Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Climate models in modal adverbials : representational practice and deep uncertainty in the IPCC summary… Roeder, Geoffrey Gilbert 2011

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2012_spring_roeder_geoffrey.pdf [ 528.42kB ]
Metadata
JSON: 24-1.0072479.json
JSON-LD: 24-1.0072479-ld.json
RDF/XML (Pretty): 24-1.0072479-rdf.xml
RDF/JSON: 24-1.0072479-rdf.json
Turtle: 24-1.0072479-turtle.txt
N-Triples: 24-1.0072479-rdf-ntriples.txt
Original Record: 24-1.0072479-source.json
Full Text
24-1.0072479-fulltext.txt
Citation
24-1.0072479.ris

Full Text

Climate Models in Modal Adverbials: Representational Practice and Deep Uncertainty in the IPCC Summary Documents  by Geoffrey Gilbert Roeder B.A. Hons., University of British Columbia, 2009  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ARTS in THE FACULTY OF GRADUATE STUDIES (English)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  December 2011 © Geoffrey Gilbert Roeder, 2011  ABSTRACT In a warming climate, policymakers require the best available information to develop the most effective responses. These responses could work to mitigate the worst consequences of climatic change or support adaption to the unavoidable ones. However, the challenges of communicating deeply uncertain information from statistical climatology to a non-technical audience are manifold. The complex uncertainties inherent in the development and study of General Circulation Models (GCMs), statistical climatology’s primary inferential tool, exacerbate these challenges. The Intergovernmental Panel on Climate Change (IPCC) is a U.N. body formed to solve this problem. Employing standardized sets of modal adverbials to indicate the certainty of a finding, the IPCC works to regularly communicate all policy-relevant uncertainties. The present thesis, analyzing the IPCC’s “Climate Change 2007: Synthesis Report” (SYR) and its attendant “Summary for Policymakers” (SPM), argues that the IPCC’s recently much-criticized treatment of uncertainty is a direct response to the combative politics of climate change. A comparison of the IPCC’s treatment of a few key findings to a number of independent analyses of the same highlights the problem. Through a linguistic and rhetorical analysis of the SYR and SPM, I argue that a number of problematic uncertainties are represented as if politically manageable. Independent literature on the reliability of long-term inference from GCMs cautions against such an interpretation. Partly motivating the IPCC’s practice (evidenced in IPCC-internal literature) is desire to control politically hostile and possibly inaccurate interpretations of the uncertainties. Nevertheless, forming adaptive policies on such highly uncertain findings could lead to ineffectual and economically wasteful infrastructure projects. Such policies could also distract from effective responses that might mitigate the worst climatic changes. Following recent sociological and historical work in the discipline of Science and Technology Studies, I argue that the IPCC’s problematic schema for the communication of uncertainty is best understood as a response to the contemporary democratic climate of distrust and distance. The IPCC’s modal adverbials, interpreted as a complex form of social and literary technology, are an attempt to erect a new and ambitious boundary around what constitutes a politically-stable scientific fact and what constitutes a scientific topic under deep investigation.  ii  TABLE OF CONTENTS ABSTRACT ................................................................................................................................. ii TABLE OF CONTENTS ............................................................................................................. iii ACKNOWLEDGEMENTS ........................................................................................................... v DEDICATION .............................................................................................................................. vi 1. INTRODUCTION .................................................................................................................... 1 1.1. General Introduction and Purpose of Study ..................................................................... 1 1.2. Theoretical Background and Overview of Argument ...................................................... 3 2. THEORY AND CRITICISM OF THE IPCC ........................................................................ 10 2.1. Sociological Approaches to the IPCC Discourse ........................................................... 10 2.2. Findings of Independent Studies of the IPCC’s Modal Adverbials ............................... 14 2.3. “Speaking-Truth-to-Power” with Quantitative Probability ............................................ 19 2.4. Subjective and Objective Probability ............................................................................. 20 2.5. Moss and Schneider’s Subjectivist-Bayesianist Definition of Probability ..................... 21 3. MODAL ADVERBIAL TREATMENT OF UNCERTAINTY IN THE SPM ..................... 27 3.1. Theoretical Underpinnings of Present Discourse Analysis ............................................ 27 3.2. Lexical and Rhetorical Analysis of SPM ........................................................................ 29 3.2.1. Likelihood Adverbials ......................................................................................... 32 3.2.2. Confidence Adverbials ......................................................................................... 34 3.2.3. Distinctions among Adverbials of Confidence and Likelihood ........................... 35 3.2.4. Summary of Findings ........................................................................................... 38 3.3. “Treatment of Uncertainty” in the AR4 Synthesis ......................................................... 40  iii  4. MODELING UNCERTAINTIES IN GLOBAL CLIMATE MODELS ............................... 46 4.1. Overview of Inference from GCMs ................................................................................ 46 4.2. GCMs and Inductive Challenges .................................................................................... 47 4.3. Political Relevancy of Modeling Uncertainties .............................................................. 50 4.4. Modeling Uncertainties and Policies of Adaption .......................................................... 53 4.5. Modal Adverbials and Deep Uncertainty ....................................................................... 55 5. THE IPCC’S MODAL ADVERBIALS IN SOCIO-HISTORICAL CONTEXT ................. 57 5.1. Private Discretion and Trust in Contemporary Democracies ......................................... 57 5.2. Organizational History of the IPCC ............................................................................... 58 5.3. History of Quantification Practices ................................................................................ 60 5.4. Conclusions and Directions for Further Study ............................................................... 63 WORKS CITED .......................................................................................................................... 65 APPENDICES ............................................................................................................................. 72 Appendix A: Organizational and Hierarchical Structure of the IPCC ......................................... 72 Appendix B: Generative Grammar and the Syntax of the Modal Adverbials ............................. 74 Appendix C: Evidence-Agreement Adverbials ........................................................................... 77 Appendix D: “Robust Findings and Key Uncertainties” in the AR4 Synthesis .......................... 78  iv  ACKNOWLEDGEMENTS I would like to thank in particular Dr. Judy Segal for her invaluable commentary on the many drafts of this essay, and Dr. John Beatty for the clarifying conversations that helped me refine my thinking on a complex and multifaceted topic. Thanks also to the many friends and family whose conversations helped me shape this work.  v  To my Erin: without your love and support during those long summer months, this work could not have been as it is.  vi  1. INTRODUCTION 1.1. General Introduction and Purpose of Study Climate change is one of the most hotly contested and politically-intractable governance concerns of the early twenty-first century. Despite robust and growing evidence of the dangers from human-induced global warming, substantive political action has yet to be taken. The recent, 2009 COP15 U.N. Climate Change Conference in Copenhagen failed to produce any binding agreement on greenhouse-gas emission caps that might replace the Kyoto protocol, due to expire next year (Vidal et al. n. pag.). By the time of the 2010 COP16 conference in Cancún, the political situation had degraded to the point that binding emission caps weren’t even on the table (Goldenberg n. pag.). These political failures have followed closely behind the November 2009 theft and public disclosure of thousands of private emails belonging to scientists working at the Climatic Research Unit (CRU) of the University of East Anglia. A number of these scientists were accused of improper secrecy and insularity. As one journalist described the situation, “Some e-mails seemed to show that leading climate scientists, who had contributed key findings to previous IPCC reports, had tried to stifle critics” (Schiermeier “Major” 261). The following increase in scrutiny upon climate scientists yielded, in early 2010, the discovery of a few mistakes and improper extrapolations from evidence in the Intergovernmental Panel on Climate Change’s (IPCC’s) Fourth Assessment Report. The problems were located primarily in Working Group II’s (WG II’s) study of the potential impacts of climate change on natural and human systems. The IPCC, however, was subjected to intense media and political scrutiny. Since the leak of the CRU emails and discovery of the evidentiary problems, the IPCC has undergone two extensive independent reviews of their communication practices for the representation of scientific uncertainty. The Netherlands  1  Environmental Assessment Agency (PBL) engaged in the first review, at the behest of the Dutch Parliament. The InterAcademy Council (IAC) engaged in the second review, at the request of the U.N. and the IPCC. Both review panels supported all of the IPCC’s key findings about the growing human influence on the changing climate. Both, however, identified a number of problems with the way the IPCC has handled the communication of scientific uncertainty, prompting calls for a massive overhaul (see Tollefson “Survive” and Schiermeier “Major”) in the IPCC’s assessment and review processes. The IPCC’s present dilemma affords a unique opportunity to study a remarkably transparent organization’s public communication of scientific uncertainty. Moreover, the independent expert reviews of the IPCC’s treatment of uncertainty provides an opportunity for communications scholars in the field of Science and Technology Studies to study the IPCC’s successes and failures in their explicitly politico-scientific project. The most salient IPCC publication at the interface of climate science and public policy is “Climate Change 2007: Synthesis Report.” This publication consists of two parts, the “AR4 Synthesis” and its “Summary for Policymakers” (SPM). The AR4 Synthesis is a collation and summary of the IPCC’s vast Fourth Assessment Report (AR4), written for non-technical readers and policymakers (IPCC “Publications and Data” n. pag.). The SPM itself is simply a summary of the AR4 Synthesis. The IPCC’s larger assessment consists of thousands of pages of materials. In it, the IPCC assess climatic change on three fronts: the physical-scientific basis of global warming, the projected regional effects of global warming, and the economic options for mitigating the warming and adapting to unavoidable changes. The final step of the IPCC’s assessment, a review of the economic options to mitigate future global warming and adapt to unavoidable changes, forms the political substance of the  2  IPCC’s mandate (IPCC “Organization” n. pag.). The IPCC reports scientific findings under uncertainty in support of their mandate. IPCC has designed and implemented, over its twentythree year history, a set of textual practices that standardizes the communication of uncertainty across all three areas of their assessment. These practices are at the heart of the IPCC’s ongoing political and evidentiary problems. The present study will engage in a discourse analysis of the IPCC’s Summary for Policymakers and AR4 Synthesis to argue that the IPCC’s recent problems can be traced to their present schema for the communication of uncertainty. Recent work in the history and philosophy of science, moreover, has shown how quantification and standardization were adopted, in the recent past, as key persuasive practices for the management of distrust and distance that attends high-stakes decision-making among a dispersed and culturallyheterogeneous democratic polity. A linguistically-informed Science and Technology Studies perspective on the IPCC’s schema for the communication of uncertainty will inform both potential solutions to the organization’s present problems and provide a case study of the political uses of quantification and standardization in the dual context of difficult uncertainties and high decision-stakes. The current form of the IPCC’s schema privileges the type of political authority that a greater quantification and standardization of uncertainty can yield over that of a careful and comprehensive reportage. The treatment of uncertainty that the garnering of such a form of political authority requires, however, interacts poorly with the many policy relevant uncertainties involved in the projection of future climate states.  1.2. Theoretical Background and Overview of Argument Moving now into a brief historical and theoretical background, the present section will lay out the major arguments of the present essay. The IPCC’s textual management of uncertainty  3  is a response to the two most salient challenges in its context of communication: the difficult uncertainties of statistical climatology and the contemporary political climate of public distrust and suspicion. The IPCC’s current textual practices for the quantification and precise communication of uncertainty, moreover, were adopted in 2000 to solve a perceived political problem of non-experts misunderstanding the quantitative level of risk (Moss and Schneider 35). The solution was a redefinition of the meaning of probability statements for the IPCC lead authors. The upshot of this redefinition is that it rendered conceptually justifiable the tacit combination of objective and subjective inductive inference in the assignment of probabilities to possible outcomes of climatic change.1 This documentary move, moreover, presupposes a particular ideal role for scientific knowledge in deliberative policy action. This role requires that scientific institutions produce precise and authoritative knowledge that can then form the sole basis of substantive policy action. Politically and pragmatically, such a set of presuppositions interacts poorly with the challenging inferential context of long-term global climate forecasting. Indeed, statistical climatology involves substantial and in many cases irreducible uncertainties in medium- and long-term projections of global warming. The IPCC, moreover, does not engage in original research on climatic change (IPCC “Organization” n. pag.). Instead, they review and summarize research that is itself produced within the relevant scientific or social scientific fields. Statistical climatology nevertheless sets the physical scientific basis for these other, interdisciplinary studies of climatic change. Statistical climatology employs ambitious and computationally complex General Circulation Models (GCMs, also called Global Climate Models) to study the global climate system (see IPCC “Regional Climate Models”). Using GCMs, climatologists work to computationally reproduce the currently-understood physical  1  I will discuss the distinctions between these two forms of probabilistic inference in greater detail in chapter 2, beginning on page twenty.  4  processes of the actual global climate. Given a sufficient structural identity between the GCMs and the real-world climate, statistical climatology will eventually be able to project, with great precision and certainty, the future states of the global climate system. However, current problems in the too-large geographical scale of the information that GCMs yield as well as known gaps in the models’ representation of the physical climate detract from the certainty and robustness of the outputs (see Oppenheimer et al.; Stainforth et al. “Decision-support Relevance”; and Stainforth et al. “Issues”). This lack of certainty forms the key challenge for a portion of the IPCC’s mandate: to review research that could support regional adaptation to unavoidable climatic changes (see Oreskes et al. “Adaptation”). Such adaptation would require geographically specific and reasonably certain projections of the magnitude and direction of climatic change that, in most cases, is unavailable in GCM outputs (Oreskes et al. “Adaptation” 1017-22). How the IPCC textually frames and represents the uncertainties behind GCM-derived projections of future climate change is therefore of paramount importance to the political mobilization of climate science in a deliberative policy decision. The Netherlands Environmental Agency and the InterAcademy Council, moreover, have, in their recent technical critiques, identified the IPCC’s assignment of quantitative probabilities2 to the possible outcomes of climatic change on the scale of human and natural systems as the most problematic feature of the reports. This is the very scale at which GCMs cannot always output sufficiently specific or certain information. The recent political failures at COP15 and COP16, as well as the recent criticisms of the IPCC’s treatment of the uncertainty, point to a serious incompatibility of certain features of the IPCC’s representational practices with the political context of climate-change  2  More specifically, they objected to portions of IPCC’s Working Group II’s contribution to the overall assessment. Please refer to Appendix A for a discussion of the IPCC’s organizational and assessment structure.  5  policy. These representational practices, moreover, have a specifiable locus in the discourse: the IPCC’s choice to regularly communicate all types and sources of uncertainty under three distinct sets of modal adverbials. Linguistically, the IPCC’s modal adverbials3 consist of three sets of formally distinct phrasal sentence-constituents that appear in such forms as “likely,” “high confidence,” or “high agreement and much evidence.” Over their twenty-three-year history, the IPCC developed and adopted the modal adverbials in order to manage the non-technical communication of the highly technical uncertainties involved in assigning probabilities to GCM outputs. The modal adverbials standardize the IPCC’s communication of all types and sources of uncertainty under these three sets of terms. In service of that standardization, two of the three sets of modal adverbials, the first involving the lexeme “likely” and the second involving “confidence,” are calibrated to a precise, quantitative scale.4 The modal adverbials are thus the mechanism for the lead authors to assign a probability to a given finding by indicating both verbally and quantitatively how certain it is. The IPCC’s use of the modal adverbials is the focal point for the problems noted by independent reviewers of the reports. The modal adverbials, moreover, were initially developed and suggested to solve an explicitly political problem. The IPCC adopted the modal adverbials as part of a larger set of guidelines in 2000 written by the public policy scholar Richard Moss and statistical climatologist Stephen Schneider. These guidelines presented a new way to think about the meaning of assigning probability to possible outcomes. Moss and Schneider suggested that, given the difficult uncertainties in statistical climatology, lead authors should interpret probability within a strictly 3  In terms of a traditional English grammar, the modal adverbials are analogous to “sentence adverbs,” or adverbs that modify an entire clause or clausal complex. I will return to a formal definition of the modal adverbials in greater detail in chapter 3 of this study. 