Logical Argument Structures in Decision-making JANE MACOUBRIE Department of Communication North Carolina State University Campus Box 8104, 204 Winston Hall Raleigh, NC 27695-8104, U.S.A. E-mail:
[email protected]
ABSTRACT: Perelman and Olbrechts-Tyteca’s practical reasoning theory has attracted a great deal of interest since its publication in 1969. Their most important assertion, however, that argument is the logical basis for practical decision-making, has been under-utilized, primarily because it was not sufficiently operationalized for research purposes. This essay presents an operationalization of practical reasoning for use in analyzing argument logics that emerge through group interaction. Particular elements of discourse and argument are identified as responding to principles put forward by Perelman and Olbrechts-Tyteca, and are viewed as fitting together in a kind of logical argument structure that is well suited to the study of practical arguments in decision-making. Both the content elements and the logical argument structure are illustrated using examples from two studies examining decision logics in public participation and jury decision-making. Advantages of this approach and proposed recognition of a new ‘filtered’ type of argument structure are discussed. KEY WORDS: argumentative case, argument content, filtered argument, logic structures, Perelman
Chaim Perelman and Lucie Olbrechts-Tyteca (1969); Perelman (1979, 1980), in their theory of practical reason, proposed that argumentation is a form of reasoning for action, i.e., the form of reasoning used for decisionmaking. They did not, however, formalize a model of practical reasoning, nor did they specify how others interested in modeling arguments should map the content ‘liaisons’ that they view as the source of an argument’s logic. This essay addresses that need, extending practical reasoning theory by operationalizing it for the study of interactive, decision-making reasoning. The essay also describes a kind of argument structure particularly well suited to discovery of a discursive logic, or the relationship between interactants’ substantive thoughts and propositions that logically leads to a particular decision. A logic, as the previous paragraph suggests, is formed by and concerned with the substantive content of a decision. The concern of this article is thus delineating and specifying how certain content elements interact to form an emergent, interactive logic. In the view presented here, the content of an argument is both the substance and the means of decisionmaking, as Perelman and Olbrechts-Tyteca proposed (Perelman, 1979, 1980; Perelman and Olbrechts-Tyteca, 1969).1 Argumentation 17: 291–313, 2003. 2003 Kluwer Academic Publishers. Printed in the Netherlands.
292
JANE MACOUBRIE
The type of argument structure discussed here addresses a dimension of argument that has been little studied and is poorly understood: complex argument structures (Jacobs, 2000). What is new in this essay is the integration of certain existing knowledge and delineation of how it can be used to locate logical argument structures in interactive discourse. Complex and interactive argument structures have component parts but also have to be understood as a whole, as an entire set of premises, reasons, and their relationships (Perelman and Olbrechts-Tyteca, 1969; Simon, 1979). Substantive logic structures are just one dimension of or one ‘garb’ for argument (Barth and Krabbe, 1982), but one that is important and understudied. The essay will proceed in three steps. Since it may not be obvious to all readers that a new way to understand the logic of argument is needed, the reasons for a new approach will first be explained. The essay then defines content and discusses specific principles of interaction and argument that can support the discovery of interactive logics. In discussing what counts as content, the reasons content must be clearly differentiated from the individual process steps of interaction will be addressed. Finally, building on those principles, the essay identifies a necessary and sufficient set of discourse and argument components that can comprise an emergent argument logic, and the content ‘liaisons’ or relationships that result in a logic. Three levels of propositional structures are identified that together can illuminate logical, interactive argument structures; the essay also presents the idea of a filtered argument. Throughout this article, the ideas presented will be illustrated with examples from (with one exception) the author’s studies (1996, 1998, 2001)2 investigating complex, interactive argument logics. This is not intended to be a full research report, but selected examples from research will be used to clarify and illustrate the type of argument structure presented. The author has so far investigated argument logics in two decisionmaking arenas: in mock and real juries deliberating criminal law charges and in citizen groups participating in forming public policy.
WHY A NEW ARGUMENT STRUCTURE?
