Steve Oswald is Senior Lecturer and Researcher in English Linguistics at the University of Fribourg. He is a founding member and vice–chair of the European Conference on Argumentation (ECA), as well as a founding member of the Collectif Romand de Recherches sur l’Argumentation (CoRReA) and the Argumentation and Language (ARGAGE) conference series. His research spans across the language sciences, cognitive science, the study of argumentation, and discourse analysis. His research trajectory led him to defend a cognitive pragmatic model of non–cooperative and manipulative communication (Oswald, 2010). During this period, he delved into the study of argumentation and rhetoric, with a particular interest in issues related to the influence of meaning on discourse effects. In recent years, he has closely examined the connections between pragmatics and argumentation, exploring their methodological intersections, shared concepts, and mutual influence to better map their close relationship (Oswald, 2022, 2023a, 2023b). His interdisciplinary approach draws on typical experimental methodologies from cognitive sciences to address classical rhetorical questions, such as the effects of persuasion. This approach rejuvenates rhetorical studies by documenting the intuitions or empirical observations of ancient rhetoricians.
Thierry Herman is a Senior Lecturer and Researcher in the French department at the University of Lausanne and in Communication and Cognition Sciences at the University of Neuchâtel. He has authored several articles on the intersection of rhetoric, argumentation, and discourse analysis, applied to the analysis of contemporary political, journalistic, and academic discourse. Some of these articles were co–authored with Steve Oswald, with the conviction that cognitive pragmatics can illuminate and explain rhetorical strategies of persuasion.
1. Thierry Herman. Steve Oswald, would you describe yourself as a rhetorician?
Steve Oswald: People often define themselves based on their training, and I’m not formally trained as a rhetorician, but I have a profound interest in rhetorical matters. Even though I don’t delve into dispositio, for instance, I can envision that studies in experimental pragmatics might shed light on these aspects. I’m intrigued by the functioning of the argumentative arsenal, to borrow Marc Angenot’s formula (2012), but my aim is to intricately describe rhetorical strategies, relying on a pragmatic approach and on experimental validation. This approach aims to avoid merely postulating the existence of persuasive effects linked to different rhetorical strategies, such as the impact on ethos of an ad hominem argument. In this sense, I would define myself as a pragmatician whose research domain intersects with that of rhetoric.
2. T.H. In essence, you advocate for a form of fruitful interdisciplinarity.
S.O. Many bridges can be built, even if occasionally irreconcilable positions emerge. Concerning rhetoric and pragmatics, one area of overlap lies in the central question of discourse effects. However – and I’m aware this is a bold statement that may elicit some reactions – I believe that the study of such effects is more psychological than linguistic. Of course, a discourse analyst can formulate several interpretative hypotheses about the effects that the use of “fœtus” or “baby” to refer to a human being in gestation may have in a debate on the issue of abortion. But studying these effects requires both a cognitive and an experimental approach: cognitive, to make assumptions about the inferences that may be contextually drawn, and experimental, to confirm (or refute) these assumptions.
This is what relevance theory has been proposing for the last twenty years. I’ve recently read a paper on metaphor processing (Carston and Yan 2023), which critically and nuancedly challenges the widely–held notion that interpreting a metaphor requires no particular processing effort compared to the effort involved in processing a literal expression. Carston and Yan’s experiments show a significantly longer variation in processing cost between a referential metaphor (one that denotes a specific entity, such the swimming children referred to below as tadpoles: “The tadpoles can get out of the pool now”) and a predicative metaphor which attributes a quality to the referent (“You, children, are little tadpoles”), even though the referential metaphor is not in the position of the grammatical subject of the sentence. Such a result is interesting from a rhetorical perspective: it allows us to consider the effectiveness or intelligibility of certain metaphors, and, in other words, the interest in using them or not in certain contexts.
