one reason: The words in this category are expected to be unanimously approved by the audience. We will then develop a flowchart of our invention to classify evaluative adjectival groups (Sect. 6) to illustrate how this flowchart can be used and why using it may be of some interest for analysing rhetorical strategies (Sect. 7-9). Finally, this will allow us to note and comment on some specific argumentative strategies relying on the use of evaluative adjectives (Sect. 10).

About Evaluation in Argumentation
The aforementioned typology (proposition of fact, value or policy) presupposes an ability to classify propositions in one of the three categories. From a linguistic point of view, which prefers to study utterances as they appear rather than a reconstruction of an argument (van Eemeren et al. 1993), it is nevertheless not as clear-cut as that. For example, a proposition of policy, "we should reject this violation of fundamental human rights", may presuppose a proposition of value, "X is a violation of fundamental human rights". Tseronis (2014) also mentions for example "detached appreciative modalities" like in this example "Unfortunately, the appointment has been rescheduled": The main sentence is a proposition of fact, but there is also an embedded proposition of value, "It is unfortunate that X". In our first example, "for a more affordable housing" is not really a proposition as such: it should be interpreted as a proposition of value ("Housing is not affordable enough"), which is a premise for an implicit policy standpoint "you should accept this initiative to have more affordable housing". Hinton states that these types "are distinguished largely by the contents of their predicates" (2021: 176, we highlight) and therefore shows that the reasoning must already be analysed, decomposed into single propositions before a decision regarding the nature of the proposition. But Hinton, following Wagemans (2016), uses this step of analysis to identify argument schemes, which is not our goal here. In order to analyse rhetorical effects, we argue that we should examine the "raw material" first, before any reconstruction.
Attention to the influence that words may have on a social or cognitive level, even if they are omitted in a set of propositions which constitute an argument, is obviously not new in argumentation. Whether it is Grize's natural logic (1990), of which we find as affiliation in the concept of the "argumentative dimension" proposed by Amossy (2005), or whether it is the work from the school of argumentation in language (Anscombre & Ducrot 1983), the impact of the choice of words is at the centre of their reflections, independently of a set of propositions. While considering these effects under the label of argumentation is probably exaggerated and problematic for the interdisciplinary dialogue (Herman, 2018), the effects of the orientation given by the words in a situation cannot be eluded in a rhetorical situation (Bitzer 1968). For example, Wierzbicka's cultural keywords (1997), the role of which in argument has been shown by Rigotti and Rocci (Rigotti & Rocci 2005;Rocci 2009), illustrate the impact of lexical choices on argumentation and reasoning. Keywords, defined as "particularly revealing of the values of a culture" (Rigotti & Rocci 2005: 903), are methodologically difficult to pinpoint 4 but "give access to fundamental beliefs, values, institutions and customs" of a culture (Rigotti & Rocci 2005: 903). While keywords, by their nature, refer rather to nouns, or even flag-bearer nouns for a community, our study aims at evaluative and axiological adjectival groups, for the reasons we mentioned earlier. Even if they do not have the same symbolic "weight" as a noun and its connotations, the idea of an expression of fundamental values for a community is nevertheless kept under consideration. We will use the concepts of Miranda and Anti-Miranda adjectives, which can be defined as culturally unanimously praised or blamed adjectives (Sect. 5), but we are sure that they should be closely related to the idea of cultural keywords.
We will observe qualifying adjectives in our corpus. As said earlier, evaluative adjectives are one of the main categories capable of determining or identifying a proposition of value, the definition of which is: "an evaluative judgment about a particular entity based on a subjective selection and weighing of assessment criteria" (Wagemans 2020). However, the evaluative adjectival groups may appear in a statement of fact or policy as well, like in this example from our corpus: "It is therefore not advisable that the supply of housing offered by housing cooperatives be developed in an excessive manner and at taxpayers' expense." If we reconstruct the argument, "advisable" helps reconstruct a statement of policy that embeds a proposition of value (manifest through the adjective: "excessive"). Once again, our main interest is to wonder if "advisable" and "excessive" have the same rhetorical effects.