Before presenting a new form of argument structure, the reason a new form is necessary should be addressed. Currently, there are four central ways to understand the logical structure of arguments: Toulmin’s argument structure (1958/1988), formal logic’s syllogism, Perelman’s liaisons (1969), and dialogue logic (Barth and Krabbe, 1982; Lorenzen and Lorenzen, 1978). These approaches have made important contributions to understanding the logic of argument, but at present, no theory or model supplies or intends to supply a systematic way to model complex arguments (Jacobs, 2000). To understand complex arguments, then, a new kind of argument struc-
LOGICAL ARGUMENT STRUCTURES
293
ture is needed, as Jacobs (2000) noted. It has also been argued for some time that the syllogism, the oldest of these four argument structures, is poorly suited for the task of explaining reasoning in ordinary decisions. Toulmin (1958) first made this point; Perelman and Olbrechts-Tyteca (1969) also asserted that arguments for action require a flexible structure, must often be based on conditional proof, and often include value choices. The syllogism, of course, dictates a fixed structure, requires conclusive proof, and excludes value justifications as part of logic. Toulmin (1958, p. 184) concluded that formal logic is ‘idealized theory . . . [we] should retain no illusions about its application to practical arguments.’ Toulmin’s (1958, 1988) model of an argument’s logical structure (Figure 1, below) is also problematic when applied to interactive argument. Toulmin’s model easily describes the logical structure of discrete claims but does not provide a way to model the overarching logic of a complex argument structure. Dialogue logic provides, in addition to Toulmin’s model, another way to look at the logic of arguments. Lorenzen and Lorenzen’s work on dialogue logic (1978) represents an effort to specify the linguistic rules that enable logical argument, such as logical constants (common understanding of terms) and logical connectors (e.g., and, or, if, etc.). Barth and Krabbe’s work on dialogue logic extended this idea (1982; also now called formal dialectics) and delineates a system of rules for the conduct of conflict-resolving discussion (formal dialectics3), with the use of logic features such as logical constants, etc. (formal dialectics2). Van Eemeren and Grootendorst’s pragma-dialectics (1984) links their own system of dialectical rules with certain functional speech acts that are highly relevant to conflict resolution, and includes rules for determining the winning argument. As noted earlier, none of these theories provides or intends to provide a means to model the coherent connections between components of complex, multi-tiered arguments. In pragma-dialectics, for example, argument is examined as an exchange of speech acts, and speech act analysis is used to discover elements of an argument that are relevant to resolving differences of opinion. Toulmin’s model (1958, 1988) works well for specific, statement-level arguments but not for complex ones, and Lorenzen and Lorenzen’s, Barth and Krabbe’s, and van Eemeren and Grootendorst’s work all theorize argument as a process, established via the specific moves made in an argument (Wenzel, 1979, 1990).
Figure 1. Toulmin’s argument structure (1988).
294
JANE MACOUBRIE
Complex arguments are usually reduced to their component pieces, Jacobs says, because there is ‘a lack of any systematic attention to how messages might or might not function to express informational assemblies . . . without systematic theoretical modeling, it looks like there is nothing systematic to model’ (2000, p. 264). Most importantly, we need a way to model complex arguments because the appropriate unit for understanding practical decision-making is the combined premises underlying a decision (Simon, 1976; Perelman and Olbrechts-Tyteca, 1969). Separating decisions into individual premises cannot explain a complex logic; the problem of uniting them into their coherent logical state is what needs addressing. A new model for understanding complex argument structures could also be important to researchers studying interactive decisionmaking. To discover logics that emerge through discourse between people, we need to understand logic without any help from organization via text structure or formal debate case structures. The problem addressed here, then, is how to model and make visible the overarching, substantive interactive argument structure that leads to a certain decision, called a decision logic.