By emphasizing the psychological and experimental perspective on discourse effects, I approach the classical question of adherence, as per Perelman and Olbrechts–Tyteca (2008), under a psychological lens. Recently, my team has investigated the question of insinuation in connection with ethos. The classical theory of insinuation, or innuendo (cf. Bell 1997), posits that insinuation is persuasive and tarnishes the reputation of its target, with more pronounced effects when pejorative content is insinuated compared to when it is asserted. Bell and Fraser (Bell 1997, Fraser 2001) argue that denying the insinuation after the fact maintains the reputational damage effect. In an experimental study of this staining effect, we (together with Tamlyn Adatto, University of Neuchâtel, and Daniel de Oliveira Fernandes, University of Fribourg) contradicted this conclusion: denying an insinuation can, on the contrary, enhance the reputation of the target of the insinuation.
This case illustrates the gist of my approach, which aims to provide insights into claims in rhetorical literature or in discourse analysis which assume the effect of certain verbal forms or argumentative patterns. Experiments reveal that what we thought we knew is often not as straightforward as we presumed and that certain common–sense assertions are poorly if at all, documented. What I’m attempting to contribute is precisely an additional source of justification for these effects that have been discussed for 2000 years in the history of rhetoric, because we now have the tools to test them. To cite a recent example, your own model of explicit ethos analysis, postulating the idea that the image of others given by a speaker modifies the ethos of the speaker himself, has been experimentally validated in work you are conducting with a member of my team, Jennifer Schumann, while also testing the Aristotelian tripartition of benevolence, competence, and virtue. This is what I find intellectually thrilling: revisiting age–old questions and improving the descriptive and explanatory adequacy of our models.
3. T.H. Let’s go back to the beginning of your journey. You wrote a thesis (Oswald 2010) focusing on non–cooperative communication and the intricate, scientifically challenging concept of manipulation. Before delving into the latter notion, could you elaborate on the origin of your interest in this research domain?
S.O. I approached the study of manipulation in discourse only indirectly. Towards the end of my Bachelor’s degree in Linguistics at the University of Geneva, my initial inclination was to explore humour. However, my thesis supervisor, Louis de Saussure, had recently organized a symposium on the concept of manipulation, during which various questions had been raised from a cognitive pragmatic perspective – a perspective in which I had received training. Indeed, building on Grice’s foundational work, which underlined the cooperative nature of communication, the discipline also delved into instances of non–cooperative communication. Having spent a part of my childhood and adolescence in Argentina, immersed in the enduring legacy of the public discourses of Juan Domingo Perón and his wife, Evita, I wanted to closely examine discourses that are likely to move people. Consequently, I defended a dissertation on Peronist discourse before embarking on a thesis focusing on non–cooperative and manipulative communication.
Beyond these biographical details, I’ve always been drawn to any communicative phenomenon in which ‘something doesn’t quite add up,’ especially from an interpretative standpoint. I have always been attracted by wordplay and puns, as well as by riddles, enigmas, and other discursive challenges that involve a hidden element – a “trick” – whose concealed nature, subsequently revealed, renders these discursive events amusing and, therefore, interesting from the perspective of the way they work. Manipulation, despite its serious moral quandaries, shares operational similarities with humour: its success hinges on keeping the mechanism the manipulator deploys for securing adherence concealed. If this mechanism were evident from the outset, manipulation would falter. This sets it apart from humour, which, while similar in its initial mechanism, necessitates the subsequent revelation of the underlying ‘trick.’
4. T.H. Rhetoric, since its inception, has been the target of criticism due to its ability to manipulate the beliefs of others – the very term 'manipulation,' associated with a discourse activity, continues to carry a negative connotation. Do you still use this notion, and more importantly, how have you defined it to overcome the challenge of scientifically examining a concept loaded with negative connotations?