4 The IPCC’s Working Groups I and II employed these scales. Please refer to Appendix A for a discussion of the IPCC’s organizational and assessment structure.  6  subjectivist paradigm of probabilistic inference.5 Such a move rationalized, for the IPCC authors, a greater quantification of otherwise-unanalyzable probabilities and their assignments among the modal adverbials. Moss and Schneider’s guidelines thereby rationalized the tacit combination of subjective and objective probabilistic inference in the assignments of probability among the modal adverbials. Moreover, Moss and Schneider’s rationale for the IPCC’s documentary move is cogent only to the extent that one subscribes to a particular role for science in policy: to provide authoritative and unambiguous information to policymakers. Within the difficult inferential context of GCM modeling, however, these assumptions have come to impede rather than facilitate the communication of policy-relevant uncertainties in the IPCC assessment reports. Moss and Schneider’s recommendations were framed as the solution to a perceived problem: the potential miscommunication of risk by scientists to policymakers. Historian of science Lorraine Daston has identified a representational practice like the one employed by the IPCC—one that backgrounds the people or subjective choices behind a scientific assessment—as a cultural practice of “aperspectival objectivity” (“Escape” 112). Aperspectival objectivity as an organizational practice indexes the absence of any biased, vitiating subjective perspective. The relevance of this practice to the IPCC becomes apparent in the work of another recent historical study. Theodore Porter extends Daston’s work in his Trust in Numbers, a cultural history of quantification and standardization practices. Porter argues that “Rigorous quantification is demanded … [when] discretion has become suspect. Mechanical objectivity serves as an alternative to personal trust” (196).6 The IPCC’s elision of the difference between subjective and  5  A subjectivist paradigm of probabilistic inference will be explored in much greater depth beginning on page twenty. 6 Porter mobilizes the term “mechanical objectivity,” which appears first in Daston and Galison’s “Image of Objectivity.” However, Porter appears to have in mind also the authority of Daston’s “aperspectival” form of  7  objective probabilistic inference in the assignment of a probability draws upon and textually garners a “mechanical objectivity” in Porter’s sense of the term. Before going any further into my argument, a brief aside on the course of the present study will help clarify my analytic approach to the IPCC. The adoption of a discourse-analytic framework in the present analysis will facilitate a study of the connections among the modal adverbials in the Summary for Policymakers and AR4 Synthesis; their rationalization and basis in Moss and Schneider’s guidelines; and the recent historical work on the cultural functions and uses of such practices. In the chapter to follow, I develop a theoretical framework to facilitate the above explication. At the core of this framework is Erving Goffman’s theory of “footing.” Goffman’s theory hypothesizes a causal relationship between strategic textual choices in a document and the imagined orientation of an author towards his or her audiences. As a methodological move, a look at the IPCC’s “footing” will focus the present study on the modal adverbials as persuasive resources of the text, as features of the “rhetoric” of the AR4 Synthesis and SPM. The rhetoric of a text is the contextually local, socially- and historically-contingent sets of textual resources that achieve a given persuasive function in the discourse. The rhetorical features of the discourse-functions of the modal adverbials, to be explored in detail in chapter 3, facilitate the standardization and quantification of subjective inductive inference. Motivated primarily by the IPCC’s need to manage trust and authority in a deliberative policy context, the IPCC’s current use of the modal adverbials garners a political authority of mechanical and  objectivity. At one point Porter writes, “Implicit already in [the] bureaucratic uses [of costs-benefits analysis] … were pressures to reify its terms, to deny the validity of human judgment, to lust after the impersonality of purely mechanical objectivity” (187). Porter moves easily between the mechanical and the impersonal (and, necessarily, aperspectival) forms of objectivity. In both cases, agentive human intervention is the impediment to true knowledge of the world. An ostensive display of its absence is the root, Porter’s study argues, of the political authority of objectivity: the authority of a true knowledge of the world (see p.196 in particular).  8  aperspectival objectivity at the expense of the policy-relevant communication of modeling uncertainties.  9  2. THEORY AND CRITICISM OF THE IPCC 2.1. Sociological Approaches to the IPCC Discourse Goffman’s seminal concept of “footing” provides a methodological distinction for interpreting the IPCC’s modal adverbials within their greater social and historical context. The footing of a speaker is, at the highest level of abstraction, the stance that she takes up in relation to her audience (see Goffman “Footing”). At the level of the text, she discursively performs and thereby enacts the power relationships that this role presupposes by her use of the strategic features. “Footing,” as a theoretical construct, thus names the sets of role and audience presuppositions that guide particular expressive and strategic choices in a text. The following materials begin to elicit the IPCC’s textual footing by considering the IPCC’s statements of selfdefinition. The nature of this footing will then be explicated further through a discussion of Moss and Schneider’s guidelines document. This discussion will be preceded by a closer look at the recent independent critiques of the IPCC in order to locate more precisely the problems in the assignments of probabilities among the modal adverbials. The present chapter will develop a sociological framework within which the IPCC’s strategic textual choices may be productively interpreted. This framework will then be applied to a discourse analysis of the SPM and AR4 Synthesis in chapter 3. The IPCC’s published statements of self-definition provide a general indication of their footing. Under the “Organization” link on their website they write, “[the IPCC] does not conduct any research nor does it monitor climate related data or parameters” (“Organization”). Rather, “thousands of scientists from all over the world contribute to the work of the IPCC on a voluntary basis” (ibid). Thus, while the IPCC does insist on being recognized as a “scientific body” (ibid) they understand their role to be primarily hortatory: to animate the work of  10  scientists worldwide. Indeed, they strive “to provide the world with a clear scientific view on the current state of climate change and its potential environmental and socio-economic consequences” (ibid). On account of this role, they make explicit demands for a particular kind of political authority. “Because of its scientific and intergovernmental nature,” their official statement reads, “the IPCC embodies a unique opportunity to provide rigorous and balanced scientific information to decision makers. By endorsing the IPCC reports, governments acknowledge the authority of their scientific content” (ibid). The IPCC makes an explicit bid for political authority on the credentials of the science that they review. The IPCC, in other words, bids for a political and deliberative authority on the cultural authority of the objectivity of science, on the cultural authority of Daston’s mechanical and aperspectival objectivity. Daston, in her “Objectivity and the Escape from Perspective,” and Peter Galison in the article he co-wrote with Daston, “The Image of Objectivity,” historicize the concept of objectivity by analyzing the different cultural functions that the term has served over the 18th, 19th, and 20th centuries. Daston and Galison were seeking to expand upon the thendominant theory of objectivity as what Thomas Nagel has called the “view from nowhere,” or what Donna Haraway colourfully rephrased as the “god trick” (“Situated” 581): the idea that objectivity is the absolute separation of the enquirer from him or herself, the adoption of the privileged and one true perspective upon the world. Daston and Galison argue, in “The Image of Objectivity,” that, over the history of the concept: mechanical objectivity became fused with other varieties of objectivity, such as the metaphysical element that makes objectivity synonymous with truth, or the aperspectival element that identifies objectivity with the escape from any and all perspectives. Each of  11  these elements has a distinct history, as well as partaking of a collective history binding them into a single concept. (123) Daston and Galison’s study suggests that, in the contemporary usage of the concept, “objectivity” can carry these three different valences simultaneously: protection against the intrusion of an individual, biased mind; the imputation of a metaphysical truth; and the escape from any single, local perspective. The IPCC’s invocation of science as a de facto authority in political governance presupposes the availability and cultural functionality of these intertwined meanings of objectivity. Because objectivity yields an unbiased truth through the absence of biased human intervention, the role of a scientific organization in governance must be that of an authoritative outsider, an adviser to, rather than a participant in, deliberative policy action. Such a role yields a particular footing for the IPCC discourse, an imagined schematization of their stance in relation to their non-technical audience. Sociologists of scientific knowledge Sheila Jasanoff and Brian Wynne have studied in detail the influence of such presuppositions on the orientations that scientific advisory bodies adopt towards their non-scientific audiences. The presuppositions undergirding the IPCC’s selfdefinition have been a typical feature of late-twentieth-century scientific advisory bodies. Jasanoff and Wynne call such a group of presuppositions the “speaking-truth-to-power” model for the role of science in policy contexts (“Science and Decisionmaking” 8). Such a model suggests that the ideal role for scientists in deliberative policy decisions is “to stand apart from politics while proffering impartial knowledge to formal policy institutions” (idem 7-8).7 Such a  7  As a schematization of the role for science in society, this model itself presupposes a sharp distinction between the spheres of political action and scientific research (idem 8). As the originator of this distinction, Jasanoff and Wynne cite Merton’s seminal 1942 “The Normative Structure of Science” in which Merton argues for the moral exceptionality of the discipline of science. This exceptionality putatively provides science and scientists with a privileged position from which to produce objective knowledge of the real world. In the intervening years, Mertonian sociology of science has fallen out of favour after empirical studies of the behaviour of scientific  12  model, moreover, poorly accounts for the role that textual practices of definition and framing by scientific organizations play in the non-technical, deliberative evaluation of “the efficacy, merits or legitimacy of competing social policies” (idem 5). As Jasanoff and Wynne remark, “[continued] efforts by both scientists and policy analysts to preserve the distinctness of their spheres of action underscores the dominance of the [“speaking-truth-to-power”] paradigm, even though participants sometimes admit that the separation, if any, is more honored in the breach than the observance” (8).8 Goffman’s “footing,” when incorporated into a “speaking-truth-to-power” framework, provides a connection between the IPCC’s textual choices and their social and political aims. The IPCC’s “speaking-truth-to-power” footing—the orientation of the organization towards their non-technical audience, expressed and enacted through the strategic choices in the document— can account for the problematic representational choices in the AR4 Synthesis and SPM. Providing the background for this account, a closer look at the independent scientific critiques of the IPCC will locate more precisely the problems with the IPCC’s assignment of probabilities among the modal adverbials. These independent critiques show that the assignment of probabilities within the modal adverbials yields an occasional over- or under-attribution of risk. Additionally, the modal adverbials provide a veneer of certainty even when the underlying literature does not always support it. This veneer is the cultural authority of objectivity in Daston, Galison, and Porter’s sense. It operates, as I will show in chapter 3, within identifiable rhetorical discourse-functions of the modal adverbials. Before entering into the discussion of the communities called into question the existence and social function of the norms (see Shapin’s Scientific Life 11319). 8 Jasanoff and Wynne refer in this quote to Vavenar Bush, adviser to President Roosevelt, who mobilized a Mertonian ideal of pure science as the grounds for effective policy advising in a 1945 report entitled Science, the Endless Frontier (Jasanoff and Wynne 7). Bush distinguished between pure apolitical science and applied, less objective science (ibid). I changed “Bush” to “[speaking-truth-to-power]” to maintain that previous line in their argument.  13  independent critiques, a look at the modal adverbials as they appear in the text will help ground the materials to follow in the IPCC’s Summary for Policymakers and AR4 Synthesis.  2.2. Findings of Independent Studies of the IPCC’s Modal Adverbials In a typical sentence, IPCC authors employ the modal adverbials as pre- or postmodifiers of a given clause to signal the epistemic status of a reported finding. One line in the Summary for Policymakers reads, “There is very high confidence that the net effect of human activities since 1750 has been one of warming” (5). In the syntax of this sentence, the italicized noun-phrase “high confidence” adverbially modifies not the copular verb “to be,” but the rightmost sentence-constituent, the subordinate “that-clause.” The central claim, “the net effect of human activities since 1750 has been one of warming,” is conditionalized by the modal adverbial which expresses what level of confidence a reader should have in it. Modal adverbials represent one of the most complex areas of English grammar and can most clearly be defined in terms of their similar functions, rather than any shared formal features. The IPCC calibrates two of the three sets to a quantitative scale. That is, if a reader were in doubt about just how high “very high confidence” is, she could consult a table on page twenty-seven and find that the modal adverbial precisely indicates “at least [a] 9 out of 10” chance that the finding is correct (IPCC “Synthesis”). The standardized and quantitatively-calibrated precision of the modal adverbials is, in some cases, problematic from the perspective of a number of independent scientific reviewers. The Netherlands Environmental Assessment Agency (or Planbureau voor de Leefomgeving; PBL) has engaged in the most detailed review of the AR4 Synthesis to date. They found a few of factual errors and a number of improper extrapolations from evidence. In a few rare cases, the PBL was unable to trace the findings back to any underlying scientific research  14  whatsoever (see PBL pp. 33-44). While the PBL found no problems with the overarching conclusions of the IPCC (44), they criticized the specificity and evidentiary basis of a number of Working Group II’s (WG II’s) use of the modal adverbials. WG II was tasked with assessing the specific regional effects of climatic change. WG II had conditionalized a number of the statements that lacked an appropriate evidentiary basis with the following modal adverbial: “Unless stated explicitly, all entries are . . . either very high confidence or high confidence” (12). “Very high confidence” is correlated, within the IPCC’s schema for the communication of uncertainty, with a specific quantitative value to indicate “at least 9 out of 10” chance of “a finding being correct” (AR4 Synthesis 27). The lead authors, at these problematic points in the text, presented their subjective confidence in highly uncertain and empirically unsupported outcomes of climatic (PBL 43) as if the findings were well-established and peer-reviewed. Another independent group of scientific reviewers criticized the IPCC for a similar problem: expressing a possible range of sea-level rise as “likely,” or “>66%,” when the range manifestly did not take into account a known and potentially very large contributor to that future rise. Michael Oppenheimer, Brian O’Neill, Mort Webster, and Shardul Agrawala, in a 2007 article in the periodical Nature, criticize the IPCC’s projection of sea-level rise by the start of the 22nd century as too low. They note that the numerical range reported in the Summary for Policymakers9 does not include a possible contribution by rapid dynamical ice-flow processes in the Greenland and West Antarctic Ice Sheets (or “WAIS”) (Oppenheimer et al. 1505). These processes, comment the authors, are well known to have contributed to sea-level rise in the past decade and the recent, observed sea-level rise can’t be accounted for without it (Oppenheimer et al. 1505). Nevertheless, the IPCC excluded these processes from the reported sea-level rise range 9  In this case, the summary the authors refer to is of WG I’s contribution to the overall assessment, not the summary of the SPM or AR4 Synthesis. The same numbers and quantified likelihood adverbial do appear in the latter summaries, however.  15  of 18 to 59 centimeters (IPCC “Summary” Physical Science 13). The authors still employed the modal adverbial “likely,” indicating a quantitative likelihood range of 66% or greater. As in the cases noted by the PBL, the quantitative modal adverbial suggests a high degree of robustness to the analysis that isn’t always supported by a closer look at the underlying research. The IPCC do allude to this uncertainty that tacitly conditionalizes the precise, quantified value of the modal adverbial. A comment in the column-title of the table that tabulates the possible range of sea-level rise reads, “Model-based range excluding future rapid dynamical changes in ice flow” (ibid). The melting of the WAIS, comment Oppenheimer et al., will undoubtedly increase the final amount of sea-level rise10 over the 21st century, rendering the policy value of the “66% or greater” modal adverbial questionable (1506). Oppenheimer et al. and the PBL show that the modal adverbials are flexible enough to lend the authority of quantitative analyses to findings even when those analyses are not necessarily well-studied enough to provide sound adaptation-policy advice. More problematic is that it is difficult if not impossible to distinguish such partially-incomplete findings from more robust ones on the basis of the information presented in the summary documents. The standardized modal adverbial “likely,” along with the quantified values reported in the table, have screened from view any assessment of the policy-relevant influence of the uncertainties. Without a study more focused on these uncertainties, even Oppenheimer et al. can only speculate about the influence of the WAIS on the possible magnitude of 21st century sea-level rise (1506). In the context of policy advising, the modal adverbials can, in their underestimation of risk, be dangerously misleading.  10  The WAIS consists of enough ice, comment the authors, to raise the sea level by 5m (1505). Oppenheimer et al. are not saying that a complete melting of the WAIS is going to occur in the 21st century. Their point is that, in the IPCC’s assessment of risk, any possibility of a contribution to sea-level rise from such a large body of ice is highly relevant to policymakers.  