PRINCIPLES OF DISCOURSE RELATED TO LOGICS
Three basic principles of interactive discourse previously delineated by others can be a foundation from which to map interactive discourse logics. These are not sufficient by themselves to permit mapping overarching logical argument structures, but they are fundamental principles from which to begin. The principles involved concern three things: how content should be defined, the specific kind of discourse structure that is most appropriate to discovering a reasoned, interactively-constructed logic, and the level of argument that should be investigated. Defining ‘content’ First of all, a specific definition of what counts as content is needed. As stated earlier, logics should be concerned with substantive content, so a clear definition of that is important. The definition of the term content that will be used here is the generally accepted one: content is the substance of a conversation, the substantive matters being discussed. Regardless of the form a statement takes, all statements speak to something, and the ‘thing’ under discussion is called the substantive content. What communication is about, researchers call the topics (Cappella, 1994; Cegala et al., 1989, 1992; Tracy, 1982, 1983), the substantive subject matter, or the substantive content of discussion (Ervin-Tripp, 1964; Foster and Sabsay, 1982). Differentiating substantive content from other aspects of discourse is
LOGICAL ARGUMENT STRUCTURES
295
essential to understanding complex argument logics, interactive or otherwise. An argument logic should deal with content, as previously mentioned, which means that content must be examined and understood as a distinct entity. Content should also be defined carefully because researchers sometimes say they are applying various forms of analysis to ‘the content.’ This loose definition of content (which seems to mean everything that is said) is problematic, especially if what is then analyzed is the form of messages (speech acts such as questions or accusations, for example). Content can be as important to know as function, as ‘I have a hangnail’ and ‘I am scheduled to die tomorrow at sunrise,’ both assertions, address radically different matters (Sykes, 1990). That form and content are two different aspects of discussion (Foster, 1986; Sykes, 1990), intertwined in interaction but still separable from it, can be illustrated by considering the following dialogue: ‘OK?’ ‘I guess so.’ ‘Why not?’ ‘I don’t know.’
Each turn above can be easily labeled as a functional move or speech act (an explanation, a query, or agreement). The substantive issue is not visible, however. Statements have substantive subjects and make functional moves; these are not the same things. It could be argued that content and interactivity are so interdependent that they should not be analyzed separately, on the grounds that issues are negotiated in interaction and particular moves affect content. Having said these things, however, we have inherently recognized substantive issues and content as conceptual things that exist in their own right, separately from each move: functional moves affect issues and content is a thing on which negotiations operate. We can and do differentiate content from other aspects of discourse, and although a particular move can affect subsequent content (insult vs. agreement, for example), content has to be understood as a separate entity just to gauge that effect. Content can be separated from particular interactive steps, the foregoing shows. Further, complex content structures must be separated from interactive steps to be understandable. The reason for this lies in the essential fact that every substantive statement makes a point about something. The substantive issue can be located in just one statement, but in complex argument, one would expect not to find each statement as a stand-alone topic, but that many would relate to larger matters under discussion. The larger matters under discussion are built from local turns or issues, then, but local turns and issues are only understandable in relation to the larger argument structure. Research has shown that people track discourse by mapping each point or localized issue onto the relevant larger issues (Keenan and Schieffelin,
296
JANE MACOUBRIE
1976; Reinhart, 1981; van Dijk, 1980). Understanding complex arguments thus requires two things: examining the content of discrete propositions and mapping them onto the more general issues. When any local proposition is mapped onto the larger issues, the turn-by-turn introduction of propositions is lost, but the larger argument frame is revealed. Such a discursive argument structure must still be considered to be inherently interactive, however, as it was constructed through interaction and is based on turnby-turn argument (rather than discourse) analysis. An example can help to make this clearer. Suppose that in a jury’s deliberation, two jurors assert that ‘his fingerprints don’t match’ and then ‘maybe the cap isn’t really his; my brothers switch caps all the time.’ These statements introduce two local issues, fingerprint match and cap ownership. Both statements are also related to a larger-scope matter, this defendant’s correct identification. Resolution of the cap or print issues, that is, are not stand-alone matters: They are important in helping to resolve the identification issue, which exists at a higher level in an issue hierarchy. Resolution of identification, of course, in turn bears on yet a higher-level issue, guilt or innocence. Without a map of the higher-to-lower substantive structure, each statement makes a proposition or introduces a topic but remains simply a stand-alone feature of discourse. Mapping local issues onto the larger issue structure is necessary, if we would understand complex argument, and that means the local points or issues have to be extracted from their turn-by-turn locale. Statements support or negate a higher-level proposition, and their effect is cumulative. Local issues combine to establish whether identification has been established, for example, and the chronological interaction structure cannot reveal the cumulative effect of various points made on a larger issue. In discourse, propositions and issues are not discussed in a linear, top-tobottom fashion, in neat packages. Simply mapping local issues or adjacent turns would not reveal the substantive framework of a complex argument; that has to be reconstructed from chronological turns. This article is interested in grasping the entire set of major premises underlying a logic, or the argument structure into which smaller issues fit. The overarching topical structure must be reconstructed from (and exists separately of) the chronological flow of issues or specific interacts. Debates over definitions and differing frames are not ignored in this approach, however. Instead, they are represented in proportion to their emphasis in the larger-frame argument. A dispute over how to define the issue will surface as an important topic in the overarching logic. The meaning of a fact is negotiated, but its relevance to a complex argument lies in the relationship to the whole. Content structures are based on but are separable from interactive moves. A close examination of each turn is still important, for two reasons. One is that the content structure can only be derived from that level of examination. As well, the complete basis of a logic relies both on understanding
LOGICAL ARGUMENT STRUCTURES
297
the larger argument structure and on understanding the types of reasons used and which ones dominate discussion. Larger-frame content structures spell out the major issues under discussion, but a reasoning structure also has an influential lower-level type of content: the kinds of reasons used to make choices. For example, saying ‘but his fingerprints don’t match,’ has fingerprint match as its substantive subject, and then ‘it’s clear he’s guilty, he confessed’ changes the subject to confession. The content of both of these justifications, however, is a substantive, physically verifiable material fact, as opposed to a principle or value statement (i.e., he shouldn’t have confessed). Since a decision based on facts is different in kind from one based on principle, this lower level of reasoning is important to understand. Together with differences in substantive topics/issues, collectively examining the use of particular justification types, especially in relation to particular topics, may help locate important differences in logics within or between groups. It is perhaps surprising that substantive content assemblies are a poorly understood and under-theorized basis of decision-making (Sykes, 1990), although to review other decision theory here is not possible. Small group decision-making theorists have certainly studied interactive argument, but to date, functional speech act analysis has most often been the object or method of these studies (e.g., Bales, 1951; Gouran and Hirokawa, 1996; Gouran et al., 1993; Hirokawa, 1983, 1985; Meyers and Brashers, 1998; Van Lear and Mabry, 1999). In the area of substantive content’s effect on group decisions, only the inference level of argument has been previously investigated (Gouran, 1983, 1986). The present project presents a way to map the cumulative substantive structure of logics that emerge through interactive discourse, a particular area of argument analysis that is under-theorized, that precedes the analysis of functional moves’ effect on a substantive logic, and that is an essential first step towards evaluation of the quality of a logic. The objective here is to describe the substantive, overarching logical structure that develops and is negotiated throughout the discourse. Discourse proposition structures Describing substantive content structures separately of interactive moves is a necessary part of discovering interactive argument logics. For this, it will also be helpful to recognize that discourse has three kinds of structure (Reinhart, 1981): functional, text-based, and propositional. In the functional structure (Reinhart, 1981), conversation builds up in sequences of functional acts or speech acts. Functional structures are generally described by identifying particular speech acts, or labeling statements as fulfilling a particular function, such as assertion, agreement, justification, or accusation (see, for example, McLaughlin, 1984). A second type of discourse structure, a text-based structure, organizes discourse in
298
JANE MACOUBRIE
