The challenges of a semantic and argumentative approach of corpus in the era of “big data”
Francis Chateauraynaud and Josquin Debaz (GSPR – EHESS, Paris)
Social science methodologies face continuous challenges emerging from the evolutions of the digital worlds. Promises and prophecies connected with the rise of big data are further accentuating this trend. Drawing on our long experience of analyzing complex issues through large evolutionary corpora, we propose a four steps paper in order to characterize some key drivers for the understanding of contemporary social processes.
“An argument is an interaction between two or more participants which involves a claim by each participant that his contention can be justified. Arguments may involve complex and lengthy sequences of steps, questions, answers, and objections. At any particular stage of an argument, however, we should be able to identify the semantic core of the argument. The semantic core is a set of propositions made up of one or more conclusions and some sets of premises.” (Walton, 2008)
How can we counteract the very idea that the mere massive accumulation of data would be enough to produce social intelligibility? Many experiments with the software Prospero put forward the argumentative and narrative dimensions of discursive productions, the need for coherent rules in the making of corpora and the development of congruent models, based on the abductive reasoning specific to pragmatism, and relevant for a certain class of complex processes.
During the last two decades, our main domain of investigation has been oriented to public issues related to health, technological and environmental risks. We paid a special attention to the critical or contentious moments, creating intense argumentative activities, and generally summarized under the term of “controversy”. Addressing the different forms of critique created an opportunity to grasp and to make intelligible the cognitive and political processes of alert, controversy and regulation – considered both at a national, European or global scale (Beck, 2008; Chateauraynaud and Debaz, 2017). Between 2006 and 2014, this research program was oriented toward the experimentation and the design of an Observatory of alerts and controversies based on socio-informatics tools1]. In the following posts, we will examine different aspects of this experience and aim to provide guidelines for the development of collaborative platforms dedicated to social research on contemporary issues.
By using digital techniques, the main objective was to aggregate and analyze large amounts of heterogeneous documents, and to nourish discussion forums in order to create moments of clarification of the complex processes underlying risk policies. Based on this recent experience of bringing together different logics of research, which epistemological norms in the social sciences describe commonly as contradictory (as quantitative vs qualitative; or social network vs discourse analysis), we now explore how computational approach to large corpus, calculated from web flows, and socio-informatics approach of controversies, evolving through non-linear critical processes can be articulated (Chateauraynaud and Chavalarias 2018). By using new digital mediations, we suggest some pathways for a learning interface intertwining human interprets and algorithms. The collective cognitive device that emerges from this cross-fertilization can create a new critical approach on how social processes unfold in digital worlds, from official sites to social media. Our strategy provides different pathways in order to follow numerous issues, avoiding knowledge fragmentation or reasoning on each type of public issue, by a particular emphasis on interdependencies. This research device could be collectively tested and made available to other research networks. In this draft, we examine or reexamine some critical aspects of this approach: argumentative theories, temporalities, computer-based discourse analysis, architecture of corpus, non-monotonic and abductive forms of reasoning…
Argumentative activities, narratives and discourse analysis
Since more than 20 years, network analysis appears as an important way to study relations between topics, authors, publications, websites etc, hence graph theory and network analysis have been adapted by mapping controversies movement. No doubt that graphs and networks can help us to sketch the multiple connections structure produced by activities, discussions or circulations2. Nevertheless network analysis obfuscate, often implicitly, the ways of narrativizing or arguing used by actors. Formally speaking, there is no controversy without a confrontation between arguments mobilized, developed, revised, criticized by individual or collective actors.
But what do we mean exactly by argumentation? An argumentation is a discourse or a device, linked or not to an ongoing action, which is organized through a disputing process – or its anticipation – in order to defend a standpoint, an opinion or a thesis, and designed to resist against relevant criticism. It is to say that an argumentation contains, at least as implicit requirement, one or many counter-argumentations. Thus, the analytical moment will focus on the forms of typification of the arguments which are used by protagonists of controversies or debates: “This is not a good argument” ; “This is an argument ad hominem” ; “His reasoning lies on totally simplistic economic arguments…” ; “It is not enough argument for…” etc.