Catherine Kerbrat-Orecchioni, a French linguist, proposed in 1980 a useful model to sort different types of adjectives. It is a way to classify in particular how the subjectivity is inscribed in language and in adjectives. In it, she distinguishes four categories: • Objective adjectives: where all traces of a speaker's presence are supposedly erased • Affective adjectives: They state a property of the object and an emotional reaction of the speaking subject in front of this object (e.g., "funny", "pathetic") • Subjective evaluative non-axiological adjectives: They imply a quantitative or qualitative evaluation of the denoted object, without any value judgment or emotional commitment on the part of the speaker (e.g., "big", "numerous" …). They are related to an idea of a standard ("big" for this kind of object). • Subjective evaluative axiological adjectives: They have a double standard-the internal standard of the class of the denoted object and the internal standard of the speaker and her/his value judgments. For example, in "the tree is beautiful", the adjective "beautiful" is evaluated in comparison to other similar trees and in respect to the speaker's judgment on beauty of the trees.
An axiological adjective implies therefore an intrinsic and an extrinsic dimension. The first one is the linguistic expression of the value: the lexical means to express a value, that is, the semantic angle. The second one is the assumption of this value by the speaker, that is, the enunciative angle and the commitment by the speaker in the value that he or she takes in charge. 5 But it seems relevant to also consider a third dimension, considering axiological adjectives from a pragmatic angle, namely the question of their reception. In Addor's example, we assume that the adjective "affordable" has a central significance, because the reception of it should be unequivocally favourable in the contemporary Swiss context. In a pro or contra debate, the fact that the contrary of some evaluative adjectives cannot be defended by the opposing side, that is, the fact that these adjectives are necessarily considered or planned as unanimously referring to either a bad or a good value, can be strategically critical. They are indeed difficult, if not impossible, to counter. When faced with an evaluative adjective X defended by one side, it appears that it is sometimes socially, politically or publicly impossible for the antagonist to plead for a symmetrical non-X. However, it is a constraint of the referendum to vote either yes or no, in other words, either for "more affordable rents" or against it. In the secrecy of the voting booth, this is not an issue. However, public figures who intend to oppose this initiative are taking the risk of an ethos of cruelty against disadvantaged people. So, they must not only justify their point of view, which is against the doxa, but should also be concerned about their image and show that they are still worthy of the community and its values. We argue that some adjectives are so widely accepted (or rejected) by a community that they may prevent a discussion during which each participant is on an equal footing. The presence of these adjectival groups pushes the antagonist to find rhetorical strategies to circumvent them. However, before looking at strategies and consequences of the evaluative adjectives, we need to identify them in a corpus.

The Swiss Political Context: Official and Unofficial Framing of the Argumentation Around a Popular Initiative
The Swiss political context and its many votes over the course of a year allow us to observe particularly interesting cases of argumentation, with multiple valuebearing propositions. In the Swiss system of semi-direct democracy, citizens have the possibility of approving or rejecting propositions of amendments to the Swiss Constitution. Any small group of Swiss citizens has the possibility to submit to the Confederation their propositions of amendments to the Swiss Constitution as well. These spontaneous proposals of the people are called "Popular Initiatives". Launching a popular initiative in Switzerland is a long process: Sometimes it can take several years. The initiative goes through various procedures before approval by the Federal Council and the Federal Chancellery. These institutions establish an official title, which neither contravenes legality nor is "marketing". 6 The opinion of the Federal Chancellery, the Federal Council, and the Parliament, but also the arguments of the initiative committee and of the opponents are included in an official brochure. Finally, this brochure is distributed and, at least three weeks later, the people vote directly for or against the initiative. Thus, in February 2020, an initiative officially named "More Affordable Housing" was put to the vote of Swiss citizens. The initiative committee required the construction of low rental housing to be encouraged. To this end, the text of the initiative, largely driven by the left-wing parties, proposed to establish a fixed quota of 10% of new constructions for public utility housing. The committee also called for the monitoring of the abusive increase in rents by private real estate companies. Their main arguments were as follows: The current supply of low-rent housing is too low in relation to demand; the less well-off population does not have the same opportunities to find housing as the wealthy population of Switzerland: They are even sometimes forced to leave housing because of rising rents; speculation and energy renovations undertaken by private real estate companies lead to high or even unreasonable increases in the price of rents. The parties rejecting this initiative were predominantly right-wing, but they were joined here by the Federal Council and the Parliament. The majority of the opponents' arguments were as follows: There is currently sufficient affordable housing in Switzerland; the initiative is likely to cost the federal government, and the taxpayer, a lot of money; the 10% quota was particularly criticized, as it does not consider the reality of demand and would be too rigid; entrepreneurial freedom is restricted, as speculation on housing could be hampered; and finally, the monitoring of rent increases in the event of energy efficiency improvements could hamper ecologically beneficial improvements (heating, insulation, etc.) (Office fédéral du logement [n.d.]; Logements-abordables.ch 2019; NON à l'initiative extrême sur le logement 2019).