16  The InterAcademy Council’s review similarly criticized IPCC lead authors for a misleading use of the modal adverbials. The IAC expressed particular concern with the IPCC’s modal adverbial expressions of confidence and likelihood. The IAC writes: authors reported high confidence in statements for which there is little evidence, such as the widely-quoted statement that agricultural yields in Africa might decline by up to 50 percent by 2020. Moreover, the guidance was often applied to statements that are so vague they cannot be falsified. In these cases the impression was often left, quite incorrectly, that a substantive finding was being presented (IAC prepublication 36). The IAC’s concern is with the appropriateness of degrees of confidence and certainty expressed by the modal adverbials. Indeed, the IAC’s choice of “falsifiability,” drawing on a Popperian philosophy of science, is telling. The word “falsified” was used in the IAC’s prepublication copy quoted above, but in the final copy, they replaced it with an alternative: “so vague they cannot be disputed” (IAC final ed. 13). The IAC’s comments suggest that certain statements to which the “guidance,” or the modal adverbials derived from Moss and Schneider’s recommendations, was applied, are inappropriate within the normative conventions of scientific reportage. The modal adverbial provides the appearance of a substantive, robust finding in each case to which it is applied, even when the finding was, in fact, “based on unpublished or non-peer reviewed literature” (IAC final 34). Both the IAC and the PBL frame these moments in the IPCC reports as either “errors” or “misuses” of Moss and Schneider’s guidelines. Given the IPCC’s careful internal assessment structure, this is surely an incomplete explanation. Each section of the expansive, three-part assessment report and each line of the IPCC’s summary documents were carefully reviewed by IPCC coordinators and review editors, as well as thousands of independent, government-  17  sponsored commentators. The approximate number of comments on drafts of the Summary for Policymakers and the AR4 Synthesis is 90,000 (Tollefson “Trust” 14). The errors and misuses would have to have been repeated consistently among these independent levels of review. The problems noted by the IAC, PBL, and Oppenheimer et al. can be better accounted for as an outgrowth of the IPCC’s orientation towards the role of science in policy. The strategic use of the modal adverbials, in the IPCC’s summary documents, presupposes that authoritative, robust warrants for policy action are the ideal way of communicating with policy makers. The numbers they present, in this imagined orientation, univocally affect the assessment of risk and adoption of policy by the policymakers. Thus, when a choice had to be made between providing a solidseeming quantitative value and not so doing, the IPCC authors erred on the side of caution (as understood within the IPCC’s footing). As the IAC, PBL, and Oppenheimer et al. show, however, the deep11 uncertainties of climate change science render the simple assignment of probabilities among the modal adverbials, in some cases, a dangerously misleading characterization of the risk. The locus of the evidentiary problems noted by independent reviewers is the quantifiability permitted in the standardized modal adverbials expressing likelihood or confidence. Moss and Schneider’s guidelines are the source of such uses of the modal adverbials. Their document redefined, for the lead authors of the IPCC reports, the meaning of the term “probability” and, in so doing, rationalized and permitted the problems noted by independent reviewers. The motivations behind the adoption of the schema and this definition of probability, moreover, were to solve a perceived problem: the potential for policymakers to misinterpret the 11  The debate over the best climate change policies, as the climatologist Stephen Schneider once commented in a 2003 testimony to a U.S. Senate committee: “is characterized by deep uncertainty, which results from factors such as lack of information, disagreement about what is known or even knowable, linguistic imprecision, statistical variation, measurement error, approximation, subjective judgment, and disagreement about structural models, among others” (“Testimony” n. pag.).  18  quantitative level of risk in particular outcomes of climatic change. This issue, however, only appears to be a problem if one presumes that policymakers rely solely on the quantitative values that the IPCC presents. Such a presumption is the hallmark of a “speaking-truth-to-power” model for the role of science in policy decisions. Indeed, while an implicit feature of the IPCC’s statements of self-definition, the “speaking-truth-to-power” orientation towards science for policy is an explicit feature of Moss and Schneider’s guidelines document.  2.3. “Speaking-Truth-to-Power” with Quantitative Probability In 2000, Moss and Schneider published their recommendations for the writing teams of the IPCC’s Third Assessment Report (TAR). These recommendations were disseminated to lead authors in the Guidance Papers on the Cross-Cutting Issues of the Third Assessment Report of the IPCC (Moss and Schneider).12 The motivation behind their recommendations, comment Moss and Schneider, is a concern that “. . . users of IPCC reports often assume for themselves what they think the authors believed to be the distribution of probabilities when the authors do not specify it themselves” (Moss and Schneider 35). Such assumptions, Moss and Schneider feared, might lead to misinterpretations of the degree of probability that the experts intended to express about particular outcomes of climatic change. This potential problem of interpretation was exacerbated by a second, IPCC-internal problem. “Some researchers,” they write, “have expressed concern that it is difficult to even know how to assign a distribution of probabilities for outcomes or processes that are laced with different types of uncertainties” (35). Moss and Schneider’s solution was to suggest a different way of thinking about the assignment of  12  While not officially vetted by the IPCC, Moss and Schneider’s paper has been the conceptual capstone of the IPCC’s schema for the communication of uncertainty in both the Third and Fourth Assessment Reports. The more recent guidelines documents for the Fourth and upcoming Fifth Assessment Report cite Moss and Schneider as their conceptual basis.  19  probabilities to outcomes of climatic change: “subjectivist Bayesianism.” Subjectivist Bayesianism, as defined by Moss and Schneider, provided an epistemological warrant for the elision of the distinction between the possible objective and subjective bases in a single assignment of probability. However, this warrant was less evidentiary than it was political.  2.4. Subjective and Objective Probability Before discussing this warrant, a distinction needs to be made between the different senses of “objective” at play in the present essay. “Objective” when applied to probabilities has a specific, technical meaning that is different from the sense in which historians of science Daston, Galison, and Porter use it. In terms of its evidentiary basis, a probability may be classified as either subjective or objective. In the physical sciences, the primary difference between the two is evidentiary, determined by how much is known about the event at the time the probability is assessed.13 The philosopher Ian Hacking explicates this evidentiary focus in his distinction between a belief-type probability and a frequency-type probability (Inductive Logic 132-3). A frequency-type probability indicates a measure of the propensity of a particular chance set-up to yield a particular outcome. Given a limited and well-defined chance set-up such as a coin toss, for example, a probability is the frequency with which a particular outcome can be expected to be observed over a long run of trials. Frequentist statistical inference is called “objective” because it measures a stable property of the coin in a particular chance set-up, analogous to its weight or its material composition (Hacking Emergence 14). Belief-type probability, on the other hand, is a more capacious concept.  13  See Mike Hulme’s Why We Disagree About Climate Change, pp. 85-7, for an IPCC insider’s perspective on subjectivist Bayesianism. Hulme’s explanation, however, deals with an idealized case of inductive inference rather than one more appropriate to the IPCC’s complex socio-political context.  20  Subjective or belief-type probability indicates the belief of a person that an event will occur given the well-understood features of a chance set-up and a best-guess estimation of the unknown features (Hacking Inductive Logic 132). Rather than relying on the results of a set of formulae, or the calculation of a frequency, a subjective probabilistic inference assigns a chance of occurrence based on private or difficult-to-formalize criteria. Belief-type probability was Moss and Schneider’s solution for those IPCC authors who didn’t know how to assign probabilities to the deep uncertainties of climate change. “[F]or most instances in the TAR,” write Moss and Schneider, “a ‘Bayesian’ or ‘subjective’ characterization of probability will be the most appropriate” (36). They suggest, in this remark, a new way for lead authors to conceptualize their reportage of uncertainty. Moss and Schneider tread lightly, here, on complex evidentiary and pragmatic issues. Evidentiary problems are at the core of the independent scientific critiques of the IPCC’s representational practice. Moss and Schneider’s definition of probability, as they present it in the introduction to their recommendations, provided a conceptual rationalization for the very problematic misuses of the modal adverbials. A closer look at this rationalization reveals its political foundation.  2.5. Moss and Schneider’s Subjectivist-Bayesianist Definition of Probability “In [subjectivist Bayesianism],” the Moss and Schneider explain to IPCC authors, “a ‘prior’ belief about a probability distribution (typically based on existing evidence) can be updated by new evidence, which causes a revision of the prior, producing a so-called ‘posterior’ probability” (36). Moss and Schneider, in this definition, have actually elided an important evidentiary difference between subjective and objective probabilities. “Bayesianism” on its own refers only to the formal method of updating an initial estimate of probability with new evidence  21  as it becomes available. Bayesianism can be either subjective or objective depending on that initial estimate of probability and its underlying assumptions. What logically and formally constitutes an objective or a subjective prior probability distribution has filled volumes of philosophical journals.14 Moss and Schneider, moreover, largely discount this literature and the important logical and evidentiary distinctions that underlie it when they imply that a “Bayesian” and “subjective” characterization of probability are coextensive concepts. Moss and Schneider mobilize the idea of a subjectivist Bayesianist interpretation of probability more than they do the philosophy underlying it. Indeed, Moss and Schneider’s definition allows for a greater standardization of the expressions of uncertainty through a quantification of individual or group beliefs: Applying the paradigm in the assessment process involves combining individual authors’ (and reviewers’) Bayesian assessments of probability distributions and would lead to the following interpretation of probability statements: the probability of an event is the degree of belief that exists among lead authors and reviewers that the event will occur, given the observations, modeling results, and theory currently available. (Moss and Schneider 36) Probability is thus defined by Moss and Schneider as the degree of an author’s (or the group’s consensus) belief that an event will occur. As a paradigm for communication, such an interpretation of probability means that any likelihood reported by the IPCC or any assessment of confidence in the AR4 Synthesis may have its evidentiary basis in an author’s (or group of authors’) belief-type probability; or, the modal adverbial may represent a frequency-type probability; or it may also express a combination of both. The IPCC’s communication of  14  Although see Hacking’s Logic of Statistical Inference pp. 208-27 for a discussion of the major positions and issues surrounding the subjective theory of probability.  22  uncertainty elides the different possible evidentiary bases of their assessment process. The evidentiary basis, under Moss and Schneider’s definition, is backgrounded behind the reportage of an author’s confidence in the finding. What, however, does this mean in terms of the specific objective and subjective evidence that the IPCC authors actually considered? Swart et al. comment just this issue of evidentiary inference in their critique of the communication practices of the Third Assessment Report (published in 2001). Swart et al.’s insider perspective on the schema provides a useful look behind the probabilities of modal adverbials to the authorial choices that assigned them: [Moss and Schneider’s] definition of likelihoods refers to “judgmental estimates of confidence”, which seems to refer to Bayesian probabilities, even though many of the lead authors used the scale as a hybrid of objective and subjective probability. For example, after having determined an ‘objective’ (that is, solely observation- and modelbased estimate of the likelihood of a significant anthropogenic climate signal) likelihood of “more than 90% chance” (“very likely”), a ‘subjective’ evaluation of the quality of the estimate (and thus of the underlying climate models) led the lead authors to lower their likelihood estimate to “more than 66% chance” (“likely”). These considerations by the lead authors have not been made explicit however, and have only come to light after careful study of the TAR WG I production process (Petersen 2006). (11) The invisibility of these subjective choices in the assessment process led directly to the representational problems noted by the PBL, the IAC, and Oppenheimer et al. At certain points, the authors employed private criteria to assign a probability to a finding. Among the modal adverbials, it is impossible to distinguish findings that relied solely on subjective criteria from  23  those that relied on a mixture.15 The IPCC, although manifestly not a scientific research group, sees its role as the provider of the most robust scientific information available. What could have been the motivation for Moss and Schneider’s subjectivist Bayesianism as a paradigm of communication? Moss and Schneider introduce their redefinition of probability with a motivating rationale for its adoption: We believe it is more rational for scientists debating the specifics of a topic in which they are acknowledged experts to provide their best estimates of probability distributions . . . based on their assessment of the literature than to have users less expert in such topics make their own determinations. (35, their emphasis) Within this rationalization, Moss and Schneider mobilize three highly value-laden concepts: “more rational,” “acknowledged experts,” and “users less expert.” These concepts reflect an implicit hierarchy of power-through-knowledge that consolidates the rights to assess uncertainty in the hands of institutionally-licensed statistical climatologists. Appropriate political action, in this hierarchy, requires an expert’s interpretation of statistical climatology that will, by itself and without consultation, identify the one true interpretation of the deep uncertainties. Because the “acknowledged experts” have the greatest access to these interpretive skills, they have an implicit responsibility to manage the perceptions of “users less expert” who might otherwise stray from the correct interpretation. The non-technical audience of the IPCC summary reports is imagined, in this footing, as the “users less expert.” For such users to make their own determinations of risk could be politically dangerous. It is more rational, Moss and Schneider 15  Statistical climatology necessarily relies on a mix of subjective and objective statistical inference. There is, of course, nothing inappropriate about such forms of inference, but they do have different consequences for policy decisions. This topic will form the substance of chapter 4 of the present essay. Please also refer to Stephen Schneider’s 2002 editorial in Climatic Change for a discussion of the role of subjective probabilities in statistical climatology (“Estimate”).  24  argue, for the IPCC lead authors to intervene a priori and ensure that the only quantitative probability assignments are licensed by the community of statistical climatologists. Moss and Schneider’s well-intentioned but mildly paternalistic redefinition for the communication of uncertainty is most cogent within a “speaking-truth-to-power” model for the role of science in policy. For a scientist not to provide a subjective probability, under such a model, would be politically irresponsible. Indeed, Moss and Schneider’s remarks suggest that the issues motivating and rationalizing the adoption of subjectivist Bayesianism are less inferential or communicative than they are political.16 These political concerns manifest in the organization’s discourse footing. This footing, moreover, is discursively enacted through particular strategic discourse-choices in the AR4 Synthesis and SPM that reify the imagined power and role relationships. The rhetorical discourse-functions of the modal adverbials are these strategic discourse-choices. In this light, the PBL, Oppenheimer et al., and the IAC were objecting not to errors in the IPCC reports, but to the rhetorical upshot of Moss and Schneider’s definition of probability. Moss and Schneider’s redefinition of probability, yielding the assignment of subjective and objective probabilities among the modal adverbials, provided a textual bridge between the IPCC’s political aims and the representation of difficult uncertainties in the SPM and AR4 Synthesis. My discussion of Moss and Schneider has, until this moment, set aside an interesting and crucial feature of their guidelines document. A public policy expert and statistical climatologist respectively, they were not blind to the potential policy problems raised in a subjectivist  16  Swart et al. also remark on the motivations behind Moss and Schneider’s guidelines. “[I]ntentionally or unintentionally,” they write, “more political considerations may have played a role. A more thorough reflection on the scientific methods and framing of problems could be seen as a threat to the status of these methods and framings and to the existing culture in the associated disciplines” (12). 16 Swart et al. suggest, here, the very socio-political explanation for the IPCC’s representational choices that my essay is exploring: the political concerns surrounding scientific authority and the history of previous, similar representational practices internal to scientific disciplines.  25  Bayesian paradigm. They made the adoption of a subjectivist Bayesian definition of probability contingent upon the production of what they called “traceable accounts.” A traceable account would identify precisely at what point subjective or private-criteria-based judgments had entered into the assignment of probability to a modal adverbial (46). These traceable accounts would be archived and accessible to any interested reader. The IPCC evidently discounted this precondition; no author provided a traceable account. The inclusion of traceable accounts, moreover, reveals an interesting tension in Moss and Schneider’s recommendations. Moss and Schneider motivate their subjectivist-Bayesianist definition of probability in the value-laden terms of a “speaking-truth-to-power” frame of science-for-policy; they redefine probability in a way that encourages the simplified communication of complex uncertainties; but they ultimately insist that absolute transparency is a precondition for the adoption of such recommendations. Such a tension between recommendations and textual realization, as well as the degree to which absolute transparency would have ameliorated the problems in the schema, will form the substance of chapter 5 of this essay. Both Moss and Schneider’s recommendations and the IPCC’s only-partial adoption of them, I argue, are their response to difficult problems of trust in a difficult political context. Turning now to a lexical and discourse analysis of the rhetorical discourse-functions of the modal adverbials, I will apply the socio-historical framework developed above to the IPCC’s Summary for Policymakers and AR4 Synthesis.  