S.O. Indeed, the prevailing connotation is negative, although not always. For instance, Chalvin (2001) introduces the term “manifluence” to describe the judicious use of manipulation – essentially, manipulating others for their own good. Confronted with the connotation–related challenges of the term, I chose not to focus on the ethical dimension but rather on the operational facet of the phenomenon. Instead of attempting to assess the morality of manipulation, I opted to investigate how the manipulative mechanism operates. While ethical questions remain fundamental, my perspective, rooted in a pragmatic, linguistic, and cognitive approach, does not prioritize them. Consequently, I now use the term less frequently, recognizing its diverse interpretations. Instead, my focus has shifted to the concept of deception, which is more precisely defined by language philosophers; deception involves inducing or perpetuating a false belief or preventing someone from holding or acquiring a true belief. This approach specifically avoids issues of intentionality: can deception occur by chance, is there an intention to harm, etc.? The Oxford Handbook of Lying (Meibauer 2018) extensively discusses criteria for definition.
- 1 The example is discussed in Maillat and Oswald (2011).
What emerges is the inextricable link between deception and the notion of truth. In cases of manipulation that caught my interest, it seemed plausible to deceive while telling the truth. Take, for example, the well–known case of Jo Moore, a former special advisor for the UK Department for Transport. In an email to the department’s press office, sent minutes after the collapse of the first World Trade Center tower on September 11, 2001, she suggested that it was “a good day to bury bad news.” This incident, which leaked to the media and sparked a scandal, illustrates, in addition to the cynical exploitation of the tragedy, a manipulation not rooted in distorting news but in the process of discouraging people from attributing importance or relevance to verifiable facts made public – partly because attention was focused on the collapse of the New York towers.1
However, the term “manipulation” presents a series of problems: etymologically, it refers to the hands, a notion retained in sleights of hand without negative connotation. It raises numerous questions about its boundaries with persuasion or argumentation... For these reasons, I tend to avoid using the term, instead maintaining a focus not on the “What” but rather on the “How.” If deception is at play, how does it function?
I think it is impossible to ascertain that a particular speech or statement is manipulative. When G.W. Bush says “You are with us or against us,” the public accuses him of manipulation, citing the false dilemma – but what evidence do I have that Bush does not genuinely hold a Manichean worldview and is not expressing a belief he sincerely considers true? We cannot peek into people’s minds, making it impossible to definitively judge. Nonetheless, we can describe the effect of the proposed alternative and contemplate how it diminishes the salience and relevance of a third position. This is why I believe that a list of criteria for defining manipulation is, if not impossible, of little interest as, in my opinion, it will not capture all occurrences of the phenomenon in discourse. One can have suspicions, make hypotheses and conjectures, albeit with varying degrees of evidence. Regardless, I do not consider defining such a discursive practice as the central goal of a pragmatic approach to what might be termed manipulation. On the contrary, comprehending the process that fosters a belief or impedes another seems attainable and scientifically intriguing, which is the orientation that underpins the approach I adopt in my analyses.
5. TH. Some concepts in cognitive pragmatics face a form of criticism, centered around a positivist view of communication. Your field uses terms like ‘successful communication’ and ‘truth–conditionality,’ which are not commonly found in descriptive approaches to discourse analysis. What is successful communication, and who judges it?
S.O. I think we need to first agree on the meanings of the terms we use. In pragmatics, we indeed refer to successful communication for a communicative exchange which results in a situation where the recipient’s representation bears a very high degree of resemblance to that of the speaker – or sufficiently high for the needs of the exchange. It is not about achieving perfect identity between representations or the exact same thought, but about ensuring a very high degree of resemblance between the speaker’s thought and that which the recipient attributes to him or her. A theory of verbal comprehension must explain how this ‘success’ occurs, and we use the term successful communication even for antagonistic or non–cooperative exchanges from a behavioural standpoint. If I call you an anacoluthon or a cercopithecoid, you will understand that I am insulting you and that my intention is to attribute negative properties to you. In this sense, we still cooperate on the level of interpretation – from this perspective, our communication is successful because we understand what each other means, even if it might fail on the level of social harmony. Therefore, there is no need for a judge to determine if understanding has taken place. On the other hand, a judgment would be necessary to determine whether what I have told you is normatively acceptable or offensive.