In practice, an argumentative sociology must articulate different levels of analysis:
- Contexts, situations and arenas in which actors deal with some argumentative constraints;
- The making of arguments as an activity around linguistic and/or semantic nodes or cores, norms of reasoning or argumentative tools for strategic manoeuvring;
- The transformation of arguments over time through a long series of redefinitions generated by disputes and controversies and by which some arguments are selected and become strengthened enough to join common representations and ordinary discourses.
The new rhetoric developed by Perelman and Olbrechts-Tyteca is a model of argumentation that collapses traditional distinction between rhetoric and dialectic. This approach is based on concepts of audience and adherence which are constantly modified by the practices of argumentative communities depending on social contexts. Then, argumentation is not necessarily oriented towards a formal type of reasoning or inference, arguers being also concerned by strategies “to elicit or increase the adherence of an audience to theses that are presented for their consent”3. In other words, an argumentation builds a connection between a thesis and an audience, producing arguments congruent with anterior ideas or judgments shared by its public. Thus, the aim of argumentation is not purely intellectual adherence, but involves incentives by creating a disposition to act. Arguers exhibit great adaptability: “depending on the circumstances, their arguments will seek different results and will use methods appropriate to the purpose of the discourse as well as to the audiences to be influenced” (Tindale, 1999).
Argumentative movements drives the revision of beliefs. A major problem for knowledge representation is how to revise a knowledge system in the light of new information that is inconsistent with current system. Another related problem is related to contractions: where some of the information in the knowledge system is taken away. Gärdenfors and Makinson have formalized the revision problem in two ways. First, two sets of rationality postulates or integrity constraints are presented, one for revisions and one for contractions. On the basis of these postulates they supposed a natural correspondence between revisions and contractions, in order to develop a more constructive approach based on the idea of an epistemic entrenchment (Gärdenfors and Makinson, 1988). In other words, actors or arguers will select the relevant facts from their knowledge system which determine their priority in revisions and contractions.
By tracking the very moments of reconfiguration in long run controversies, we can describe the drivers and constraints by which a process of reordering of epistemic entrenchments. In many debates, when they are oriented towards consensus and cooperation, believing in genuine process of discussion, players will try to find a compromise, by using multiple tactics to close as soon as possible the discussion and to avoid a fatal dispute(“we will not argue on this point,” “it would lead us too far”). In the case of a dissensus orientation, the figure that Lyotard describes under the concept of “différend” (deep disagreement) (Lyotard, 1983), leads to a defense crystallization in order to reduce the views of others and to literally bomb opponent’s arguments so that he or she could not respond, aiming to reduce possibilities of counter-argumentation. In both cases, arguing involves faculties of action and emotions whose mode and intensity depend on resistance of the different actors to the revision process.
Depending on the type of materials and of the research questions, it is relevant to discuss the theoretical divide between argumentation activity and narrativization – even if in the two cases many basic speech acts are involved (Smith, 2017). The distinction between narratives/narrations and arguments/argumentation is important to identify the regimes of discourse4 by which the authors-actors (i.e. the contributors-and-interpreters of the texts) try to convince, to influence or to affect different publics – potential allies, authorities, public opinion etc. Arguers rarely succeeds in convincing opponents (Angenot, 2008). It appears that arguments necessarily suggest or refer to counter-arguments, whereas narratives do not necessarily come packaged with a counter-narrative. As long as we have explored text corpora, we can sustain the idea of a tangible difference between argumentative moves and narrative structures lying on the degree of explicitation of the reasoning rules and conceptual definitions, and of the degree of presence of elements of counter-discourse. Another criterion can be based on the presence of operations aimed to deconstruct other discourses. That is to say that the key elements of distinction lay on the use of meta-argumentation or meta-discourse – which is less frequent in narratives. But, in practice, the formal distinction is purely analytical, since it is also true that in most cases, discourses are made with a hybridization of narratives and argumentations.
In what conditions new arguments could appear and could be transformed in common features, informing the design of collective representations or standardized devices? By comparing many issues, we observed a circular property which describes a social learning process: through disputing trials, common grasps based on tangible assertions, outgoing from collective tests, are gradually embedded in ordinary practices and social representations. On the basis of these new social resources, new controversies and conflicts arise… Involving into an argumentative process implies to put at risk one’s basic beliefs: first, one has to deal with others’ beliefs; a second reason lays on the fact that elements drained by different arguments come into contradiction with the principles underlying beliefs and fundamental values.