26  3. MODAL ADVERBIAL TREATMENT OF UNCERTAINTY IN THE SPM 3.1. Theoretical Underpinnings of Present Discourse Analysis The influence of Moss and Schneider’s subjectivist-Bayesianist definition of probability is most salient within a discourse analysis of the modal adverbials in the IPCC’s Summary for Policymakers. Since the SPM is the IPCC’s primary interface with policymakers, politicians, and the general public, I treat it in the following analysis as the upshot of the organization’s schema for the communication of uncertainty to a non-technical audience. Because of the scarcity of information on or about uncertainties in the SPM, I follow the logic of the IPCC’s documentation and turn, in the second half of this chapter, to the larger “Synthesis for Policymakers.” Presenting and applying a theoretical framework of generative grammar17 and functional linguistics, the following analysis will highlight both the forms and functions of the modal adverbials. Such an analysis will reveal the resources and gaps in the explanations provided for non-technical readers. It will also explicate the rhetorical discourse-functions of the modal adverbials that textually enact the power-hierarchy and role presuppositions that undergird Moss and Schneider’s subjectivist Bayesian definition of probability. In chapter 4, I will compare the treatment of uncertainty in the AR4 Synthesis and SPM to that of two IPCC-independent groups of scientific commentators. In the AR4 Synthesis and SPM, designed for non-technical, policyoriented readers, the modal adverbials render many of the policy-relevant uncertainties of statistical climatology heavily backgrounded in the majority of cases and entirely tacit in others. The IPCC’s modal adverbials express the grammatical category of “modality.” Modality, explains functional linguist M.A.K. Halliday, “refers to the area of meaning that lies between yes and no – the intermediate ground between positive and negative [clausal] polerity” (332). Within 17  Please see Appendix B for a discussion of my use of generative grammar in the present chapter and a brief analysis, in those terms, of the syntax of the modal adverbials.  27  this area of English grammar, there are many different shades of meaning and modes of expression. In the framework of functional linguistics, the IPCC’s modal adverbials18 signal a kind of modality that Halliday calls “modalization” (“Diversity” 335). Modalization, Halliday remarks, is an “epistemic modality” (ibid) because it signals the both the non-factual status of a sentence and the degree of probability for its occurrence. In the IPCC discourse, modalization functions as what Halliday calls a “speaker’s comment,” where a “speaker associates with the thesis an indication of its status and validity in his own judgment; he intrudes, and takes up a position” (ibid). The modal adverbials, despite their variations in form, each express this same general language-function. The modal adverbials stake out positions on the epistemic status of the findings that they conditionalize. Implicitly, the modal adverbials suggest the level of confidence that a reader should have in the claim by providing assigning a quantitatively-correlated probability. The modal adverbials background the types, sources, and individual experiencers of the uncertainty. Although functionally “speaker’s comments,” they rhetorically “objectify” the findings in the most literal sense possible: by separating them from any grammatical, subjective locus of experience. The discussion to follow will focus primarily on the large-scale features of the IPCC discourse, at the level of the entire clause-complex or greater. It is at this larger scale that the evidentiary problems created by Moss and Schneider’s solution to the communication of uncertainty are most evident.19 The following discussion will catalogue the instances of the modal adverbials in the order that they would be encountered by a first-time reader of the 18  My use of the label “adverbial” in the following discussion (rather than “adverb phrase”) reflects the variations in form but identity of expressive function among the IPCC’s modal adverbials. All the modal adverbials express epistemic modality. Rather than solely adverb phrases (AdvPs), the IPCC authors also use noun phrases (NPs), prepositional phrases (PPs), and adjective phrases (APs) to modalize sentences. 19 Nevertheless, at the fine-grain level of the syntax, the modal adverbials express a similar, objectifying rhetorical function. Please refer to Appendix B for a discussion of this topic, along with a brief methodological discussion of generative grammar.  28  document. Meaning, from a text, is necessarily construed by a reader through both spatial and temporal form. A rational reconstruction of the experience of such a first-time reader who reads linearly through the document will thus reveal otherwise-invisible rhetorical features of the information structuring in the SPM. The following analysis will argue that the modal adverbials are an ideationally truncated but rhetorically rich feature of the IPCC discourse. The distinctions among their subjective and objective inferential bases rely entirely upon their contextualization and discussion within the text SPM. This contextualization, rather than tracing out these different inferential bases, actively identifies the two among the different rhetorical discourse-functions of the modal adverbials.  3.2. Lexical and Rhetorical Analysis of SPM The IPCC authors provide readers with the first and only explanation of the uncertainties behind the modal adverbials in a footnote to the first occurrence of a modal adverbial, on page two. “Words in italics,” a footnote reads, “represent calibrated expressions of uncertainty and confidence. Relevant terms are explained in the Box ‘Treatment of uncertainty’ in the Introduction of this Synthesis Report.”20 This explanation, within the self-contained SPM, suggests that the modal adverbials are self-evident features of the discourse. Moreover, by only labeling them “calibrated,” the text underscores their precision and specificity. A number of important features of the modal adverbials, however, go unexplained. To what have the terms have been calibrated? What types or sources of uncertainty are guiding the assignment of probabilities to findings by means of the modal adverbials? 20  On page nine of the SPM, another footnote indicates that a particular group of findings originate in “… expert judgement of the assessed literature … considering the magnitude, timing and projected rate of climate change, sensitivity and adaptive capacity.” Although the sentence to which this footnote is appended is modified by the adverbial likely, the footnote refers the findings themselves, rather than the modal adverbial which modalizes the findings.  29  A reader is free, of course, to skip past the twenty-two page SPM and revise her understanding of these terms by following the directions in the footnote, by reading the explanation at the start of the AR4 Synthesis. The AR4 Synthesis, however, as I will explore in the next section, is similarly opaque as to the influence of the uncertainties on the modal adverbials. For the reader to jump past the SPM suggests, moreover, that the reader does not take seriously the claim on the title page of the SPM, that, “this summary . . . represents the formally agreed statement of the IPCC concerning key findings and uncertainties contained in the Working Group contributions to the Fourth Assessment Report” (1). Each line of the Summary for Policymakers was carefully vetted and reviewed not just by the authors, but by independent editors and even representatives from U.N. member governments. The extent to which the IPCC addresses uncertainties in the SPM is, therefore, representative of the IPCC’s schema for the communication of uncertainty. In light of this, what resources do the authors provide for firsttime, non-technical readers? The authors explain one form of uncertainty. This strictly quantitative type of uncertainty has little to do with the deep uncertainties involved in projecting the future consequences of climatic change. The authors use the lexeme “uncertainty” in two distinct denotative senses. The first use of the word “uncertainty” is in the first footnote to page two, a footnote immediately preceding the other footnote discussed above. In this immediately prior footnote, the authors define the measurement error that obtains within a measurement of global sea-level rise. “Numbers in square brackets,” write the authors, “indicate a 90% uncertainty interval around a best estimate, i.e. there is an estimated 5% likelihood that the value could be above the range given in square brackets and 5% likelihood that the value could be below that range” (SPM 2). The authors provided this footnote for the sentence beginning, “Global average sea level has risen since 1961  30  at an average rate of 1.8 [1.3 to 2.3] mm/yr” (2). The quantitative confidence intervals, in square brackets in the sentence, do indeed specify the confidence that a person may have in a claim about the past influence of climatic change. Such a confidence is qualitatively different from any modal adverbial that, even when it does employ the lexeme “confidence.” By the confidence interval, the authors are indicating that there is only a very small likelihood that the actual sealevel rise was outside of the range of numbers they present. In less compact language, the quantitative expression may be unpacked as follows. There is only a five percent likelihood that the actual, historical sea rise was below the lower of the two numbers in square brackets, and only a five percent likelihood that the actual sea rise was above the largest number.21 The authors carefully account for this conventionally quantitative form of uncertainty. Indeed, they were no doubt simply following the conventional technical and vocabulary choices in the field of statistics. Nevertheless, the first footnote to page two is both the first and the only point at which the authors explain their technical and multiple uses of the lexeme “uncertainty” in the SPM. Rather than disambiguate such a strictly quantitative source of uncertainty from the additional types and sources that conditionalize the other findings, the authors leave the two senses undifferentiated. In so doing, they identify confidence intervals, which express a quantitatively-precise and well studied type of uncertainty, with the three sets of other, un-discussed types. By means of this explanatory gap, the authors imply that the uncertainties that have led to the assignment of probabilities among the modal adverbials are easily quantifiable and rigorously understood. Turning now to a catalogue and discussion of the modal adverbials in the SPM,22 the additional rhetorical discourse-functions draw on and amplify  21  Please see, for a more detailed and formal explication of confidence intervals, Ian Hacking’s Logic pp. 159ff. The following discussion focuses only on the two sets of modal adverbials that have been correlated to a quantitative scale. Please see Appendix C for a discussion of the rhetorical discourse-functions of the third set of modal adverbials. 22  31  this initial identification of uncertainty with strict quantification. Equally conditionalizing claims about past and future time periods that are under different types and sources of uncertainty, the authors suggest through their uses of the modal adverbials that the uncertainty under discussion is stable, quantitative, and politically manageable.  3.2.1. Likelihood Adverbials The first modal adverbial that would be encountered by a reader who reads linearly through the document is in a sentence on page two of the SPM. The sentence reads, “Globally, the area affected by drought has likely increased.” This first set of modal adverbials is distinguished from the others by its use of the lexeme “likely.” The authors use a likelihood adverbial thirty-eight more times in the SPM. Twenty-six have the form “likely”; twelve are “very likely”; two are “very unlikely”; one is “more likely than not”; and one is “virtually certain.” Although this last form doesn’t include the lexeme “likely,” it follows a pattern in the text of an epistemic “speaker’s comment” upon the degree of likelihood. While the form of this modal adverbial is regular, the sentential contexts in which it appears are variegated. Nevertheless, these contexts evince a certain pattern of use. A reader of the document who was attempting to understand the uncertainties behind the modal adverbials might pick up on this pattern. The uncertainties that influence the assignment of probability to the findings, however, must be different. The variety of sentential contexts confounds a simple interpretation of the influence of the uncertainties on the authors’ assignments of probability. For example, the authors use identical likelihood adverbials to modalize sentences that express either past- or future-time reference. Sentences like the following express past-time reference: “It is likely that: heat waves have become more frequent over most land areas, the frequency of heavy  32  precipitation events has increased over most areas, and since 1975 the incidence of extreme high sea level has increased worldwide” (SPM 2). The meaning of such sentences is “forensic” in that the authors present a claim, under uncertainty, about the past presence and continuing consequences of climatic change. In the first six pages of the SPM, the authors only employ likelihood adverbials in the context of forensic claims. Beyond the first six pages, the authors use likelihood adverbials to modalize a different type of claim: “predictive.” In such sentences, the authors make a claim about the future course or potential consequences of global warming under uncertainty. The first use of a likelihood adverbial to modify a predictive sentence occurs on page seven of the SPM: “Continued [greenhouse gas] emissions at or above current rates would cause further warming and induce many changes in the global climate system during the 21st century that would very likely be larger than those observed during the 20th century.” While predictive sentences must involve different or additional types of uncertainty than forensic sentences (given that future events involve an irreducible element of unpredictability, for one), the reader is not given any resources for distinguishing among these uncertainties. Forensic and predictive sentences are treated as if they are both conditional upon a measurable, conventionally quantitative type of uncertainty that the lead authors have analyzed and taken into account in their assignment of a likelihood to the finding. This imputation of unanimity among the sources and types of uncertainty extends also to the second set of adverbials that a reader of the SPM would encounter, the confidence adverbials.  33  3.2.2. Confidence Adverbials The second type of modal adverbial provides a “speaker’s comment” on the authors’ confidence in a finding. Formally, this adverbial occurs as noun-phrase with the head-noun “confidence” modified by either of the adjectives “medium” or “high.” The adjective “high” is sometimes intensified by the adverb “very.” The authors first use this terminology on page two in the sentence beginning, “Changes in snow, ice and frozen ground have with high confidence increased the number and size of glacial lakes . . . .” IPCC authors use confidence adverbials twenty-two more times in the SPM. Three instances are “very high confidence”; eight are “high confidence”; four are “medium confidence”; four are the comparative form “higher confidence”; and three others are the word “confidence” fulfilling a variety of nominal functions in their clausal contexts. These nominal functions, however, nevertheless express a “speaker’s comment” on the ideational content of the sentence. Confidence adverbials, like likelihood adverbials, modify both forensic and predictive statements from statistical climatology. The authors present no discussion of how their confidence is related to the types or sources of uncertainty in particular forensic or predictive claims. Moreover, the largest gap in the SPM’s explanatory resources is the distinction between confidence and likelihood. At first glance, the two modal adverbials appear to cover the same type of “speaker’s comment,” an epistemic assessment of the probability that a statement will occur. Likelihood and confidence adverbials appear mutually interchangeable in these cases. If the author has “high confidence” in a claim about a future rise in temperatures, she would assumedly also assess it as “very likely.” The authors provide no explanation as to what extent confidence and likelihood cover the same sources and types of uncertainty. This ambiguity between confidence and likelihood in the SPM, moreover, cuts to the core of the rhetorical  34  discourse-functions of the modal adverbials in the SPM, as well as the influence of Moss and Schneider’s subjectivist-Bayesianist definition of probability upon the IPCC’s schema for the communication of uncertainty.  3.2.3. Distinctions among Adverbials of Confidence and Likelihood In order to explore this ambiguity, I will explore, in this section, the distinctions available to a reader23 between the confidence and likelihood adverbials in the SPM based only on the resources provided to readers. One of these resources is the successive collocation of confidence and likelihood adverbials on particular findings. Within three sentences, that is, the lead authors provide an assessment of both confidence and likelihood. In the first, on page eight, the authors modalize a finding with a likelihood adverbial. In the following clause, conjoined to the first by a semicolon, they explain why they can’t modalize the finding with a likelihood adverbial because their confidence in it is too low. “Regional scale changes include,” reads the sentence, “[a] likely increase in tropical cyclone intensity; less confidence in global decrease of tropical cyclone numbers” (8). This sentence suggests that a likelihood adverbial requires a robust evidentiary basis. Only once a particular level of confidence about such an evidentiary basis has been reached will the authors assign it a likelihood adverbial. Such a “degree-of-probity” reading of the distinction between confidence and likelihood adverbials finds support in a number of other sentences in the SPM. The authors use the confidence adverbial, in these cases, to express a “speaker’s comment” on the probabilistic chance that a statement will occur or has occurred, as in the following example: “There is also high confidence that many semi-arid areas . . . will suffer a decrease in water resources due to 23  Discourse analysis posits the meaning of a text as construed over time as well as within a textual space. Correlating my present discourse analysis with the rationally-reconstructed experience of a reader highlights the temporal features of the rhetorical discourse functions of the modal adverbials.  