Regarding cases of deception or manipulation, we must posit different levels of communication, making their description more complex. Intentions that are made manifest on the surface do not include the intention to manipulate, which must remain concealed. While the intention to manipulate stays hidden, communication is still considered successful in that an intended content has not only been made manifest but also recognized and identified.
However, if we approach communicative success from a rhetorical perspective, the success of an exchange should be reflected in the adherence of minds to the expressed ideas or in a form of readjustment of the beliefs that were held before the exchange. This includes a perlocutionary dimension that is not necessarily implied by the act of understanding. One can perfectly well understand a persuasive speech without adhering to its claims – a dimension that pragmaticians have typically neglected.
6. T.H. What about truth–conditionality? Rhetoric does not typically work with the notion of truth.
S.O. Cognitive theories in the language sciences are the heirs of theories developed in the philosophy of language, which focused on linguistic meaning (Frege, Carnap, the early Wittgenstein, Russell, the Vienna Circle, etc.) – a movement that led to modern semantics and a type of modern pragmatics. Austin, Searle, and Grice, on the other hand, were part of a group that has been called the philosophers of ordinary language, even though they maintained ties to the formal tradition. Consequently, the influence of research on truth–conditionality is significant, with the idea that the meaning of a statement is defined by its truth conditions. This means that certain conditions in the world must be satisfied for the statement to reflect a true state of affairs and, therefore, to have meaning.
However, meaning is not always truth–conditional. Of course, truth–conditionality applies to the literal declarative meaning of a statement like “this is a table,” but a question, for example, cannot be evaluated in terms of truth or falsity. Other phenomena, related to implicit meaning, also do not contribute to truth conditions. For instance, the statement “I have an appointment later” in response to “Do you want to grab a coffee?” implies the response “No, I do not want to have a coffee,” but the truth conditions of the inferred proposition have nothing to do with the truth conditions of the sentence about the appointment that the speaker claims to have in the statement that was actually formulated. In other words, conversational implicature does not contribute to the truth conditions of the utterance that communicates it.
Furthermore, several sub–domains of meaning are non–propositional and thus escape evaluation in terms of the truth or falsity of a proposition in subject–predicate form. Ineffable things, like the taste of grilled chicken, or emotional reactions – referring to pathos – which are certainly laden with meaning and able to generate thoughts, are not propositions. Being hurt by a comment has meaning, but one that is challenging to approach with the tools of truth–conditional approaches. Similarly, we can certainly observe a difference, which we cannot quite quantify, between the statements “My wife is very nice” and “My wife is an angel.” While there is obvious semantic overlap between both statements, I lose something by saying “very nice” and not “angel.” This loss is not truth–onditional and is not expressed in propositional form, but it remains linked to linguistic meaning, as metaphorical expressions undoubtedly activate a broader or richer conceptual space than the attributive expression alone. In other words, the notion of meaning encompasses much more than truth–conditionality.
7. T.H. This is very interesting, and I see in it an echo of ways to influence others through language; I am thinking especially of Jean–Blaise Grize’s “éclairages (lighting)” (1996) or Ruth Amossy’s “argumentative dimension” (2000). Does pragmatics now aim to go beyond the sole question of comprehension?