Let us take the example of the “Warning to Humanity” written by a group of scientists – the Union of Concerned Scientists (Ripple et al. 2017). In 1992, with the organization of the first earth summit in Rio de Janeiro, a battle took place within the scientific field, opposing rationalists (producing the “Heidelberg Appeal”,5 a sort of warning against anti-scientific movements using “ecological rhetorics” and “new age eschatology”) and scientists involved in the production of evidence of “current, impending, or potential damage on planet Earth involving ozone depletion, freshwater availability, marine life depletion, ocean dead zones, forest loss, biodiversity destruction, climate change, and continued human population growth”6. The rationalist reaction contested the idea that progress was responsible of environmental destructions. On the other side, scientists claiming their independence and freedom of speech in their manifesto, cautioned that “a great change in our stewardship of the Earth and the life on it is required, if vast human misery is to be avoided”, using many early signals that “humans were on a collision course with the natural world”. 25 years after, the idea that fundamental changes are urgently needed to avoid very negative consequences is sustained by international organizations and most of the governments in the world7].
By studying this kind of global issue, we can measure the strength of the different argumentative “dispositifs” by following the trajectory of actors and arguments, and look precisely to the discursive strategies used to produce the good “assemblage” between narrative and argumentation. Adverbs and modalities are key analytical elements to observe the procedures used by actors. Let us take a short example:
“Since 1992, with the exception of stabilizing the stratospheric ozone layer, humanity has failed to make sufficient progress in generally solving these foreseen environmental challenges, and alarmingly, most of them are getting far worse. Especially troubling is the current trajectory of potentially catastrophic climate change due to rising GHGs from burning fossil fuels […], deforestation […], and agricultural production – particularly from farming ruminants for meat consumption […]. Moreover, we have unleashed a mass extinction event, the sixth in roughly 540 million years, wherein many current life forms could be annihilated or at least committed to extinction by the end of this century.” (Ripple et al. 2017)
Giving a power of expression to a certain list of facts supposes a concatenation, in the same discourse, of fragments of narrations and moments of argumentation – in the present case, the rhetorical manoeuvre seeks to avoid producing a discourse which could be read as a typical “prophecy of doom”. Thus, the use of adverbs, modalities, markers of prudence, showing a good understanding of uncertainties, answers to a strong argumentative constraint. A good strategy for argumentative analysis is to take seriously the techniques by which protagonists themselves perform the task to identify, classify and evaluate arguments. Argumentative sociology provides powerful analytic grids to detect what kind of arguments or counter-arguments an actor takes in charge and what kind of argumentative movement is produced in interactions or monologic productions as texts and discourses. The presence of key indicators like “argument”, “ claim ”or, “problem”, critical attributes, comparative marks, signs of agreement or disagreement, temporal modalities and key adverbs, and many other linguistic tools, help to find and to analyze argumentative activities.
As with the previous example of “global warning”, looking closely to the production of discourse allows us to understand how actors deal with the question of reversibility or irreversibility. How do they identify, represent and argue the presence of a turning point? How do they try to avoid the tipping point upstream or, afterwards, how do they interpret the causality and consequences, especially after a marking event? According to Andrew Abbott, the concept of turning point is a « narrative concept ». For him, “what makes a turning point a turning point rather than a minor ripple is the passage of sufficient time ‘on the new course’ such that it becomes clear that direction has indeed changed”, and “what makes the trajectories trajectories is their inertial quality, their quality of enduring large amounts of minor variation without any appreciable change in overall direction or regime.” (Abbott, 2001) If we must take seriously the turning points and the different phases before and after their occurrence, we have to design analytic grids able to deal with contingencies, sudden ruptures, surprises, uncertainties and variations.