35  climate change” (8). Likelihood adverbials express this same epistemic modalization of a finding under uncertainty, as in the following example: “Regional scale changes include . . . [a] very likely increase in frequency of hot extremes, heat waves and heavy precipitation” (6). Based on the SPM alone, a careful reader might conclude that both confidence and likelihood adverbials can express the degree of probability that a finding will occur. When the authors reach a certain unspecified level of confidence, they will assign a likelihood adverbial to the finding. Confidence, then, appears to be an assessment of likelihood in a less robust evidentiary context. The consistency of this apparent rationale of use between the two adverbials, however, is vitiated by the next collocation. “There is medium confidence,” the SPM reads, “that approximately 20 to 30% of species assessed so far are likely to be at increased risk of extinction if increases in global average warming exceed 1.5 to 2.5°C (relative to 1980-1999)” (13, 19). In this sentence, the authors modalize the core finding with both a confidence and a likelihood adverbial. While an atypical sentence, it is found twice in the SPM and three times in the Synthesis proper (on pp. 48, 54, and 64). In this deeply-hedged and politically-cogent statement, a likelihood adverbial modalizes a finding that is already modalized by a confidence adverbial. The authors have used the standardized modal adverbials to express a “speaker’s comment” of confidence in their own “speaker’s comment” of an assessed likelihood. Such double modalization detracts from any simple interpretation of the confidence adverbials as an expression of probabilistic chance of occurrence, because, in this often-repeated claim, the authors have already expressed that chance of occurrence through a likelihood adverbial. The doubly-modalized sentence suggests a different explanation of the difference between the two, seemingly similar “speaker’s comments.” Likelihoods, under this new rationale of use, have been calculated independently of the authors, who will sometimes have confidence  36  in those calculations and sometimes will not. The first explanation, confidence as a “degree-ofprobity” in a less robust evidentiary context, is inconsistent with this new interpretation. The “degree-of-probity” interpretation presupposes that the IPCC authors had themselves engaged in an unreported analysis of the probability that a finding will occur. This unreported analysis led them to a conclusion about the probability of a finding’s occurrence. Because of some opaque criteria, perhaps evidentiary, the authors sometimes express this conclusion as a measure of their confidence in the finding, and other times as a measure of likelihood. The doubly-modalized sentence, however, challenges the “degree-of-probity” reading. The doubly-modalized suggests that the IPCC authors have confidence in the validity of the methodology that yielded a probability that they themselves did not calculate. Their confidence, in second, “confidence-inmethods” rationale of use, is in the methods and the evidence that yield an independentlycalculated likelihood. For the sake of clarity in conceptual delimitation, the two readings shouldn’t both be right. The lexeme “confidence” would, in that case, fail to distinguish among different policyrelevant types of uncertainty in different evidentiary contexts. In the “degree-of-probity” interpretation, their confidence is just like a statement of probability: it assesses a chance of occurrence. In “confidence-in-methods” case, their confidence is like a gradation of quality: it assesses how good the methods and analysis were that led to a probability. The “confidence-inmethods” case fits better with the denotative meaning of confidence, “The mental attitude of trusting in or relying on a person or thing; firm trust, reliance, faith” (OED). The “degree-ofprobity” interpretation should, after all, be sufficiently covered by the likelihood adverbials. It turns out, in fact, that both readings are correct. At times, the authors use confidence as a proxy for likelihood. At other times, they use it to assess the underlying methods that led to an  37  assignment of likelihood. In the SPM itself, readers of the report do not have enough information to determine why the IPCC authors used the confidence adverbials in the way that they have.  3.2.4. Summary of Findings A reader finds two distinct sets24 of modal adverbials that assign a probability to a finding. The authors have left large gaps in the explanatory resources they provided for the comprehension of this assignment and the policy-relevant uncertainties that constrain it. The authors treat uncertainty as a simple concept that admits of easy, quantitative assignment among the modal adverbials. Moreover, the authors do not define what they mean by “uncertainty” except in one case: measurement error. The authors carefully define what they mean by uncertainty in measurement error while leaving the other uncertainties untouched, eliding the differences among them. Uncertainty appears to admit of robust quantification among the assignments of probability by means of the confidence and likelihood adverbials. Finally, the authors use the confidence adverbials inconsistently in the SPM. At times, they indicate a “degree-of-probity” speaker’s comment on the probabilistic occurrence of a finding. At other times, they provide a “confidence-in-methods” comment on the validity of the inferential methods that led such a distinct likelihood. The rhetorical features of the modal adverbials thus far identified are an outgrowth of Moss and Schneider’s subjectivist Bayesian definition of probability. The slippage25 between confidence and likelihood is the most evident result. When “probability” is defined as the degree 24  There is a third set of modal adverbials in the SPM that does not purport to assign a probability to a finding, but instead assessed the quality of evidence for a finding and the finding’s support in the literature. Its range of application supports the rhetorical discourse-functions noted above. The range of application of this set is not immediately relevant to the present discussion. For an analysis, please refer to Appendix C. 25 The IAC, and Swart et al. each noted this potentially confusing use of confidence as a proxy for likelihood when, in fact, a more accurate likelihood adverbial was already available for just such an assessment (see IAC final ed. p.5 and Swart et al. p.18).  38  of belief held by an IPCC author that an event will occur, quantified confidence is, for the intents and purposes of the IPCC, coextensive with a robust, frequentist probability. Rather than accounting for or explaining the evidentiary or inferential choices behind the assignments of probability, the modal adverbials act as rhetorical intensifiers and highlighters of scientific authority: they tag as scientific and thereby underscore the persuasive scientific authority of particular findings that name either the evidence for or the consequences of climatic change. The ideational content of the modal adverbials’ epistemic “speaker’s comment” is limited but their functions as rhetorical features of the discourse are salient: to mark particular findings as authoritative by their origin in rigorous and quantitative scientific inference. The modal adverbials associate, with a finding, the authority of an objective scientific study. The syntax and discourse-level functions of the modal adverbials in the Summary for Policymakers heavily background the uncertainty. Taking the syntactical and rhetorical foreground are the modal adverbials. They bound and define what the authors believe should form the basis for sound policy decisions by representing those findings as quantitative and robust. Any reader trying to understand the policy-relevance of the uncertainties would have quickly moved past the SPM to the AR4 Synthesis itself. The resources in the AR4 Synthesis, however, achieve similar rhetorical discourse-functions as the modal adverbials in the SPM. As a sample of this treatment, I will explicate the rationale and explanation of the terms provided for readers on page twentyseven of the report. This explanation amplifies the associations of the modal adverbials with a frequentist, quantitative, and robust objective study.  39  3.3. “Treatment of Uncertainty” in the AR4 Synthesis The IPCC explains their use of the modal adverbials in a brief aside on page twentyseven of the AR4 Synthesis entitled “Treatment of Uncertainty.” This one-page rationalization gives readers a compact description of the schema and explains why the authors adopted the three different sets of modal adverbials. Although the AR4 Synthesis is the more comprehensive of the two documents, little substantive explanation of the schema for the communication of uncertainty is provided. Indeed, the SPM includes many in-text citations of the AR4 Synthesis for the findings that it summarizes. The AR4 Synthesis itself includes citations to the comprehensive, longer report. The AR4 Synthesis is thus the nexus between the front-line Summary for Policymakers and the much longer assessment reports from which the findings were taken. As such a nexus, the AR4 Synthesis plays a central role for readers who are evaluating the IPCC’s own assessment process and trying to determine both the nature and policy-relevance of the uncertainties. The explanation in the AR4 Synthesis suggests that, for lead authors, there were two primary criteria of choice: the type of uncertainty and the degree to which it may be quantified. The inferential bases of the probabilities are neither mentioned nor contextualized. The three terminologies, as framed by the AR4 Synthesis, express a rising scale of specificity, quantifiability, and expertise. The authors begin their explanation by correlating the type and quantifiability criteria with the working-group divisions in the IPCC of the overall assessment. “The nature of data, indicators and analyses,” write the authors, “used in the natural sciences is generally different from that assessing technology development or the social sciences. WG I focuses on the former, WG II on the latter, and WG III covers aspects of both” (27). The authors explain that the choices among the three forms of language arose from “both the nature of the information  40  available and the authors’ expert judgment of the correctness and completeness of scientific understanding” (27). In this introductory statement, the authors cite criteria of choice that are both subjective and objective, both from the subjective discretion of experts and from the objective nature of the data. As the authors of the AR4 Synthesis continue their explanation of the schema, they map this subjective and objective conceptual distinction onto a scale from qualitative to quantitative. The first set of modal adverbials, employed solely by WG III, is framed as strictly qualitative. “Where uncertainty is assessed qualitatively,” the report reads: it is characterized by providing a relative sense of the amount and quality of evidence (that is, information from theory, observations or models indicating whether a belief or proposition is true or valid) and the degree of agreement (that is, the level of concurrence in the literature on a particular finding) (27). The terminology is labeled “self-explanatory” and said to occur in the range “high agreement, much evidence” through “medium agreement, medium evidence; etc.” (27). The AR4’s explanation of WG III’s modal adverbials tags them as relative and qualitative. As the initial point on a rising scale of quantitative and objective assessments, the evidence-agreement adverbials provide the baseline against which the other two modal adverbials will be compared. The authors note that the second set of modal adverbials was used primarily by WG II. “Where uncertainty is assessed more quantitatively using expert judgement of the correctness of the underlying data, models or analysis,” the text reads “the following scale of confidence levels is used to express the assessed chance of a finding being correct” (27). The range of this terminology is said to be from “very high confidence at least 9 out of 10” to “very low confidence  41  less than 1 out of 10” (27). The form of this second set of modal adverbials suggests that when an author has confidence in an outcome, the outcome may be quantified as a measure of chance. Indeed, the slippage between a “degree-of-probity” usage of a confidence adverbial and a “confidence-in-methods” usage is here reproduced in the IPCC’s explanation. If an author expresses her degree of confidence that a finding is “correct” in its underlying data, models, or analysis, then she does so within a scale of chances from “1/10” to “9/10.” Her confidence in the underlying data, models or analysis, yields a chance of occurrence on a scale from 0.1 to 0.9, or 10% to 90%.26 The evidence-agreement adverbials, by contrast, are strictly qualitative and entirely subjective within an author’s “relative sense.” The confidence adverbials, however, although they still originate in an authors’ assessment, come from an expert assessment and express a quantified chances of occurrence. The IPCC identifies “chance of occurrence” and “expert assessment” without explanation. This reinforces the mapping of the subjective-objective binary onto a scale of quantifiability: the confidence adverbials, arising from expert judgment but quantitatively calibrated, are applied by the authors to probabilities of occurrence in the objective world. Their “speaker’s comment” carries a greater objective weight than the subjective evidence-agreement adverbials. The authors reinforce this mapping of the subject-objective distinction onto a rising scale of quantifiability by the third and final set of modal adverbials they explain. These are the likelihood adverbials, the most quantitative of the three sets. They were employed primarily by WG I. WG I assessed the physical scientific basis of climate change, and thus carry the greatest association with an objective scientific study exogenous to the discourse. “When uncertainty in 26  To a reader familiar with the methodologies of statistical inference, this movement in my argument from the decimal 0.1 to the percentage 10% might seem inappropriate. I have extrapolated an assessment of probability from this decimal value by the IPCC’s use of the phrasing “chance of occurrence,” which implies a one-time probability of occurrence. I doubt that the intended non-technical readers of the AR4 Synthesis would make a conceptual distinction between 1/10 chance and 10% probability.  42  specific outcomes is assessed using expert judgment and statistical analysis of a body of evidence (e.g. observation results),” write the authors, “the following likelihood ranges are used to express the assessed probability of occurrence” (AR4 Synthesis 27). Each of the likelihood adverbials is calibrated to a percentage range. The range spans a scale from “virtually certain >99%” to “exceptionally unlikely <1%” with a midpoint of “about as likely as not 33% to 66%” (ibid). Whereas the confidence adverbials express a quantitative “chance of occurrence,” a likelihood adverbial is framed here as originating solely in an objective study of the natural world, or a “statistical analysis of a body of evidence” (ibid). Likelihood adverbials appear to express frequentist or objective probabilities that name the likelihood that a well-understood, well-defined event will occur. None of the three sets of modal adverbials mention or explain their potential origin in subjective criteria of inference, nor do they discuss the guiding and unusual definition of probability that underlies them. By the absence of such explanations, the authors screen any subjective choices from view and instead direct the readers’ attention towards the authority of an objective scientific study. To recap, the authors rationalize the modal adverbials by mapping them onto a rising scale of quantifiability and objectivity. The first set of modal adverbials indicates to a reader that the finding is qualitative, meaning that it provides only a qualitative comment upon the amount and quality of evidence rather than a quantified expert comment. WG III’s modal adverbials, that is, give a “relative sense,” rather than an “expert judgment.” Of the three, this first terminology is the only one that doesn’t mobilize the notion of expertise; the authors locate the other two terminologies specifically in an “expert assessment.” At the top of the scale, WG I’s likelihood adverbials are “assessed probabilities of occurrence” based on “expert judgment and statistical analysis of a body of evidence” (27). In the middle is the nebulous and inconsistently-used  43  confidence adverbial. “More quantitative” than a “relative sense,” but less formal than a statistical analysis, the confidence adverbials appear to allow the authors to comment on a probabilistic chance of occurrence without engaging in a formal analysis. The confidence adverbials, employed problematically by WG II, are the point at which Moss and Schneider’s subjectivist-Bayesian elision of the evidentiary difference between subjective and objective probability is most visible. The PBL and IAC heavily criticized WG II’s use of the confidence adverbials. However, as Oppenheimer et al.’s article shows, the likelihood adverbials can also elide the sometimes-problematic discretionary and private basis of the subjective probabilities. Among the modal adverbials more generally, there is a crucial mismatch between Moss and Schneider’s rationalization of the schema and the rationalization given to readers in the AR4 Synthesis. Specificity, quantifiability, and expertise, in the context of the truncated communication of uncertainties in the SPM, achieve little in the communication of policy-relevant uncertainties. Instead, the explanations in the AR4 Synthesis undergird the scientific authority of the confidence and likelihood adverbials by mapping them onto the hallmarks of an objective scientific study: a quantitative expression with an expert locus. As in the SPM, the meaning of the modal adverbials is determined by the degree to which they are explained and situated within the uncertainties of statistical climatology. Additional resources27 in either document could have corrected the framing and fill in the gaps in the resources provided for readers. A turn towards the IPCC-independent scientific literature provides a very different picture of the policy-relevant uncertainties of statistical climatology. This literature will provide a contrast the IPCC’s modal-adverbial treatment of uncertainty, the IPCC’s assignment of 27  In the AR4 Synthesis, however, the other, non-modal-adverbial textual resources actually amplify the rhetorical functions of the schema so far noted. Please see Appendix D for a sample of such a resource, an analysis of the final section of the AR4 Synthesis entitled “Key Uncertainties, Robust Findings.”  44  probability rationalized by Moss and Schneider’s subjectivist-Bayesianist definition of probability. It will also display the extent and the policy-relevancy of the deep uncertainties involved in climate modeling and projection. After this discussion, I will return to the key question I left unanswered before the lexical and discourse analysis. Moss and Schneider required the inclusion of traceable accounts in the use of their definition of probability. Moreover, all the independent reviewers (the IAC, PBL, and Oppenheimer et al.) of the IPCC agree that even findings under deep uncertainty could be useful to inform a deliberative policy decision. Why would the IPCC authors, intimately aware of the uncertainties in statistical climatology and their policy-relevance, discount Moss and Schneider’s precondition? An answer to this question forms the substance of chapter 5. I will propose that the IPCC’s truth-to-power orientation towards science for policy, their subjectivist-Bayesianist definition of probability, and the rhetorical functions of the modal adverbials, are an inapt response to the difficult political and inferential challenges that the IPCC faces.  