S.O. Yes, and poetry is a typical example: not everyone sees the same thing in a poem, and we have probably all asked our literature teachers: ‘How do we know what the author had in mind when they wrote this?’ Differences of interpretation are unavoidable. A theory of communication should be able to explain these things, which is why a whole movement now led by people like Elly Ifantidou, Tim Wharton, Louis de Saussure (2015), among others, is interested in what lies beyond the boundary of what is plausibly attributable to a speaker but is nevertheless communicated – weak implicatures, for example. Dan Sperber and Deirdre Wilson themselves have published an article (2015) entitled “Beyond Speaker’s Meaning,” which considers elements of meaning that lie beyond intended meaning. This means going beyond the aspects of understanding and recognizing intention to include facts that are unclear as to intention but are indeed present, as in poetry. The challenge is to find a model that explains the intentions made manifest to account for the possible presence of less obvious intentions or unintended effects of meaning. It is therefore a form of openness to interpretation that interests discourse analysis (some in pragmatics, such as Manuel Padilla Cruz [2016], refer to this as hermeneutics). The difference is that we are allowed to discuss interpretations but, in a way, equipped by cognitive science; we are allowed to say, for example, that “given this context, this interpretation is more plausible than that one.” These interpretations remain conjectures, but they can be discussed, and, above all, justified. In some respects, such an approach ties in with my work on the identification of implicit premises in argumentative discourse, in which I proposed a model, inspired by relevance theory, making it possible to say, for example, that in a given context, such and such a premise is more likely to correspond to the speaker’s intentions than such and such other (Oswald 2016).
8. T.H. In your statements, you often mention rhetorical effect; how do you envision this notion?
S.O. Rhetorical effect is an extremely vague catch–all noun phrase – which I define as an effect that is triggered by an utterance in a communicative situation and that affects what happens downstream in the exchange, particularly in argumentative contexts. The effects of logos, pathos, and ethos are rhetorical effects; they influence others’ perception, such as thinking that the speaker is skilled, or our reactions, for example, when we feel compelled to remain silent or counter–attack. These are effects on the exchange, on our perception of the world (including our perception of the speaker), on mental and emotional states; in short, a set of elements, of which, unlike stylistic figures, there is no inventory – a set that deserves to be investigated and documented. For example, in Fribourg, we have just completed a study to determine whether insinuation could be advantageous in personal attacks, particularly whether it could have effects on the insinuator’s ethos, such as appearing more trustworthy. It seems to be the case, according to our findings.
9. T.H. Unlike many works in rhetoric and discourse analysis, your recent projects give research an experimental dimension. Given that discourse analysis and rhetoric are interdisciplinary fields, and that the debated notions of persuasion or simply influence are also multifactorial, don’t you have the impression that it is difficult to conduct experiments in such a field to isolate one factor?
- 2 Imagine the following dialogue between A and B, two friends who attend a car show and discuss certa (...)
S.O: Yes, and we have encountered problems for several reasons; first and foremost, because the experimental approach must necessarily adopt a reductionist perspective. In a questionnaire, you can only test one condition at a time; you change a single element within items that remain identical for the rest. For example, to test whether the use of insinuation in an ad hominem is rhetorically advantageous, we need to compare it with an equivalent version, in terms of meaning, that does not involve insinuation (such as an assertion with the same content), in order to establish two distinct conditions. What is more, we need to design a scenario in which the only variable is this insinuation/assertion pair – the rest of the scenario must remain unchanged. This is the only way to ensure that, in the event of different (and statistically significant) responses from experimental participants in each condition, this difference is indeed due to the use of insinuation, since this is the only element that has changed between the two conditions.2 Such a project requires meticulous, fine–tuned work in terms of semantic and pragmatic resources, with effects that are unlikely to be spectacular. Aspiring to attain a result like “if you insinuate, you have a 90% chance of persuading” is unrealistic. These experiments show that several parameters, such as background beliefs and political preferences, cannot be controlled, even if there are ways of more or less neutralising their influence (for example, by developing items unrelated to political issues or ideology, or by using scenarios involving anodyne everyday activities). One of the challenges, therefore, is to manage the potential impact of uncontrollable factors – this is the rationale behind the reductionism I mentioned – without making our scenarios entirely artificial. For the time being, we have been able to observe that differences are emerging all the same, and this is encouraging, because these differences can fuel our attempts at explanation. Crucial work goes into creating the items we submit to the panel of respondents – these items must seem natural, but not too natural either, to avoid ‘noise’ linked to attitudes that our material might trigger involuntarily, such as the activation of stereotypes or negative attitudes on the basis of the use of certain terms (for example, we can imagine that the use of language that is too formal, or on the contrary too informal, in our scenarios, might be able to bias responses); it is challenging to ensure that the items produced are 100% adequate in this regard. This creative process, based on a linguistic intuition of what is “natural”, is nevertheless one of the only ways of documenting effects, and it is always important to test the acceptability, or naturalness, of our scenarios before launching an experiment on their rhetorical properties.