yet a different way, through organizational features such as sentences, paragraphs, adjacency pairs of utterances, and so forth. The propositional structure of discourse, however, is most relevant to discovering substantive argument structures that emerge through interaction. The propositional structure of talk (Reinhart, 1981) refers to interactants’ topical references, which are sometimes implied in conversation. For example, ‘But this is really important’ or ‘I don’t understand’ inherently refer to and propose ‘This discourse is about X’ (the subject, problem, or issue). In interaction, propositional references (‘This is about fingerprints’) introduce or continue a topic and are best understood as existing within topic hierarchies. Topics are, in the most general sense, an entity about which something is said (Tracy, 1982), and they are also either local (what a particular turn or sentence is about), or global (the larger issue to which the local topics relate) (Tracy, 1983; Reinhart, 1981). Interaction has, in short, a topical macrostructure under which specific localized topics cohere. Conversation’s coherence is a function of a local unit’s relatedness to the macropropositions, ‘this conversation is about X,’ and accumulated statements proposing a view of X (Keenan and Schieffelin, 1976; Reinhart, 1981; van Dijk, 1980). People comprehend conversations not by tracking the local topics, but by understanding the more general, global topics (Planalp and Tracy, 1980). The substance of discussion is best understood as a topic hierarchy, with issues/topics at higher and lower levels, from the most general (global) to the most specific (local) (Foster and Sabsay, 1982, McLaughlin, 1984; Tracy, 1983). Suppose in conversation that someone says ‘I went to Miami on vacation and saw Aunt Margaret on the beach!’ Aunt Margaret on the beach could not be understood, would not fit into the discourse context, without reference to the larger subject of my vacation. Dialogue is ordered, in other words, by nested sets of topical hierarchies, whether in interaction or text; the most general topics are at the top of the hierarchy and within them are the more specific, local topics (Reichmann, 1978). To locate the coherence of a complex substantive argument will require identifying the global topics and propositions on them, making these two of important ingredients in a logical propositional structure. The appropriate sense of ‘argument’ Recognizing that substantive content is oriented around global topics and topic hierarchies, it follows that the sense of argument most relevant to understanding decision-making is the argumentative case (Rieke and Sillars, 1997). This derives from understanding that in decision-making, the local/global topical hierarchy is more than merely a property of discourse. In a tenure decision, for example, the most general topics might be A’s
LOGICAL ARGUMENT STRUCTURES
299
papers, A’s books, and A’s teaching. In fact, the decision topic of tenure is literally devoid of content without its specific global topics (books, papers, etc.); propositions on these global topics are actually the basis for a decision on tenure. (See Figure 2 for a graphic example display.) Individual statements gain their coherence from alignment under global topics (Planalp and Tracy, 1980); the justifications given in discussion inherently either support or deny a global proposition and eventually ‘add up to’ the end decision, or make an argumentative case (Reike and Sillars, 1997). An argumentative case is related to Wenzel’s (1979, 1990) sense of argument as product or logic, but since in Wenzel’s conceptualization the product of argument may be just a particular conclusion, the argumentative case is similar in type but fundamentally different in scale. The most appropriate sense of argument involved in complex logics, the argumentative, case refers to an overarching, combined argument. This is the sense of argument that seems to be meant when someone says ‘What are your points? What is your argument?’ The argumentative case subsumes individual statements, or argument as process, i.e., local propositions (Wenzel, 1979, 1990) or specific rhetorical strategies such as definitions, analogies, tropes or figures of speech (e.g., Warnick and Kline, 1992). In interaction, the argumentative case is also not necessarily related to the chronological steps of argument. Specific interactive moves delineate argument as procedure (Wenzel, 1979) (or dialectic, 1990), such as is addressed in dialogue logics (Lorenzen and Lorenzen, 1978; Barth and Krabbe, 1982) and in the pragma-dialectic approach to understanding logic (van Eemeren and Grootendorst, 1984). The argumentative case, as it exists across an interaction, is oriented around the global questions or issues whose resolution ‘adds up’ to or logically leads to a decision.
LOGICAL ARGUMENT STRUCTURES IN INTERACTIVE DISCOURSE
The previous section introduced the idea that to discover interactive logics, we need to recognize the substantive content as a global/local topic structure. The global topics are critically important, since their description inherently displays the higher-level substance of a decision-making dis-
Figure 2. Example of global and local topic hierarchy.