“Unexpected connections can make new things come into being. New technologies, new economies, new identities and political visions: futures of all sorts are forged in the contingencies of strange connections. At a time when our future seems foreclosed in the narrow channels of corporate expansion, on the one hand, and clashing state and popular terrorisms on the other, we might look for our best hopes (as well as our inchoate terrors) in the possibilities of something different. There are futures about which we have never even dreamed.” (Tsing and Pollma, 2005)
If the global discourse of the coming catastrophe seems to gain strength and legitimacy, environmental issues are entering a new period shaped by a number of on-going processes from international to local. Through our series of corpus, we can observe an increasing tension between the expansion of environmental standards and norms on the one hand, and the emergence of multiple sources of conflict on the other. In fact, every controversy is based on a confrontation of visions of the future. More: It creates new versions of possible or probable future and produces a feedback on the tools and models of prediction or forecasting – even sometimes on science fiction narratives. Thus, we must integrate specific tools in our argumentative analysis, by defining some discursive indicators of the different regimes of anticipation, prediction or prophecy.
The future is inherently unpredictable, even for the coming years or decades, leaving actors to deal in terms of probabilities, uncertainties, and varying levels of confidence. Thus, the different regimes of enunciation of the future are crucial to investigate complex dynamics with an argumentative approach. We can build a large repertoire of ways of considering future, both in epistemic and in axiological terms. Since the beginning of the 21th century, the literature concerning future studies is constantly increasing. Some authors like Daniel Gilbert and Timothy Wilson identified four kinds of errors in thinking about future: generalization of scenarios based on extreme events leading to neglect ordinary or silent processes; essentialization of main characteristics, leaving out details about how the simulated future is to be achieved; focalization on a few select moments of the future; and decontextualization, because the context in which we form expectations about the future may well not be the same in which they are realized8].
Recent developments in the sociology of controversy have led to following the transformation of actors and arguments over a long series of events and confrontations. To underline the heuristic virtues of the concept of argumentative trajectory, we have put the focus on the emergence of future representations in argumentation, sometimes strong at the beginning, but increasingly controversial, sometimes weak or in the margins but more and more taken seriously by actors. Without developing here deep examples, let us cite two contrastive cases:
The first case study focused on the use of climate change to justify new nuclear power developments, which followed the global alert raised by the successive reports of the Intergovernmental Panel on Climate Change (IPCC) and the Kyoto Protocol (1997). It was long and difficult for the anti-nuclear movement to elaborate convincing and strong counter-arguments against this new strategy of justification. The shock between the different visions of the future is intense since many years. Fukushima has clearly break the calculation space shared by the nuclear industry and the authorities, providing a new legitimacy to actors defending the complete ban of nuclear energy – even if the balance of power varies a lot from one country to another.
The second example is based on the shift in toxicology and risk assessment concerning low dose and cumulative exposure with endocrine disruptors. The clear acceleration of the endocrine disruptors issue in the public arenas, especially in 2010 and 2011, around the Bisphenol case, in the US and in Europe, has produced a transformation of the sets of actors and arguments, changing the ways of presenting the past, the present and the future, and the urgency to regulate more strictly chemicals.
Our argumentative sociology program shares many dimension with the approach developed by Reiner Keller in SKAD (Sociology of Knowledge Approach to Discourse) (Keller, 2012). What is the historical trajectory (the emergence, presence and disappearance) of discourses and the way they change through time and space? The types of definitions and knowledge performed by discourse in given social contexts are common points examined by the sociology of public problems, the critical discourse analysis approach, SKAD and the pragmatics of controversy. Ways of arguing, rhetorics and narratives are key drivers in the making of hegemonic discourses or power/knowledge regimes. But we must avoid considering discourses as stabilized frameworks. And this, for three reasons: first, we have to consider the possible shifts or perturbations produced by marking events or tipping points (9/11 attacks, Fukushima, Trump election, etc); second, we have to understand the types of anchorage of discourse in sensitive and physical world through what Reiner Keller calls “infrastructures of discourse production”, shaping the dynamics of discursive meaning making – as corpus analysis does not mean ignoring the materiality of social practices; third, even in a totalitarian state or authoritative government, a hegemonic discourse allows some actors to produce a counter-discourse, often engaged provisionally in micro-worlds, by which, depending on balance of power and social mobilization, they can prepare and organize the challenge of the dominant representation and impulse change in institutions (Keller, 2018).
Abbott (2001) On the concept of turning point. In: Time matters: on theory and method. Chicago: University of Chicago Press: 245-248.
Angenot M (2008) Dialogue de sourds: traité de rhétorique antilogique. Essai. Paris: Mille et une nuits.
Beck U. (2008) World at Risk. Cambridge (UK): Polity Press.