A certain amount of time (about eight weeks) elapses between the moment when the Swiss people are informed of the forthcoming vote and the moment when the votes are counted. During this period, besides the text of the official brochure, Switzerland's political parties lead their campaigns and present their standpoints on the initiative, on various platforms. In doing so, they may have recourse to a different, unofficial, framing of the proposal (e.g., "l'initiative extrême sur les logements", or "the extreme housing initiative") and more "marketing" or "subjective" arguments. In this context, posters, leaflets, but also websites specifically dedicated to the "pros" or "cons" of the vote emerge and since they explicitly have pages that offer lists of key arguments ("Argumentaires" in French), they constitute an interesting corpus in which we can find multiple value-bearing propositions. Therefore, we collected our data on the website "Logements-abordables.ch" (2019) created by the Pro-Party during the campaign, and on the website "NON à l'initiative extrême sur le logement" (2019), created slightly afterward by the opposing party. More specifically, our data are composed of two webpages for each political party: the home page and the arguments page. On the Pro-Party's website, we found 63 adjectives (29 of them from the front page, 34 of them from the arguments' page), and on the Con-Party's website, we gathered a total of 207 adjectives (24 of them from the front page, 183 of them from the arguments' page). It is not a broad corpus but it is enough to test the accuracy of our methodological process: Introducing a tool for analysing such a corpus is the main goal of this paper.

What Kind of Adjectives? Our Corpus
Based on the typology of adjectives from Kerbrat-Orecchioni (1980), we first notice that, even if the objective adjectives were numerous in our corpus (we gathered 150 of them), they play no role in an argumentative approach of discourse: They are used to name some referents, such as "National Council", "Scientific study" or "Middle class".
The subjective adjectives are far more interesting. We notice that among our 269 adjectives, there is barely one affective adjective, located on the Con-Party's front page: "all united". 7 We also found 28 evaluative non-axiological adjectives and 90 evaluative axiological adjectives. These 119 adjectives (Table 1) were the best candidates for our analysis on how subjective and VBWs may impact the debate or constitute a rhetorical strategy for the parties. In such a context of political argumentation, subjective adjectives are rather polarized between, on the one hand, negative adjectives criticizing the opponent and, on the other hand, positive adjectives developing the "proponent's" point of view, specifically when they offer values and hierarchy of values. The number of footnotes in the following table gives witness that this identification of adjectives or adjectival groups is rather challenging because of the context dependency of our interpretation.