45  4. MODELING UNCERTAINTIES IN GLOBAL CLIMATE MODELS 4.1. Overview of Inference from GCMs A brief discussion of the inferential techniques of global climate modeling will clarify the following discussion of the policy-relevance of the uncertainties in statistical climatology. “Global Warming” as a physical process refers to an observed increase in a statistical average called the “Global Annual Mean Surface-Air Temperature Index”28 (GAMSAT). The GAMSAT is the average of the mean-annual-temperature measurements from the different regional climates29 across the planet. It functions for scientists as a useful abstraction, a single metric that reveals a similar change in temperature that might otherwise be lost in the details of local weather patterns (Hulme 8). Climatologists have modeled physical processes that contribute to the GAMSAT, along with other relevant global climate processes, in computational programs called Global Climate Models (GCMs, also called General Circulation Models). The design of the most technically advanced30 GCM, the “HadCM3” (named after the UK Met Office Hadley Centre for Climate Change) combines well-studied regional climate models with the well-studied global physical processes. Through GCMs, scientists attempt to account for the recent observed warming trend in the GAMSAT (as well as the many other observed changes that are attributed to global warming). Indeed, the HadCM3 and other, simpler models allow climatologists to run experiments on a virtual representation of the currently-understood physical climate processes (Viner n. pag.). Such a technically innovative inferential tool permits climatologists figuratively and literally to “scale down” the global climate to a spatially and temporally manageable 28  See Hulme p. 8 ff for an introduction to this metric and the Goddard Institute for Space Studies for a more technical and detailed discussion. 29 A regional climate or “climate type” is a large geographical area that has similar weather patterns and temperatures (Hulme 8). These similarities result in similar kinds of natural vegetation, which form the basis of the most commonly used classification (Koeppen classification) (Hulme 8). 30 In the sense that it incorporates the most model components and yields the most detailed outputs. See the University of East Anglia’s Climatic Research Unit’s website for more detail on GCMs (Viner n. pag.).  46  representation. GCMs, however, have a number of fundamental limitations on how well they can projective and predict the behaviour of the actual global climate. In terms of the IPCC’s modal adverbials and Moss and Schneider’s definition of probability, a relevant question is to what extent subjective choices form a basis for the output of GCMs. Sheila Jasanoff and Brian Wynne, in a review of earth-science modeling techniques, explain that: The selection of end-points for modeling (e.g., climate warming, forest damage or soil erosion) may call for assumptions by scientists. Even when the end-points of interest to policy are clear, scientists exercise judgment about what precise model output variable (or combination of variables) adequately represents these end-points. (“Science and Decisionmaking” 61) GCMs thus require numerous subjective judgments to yield the likelihoods and value-ranges that the IPCC reports. GCMs are a complex form of Hacking’s belief-type inductive inference, a way of extending a particular scientific hypothesis beyond the empirical evidence available to assess how likely an event is to occur. Thus, each quantitatively-calibrated likelihood and confidence adverbial derived from a GCM is conditional upon a set of assumptions that are often themselves unquantifiable. Such assumptions lead to a number of difficult challenges in the projection of future states of the climate.  4.2. GCMs and Inductive Challenges Indeed, statistical climatologists David A. Stainforth, M.R. Allen, E.R. Tredger, and Leonard A. Smith in an article entitled “Confidence, Uncertainty and Decision-Support  47  Relevance in Climate Predictions,” express a number of worries about the use of GCMs31 as predictive tools for effective climate change policies. “Complex climate models,” they write, “. . . cannot be meaningfully calibrated because they are simulating a never before experienced state of the system” (Stainforth et al. 2145). That is, despite scientists’ best efforts, no model can take into account possible physical processes outside the experience or the imagination of the individuals and groups who create the models. This basic inductive problem becomes exacerbated as the complexity of a given system increases. Nevertheless, as a different group of specialists commented, “the raison d’être of numerical modeling is to go beyond the range of available analytical solutions” (Oreskes et al. “Verification” 642). Models, in other words, are neither nothing more nor anything less than very intricate scientific hypotheses (idem 643). Complex computational models are an ambitious attempt to extend inductive inference to physical processes and systems that aren’t yet fully understood in terms of physical and causal relationships (idem 642). However, both scientists and end-users will sometimes treat such models as if they were microcosms of the global climate rather than computational, provisional reconstructions of it (ibid). This problem surfaces in the technical problems of “calibration” in GCM modeling, and has serious consequences for the IPCC’s reportage of highly specific model outputs. “Calibration” refers to a computational “tuning” of the components, the parameters, and the initial values of a complex model in the earth sciences (Oreskes et al. “Verification” 643). Calibration involves changing a GCM so that its output matches the actual, observed results in the real climate (Stainforth et al. 2147). Such a calibration is a desirable and effective feedbackmechanism for improving the forecasting skill of any of the short-term processes studied by 31  They actually refer in their article to the use of large ensemble runs of these models. Unfortunately, I don’t have the space here to do a more comprehensive review of the innovative statistical techniques of Global Climate Models. For the sake of simplicity, this essay will refer to such model ensemble runs informally as inference from GCMs.  48  climatologists (ibid). Scientists will use a GCM to make a short-term forecast and then calibrate the model to match the event when it actually occurs (ibid). The many long-term climate processes modeled in GCMs, however, can’t be confirmed in this way (ibid). Scientists can’t measure the changes in these long-term processes over within the relatively-short scale of time for which the forecast is needed. Scientists must employ, in order to calibrate these model components, the same paleoclimatological data on the basis of which they were initially formed (ibid). Scientists are stuck, in this case, in a closed inferential loop: all the data used to calibrate these long-term processes (thereby producing a better match to the actual global climate) are already “in-sample,” or a part of the model design itself (ibid). Confirmation, for such a model, is no more than a test of its internal consistency (ibid). “In-sample confirmation” entails that there is no way to verify that complex models in the earth sciences are accurate for long-term projections (Oreskes et al. “Verification” 643). The confirmation of computational models, however, is sometimes treated as if it were its verification (ibid). For GCMs and the IPCC, this can have heavy consequences in the sphere of policy action. Stainforth et al. explain: The severity of model inadequacy suggests a more qualitative interpretation than one might wish. In particular, it is not at all clear that weighted combinations of results from today’s complex climate models based on their ability to reproduce a set of observation can provide decision-relevant probabilities. Furthermore, they are liable to be misleading because the conclusions, usually in the form of [probability distribution functions], imply much greater confidence than the underlying assumptions justify; we know our current models are inadequate and we know many of the reasons why they are so. (2158)  49  Stainforth et al. notes the very problem that the rhetorical functions of the modal adverbials promote. A quantitative, highly specific expression can imply a much greater certainty than the underlying evidentiary basis pragmatically justifies. Stainforth et al. underscore that, as scientific hypotheses, GCMs do have a very important role to play in policy advising (ibid). They suggest, however, that because of structural and value uncertainties in the models, a quantitative expression for these model outputs can be misleading (ibid). Moss and Schneider’s subjectivist Bayesianism, however, extended the justifiable range of these quantitative expressions to even less well-defined outcomes of climatic change by redefining probability for the lead authors. In the discourse of the SPM and AR4 Synthesis, such modeling uncertainties are minimally communicated and heavily backgrounded. To some statistical modelers involved “on the ground,” however, the uncertainties are so extreme that they worry about the accuracy of any quantitative expression for policy purposes. The pertinent question at this stage is the degree to which these uncertainties are, in fact, relevant to the IPCC’s project of informing policy decisions. The next group of authors provides the answer.  4.3. Political Relevancy of Modeling Uncertainties A 2010 article, jointly written by Naomi Oreskes and statistical climatologists David A. Stainforth and Leonard A. Smith, discuss the potential for the use of results from GCMs for two different policy choices: adaptation, or mitigation. “[W]hile climate models consistently suggest that the mean global temperature of the planet will rise,” they begin, “mean global temperature is not what any one person, state, or nation will be adapting to” (1013). Setting aside the worrying concerns about the adequacy of the models for skillful, long-term statistical analysis, the very model outputs themselves are not yet precise enough to be useful for adaptation purposes.  50  Oreskes et al. point to recent claims by some groups of politicians that adaptation will be the most cost-effective way of responding to global warming. Such claims, note the authors, are being used to justify withdrawing support for policies that would mitigate greenhouse-gas emissions. The authors comment that, “This implies . . . [that] we have a good handle on the implications of change on the scale at which human decisions are made and actions are taken” (1017). However, for adaptation to effectively counter the detrimental effects of climatic change, a region’s limited economic resources must be invested in the best possible infrastructure improvements. This presupposes that scientists have very specific and reasonably certain information about local climatic changes, in the near future. “Do we?” ask Oreskes et al. (idem 1017). “No,” they contend. Oreskes et al. pinpoint a number of limitations on the present state of GCM modeling that reduce the certainty and specificity of the projections. The most pressing problem is the too-large resolution of the model outputs, the “Scale Gap” problem (idem 1019). The output of the most advanced climate model, the HadCM3, has a geographical resolution of 2.5° latitude by 3.75° longitude (Viner n.pag.). This resolution yields model outputs that cover rectangular areas of the world between (approximately) 100 and 500 kilometers in size (IPCC “Regional Climate Models” n. pag.). Such a resolution means that any research based on the output of the HadCM3 yields information which has a high degree of agreement32 predominantly for continental-scale processes like the global mean temperature index (Oreskes et al. “Adaptation” 1013). This information is inadequate for adaptation-policy changes by local governments. In sub-continental-scale climate processes (on the level of cities or regional districts), they often show a less than 66% agreement for the projected results (idem 1017). This  32  Agreement, that is, among different runs of the same computational model (General Circulation Model) using a wide range of initial conditions and parameters deemed likely by the researchers.  51  means that, for example, in summers in the North American Midwest in the years 2090 to 2099, the models project both more rain and less to varying magnitudes (ibid). In the case of precipitation, “the models do not just disagree about the magnitude of the [local climate] change, they fail to agree on its direction” (ibid). Because the models do not yet provide consistent information on the scale at which local governments may act, they cannot currently indicate, to those local governments who operate at that scale, what the best course of adaption-policy would be. Moreover, Oreskes et al. worry about the practicality of making specific adaptive decisions on the basis of GCM outputs the isomorphism (or the sufficient structural identity to the actual, experienced climate) of which cannot be known before the fact. “Model ensembles,” write Oreskes et al., “that explore a wide range of physically plausible input conditions may help us to capture a range of physically plausible outcomes, but they will not correct systematic bias, error or distortion, or unresolved interactions and feedbacks, or account for missing processes” (1019-20). Despite the valuable work being done in climatology to correct for the known inadequacies in the models, the known gaps, as Stainforth et al. comment above, reduce the reliability of highly specific model outputs. In terms of adaptive policies, Oreskes et al. comment that the models are also unable to account for possible extreme changes in climate (idem). Imaginable extreme events in response to a sudden warming of the global climate include “a major dieback of the Amazon, or a sudden increase in release of stored greenhouse gases from arctic permafrost” (1014). Such events “perhaps unlikely, but potentially grave,” further reduce the value of a highly specified, probabilistic risk assessment (1014). Nevertheless, the current state of climate science has few resources to model such sudden changes, much less assign  52  quantitatively-calibrated likelihood or confidence adverbials to them (Oreskes et al. “Climate Models” 1014). To recap, predictive climatology relies on complex and technically intricate Global Climate Models (GCMs). These permit scientists to run virtual experiments on the global climate that wouldn’t be possible in the real world due to the long time-scales involved in physical climate processes. A number of politically-difficult uncertainties obtain in the political use of information from GCMs. Stainforth et al. discuss a number of these uncertainties that originate in questions of the adequacy of these models in terms of their isomorphism with the actual global climate. For a number of the long-term climate processes involved in these models, climatologists lack appropriate empirical data to confirm the adequacy of the computational representation of the actual, real-world physical processes. Questions of model adequacy therefore complicate the political use of such models for highly specific adaptation-policy advising. Oreskes et al. 2010 argue that GCM outputs, both because of questions of model adequacy and the too-large geographical scale of the information to which the models are currently limited, is insufficiently certain to form the basis of adaptation policies. Although the information from these large-scale GCM projections indicates with great certainty that global warming is happening and must be addressed quickly, the most effective policies that could be chosen on the basis of that information are those that mitigate current greenhouse gas emissions.  4.4. Modeling Uncertainties and Policies of Adaptation GCMs do not exhaust the inferential methodologies of climatological research. The most pertinent question at this stage is to what degree the IPCC relies on the numerical output of  53  GCMs in their formation of their quantitatively-calibrated “speaker’s comments” on confidence and likelihood. Oreskes et al. (2010) comment on the use of GCMs by the IPCC. “The IPCC authors,” they write, “draw on [non-GCM] evidence to refine their projections of regional scale change” (1023). The IPCC’s additional techniques work to “scale down” the GCM outputs to geographical grid-squares of 50kms or less (IPCC “Regional Climate Models”). These scaleddown model outputs, however, continue to reproduce the same inductive and scale problems of the original GCMs (idem 1024). The 100 to 500 kilometer scale of the model output cannot translate with sufficient certainty to the scale of regional action (ibid). “The information we have,” Oreskes et al. remark, “is simply not on the same scale as the information we need” (ibid). Effective adaptive policies on the part of local governments requires a level of specificity in the outputs of GCMs not yet obtainable in the majority of cases. Does this mean, though, that the IPCC must restrict its statements to certainties? Neither Oreskes et al. 2010 nor Stainforth et al. 2007 think so. They both agree that even highly uncertain information garnered from statistical climatology can be valuable to inform the political decision-making process. The IPCC’s modal adverbials, however, attempt to determine the decision-making process by assigning a single, numerical value for likelihood or confidence. The problem is that in some cases, a quantitative expression misrepresents the certainty of the claim, just as the PBL, the IAC, and Oppenheimer et al. notes through their independent research. The IPCC’s mandate, to provide policy-makers with specific scientific findings that would support both mitigation and adaptation, is unavailable within the deep uncertainties of statistical climatology. A better communication of these uncertainties could, nevertheless, reveal which findings are relatively robust; which ones rely to a great degree on subjective judgments; and which ones discount known processes that can’t yet be incorporated into statistical models. Transparency, as all the  54  expert reviewers have noted, would be the key to such effective policy advising. Transparency, however, is precisely what the SPM, the AR4 Synthesis, and the modal adverbials lack. Having now explored the conceptual basis of the schema for the communication of uncertainty, a relevant sample of its rhetorical features in the documents, and its sparse representation of the actual uncertainties of statistical climatology, I return to an as-yet unanswered question. Why did the IPCC discount Moss and Schneider’s precondition of providing traceable accounts of the subjective and objective mix of inductive inference that yielded the modal adverbials?  4.5. Modal Adverbials and Deep Uncertainty It goes without saying that the IPCC is not a politically naïve or ill-intentioned organization. The IPCC reviewers, moreover, are themselves intimately aware of modeling uncertainties. Moss and Schneider went so far as to make their recommendation for the adoption of subjectivist Bayesianism contingent upon the preparation of “traceable accounts” of the decision steps that led to the assignment of probabilities among the modal adverbials (Moss and Schneider 46). These accounts would describe how the estimates were constructed that describes the writing team’s reasons for adopting a particular probability distribution, including important lines of evidence used, standards of evidence applied, approaches to combining/reconciling multiple lines of evidence, explicit explanations of methods for aggregation, and critical uncertainties. (Moss and Schneider 37) Moss and Schneider believed that the inclusion of a traceable account would render subjectivist Bayesianism a politically-neutral methodology in the context of deliberative and open democratic politics. Undoubtedly, a traceable account would have ameliorated the evidentiary  55  problems with the modal adverbials that the PBL, IAC, and Oppenheimer et al. noted. If the lead authors of the SPM and AR4 Synthesis had provided a traceable account of their decision steps, an independent review of the evidentiary basis of the findings would have been unnecessary. Rather than burying the actual decision steps that they made, the lead authors could have provided an argument for why the outcome they tagged with a modal adverbial had been deemed “likely” by them. The IPCC’s use of modal adverbials, however, presupposes that a quantitative, institutionally-licensed assignment of probability is the ideal method for researchers to contribute to a deliberative process. Traceable accounts lack cogency and necessity in the “speaking-truthto-power” model of the role of science in policy decisions. Scientists are imagined, in such a morel, as consultants to, but not participants in, a deliberative process: traceable accounts, in such a footing, could only distract from the univocality of the probability assignment. Robust and adaptation-policy-relevant probabilities, as the IPCC’s work has shown, can be difficult to obtain under the deep uncertainties of GCMs. Moss and Schneider’s remarks and the IPCC’s “speakingtruth-to-power” paradigm interact poorly with the inferential context of statistical climatology. In light of the IPCC’s and Moss and Schneider’s knowledge of the many uncertainties in statistical climatology, the IPCC’s modal adverbials can best accounted for as a response to the social and political context that precipitated them. The key to these social and historical contexts, moreover, can be elicited from the very rhetorical functions of the modal adverbials in the discourse of the SPM and AR4 Synthesis.  