Chateauraynaud F (ed.) (2014) Une pragmatique des alertes et des controverses en appui à l’évaluation publique des risques. Paris: ANSES (December).
Chateauraynaud F and Chavalarias D (2018) L’analyse des grands corpus évolutifs et la sociologie pragmatique des controverses. Croiser les méthodes face aux transformations des mondes numériques. Sociologie et Sociétés. Forthcoming.
Chateauraynaud F and Debaz J (2017) Aux bords de l’irréversible: sociologie pragmatique des transformations. Paris: Pétra.
Chateauraynaud F and Doury M (2010) The Collective making of temporal Aspects in Public Debates. In: Proceedings of the 7th Conference of the International Society for the Study of Argumentation. Amsterdam.
Foucault M (1971) Orders of discourse. Social Science Information 10(2): 7–30. DOI: 10.1177/053901847101000201.
Foucault M (1982) The archaeology of knowledge. New York, NY: Pantheon Books.
Foucault M and Kriegel B (eds) (1975) I, Pierre Rivière, having slaughtered my mother, my sister, and my brother: a case of parricide in the 19th century. 1st American ed. New York: Pantheon Books.
Gärdenfors P and Makinson D (1988). Revisions of knowledge systems using epistemic entrenchment. In: Proceedings of the Second Conference on Theoretical Aspect of Reasoning About Knowledge. San Francisco: Morgan Kaufmann Publishers Inc. 83–96.
Keller R (2012) Doing discourse research: an introduction for social scientists. Los Angeles: SAGE.
Keller R (2018) The Sociology of Knowledge Approach to Discourse. An Introduction. To be published.
Lyotard J-F (1983) Le différend. Collection ‘Critique’. Paris: Editions de Minuit.
Perelman C and Olbrechts-Tyteca L (1969) The new Rhetoric: a treatise on Argumentation. Notre Dame: University of Notre Dame Press.
Roe E (2013) Making the most of mess: reliability and policy in today’s management challenges. Durham ; London: Duke University Press.
Ripple WJ, Wolf C, Newsome TM, et al. (2017) World Scientists’ Warning to Humanity: A Second Notice. BioScience 67(12): 1026–1028. DOI: 10.1093/biosci/bix125.
Smith S (2017) Using speech acts formulas to describe the narrative structure of affaires. Socio-informatique et argumentation. April, 27.
Tindale CW (1999) Acts of arguing: a rhetorical model of argument. SUNY series in logic and language. Albany, N.Y: State University of New York Press.
Tsing Anna and Pollma Elizabeth (2005). Global Futures: The Game. In: Rosenberg D and Harding SF (eds). Histories of the Future. Durham: Duke University Press.
Walton DN (2008) Informal logic: a pragmatic approach. 2nd ed. Cambridge ; New York: Cambridge University Press.
- Our main partner along this research was the French national agency for health safety (ANSES), see the conclusive report: (Chateauraynaud, 2014).
- Networks of entities (and others…) computed by Prospéro can be transferred for analysis and spatial visualization to specialized software (Pajek or Gephi).
- (Perelman and Olbrechts-Tyteca), 1969. For recent discussions of Perelman’s theory of argumentation, see (Tindale, 1999; Chateauraynaud and Doury, 2010).
- Different notions of Discourse are engaged in the French Social Sciences, between narrativism, Foucault and argumentative theories. According to Reiner Keller, a permanent interlocutor on the relationships between sociology and discourse, even in Michel Foucault, there are different concepts of discourse: discourses as practices which constitute objects (Foucault, 1982); discourses as ordered ways of speaking the truth (Foucault, 1971); discourses as power struggles on the definition of phenomena or situations (Foucault and Kriegel, 1975).
- It is now clear that “Heidelberg Appeal” was created and promoted by polluting industry lobby. What interpretation of the text signatories really endorsed is less obvious, as a part of them belong also to UCS.
- A petition signed by more than 1700 “independent scientists”, including the majority of living Nobel laureates, penned the 1992 “World Scientists’ Warning to Humanity”.
- A corpus was dedicated to the public discourse produced around the Rio+20 summit (2012), and the United Nations final declaration « The Future we want ».
- For an extended commentary of these aspects, see (Roe, 2013).