Another difficulty lies in the fact that the fields of rhetoric and argumentation are only now beginning to take an interest in these experimental methodologies, which are sometimes not easily received at conferences on argumentation, either because of the legacy of disciplinary divisions – argumentation, discourse analysis and philosophy do not have an experimental tradition – or because of a disciplinary mistrust of reductionist approaches of this kind. Paradoxically, this caution is fuelled by the requirement for absolute clarity that must accompany any rigorous experimental study, which opens the flank to criticism: rigour in form and transparency in the way we report what we have done expose us in conferences to classic questions, which do not necessarily question the study itself or its results, but rather the choices made and the notions used: “Why this criterion and not that other criterion?,” “What do you make of this scenario, which seems to correspond to your definition of the phenomenon in question?” In this way, we are more exposed to criticism of what we have not done than to examination of what we have presented.
10. T.H. In the experimental realm, it is challenging, if not impossible, to use authentic examples. In the works of relevance theorists, forged examples are also commonly encountered. However, your work and that which we have co–authored also provide an opportunity to analyze existing cases – how do you navigate between experimental data, forged examples, and authentic cases? How do you position yourself in relation to the data?
S.O. In an experiment, it is impossible to take as an example a political debate that is too contextualized, when the reductionism I mentioned earlier is paramount. “Contextomy” is difficult! Forged examples can be used for experimental research, but they have to be put to the test. They are the product of our introspection, but we pre–test them with a panel of participants to ensure that they are interpretable and unproblematic. We can also use established examples and adapt the wording of the test items. For example, a syntactic routine such as “You would have difficulty distinguishing A from B”, which implies “you are incompetent”, can be tested by substituting different nouns for A and B: “You would have difficulty distinguishing a fork from a knife” and so on. Entirely fabricated examples can be a valuable resource, depending on the phenomenon under investigation. And I think it is less problematic than you might think. For example, understanding and observing the conversational implicature of ‘even’ with a sentence like ‘Even Ted liked this movie’ will not be a problem for anyone, even if I have just made up the statement.
However, some pragmatic or rhetorical effects that we want to analyse lend themselves better than others to experimentation on artificial data. In my AMoRe project, which deals with the persuasive effects of rephrase, we try to do both: we start with examples taken from our corpus and adapt them to turn them into experimental items, but we also carry out experiments on forged data to identify a phenomenon that seems important in order to have results, so that the team working on the dataset can in return do something with them. Referring to our joint work (i.e. Oswald and Herman 2016, Herman and Oswald 2022), whether in the case of a documentary on the moon landing or Didier Raoult’s answers to David Pujadas’s questions, we have on the other hand worked on authentic data to illustrate rhetorical aspects. So, I do not favour one type of data over another, but I try to consider the type of data that is relevant to the research questions that arise.
11. T.H. You are interested in concepts that are extremely difficult to approach scientifically – I am thinking of dogwhistles and insinuation. Can you provide a brief definition of these various types of implicit communication, explain how they can be approached scientifically, and elaborate on their rhetorical dimension?