In addition to the frontpages, we chose to examine "argumentaires" (in French) which offer, in a more developed way than on the homepages, all the arguments converging towards a policy claim: "You should accept/reject the initiative." The deliberative status of this issue explains why we want to study adjectives relevant to it and how opposing parties respond to each other. Some adjectives of our list appear to be "questionable" or "arguable", but we wanted to take a closer look at adjectival groups that can be described as unquestionable. Such adjectives are indeed loaded with such strong social values that it becomes impossible to counter them publicly. Compared to the former, which tend to open or continue the exchange of arguments, the latter tend to close or to prevent discussion in these terms. We have also chosen to focus our attention on these adjectives, because they appear to be a central focus in the argumentative strategy of the Con-Party: If these adjectives are not salient from a quantitative point of view, they are qualitatively highlighted by the presentation of the website (Fig. 1):

Miranda and Anti-Miranda
It was striking that some adjectives could be countered by their antonyms in the context of the rhetorical situation, while others only offer publicly indefensible antonyms. To take a blunt example, an anti-racist policy can be easily defended in public, while pleading for a racist policy is not conceivable. The same applies for to "Good [social] mix" in our data: Indeed, claiming to be "against" co-education would be xenophobic and publicly unspeakable. The symbolic weight of these words that reveal the deep values of a community had to be considered, and that is why we refer to cultural keywords (see Sect. 1, above). Now, German linguists (Dieckmann 1975;Strauss & Zifonun 1986;Burkhardt 1998;Girnth 2015;etc.) analysing political discourse have been, since the mid-1970s, precisely identifying categories of keywords (and thus not only adjectives). Their typology has not to our knowledge been taken up or noted outside German-language publications. Girnth (2015: 88) mentions the difficulty of fully grasping the ideological dimensions of the language and points out that a term such as "democracy" denotes a political system, based on the characteristics of the referent, but also contains, at the connotative level, features of evaluative and deontic dimensions: a "great good that must be preserved". We have chosen for this analysis the clear expression of a position through subjective adjectives, and we use them to describe the following notions: Miranda and Anti-Miranda words (Dieckmann 1975: 49),-also called watchwords (Leitwörter) by Strauss and Fig. 1 Screenshot of the arguments' page on the Con-Party's website (the headlines of the three columns are uniquely composed of adjectives: rigid, bureaucratic, costly, statist, uncertain, and counterproductive)

When Evaluative Adjectives Prevent Contradiction in a Debate
Zifonun (1986: 100) or high-value words (Hochwertwörter) by Burkhardt (1998)the Stigmawords (Stigmawörter) and the Flagwords (Fahnenwörter). 8 These notions deserve, in our opinion, to be integrated into the study of political argumentation in a more visible way. They are, in fact, promising for the study of debates and dissent. According to Girnth (2015: 91), they condense and simplify reality; they also convey an emotional appeal which is quite important in a context of political debate; finally, they express an idea which we could not find in other theories of evaluation in language: They are expected to be unanimously received. Let us first note that the Miranda, Anti-Miranda, Stigma-and Flagwords are only nouns-hence the affiliation with the "cultural keywords", but we argue here for an extension of this model, since it is absolutely possible to consider relevant adjectives with the same perspective. Nevertheless, their symbolic nature of a "flag" is probably diminished by the fact that, syntactically speaking, the adjective has a less central, less visible position than the noun, if we consider them in a "figure-ground" perception of the syntax.
The Mirandum ("which is to be admired") is "an ideological linguistic expression which has a positive evaluation for the members of a community" (Girnth 2015: 93), that is to say a large community, like a universal audience in Perelman's and Olbrechts-Tyteca's words (1958): "Naturally, the value of this unanimity depends on the number and quality of those expressing it. Its highest point is reached when there is agreement of the universal audience. This refers of course […] to universality and unanimity imagined by the speaker, to the agreement of an audience which should be universal" (1971 [1958]: 33). A little further on, they insist on the role of values in argumentation, stating that "particular values seem to fade before the universal value [e.g., the Good, the True, the Beautiful] it determines" (1958: 76). We argue that Miranda words and adjectives are similar to philosophers' universal values in their argumentative roles, but a bit different regarding their universality, since they are dependent on a sociological context: No one in our democratic countries today can argue against freedom or peace, but these values can be disputed in other contexts. According to Dieckmann, Anti-Miranda, on the other hand, are the "despicable, fought and undesirable words" (1975: 49), which are unanimously rejected and cannot be defended in an open public speech, like "racism" or "dictatorship". These values and anti-values are not universal, but neither are they particular or disputable. They are part of a slowly but steadily evolving democratic political and sociological environment, with its taboo words, role models and its respected symbols. In brief, the Miranda words are expected to imply, in context, a total admiration and respect whereas Anti-Miranda words are expected to be unanimously rejected. Miranda and Anti-Miranda noun phrases do not leave the place for debate, since it is a priori impossible to find opponents who accept to challenge them publicly. The values that they express are close to absolute values, which cannot be compromised: It will be difficult to argue in favour of a "slightly racist" law or to argue for a "partial peace" or a "partially fair justice".