56  5. THE IPCC’S MODAL ADVERBIALS IN SOCIO-HISTORICAL CONTEXT 5.1. Private Discretion and Trust in Contemporary Democracies Beginning at the syntax-level and extending through the discourse, the modal adverbials express three primary rhetorical functions in the discourse of the SPM and AR4 Synthesis. They render the subjective choices among difficult structural and value uncertainties in GCMs quantitative by mapping them onto a scale of chance or probability; they render them objective by eliding the differences between the inferential bases of the findings; and they render them standard across all the many structural and value uncertainties of GCMs. Moss and Schneider’s subjectivist Bayesianist definition of probability allowed IPCC lead authors to “stand apart from politics while proffering impartial knowledge to formal policy institutions,” as Jasanoff and Wynne write about the “speaking-truth-to-power” orientation towards science for policy (Jasanoff and Wynne 7-8). By pushing the private discretionary decisions among the assignment of probability to particular outcomes of climatic change into the background, the IPCC could purify the bias from a set of statements around which, perhaps, substantive policy action could form. The IPCC was trying to manage the political use of climate change by their careful representation of the deep uncertainties within it. Quantification, objectification, and standardization, moreover, are three closely related and value-laden cultural practices. Cultural historian Theodore Porter has argued that such practices have been mobilized by organizations to garner a cultural authority of a mechanical and aperspectival objectivity as a response to difficult problems of trust and assurance in the social context of a dispersed and heterogeneous democratic polity. This is the very contemporary context of the IPCC assessments. The challenges of “personal and public knowledge, of trust and suspicion” that arise in an American57  style democracy (Porter 200) are compounded by the complex inferential context of statistical climatology.  5.2. Organizational History of the IPCC The IPCC was born into conflict. Naomi Oreskes and Erik Conway’s 2010 Merchants of Doubt provides a historical commentary on the politico-scientific conflicts surrounding the IPCC’s inception and early history. Their research underscores the unavoidable interpenetration of scientific and political concerns in the early history of the IPCC. The IPCC and its members have, since the organization’s formation in 1988, fought many public battles with a number of politically-powerful scientists who opposed the political management of climatic change.33 Bill Nierenberg, one of the scientists that Oreskes and Conway identify as an ideologically-motivated opponent of climate change management, headed a 1983 National Academy of Sciences review of the science and economics of climate change (idem 177). While the final synthesis report of that review recognized the strong physical evidence of global climate change in the near future— change that could seriously damage present human and natural systems (idem 177 ff.)—it discounted all the potential political solutions as economically infeasible (idem 183). Specifically, the review concluded that “the myriad of individual incremental problems [from climatic change] take their place among the other stresses to which nations and individuals adapt” (Nierenberg qtd. in Oreskes and Conway 180). Recognizing the potential threat to human  33  Naomi Oreskes and Erik Conway accuse, in their chapter “The Denial of Global Warming,” Fred Seitz, Robert Jastrow, Bill Nierenberg, and Fred Singer of deliberately misleading publics and policymakers on the available science of global warming (213-4). In certain cases, the Oreskes and Conway bring forward strong support such as blatant misrepresentations of evidence by the Marshall institute (189). At other times, Oreskes and Conway argue that the four men should have known better and were obliged to act differently given their scientific training (190). Oreskes’ and Conway’s book is an epideictic genre of history, openly concerned with the appropriate attribution of praise and blame within its historical narratives.  58  interests from accumulating carbon dioxide in the atmosphere, Nierenberg’s authoritative report nevertheless insisted that further “research, not policy action was necessary” (idem 181). Nierenberg, and the other opponents of the political management of climate change in Oreskes’ and Conway’s history, continued to produce institutional challenges to the political management of greenhouse gas emissions. The year before the IPCC’s 1991 First Assessment Report, a conservative think-tank called the Marshall Institute published a small book entitled Global Warming: What Does the Science Tell Us? (Oreskes and Conway 186). In the book, the Marshall Institute misrepresented certain graphs and data from an influential early Global Climate Model study. They used these subtle misrepresentations to support their argument that all the global warming thus far observed could be explained by a small, natural increase in the intensity of solar radiation (Oreskes and Conway 187). Stephen Schneider himself, in fact, addressed this argument by pointing out the Marshall Institute’s inconsistent interpretations of all the evidence.34 Climatologists within and contributing to the IPCC were forced, early on in the organization’s history, into a defensive posture in the American public and political sphere. The attacks from institutionally-grounded opponents employed the deep uncertainties of the science to argue against policy action. Such attacks continued throughout the 1990s. By the early 2000s, Oreskes and Conway argue, “Scientifically, global warming was an established fact. Politically, global warming was dead” (215). Moss and Schneider’s recommendations, published in 2000, came directly out of this politically-embattled decade. Theodore Porter takes as his case studies very similar, politically embattled organizations. He identified the historical contingencies that gave rise to the  34  Schneider argued that if a small change in the sun’s warming contribution to the global climate system was enough to account for all the observed warming, then on what grounds could one discount the massive change in the warming contribution to the global warming system by carbon dioxide (Oreskes and Conway 189)? The Marshall Institute’s position was manifestly inconsistent with the data and theories available for interpretation (ibid).  59  contemporary “reliance on numbers and quantitative manipulation” as both self-sufficient and fully legitimate warrants for scientific and social assent (Trust ix). Porter argues that practices of quantification have functioned historically as a “social technology” that facilitates the translation of locally-produced knowledge “beyond the boundaries of locality and community” (ix). Such practices have done so in the past, he shows, foremost as an objectifying strategy. Quantification and standardized decision-rules allowed politically-weak organizations to substitute aperspectival, mechanical rules for any biased individual or group as the source of particular knowledge-claims (196). In so doing, these organizations drew on the cultural authority of Daston and Galison’s mechanical, aperspectival35 authority (Porter 85). Porter argues that such strategies of quantification are precipitated by a democratic political “economy of personal and public knowledge, of trust and suspicion” (200). The social and political conditions within which the IPCC adopted subjectivist Bayesianism and its modal adverbial presentation were, as Oreskes and Conway show, mired in just such issues of public trust in science and suspicion of individual biases. Porter’s research into the organizational adoption of strict quantification practices is worth reviewing as a historical analogue to the IPCC’s current representational practices and political dilemmas.  5.3. History of Quantification Practices Porter grounds his study in two comparative case studies: between nineteenth century actuaries in Britain and accountants in twentieth-century America; and between the nineteenth century Corps des Ponts-et-Chaussées in France and the twentieth century American Army Corps of Engineers. In both case studies, Porter found that the nineteenth-century European  35  Please see footnote six for my use of Daston and Galison’s mechanical and aperspectival objectivity in relation to Porter’s monograph.  60  organizations had little use for practices of quantification when they were called upon to justify the authority of their organizational recommendations. In Britain, the Institute of Actuaries drew on the Victorian sense of moral order, still sufficiently hierarchical to legitimate the “gentlemanly” discretionary expertise of actuaries (101). In France, the Corps des Ponts was strengthened politically by the class-like authority they maintained as a self-styled technocratic elite (Porter 115). Both European 19th-century organizations were politically stable because they could draw upon the vestiges of a class-based, social authority. Trust and assurance, for these 19th century European organizations, were available cultural resources for reasons exogenous to their assessment practices. Very similar twentieth-century American organizations, on the other hand, faced a political environment inimical to public decisions made solely on the basis of private knowledge (148). American accountants began to standardize their decision-rules in the 1930s in the face of threats by the newly-minted Securities and Exchange Commission (SEC) who planned to impose their own set of practices (94); a powerful governmental agency had forced their hand. Similarly, the American Army Corps of Engineers initially had made public-works decisions on the basis of informal cost-benefit analyses, just like the French Corps des Ponts. As powerful utilities and competing government agencies began to challenge the Army Corps decisions on the basis of independently-produced costs-benefit analyses, the Army Corps began to standardize their decision rules more and more (149). They replaced their expert discretionary practice by a show of reproducible and standardized rules (98). Similar political forces, well-defined in Oreskes’ and Conway’s book, precipitated the IPCC’s adoption of subjectivist Bayesianism and modal adverbials.  61  Porter extends his historical case-studies to scientific disciplines, finding that “the push for rigor in the disciplines derived in part from the same distrust of unarticulated expert knowledge and the same suspicion of arbitrariness and discretion that shaped political culture so profoundly in the same period” (Porter 196). Inasmuch as “rigorous quantification is demanded in these contexts because subjective discretion has become suspect,” quantification provides the appearance and, more importantly, the authority of aperspectival objectivity (90). More interestingly, quantification practices in the sciences gained prominence in the same sociopolitical context of a weak community challenged by powerful outsiders—a similar context as the one into which the IPCC found itself at its inception. The scientific disciplines that pioneered techniques of statistical inference were applied sub-fields like parapsychology and therapeutics (211). Because parapsychology and therapeutics, the newest subfields of the already politicallyweak disciplines of psychology and medical research, were not yet institutionally established, they required methods that publicly displayed objectivity to the greatest extent possible. “The ideal,” Porter writes of such quantification practices, “is a withdrawal of human agency, to avoid the responsibility created by active intervention” (196). By shifting the apparent grounds of their claims away from human agency, politically-weak scientific disciplines are able to make their claims as “innocent as nature itself” (196). Objectivity is persuasive among a geographically dispersed and culturally heterogeneous polity because it indexes an absence of vitiating, individual biases. Quantification practices like subjectivist Bayesianism can assign probabilities to events on the basis of opaque rationales, the transparency of which depends upon their contextual presentation. Among their modal adverbials, the IPCC makes no distinctions between belief-type and frequency-type probability. The distinctions among the modal adverbials, in the SPM and AR4 Synthesis, are the degree of  62  quantifiability, specificity, and thus, implicitly, objectivity in Daston’s aperspectival and mechanical sense. Moreover, the rhetorical discourse-functions of these modal adverbials actively elide the difference between subjective and objective probabilities and thereby allow the IPCC’s modal adverbials to draw on the persuasive authority of aperspectival and mechanical objectivity. The organizations in Porter’s study employed quantification practices to manage problems of trust and authority among an American democratic polity inimical to private expertise. So too have the IPCC employed a representational practice that quantifies subjective confidence and discretionary likelihoods as simple modal adverbials to manage a difficult and embattled political context of distance and distrust. The modal adverbials render the subjective assessments of the IPCC reviewers, as Porter might say, almost as innocent as nature itself.  5.4. Conclusions and Directions for Further Study Porter’s frame of a social technology, moreover, suggests a further direction of enquiry for scholars in Science and Technology Studies. It provides a methodological distinction with which to separate out the tangled threads of socio-political context, scientific inference, and rhetorical function that together can account for the IPCC’s adoption of the modal adverbials. This methodology suggests a valuable direction for further research into the IPCC’s schema for the communication of uncertainty. Porter’s use of the term “social technology” cites the seminal, interdisciplinary research of sociologist Steven Shapin and historian Simon Schaffer. Shapin and Schaffer’s pioneering work in their 1985 Leviathan and the Air-Pump argues that the rhetorical discourse-functions of a representational practice form just one component of a scientific institution’s larger set of practices that, embedding and reinforcing each other, separate matters of concern from matters of fact. 63  The production of a matter of fact, Shapin and Schaffer argue, involves a tripartite material, social, and literary set of interconnected practices. These three sets of practices function to scrub away the labour of a fact’s construction; separate it from the locality of its production; and inscribe it by a set of rhetorical discourse-functions that facilitate its transition into other laboratories. Together, these three technologies translate a locally stable, laboratory phenomenon into a universally-stable scientific fact. The three technologies erect a discursive wall around what constitutes a matter of fact and what constitutes a questionable matter for wider interpretation. In this framework, the IPCC’s modal adverbials are a literary technology. They achieve what Porter has called the social technology of objectification through quantification, one component of Shapin and Schaffer’s tripartite technology. As a literary technology, the modal adverbials objectify the subjective judgments that are their basis. They embed a social technology of subjectivist Bayesianism, a speaking-truth-to-power rationale, and a set of methods for the objectified interpretation and communication of the deep uncertainties of GCMs. GCMs are themselves a material technology, a complex, new inferential tool that works to represent the behaviour of the global climate through sets of computational functions. Together, the IPCC’s triple technology works to enact an ambitious new boundary between matters of fact and matters of concern. This new and politically unstable border encircles the deep uncertainties of statistical climatology, old uncertainties that are as philosophically complex as they are politically unmanageable.  64  WORKS CITED Brinton, Laurel J. The Structure of Modern English. Philadelphia: John Benjamins Publishing, 2000. Print. Bush, Vavenar. Science, the Endless Frontier: a Report to the President on a Program for Postwar Scientific Research. Washington, D.C.: United States Office of Scientific Research and Development, 1960. University of Michigan Library, n.d. Web. 16 July 2011. <http:// hdl.handle.net.ezproxy.library.ubc.ca/2027/heb.01131.0001.001>. Chomsky, Noam. The Minimalist Program. Cambridge, M.A.: Massachusetts Institute of Technology Press, 1995. Print. ---. Syntactic Structures. The Hague: Mouton de Gruyter, 1957. Print. “Confidence.” Def. 1. The Oxford English Dictionary. 2nd ed. 1989. Web. Daston, Lorraine. “Objectivity and the Escape from Perspective.” The Science Studies Reader. Ed. Mario Biagioli. 1992. Print. Daston, Lorraine, and Peter Galison. “The Image of Objectivity.” Representations 40 (1992): 81-128. Print. Goddard Institute for Space Studies. “GISS Surface Temperature Analysis (GISTEMP).” National Aeronautics and Space Administration. 28 Jan. 2011. Web. 19 July 2011. <http:// data.giss.nasa.gov/gistemp/>. Goffman, Erving. “Footing.” Semiotica 25.1 (1979): 1-30. Print. Goldenberg, Susan. “Cancún Agreement Rescues UN Credibility But Falls Short of Saving Planet.” Guardian 12 December 2010. Web. 21 July 2011. <http://www.guardian.co.uk/ environment/2010/dec/12/cancun-agreement-rescues-un-credibility>. Hacking, Ian. An Introduction to Probability and Inductive Logic. Cambridge, U.K.; New York:  65  Cambridge University Press, 2001. Print. ---. Logic of Statistical Inference. Cambridge: Cambridge University Press, 1965. Print. ---. The Emergence of Probability. 2nd ed. Cambridge: Cambridge University Press, 2006. Print. Halliday, M. A. K. “Functional Diversity in Language as Seen from a Consideration of Modality and Mood in English.” Foundations of Language 6.3: 322-61. Print. Haraway, Donna. “Situated Knowledges: The Science Question in Feminism and the Privilege of Partial Perspective.” Feminist Studies 14.3 (1988): 575-99. Print. Hulme, Mike. Why We Disagree About Climate Change. Cambridge: Cambridge UP, 2009. Print. InterAcademy Council (IAC). “Chapter 3: IPCC’s Evaluation of Evidence and Treatment of Uncertainty.” Climate Change Assessments, Review of the Processes and Procedures of the IPCC. Prepublication ed. 27-38. 5 Nov. 2010. Web. 27 July 2010. <http://www.ipcc.ch/pdf/IAC_report/ IAC%20Report.pdf>. PDF file. ---. “Chapter 3: IPCC’s Evaluation of Evidence and Treatment of Uncertainty.” Climate Change Assessments, Review of the Processes & Procedures of the IPCC. Final ed. 27-41. 1 May 2011. Web. 30 Aug. 2010. <http://reviewipcc.interacademycouncil.net/report.html>. PDF file. Intergovernmental Panel on Climate Change. “Annexes: (1)Glossary, (2)Authors, (3)Reviewers, (4)Acronyms.” Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Ed. S. Solomon, D. Qin, M. Manning, Z. Chen, M. Marquis, K. B. Averyt, M. Tignor and H. L. Miller. Cambridge: Cambridge UP, 2007. Print, web. 18 Jan. 2011. <http:// www.ipcc.ch/pdf/assessment-report/ar4/wg1/ar4-wg1-annexes.pdf >. PDF file. ---. “The Applicability of Regional Climate Models at the Scale of Small Island States: Statement by IPCC Working Group I, June 2000.” Intergovernmental Panel on Climate Change. World  66  Meteorological Organization and United Nations Environment Programme, June 2000. Web. 3 May 2011. <http://www.ipcc.ch/graphics/speeches/statement-ipcc-june-2000.pdf>. PDF file. ---. “History Brochure.” Intergovernmental Panel on Climate Change. World Meteorological Organization and United Nations Environment Programme, Dec. 2004. Web. 10 Oct. 2009. <http://www.ipcc.ch/pdf/10th-anniversary/anniversary-brochure.pdf>. PDF file. ---. “Organization.” Intergovernmental Panel on Climate Change. World Meteorological Organization and United Nations Environment Programme, n.d. Web. 7 Dec. 2009. <http://www.ipcc.ch/ organization/organization.htm>. ---. “Publications and Data.” Intergovernmental Panel on Climate Change. World Meteorological Organization and United Nations Environment Programme, n.d. Web. 19 July 2011. <http://ipcc.ch/publications_and_data/publications_and_data.shtml> ---. “Summary for Policymakers.” Climate Change 2007: Synthesis Report. Contribution of Working Groups I, II and III to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Ed. Bernstein et al. (Core Writing Team), R. K. Pachauri, and A. Reisinger. Geneva, Switzerland: IPCC, 2007. 1-22. Print, web. 1 Dec. 2010. < http://www.ipcc.ch/pdf/assessmentreport/ar4/syr/ar4_syr_spm.pdf>. PDF file. ---. “Summary for Policymakers.” Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Ed. S. Solomon, D. Qin, M. Manning, Z. Chen, M. Marquis, K. B. Averyt, M. Tignor and H. L. Miller. Cambridge: Cambridge UP, 2007. Print, web. 18 Jan. 2011. < http://www.ipcc.ch/pdf/ assessment-report/ar4/wg1/ar4-wg1-spm.pdf>. PDF file. ---. “Synthesis Report.” Climate Change 2007: Synthesis Report. Contribution of Working Groups I, II and III to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Ed.  67  Bernstein et al. (Core Writing Team), R. K. Pachauri, and A. Reisinger. Geneva, Switzerland: IPCC, 2007. 23-74. Print, web. 19 Sept. 2009. <http://www.ipcc.ch/pdf/assessmentreport/ar4/syr/ar4_syr.pdf>. PDF file. Jasanoff, Sheila. “A New Climate for Society.” Theory, Culture, & Society 27: 233-53. Print. Jasanoff, Sheila and Brian Wynne. “Scientific Knowledge and Decision Making.” Human Choice & Climate Change. 4 vols. Ed. S. Rayner and E. Malone. Columbus: Battelle Press, 1998. Print. 1: 1-112. Kelly, Mick. “2: The Causes of Climatic Change.” Climatic Research Unit Information Sheets. University of East Anglia. Nov. 2000. Web. 13 Jan. 2011. <http://www.cru.uea.ac.uk/cru/info/ causecc/>. MacKenzie, Donald. Inventing Accuracy. Cambridge, M.A.: MIT Press, 1990. Print. Moss, R.H. and Schneider, S.H. “Uncertainties in the IPCC TAR: Recommendations to Lead Authors for More Consistent Assessment and Reporting.” In Guidance Papers on the Cross Cutting Issues of the Third Assessment Report of the IPCC. Ed. R. Pachauri, T. Taniguchi, and K. Tanaka. Geneva: World Meteorological Organization, 2000. Print, web. 16 Sept. 2009. <http:// stephenschneider.stanford.edu/Publications/PDF_Papers/UncertaintiesGuidanceFinal2.pdf>. PDF file. Netherlands Environmental Assessment Agency (PBL). Assessing an IPCC Assessment: An Analysis of Statements on Projected Regional Impacts in the 2007 Report. Netherlands Environmental Assessment Agency. 5 July 2010. Web. 1 May 2011. <http://www.pbl.nl/ sites/default/files/cms/publicaties/500216002.pdf>. PDF file. Oppenheimer, Michael, Brain C. O’Neill, Mort Webster, and Shardul Agrawala. “The Limits of Consensus.” Science Magazine 14 Sept. 2007: 1505-6. Print.  68  Oreskes, Naomi. “Science and Public Policy: What’s Proof Got To Do With It?” Environmental Science & Policy 7.5 (2004): 369-83. Print. ---. “Beyond the Ivory Tower: The Scientific Consensus on Climate Change.” Science Magazine 3 Dec. 2004: 1686. Print. Oreskes, Naomi, David A. Stainforth, and Leonard A. Smith. “Adaptation to Global Warming: Do Climate Models Tell Us What We Need to Know?” Philosophy of Science 77.5 (2010): 1012-28. Print. Oreskes, Naomi and Erik M. Conway. Merchants of Doubt. New York: Bloomsbury, 2010. Print. Orekes, Naomi, Kristin Shrader-Frechette, and Kenneth Belitz. “Verification, Validation, and Confirmation of Numerical Models in the Earth Sciences.” Science Magazine 4 Feb. 1994: 641-646. Print. Palmer, Frank. “Modality in English: Theoretical, Descriptive and Typological Issues.” Modality in Contemporary English. Ed. Roberta Facchinetti, Manfred Krug, and Frank Palmer. New York: Mouton de Gruyter, 2003. 1-17. Print. Topics in English Linguistics 44. Petersen, Arthur C. Simulating Nature: a Philosophical Study of Computer-Model Uncertainties and Their Role in Climate Science and Policy Advice. Antwerp: Het Spinhuis, 2006. Print. Popper, Karl R. The Logic of Scientific Discovery. New York: Basic Books, 1959. Print. Porter, Theodore M. Trust in Numbers: The Pursuit of Objectivity in Science and Public Life. Princeton, N.J.: Princeton University Press, 1995. Print. Price, Don K. The Scientific Estate. Cambridge, M.A.: Harvard UP, 1965. Print. Schiermeier, Quirin. “IPCC Flooded by Criticism.” Nature 4 Feb. 2010: 596-7. Print. ---. “Major Reform for Climate Body.” Nature 19 May 2011: 261. Print.  69  Schneider, Stephen H. “Can We Estimate the Likelihood of Climatic Changes at 2100?” Climatic Change 52 (2002): 441–451. Print. ---. “Testimony of Dr. Stephen Schneider.” Hearing on the Case for Climate Change Action.” U.S. Senate Committee on Commerce, Science and Transportation. 1 Oct. 2003. Hosted by Stanford University. Transcript. Web. 1 Mar. 2010. <http://stephenschneider.stanford.edu/ Publications/PDF_Papers/SchneiderTestimony2003.pdf>. Shapin, Steven. The Scientific Life. Chicago: University of Chicago Press, 2008. Print. Shapin, Steven and Simon Shaffer. Leviathan and the Air-Pump: Hobbes, Boyle, and the Experimental Life. 1989. Print. Stainforth, D. A., M. R. Allen, E. R. Tredger, and L. A. Smith. “Confidence, Uncertainty and Decision-Support Relevance in Climate Predictions.” Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences 365.1857 (2007): 2145-2161. Print. Stainforth, David A., Thomas E. Downing, Richard Washington, Ana Lopez, and Mark New. “Issues in the Interpretation of Climate Model Ensembles to Inform Decisions.” Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences 365.1857 (2007): 2163-2177. Print. Swart, Rob, Lenny Bernstein, Minh Ha-Duong, and Arthur Petersen. “Agreeing to Disagree: Uncertainty Management in Assessing Climate Change, Impacts and Responses by the IPCC.” Climatic Change 92 (2009): 1-29. Print. Tollefson, Jeff. “Climate Panel Must Adapt to Survive.” Nature 2 Sept. 2010: 14. Print. ---. “An Erosion of Trust?” Nature 1 July 2010: 24-6. Print. Vidal, John, Allegra Stratton, and Suzanne Goldenberg. “Low Targets, Goals Dropped: Copenhagen  70  Ends in Failure.” Guardian 19 Dec. 2009. Web. 21 July 2011. <http://www.guardian.co.uk/ environment/2009/dec/18/copenhagen-deal>. Viner, David. “8: Modelling Climate Change.” Climatic Research Unit. School of Environmental Sciences. University of East Anglia. June 2000. Web. 13 Jan. 2011. <http:// www.cru.uea.ac.uk/cru/info/modelcc/>.  71  APPENDICES Appendix A: Organizational and Hierarchical Structure of the IPCC The IPCC is a geographically dispersed and hierarchically complex organization. It coordinates the work of hundreds of expert volunteers who contribute to comprehensive assessment reports that have been published, to date, every four to seven years. Their administrative oversight structure consists of a thirty-one-person appointed bureau, elected in and overseen by a plenary session of representatives from U.N. member nations. The bureau operates in conjunction with an eleven-member secretariat. The bureau and secretariat supervise the coordinating lead authors (CLAs) of each section of the distinct working group (WG) reports. Additionally, a Technical Support Unit (TSU) is assigned to assist each WG. The fourth and most recent assessment report (AR4) consists of the individual reports of each WG that were summarized and collated by the authors of the AR4 Synthesis. Each of the three WGs evaluates a separate feature of the climate change problem. WG I assessed the physical-scientific basis of global warming and its contribution consists of eleven chapters; WG II assessed the vulnerabilities of social and economic systems to climatic change and its contribution consists of twenty chapters; and WG III assessed the economic potential for mitigating climatic change or adapting to unavoidable changes. WG III’s contribution consists of thirteen chapters. Between twelve and fifteen people wrote the content of each of these chapters.36 They were overseen, in turn, by two or three coordinating lead authors. Each chapter had a further two or three review editors. These individuals were nominated by governments and organizations on the IPCC’s request and selected by the bureau of the relevant Working Group. In total, six hundred and nineteen different authors contributed to the complete AR4, excluding the bureau, 36  The numbers I report here are generalizations based on the AR4 WG I’s appendix and may be taken as representative of the other working groups’ chapters.  72  secretariat, and hundreds of government-sponsored reviewers whose thousands of comments were taken into account in the production of the final copy. The final Fourth Assessment Report consists of over three thousand pages of text. These thousands of pages were summarized in by AR4 Synthesis in only thirty-nine.37 In the Summary for Policymakers, the AR4 Synthesis was itself summarized in a mere twenty-one pages. The IPCC’s complex organizational and assessment structure is an attempt to manage both the complexities surrounding the detection, attribution, and prediction of climate change, and the need for communicating those complexities in a politically-embattled policy context.  37  Page count excludes the title pages and front matter.  73  Appendix B: Generative Grammar and the Syntax of the Modal Adverbials The rhetorical discourse-functions of the modal adverbials are expressed at the syntax level as well as the discourse level. For example, one modalized sentence in the Summary for Policymakers reads, “The MOC [meridional overturning circulation of the Atlantic Ocean] is very unlikely to undergo a large abrupt transition during the 21st century” (14). In this example, the modal adverbial is an adverbial phrase (AdvP) with the head-adverb “unlikely.” The authors modalize their claim about the MOC by positioning the modal adverbial after the copular verb “to be,” pre-modifying the subject-complement that begins with “to undergo.” The modal adverbial comments upon the epistemic likelihood of the predicate, itself a comment upon the topic of the “MOC.” The AdvP, in this case, functions as a disjunct modal adverbial (or a “sentence adverb,” in terms of traditional grammar). The AdvP adverbially modifies the content of the infinitival clause by assessing the likelihood of its occurrence. In so doing, it modalizes the sentence. Syntactically, the modal adverbial is optional. To remove it would not render the sentence ungrammatical, as the following paraphrase shows: “The MOC is to undergo a large abrupt transition during the 21st century.” The modal adverbial comments upon the likelihood that the verbal process, located in the predicate of the sentence, will or will not occur. Other modal adverbials can be NPs, as in the following example: “There is high confidence that some hydrological systems have also been affected” (SPM 2). The syntax of this sentence is more complex because of what descriptive linguists call a “there-insertion transformation” at the subject position. A brief explanation of generative grammar will be helpful to clarify the rhetorical functions that this transformation achieves. In order to account for the unlimited variations of grammatical sentence constructions, Noam Chomsky proposed, in the mid-twentieth century, a research program of “generative grammar” (see Syntactic  74  Structures). Generative grammar distinguishes between two sets of finite rules that, together, yield the unlimited variation in possible sentences. Although since abandoned by Chomskyean linguists in favour of a different set of basic formal distinctions,38 the early theory of generative grammar and some of its subsequent changes have been adapted by linguist Laurel Brinton (see p.163) into a theory that suits empirical research into the formal and functional features of the English language. It is Brinton’s adaptation of generative grammar that I here employ to explicate the complex forms and functions of the modal adverbials. The two sets of finite rules are phrase structure rules and transformation rules (Brinton 163). Phrase structure rules are constitutive of a comprehensible, base meaning in a sentence. Phrase structure rules yield a basic root form of the meaning in a sentence. This root form, or deep structure (D-structure), is always simple, active, positive, and declarative. D-structure is the common, base semantic meaning among a related set of grammatical and lexical elements (Brinton 163) that can, by the application of the second set of rules, yield the many possible variations in surface structure (Sstructure) (see Chomsky’s Syntactic Structures for a formal explication of his basic distinction). Transformation rules yield an S-structure. They “rearrange, add, and delete elements” in the D-structure (Brinton 164). Different transformational rules can yield different sentencesemantics and, thus, different rhetorical discourse-functions through the many possible options for structuring the core information. In this case, the “there-insertion transformation” substitutes the dummy-pronoun “there” to empty the subject position of definite, animate reference. Such a nominal element would otherwise have fulfilled the thematic role of a grammatical “experiencer” in the sentence, a person or group who experiences “high confidence” in the scientific finding. In terms of the sentence semantics, this transformation-rule structures the information in such a way  38  See Chomsky’s introduction to his 1995 The Minimalist Program (1-11).  75  as to produce a “focusing” or “topicalizing” effect on the modal adverbial itself (Brinton 298).39 The modal adverbial, not the grammatical experiencer of the high confidence, becomes the foregrounded topic of the sentence. As a rhetorical discourse-function, the “there-insertion transformation” that yields the S-structure detaches the “high confidence” from any subjective experiencer, syntactically “objectifying” it by freeing it from the grounding locus of any subjective person or group. The preceding two examples provide a sample of the formal features and sentence semantics of the modal adverbials. At such a fine-grain level of linguistic detail, however, the evidentiary problems noted by the PBL, IAC, and Oppenheimer et al. are entirely invisible. Nevertheless, the way the authors have structured the information in the NP sample above expresses a notable persuasive function. The authors topicalize the modal adverbial by backgrounding the grammatical experiencer of the “high confidence.” To treat this sentencelevel choice as simply a typical feature of the scientific-reporting genre would discount its role as a rhetorical feature in the IPCC’s non-technical document, written not for scientists but a general readership. This finding, so structured syntactically, is a salient point of interface between the readers of the report and the underlying scientific literature.  39  Brinton’s approach, or Chomsky’s or Halliday’s, is one of a group of possible approaches to the English language. I use it here not to argue for its universal validity but to mobilize its efficacy in this context: sorting through the sentence-level and discourse-level semantics of the IPCC’s modal adverbials. Brinton’s grammar is an Englishspecific adaptation of generative grammar that pares the theory down to a sufficient level to account for the variations in contemporary English.  76  Appendix C: Evidence-Agreement Adverbials The authors use one final set of modal adverbials in the SPM. These adverbials would be encountered last by a reader of the SPM. They indicate the level of agreement on a statement and the amount of evidence that supports it. The first use of this agreement-evidence adverbial occurs on page seven of the SPM in the following sentence: “There is high agreement and much evidence that with current climate change mitigation policies and related sustainable development practices, global GHG emissions will continue to grow over the next few decades” (IPCC “Summary”). Only seven instances of this adverbial occur in the SPM. All occur as a conjunction of two phrases, the first assessing agreement among the authors on a finding and the second the quality of evidence for a finding. The form of the first phrasal constituent is in every case high agreement. The forms of the second phrasal constituent include five instances of much evidence, and two instances of medium evidence. These adverbials in all cases modify clauses that involve political, industrial, or individual human choices. The relevance of these adverbials to the rhetorical discourse-level functions of the modal adverbials is one of contrast: by modalizing statements that involve high levels of subjective choice, the agreement-evidence adverbials imply that confidence and likelihood are free of such choices.  77  Appendix D: “Robust Findings and Key Uncertainties” in the AR4 Synthesis One other documentary feature of the AR4 Synthesis putatively addresses policy-relevant uncertainties. The final section, “Robust Findings and Key Uncertainties,” begins by defining what counts as a robust finding and what counts as a key uncertainty. “As in the [Third Assessment Report],” the lead authors write: a robust finding for climate change is defined as one that holds under a variety of approaches, methods, models and assumptions, and is expected to be relatively unaffected by uncertainties. Key uncertainties are those that, if reduced, could lead to new robust findings. Under these definitions, the findings that are “robust” would assumedly be most relevant to policy makers interested in adapting to climatic change or mitigating greenhouse gas emissions. The uncertainties that are “key” would be of equal interest, but only to those readers interested in identify and funding further research that would then yield more “robust findings.” This final section, a chance for the IPCC authors to explain why they have assigned probabilities among the modal adverbials as they did is, in fact, only two pages long. The previous section, by contrast, is 16 pages long. Rather than addressing how these key uncertainties influence the selection of the modal adverbials, the authors simply mention the uncertainties and focus primarily on how more funding could shift them into the “robust” category. The uncertainties in the range of findings to which Oppenheimer et al. objected, the IPCC’s reported range of “likely” 21st century sea-level rise, is addressed in the “Key Uncertainties” section. It serves as a representative sample of the type of explanation that this section of the AR4 Synthesis provides. “Future changes in the Greenland and Antarctic ice sheet mass,” write the lead authors, “particularly due to changes in ice flow, are a major source of  78  uncertainty that could increase sea level rise projections” (AR4 Synthesis 73). Rather than acknowledging that the numerical range they reported might be a misleading characterization of the actual risk of sea-level rise when they reported it, they have cautiously noted, at the end of the document, that such uncertainties “could increase sea level rise projections” (ibid). The authors implicitly support the range of values that they previous had modalized with the likelihood adverbial “likely,” or a greater than 66% probability. Repeating Oppenheimer et al.’s claim, how can the sea-level range be “likely” when the processes that will almost certainly contribute to a greater increase aren’t included because they can’t be quantified? The modal adverbials display a primarily rhetorical rather than ideational function in this and the previous cases discussed. To display the authors’ subjective choices that led them to assess the findings as “likely” would have foregrounded the uncertainties in the assessment. Such a move would vitiate the opaque and quantitative mode of expression that the IPCC assumes it needs within its “speaking-truth-to-power” footing.  79  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0072479/manifest

Comment

Related Items