S.O. As a pragmatician, I try to understand how the meaning of an utterance can be different from the meaning of the words used to convey it. But implicit meaning is not monolithic. Furthermore, while Grice has already demonstrated that there are several types of implicature, we can also debate the exact scope of explicit meaning: take “I’m tired”, for example; does it mean “tired to the point of not wanting to do sports but not to the point of giving up a walk”, or “tired to the point of not wanting to go out at all”? In either case, it seems to me that the exact degree of tiredness that the speaker is asking us to infer is part of the explicit meaning of the utterance. If, when tidying up after a party, I say to you, “Can you put the empty bottles in the recycling bin?”, it is clear to me that if there is the equivalent of a finger’s worth of beer left in a bottle, I consider it to be empty. If you only put away bottles that have literally dried out, I will interpret this as a desire to make fun of me and, in such a case, you will probably agree with me that “almost empty” or “empty to the point where you can no longer keep the contents of the bottle” corresponds in this case to the explicit meaning of my saying “empty bottles”, even though this meaning is, in fact, absent from the linguistic material I have used. Thus, the question of implicit meaning is not necessarily easy to define.
However I am particularly interested in implicit meaning that is used for rhetorical purposes. I consider implicit meaning in the broadest sense to be any content communicated by an utterance, but which is not encoded in the verbal material. Insinuation is classically defined as the implicit ascription of a property that is negative from a normative point of view, or of a property that one wishes to keep silent; one never insinuates positive content, unless its formulation is problematic for some reason. Moreover, insinuation can typically be denied, more or less plausibly. Let us take an example: Laszlo and Nina each bring a salad to a potluck supper, and the host, afterwards, compliments Nina in front of Laszlo by saying: “Your salad was exquisite, Nina”. This could imply that Laszlo’s salad was not good, and insinuate that Laszlo is a bad cook. As it turns out, insinuation can be followed by a plausible denial, such as “I did not mean that your salad was bad, Laszlo, I just wanted to point out that Nina made a great salad, because I know she is taking cooking lessons at the moment and I wanted to encourage her”. From a rhetorical point of view, then, it is an interesting resource, since it allows you to communicate problematic content without being accountable for having communicated it.
The concept of dogwhistle was popularized during the American presidential race between Bush and Dukakis in the 1980s; it refers to whistles that only dogs can hear. Transposed to the political arena, the term designates messages that have a specific meaning for one part of the audience but remain inaccessible to another. Examples include inappropriate or dirty jokes in cartoons, aimed at adults but unnoticeable to children. In American politics, talking about “inner city crime”, for example, means targeting crimes perpetrated by members of the black community. When testing a description with a “violent crime” variant against a ‘violent inner city crime’ variant (cf. the work of Hurwitz and Peffley 2005), test subjects tend to favour repressive measures over preventive ones in the latter case. Such a study suggests that the dogwhistle operates unconsciously and has a real impact on concrete decision–making; a similar observation had already been made in relation to metaphor (Thibodeau and Boroditsky 2013).
Another related phenomenon is the use of fig leaves, which consist of lightly covering up what is unacceptable to display in public: “I’m not racist, but...” followed by a racist comment, “I’m not saying we should do it, but I wonder if...”, “It’s just a question, but...”. These are expressions that, on the surface, allow the speaker to absolve himself or herself of the intention of having said something that might be challenged, without refraining from saying something discriminatory that will not be well received. The rhetorical strategy of mentioning statistics, for example, often comes with figleaves: “60% of criminals are black. I am not racist, figures say so” are statements that nonetheless propagate a discriminatory belief (see Bräuer 2023). Presuppositions may also be exploited for rhetorical purposes, in particular through what is known as presuppositional accommodation: if I say, for instance, that my sister has two children and my interlocutors are unaware of my sister’s very existence, they learn this information by accommodating the presupposition. It is conceivable that unscrupulous speakers might communicate problematic information in this way, since what is presupposed is less likely to be critically examined (see Lombardi Vallauri 2021, among others). This typology of implicit content is at the core of my IMAFUN project, funded by the Swiss National Science Foundation, and which aims to document rhetorical effects triggered by some of them: insinuation to promote the speaker’s ethos, presupposition to insert problematic information (logos), and so on. The idea is to examine different types of implicit meaning as rhetorical resources used to trigger certain types of effect.