However, these concepts, fairly intuitive and not clearly defined by the cited authors, remain values for society as a whole at a given time. It seems to us more heuristic to consider a form of spectrum of acceptability for these categories. For example, "increasing security" tends to become a Mirandum expression since the terrorist attacks of the 1990s and 2000s, but it is still possible to plead for less security than that intended by a specific law, whereas it is difficult to plead for a slightly less anti-racist law. Some Miranda are therefore linked to the whole social context; others must be limited to a more precise and restricted context. The extent to which evaluative or value-bearing expressions are scalable in context is a key point for the methodological process we need to identify and analyse Miranda and Anti-Miranda.
In contrast to Miranda words, which are supposed to be unanimously praised in a community that includes both supporters and opponents of an issue, Flagwords and Stigmawords are also unanimously adopted, but inside a subgroup (excluding the opponents) and as a core element of its ideology. Each political party develops its ideology and distinguishes itself from other ideological currents of thought. According to Strauss and Zifonun (1986: 116), words that are claimed by a current of thought are the "Flagwords" and those that reject the opposing ideology the "Stigmawords". The Flagwords of a political party are often the Stigmawords of another party. Girnth gives the example of "pacifism" for the German political reality of the 1980s as a Stigmaword for right-wing politics and a Flagword for the left-wing activists of the time.

Towards a Methodological Process
Going back to our corpus, Kerbrat-Orecchioni's proposal for classifying adjectives helps us identify axiological adjectives, but some examples were difficult to sort, since they were not axiological in language but in context. Of course, one could wonder if the strategy of the pro-and con-parties is "just" to use more or less evaluative and axiological adjectives in proportion of the total of adjectives. However, in both tables below (Tables 2 and 3), looking at the percentage of adjectives to the word totals, or the percentage of evaluative adjectives to the number of words, there is no truly significant result. We may notice that there is a slightly higher proportion of evaluative adjectives in the Pro-Party in proportion to the number of adjectives (58.73%) than in the Con-Party (39.8%); although the number of adjectives is very slightly less in the Pro-Party text (9.18%) than in the Con-Party text (10.5%). And yet this difference of proportion is not salient and the rhetorical effect of the use of evaluative adjectives does not seem to be due to their quantity, but rather should be studied from a qualitative point of view.
Using only Kerbrat-Orecchioni's classification, and based on these statistics, we can neither infer a communicative strategy specific to the Pro-Party or the Con-Party, nor explain our prima facie intuition. Therefore, we needed another methodological process to identify and sort the evaluative and axiological adjectives in our corpus to determine if these evaluative adjectives could be considered as "Miranda/

When Evaluative Adjectives Prevent Contradiction in a Debate
Anti-Miranda" or "Flagwords/Stigmawords" respectively. Considering the definitions given by the German political discourse analysts as well as the nuances we mentioned above, we try to build a flowchart to help us classify these adjectives that may fit with the definitions. Now, classifying subjective adjectival groups into the German categories is also a difficult step, because no clear-cut criteria are offered in the above-mentioned literature to identify and sort these kinds of words. The same difficulty has been noted above for Wierzbicka's cultural keywords (1997).
An analyst interested in rhetorical strategies used to defend an idea or to counter it should wonder if the analyzed adjectives are already a matter of controversy, or if they could trigger a controversy in this context. This question is the first step of the flowchart. It appears firstly that some axiological adjectives were used only by one party and are not at-issue in the controversy. In our corpus, the need for "clean air", for example, is mentioned by the initiators, but is not at-issue. We discard them as non-relevant for the debate.
Then, since Miranda is opposed to Anti-Miranda and Stigmawords to Flagwords, the next question should separate adjectives according to their polarities: negative (Stigmawords & Anti-Miranda) or positive (Flagwords & Miranda) adjective. Again, some adjectives are negative only in their context of enunciation, even if it can be positively connoted in language: "Powerful real estate lobby" is negative for the initiators, since it echoes a handful of powerful people against whom they have to fight, even if in language, "powerful" can be positively connoted.