12. T.H. Your career path reveals your interest in and work on fallacies, a notion that seems to invoke a normative idea of judgment on good and bad argumentation. For some, this approach goes beyond the role of a linguist, typically rooted in a descriptive dimension, with the potential exception of Critical Discourse Analysis scholars, who embrace an ideologically committed perspective. How do you position yourself in this debate between the normative and the descriptive?
S.O. Once you take an interest in reasoning, and consider that some lines of reasoning are more fragile than others, it seems to me possible to disagree with an argument, which requires justifying why. But I make a difference between a normative character and a prescriptive character giving do’s and don’ts. I am comfortable with normativity, but it does not necessarily imply a prescriptive stance or an a priori normative position. It is well known that there are perfectly reasonable versions of what tradition classifies as fallacies: an appeal to authority can be perfectly justified in circumstances where only an authority is in a position to render a verdict, just as a personal attack can be justified in cases where the personal dimension in question is relevant to decision–making. I believe that the issue also depends on the type of proposition being defended – a fact, a value or a proposal for action, to use a well–known tripartition. So, “Everyone says the earth is round, so the earth is round” is not a strong argument, but “All my friends have told me that this person is trustworthy” is a good reason to think that this is indeed the case. When discussing the quality of argumentation, it is challenging to entirely do without the normative dimension. But the aim of such a study is not to conclude that a person who is guilty of using fallacies does not know how to argue: it is just a fact to say that this person commits fallacies. This is where I see a descriptive dimension: it is possible and reasonable to expose and justify that, according to such and such a model, this speech develops such and such a fallacy without falling into easy judgment or bursting out “Boo! It’s a bad speech!”
13. T.H. Another ongoing project (AMoRe), carried out with a team in Poland, leads you to consider the concept of reformulation from a rhetorical perspective. Until now, it was considered more as a linguistic process only, but you have shown (Koszowy et al. 2022, Younis et al. 2023) that reformulation is certainly associated with rhetorical effects, isn’t it?
- 3 In this study, we contrasted rephrased statements (e.g., “You should buy this chair because it is s (...)
S.O. Indeed. We showed that reformulations were perceived as more persuasive than statements without reformulation (for example, a statement in which the speaker rephrases by specification such as “It is shameful to behave in this way. Our country also thinks it is shameful” is perceived as more persuasive than a non–rephrased version of the same statement, such as “It is shameful to behave in this way. That is what I can tell you about that behaviour”) and that while individuals are sensitive to the rhetorical effect of reformulation, they perceive it more as a form of paraphrase, a redundancy of content, than as an argumentative strategy (Koszowy et al., 2022). We are also trying to see whether reformulation has a decisive effect: in an experiment, for example, we proposed a choice between two products, and tested which product participants would prefer, by varying the presentation of these products between one simple piece of information, strictly repeated information, reformulated information, or information accompanied by other neutral information (without reformulation).3 It appears that reformulation has a persuasive effect compared to the other conditions. We are also exploring whether reformulation has an impact on the ethos of the reformulating speaker, which seems to be particularly significant in instances of reformulation by specification involving expert knowledge or figures. While it is often assumed that reformulation is devoid of inferential effects, akin to mere repetition, our experiments reveal that it is not quite as simple as that.
14. T.H. How do you envision your future research? What projects could further contribute to making the intersection between cognitive pragmatics and rhetoric both practical and compelling?
S.O. My research will continue to employ the experimental paradigm, as many areas remain to be explored in this approach to the study of the rhetorical impact of different pragmatic and linguistic resources. I also intend to intensify collaboration on the computational side, because today’s world has to take artificial intelligence into account. However, automated linguistic models have obvious limitations when it comes to pragmatic effects: the identification of irony, metaphor and unstable implicit meaning, for example, are notable pitfalls of artificial intelligence. Nevertheless, I believe there are ways of developing reflections and models to shed light on these grey areas. However, I still have a major interest in argumentation; indeed, I consider myself more a member of the argumentation community than that of linguists, but I would like to bring disciplines and approaches into dialogue to improve the explanation of phenomena of meaning that fascinate me, and that, above all, are not self–evident.