The next step is a question that helps identify Miranda adjectives. For the analyst treating an evaluative axiological positive adjective X, a good way to determine if it is a Mirandum is asking whether the opponent may openly and publicly defend a symmetrically opposing point of view (non-X) in the context. If the answer is "no", it can be labelled as a Mirandum word. If the answer to the first question was "yes", then the analyst needs to wonder if the opponent may defend non-X without alienating his or her own party, in the sense that defending such a point of view can be risky for the cohesion of the party. If the answer to this second question is "no", then this adjective is a Flagword for the protagonist and the antagonist cannot defend it without taking the risk of "betraying" his or her own ideology. If the answer is "yes", it means that defending the positive evaluative word X or non-X is unproblematic, debatable, as well for the community as a whole as for the specific community of a party. Such adjectives can therefore be considered as "arguable". For the adjectives with a negative polarity, the first question needs to be about the assumption of the negative polarity by the opponent. It can indeed occur that the opponent defends, maybe not with the same loaded words that the proponent would use, something that is worded here negatively but is not considered as unanimously blamable. For example, someone can argue that it is perfectly reasonable to defend "powerful real estate lobbies" for economic reasons. Negatively described by one party, these adjectives are in line with the party's philosophy and can be socially and publicly claimed. They are Stigmawords for a party that are praised and embraced by another one (probably in other wording). If the opponent does not embrace the negative adjective, then a second question should make a difference between unanimously rejected or potentially conceded adjectives. "Could the opponent concede a weakened version of the adjective" helps the analyst identify if the matter can be disputed or not. We propose the test of the insertion of "to a certain extent" to help in this task. Indeed, it seems possible for one party to argue for "a rigid initiative to a certain extent", while it is impossible to publicly plead for an "unfair initiative to a certain extent". If it is impossible for the opponent to concede the negative adjective even in weakened words, then it should be considered as an Anti-mirandum adjective. Otherwise, we have only regular debatable adjectives such as "a longer process", "lengthy discussions about the application of the law", etc.
These sequences of questions are reproduced in the following flowchart (Fig. 2). This typology, which should be used and tested on many more examples, may be an interesting scientific tool to analyse corpora to get a picture of the field of possible and impossible arguments. It could be important to observe when the argument stops because it arrives in a minefield of Miranda and Anti-Miranda words, what the obstacles of the argument are, and the rhetorical strategies to bypass them.

Application of the Flowchart on our Corpus: Repartition and Ratio
The application of the flowchart to each adjective in our corpus makes it possible to observe their distribution from a quantitative point of view. The result of this application is shown in Tables 4 and 5: The first thing we notice is the very low quantity of arguable adjectives in the Pro-Party (only two, which represents 5.4% of their evaluative adjectives). Our interpretation is that the Pro-Party relies on the "Miranda" strategy or the "obviousness" strategy. Their strategy would be to act as if the values carried by the initiative were not questionable, since they are universal and obvious. This assumption is supported by the fact that the party compares, on its website, the need for accessible housing with the need for fresh air.
The second thing we notice is the very high number of Miranda words among the Pro-Party evaluative adjectives (23, which represents 62.16% of the total). It is also interesting that they are exclusively positively polarized. Our interpretation is that the Pro-Party's argument is designed for voters, with the aim of reaching as many people as possible. Since a citizen's initiative must collect "only" one hundred thousand signatures, it is important to give the impression that it concerns everyone. Otherwise, the initiative risks being accused of concerning only those who launched it, this is to say, a very small part of the Swiss population. Therefore, using the Miranda word category is indeed an interesting way to involve many people. Using Anti-Miranda words would only target those who oppose the initiative. It would risk polarizing opinion, which is rather detrimental to this initiative. Using the "obviousness" effect of the Miranda words and pretending opponents do not exist seems to be the rhetorical strategy of the Pro-Party. The third thing we notice is the higher percentage of debatable adjectives among the Con-Party (19 arguable adjectives, which represents 23.17% of the total) in comparison to the Pro-Party, the majority of which are negatively polarized (5 arguable adjectives with positive polarity, against 14 adjectives with negative polarity). In our opinion, the strategy of the Pro-Party forced the Con-Party to attack the "obviousness" effect of their massive use of Miranda. Therefore, the Con-Party chose to attack some precise points of the initiative, to recreate the debate, which appeared to be unnecessary.
Finally, we can notice that the high percentage of the Con-Party's Miranda and Anti-Miranda words (41.46% of the number of evaluative adjectives) is mainly constituted of Anti-Miranda words (26, which represents 31.7% of all their evaluative adjectives). If the Pro-Party choose the "obviousness strategy" and did not use any Anti-Miranda, the Con-Party maybe tried to turn some voters against the initiative. The Con-Party didn't "just" reopen the debate: It chose to present this initiative as something that wasn't even "admissible" to vote for, in a value-bearing sense, for the initiative seems to miss its purpose ("counterproductive"; "not necessary", "inequitable", "nonsensical", etc.).

Analysis of Miranda Examples
The name of the initiative was "for more affordable rents" and was a major impediment for the Con-Party, as affordable rents can be considered as a Mirandum. In the Swiss context, because rents are very expensive in this country, no one can reasonably argue for non-affordable rents. As a matter of fact, the Con-Party very rarely used the official name of the initiative and talked about "the extreme initiative on housing". The rhetorical strategy of renaming should be emphasized, because it replaces a Mirandum word with an Anti-mirandum one: "extreme". We hesitated to categorize "extreme" as an Anti-mirandum word, but we believe that the Pro-Party would deny the adjective and would not accept to plead for an "extreme initiative to a certain extent". They constantly advocate the obviousness and fairness of their propositions.
On an argumentative level, these adjectives are so strongly connoted with universal values that they orient the policy claims associated with them: The initiative should be accepted because no one can vote against "more affordable rents" or the initiative should be rejected because no one should accept extreme proposals. This is probably why "affordable" is considered as "catchy" by the opponents. Yet, they found an astute way to circumvent the Mirandum adjective "affordable". It seems to us particularly interesting to see how these adjectives create an asymmetry between parties.
Another interesting example is the use of Miranda words by the Con-Party. In the sentence "Housing is a special property", the adjective "special" denotes why housing is not a good like any other. In fact, it concedes that housing needs a special regulation on it. However, the Con-Party rejects the "straitjacket" and tedious aspects of the project. The Con-Party has been "forced" to recognize the Mirandum word: It is impossible to say to the public, without alienating them, "Housing is a good like any other, therefore we can speculate on it", especially since for most Swiss inhabitants, housing rents represent between 20 and 30% of the monthly salary. We see here agreements on some Miranda, but the wording is very neutral, to say the least: "Special" in not as strong as "incomparable" or "one-of-a-kind".

Analyses of Flagwords and Stigmawords
In our corpus, we find no occurrence of Flagwords. The strategy of the Pro-Party is to avoid being seen as a politically situated movement; they need to persuade the audience beyond their own side. Therefore, its strategy is to plead for a common good and common values. Their requests are supposed to be natural and obvious. Thus, they use the universalizing scope of the Miranda, and avoid any form of ideological division. The Pro-Party evokes for example that an affordable rent is a basic human need as well as fresh air is. The Con-Party has no interest in arguing for its liberal ideology (i.e., stating that inequalities between rents are unavoidable), because this is precisely what the initiative blames them of (they use Stigmawords in this respect). The Con-Party will not reply to the accusation by Flagwords arguing for neo-liberal values. Its strategy is different: The opponents are targeting the practicalities of the initiative or the administrative burden it will cause (for instance, by saying the initiative would cause "endless discussions", would be "expensive", or that it uses "disproportionate way[s]"). Thus, they address the idealism of others by highlighting their own pragmatism. The rhetorical strategy can be summed up by this motto: "They dream, but we are in the reality." Stigmawords, by contrast, are often used in our corpus. But it is essential to understand here that the opponent has no interest to debate these words. Indeed, neither party is responding to these stigmatizing labels in our corpus. For the already mentioned example of "Powerful Real Estate Lobby", it would be counterproductive for the Con-Party to argue that "this lobby is not as powerful as that", since giving power to the wealthiest people is typically an ideology that the left-wing parties denounce. The same goes for the "bureaucratic and centralizing initiative": It is not a problem for the other ones, who plead for more regulation in the country. So, here, neither party needs to circumvent the adjectives; their main interest is to ignore them.

Anti-Mirandum Example
The use of the adjective "inequitable" in a headline on the Con-Party's arguments page is particularly interesting because it attacks the heart of the ideology of the Pro-Party. If the inequity were true, it would show the inconsistency between the ideology of the Pro-Party and what the initiative wants, using an argument from commitment: "a is committed to proposition A (generally, or by virtue of what she said in the past). Therefore, […], a should support A" (Walton 2006). The Pro-Party, known to be left-wing, was committed to promote equity and cannot fail in this matter.
The counterargument points out that, in actual public utility rents, a majority of less well-off people cannot be found ("only one quarter of the occupants of [lowrent] housing is in the less well-off 20% of the population"). The Con-Party underlines this information in bold letters on its website, which reveals the significance of this argument substantiated by the fact that it is their only paragraph using statistics (Fig 3).
Our main objective was to highlight how some axiological and evaluative adjectival groups play an important role in a political debate the outcome of which is a "yes" or "no" vote, with the idea to study a "strategic effect of including appreciative modalities in the argumentative discourse" (Tseronis 2014: 22, our translation). Contrary to some debates where one has to form one's own opinion on a societal issue, the "yes or no" vote materializes a polarization between two parties and that means that the appropriation of a Mirandum word by one party complicates the positioning and the rhetoric of the opposite party. These adjectives often present a strong connection with values that can appear as "universal" or strongly high-valued in a certain sociological context. Considering the argumentative roles of such highly respected values or highly hated anti-values is important precisely because Miranda and Anti-Miranda words may block the debate. Putting forward Miranda expressions in a deliberative debate as reasons to vote for or against a proposition of policy is an interesting move from a strategic point of view, especially when the official name of the initiative is "loaded" with a Mirandum as we have seen here. Moreover, as we have shown, Miranda and Anti-Miranda are very polarized and not gradable and this positive or negative polarity is enough to trigger an implicit argument, especially in a deliberative context, in favour of or to the detriment to an implicit claim. The semantic content of the Mirandum triggers the pragmatic effect. Fig. 3 Screenshot of the argument about "inequity" on the Con-Party's website In a context of polarized political debate, we often observe the emergence of adjectives that prevent certain arguments or points of view from being countered, precisely because of their very strong connection with either partisan ideology or universal values. We have seen that the initiative for more affordable rents has a name that has been problematic for the Con-Party. This labelling strategy shaped a rhetorical strategy based on the obviousness of the project, which has been countered by qualifying this project as an extreme one. It was therefore particularly significant to see, at least for Miranda and Anti-Miranda words, the rhetorical strategies used to circumvent this impossibility of countering Miranda or Anti-Miranda words in the "More affordable housing" initiative (cf. Sects. 8-10).
In a need for a more fine-grained analysis of these adjectives, the categories of Miranda, Anti-Miranda, Stigmawords and Flagwords seem useful and need to be better integrated in a study of argumentative and rhetorical strategies. To our knowledge, they have not been taken up or noted outside German-language publications. We have offered a process to identify and sort these words in order to shape a tool for a discourse analysis on controversial debates. Many details can be refined, but the methodological flowchart and the tests that we have provided may be promising for further studies. We should, however, specify that these categories are not easy to use and that some cases are puzzling, and difficult to determine. It should be highlighted that a deliberative context is probably the better one to use these categories, because deliberative situations are about decisions regarding to policy statements. Our corpus gives many arguments which are finally mobilized for or against a policy proposition, which is here materialized by a vote for or against the initiative. 9 Despite these limitations to a deliberative context, which is the only one we studied with these categories and this methodology, we hope to have shown why it is interesting to analyse on a rhetorical level these evaluative and VBWs and how they explain some rhetorical strategies by both parties in a controversial debate.
Funding Open access funding provided by University of Neuchâtel.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.