300
JANE MACOUBRIE
cussion. We could call the global topic structure propositional structure 1 (PS1), as it sits at the highest level of a discourse. Propositions on these global issues represent another propositional level (PS2) and form one basis for a decision. The lowest level of an interactive argument structure is the positive or negative local propositions (PS3). Finally, the links between global issues, global propositions, and different types of reasons delineate an argumentative case that emerges in discourse. Together there appear to be four elements of interactive discourse that can be systematically examined to reveal the logic of decision-making argument. These are: 1) global decisional topics (PS1), 2) decisional topic propositions (PS2), 3) the types of justifications or reasons that assert facts or values (PS3) and add up to propositions on the decisional topics, and 4) global proposition relationships, whether convergent, dependent, or filtered (the latter will be explained shortly). If a specific argument acts as a logic, as well, one would expect the decision made to logically follow from the global topic propositions. The particular integration of these components is original to the author, although it draws on work by others. The structure of an interactively created decision logic, in this view, has two tiers. The upper tier is formed by decisional topics, propositions on them, and links between them. Choices on these are supported by justifications (facts or principles, for example), so that both the lower level of logic (justifications) and the upper level (global decisional topic propositions) are relevant to understanding the basis for choices and/or substantively different logics. Decisional topics It will be argued here that the term decisional topics should be used to identify the global, substantive issues relevant to a particular decision. Using this specialized term solves a number of problems. First, not all global topics may be substantive ones (others may be social, etc.), and the term decisional topics differentiates substantive concerns from others. Second, many other terms already in use are overly general. Issues, for example, is a term often employed to mean arguables in discourse, or any point of disagreement (Inch and Warnick, 1997), and in stasis analysis (Davis, 1981; Kline, 1979) includes all definitions, questions of fact, circumstances, or procedural issues (Wenzel, 1979). Other terms, such as themes, criteria, and goals also are overly general. Most problematically, none of these terms can differentiate local from global concerns, or social from substantive concerns, and so make discovering a substantive logic more interpretative and idiosyncratic than is desirable. It is preferable to identify decisional topics as the substantive global topics of interest, so that the object of study is specific, research results can be systematic and reliable, and global and local topics are not confounded. Global topics can be reliably identified in discourse, as
LOGICAL ARGUMENT STRUCTURES
301
documented by studies examining subjects’ topic identification and intercoder reliability checks (Cegala et al., 1992; Planalp and Tracy, 1980; Tracy, 1982). Reasons or justifications Decisional topics (PS1) identify ‘this is about X’ or what a discourse is about, and then views or propositions on those global issues create a different propositional level (PS2) that exists in decision-making. In addition, the PS2 propositions made can also be based on different types of justifications (PS3), or propositions that support PS2 views on X or Y. As Perelman (1982) asserts, similar conclusions may be motivated differently, and the nature of these differences represent different reasoning paths. Even when the decisional topics are the same and the same final decision has been reached, the basis of choice by either individuals or groups may have varied, either in different propositions or types of justifications used. After identification of global decisional topics, there are two critical sources of an interactive argument’s logic: the propositions made at the decisional topic level (PS2) (i.e., yes on books and yes on articles), and the justificational propositions (PS3, such as facts and principles) used to develop them. Theorists have previously differentiated kinds of reasons such as moral obligations, social pressure, and authority (Rieke and Sillars, 1997), and these form a different kind of basis for a decision. To date, in the author’s research (1996, 1998, 2001) eight types of justifications or reasons are sufficient to understand the logics found. The first five – material facts (or particulars), principles, values, situations, and consequences – are drawn from Perelman and Olbrechts-Tyteca’s work (1969). Three other kinds of justifications – needs, speculations, and experiential judgments – have been useful in delineating with precision the complete structure of both public and jury arguments, although they also have not to date been a dominant force in any logic found so far. Definitions of these eight justification types are provided in Table I; the examples given in discussion, below, will be drawn from the author’s studies of interactive argument in decision-making group interaction. Material facts (or particulars, in Perelman’s terms), or concrete, verifiable evidence, are one key type of justification or reason. Facts, like other types of justifications, inherently make a local proposition on a decisional topic. Saying ‘All the health studies over the last 50 years show not one person ever became seriously ill from this product’ presents a fact derived from research. ‘He said he was at the beach,’ if said by a juror in deliberation, counts as a fact because it can be physically verified. (The truthvalue of any such ‘fact’, of course, is a different matter.) Toulmin (1988), after Ryle (1949/1976), distinguishes six logical types of fact statements or assertions, such as stated facts about the past, present, or moral judgments.
302
JANE MACOUBRIE
Table I. Justification types (PS3) 1. 2. 3. 4. 5. 6. 7. 8.
Material Facts: concrete, physically verifiable evidence Principles: decision rules other than moral values Values: good/bad, right/wrong moral decision rules Situations: circumstances indirectly related to the substantive problem Consequences: results of actions taken by deciders Needs: civic or personal needs as decision rules Experiential Judgment: personal experience Speculation: hypothetical, speculative statements
In this project, however, it is preferable to define ‘fact’ more simply, as concrete, physically verifiable evidence. Under that definition, the material facts in a legal case are those presented at trial, and the material facts in a business decision are those available for physical verification. Whether people are generally rational in decision-making, of course, and what uses they make of facts versus principles (etc.), is an interesting question. Principles, defined here as decision rules other than moral rules, would be apparent in statements such as ‘it has to be beyond reasonable doubt’ or ‘there is a risk of 4 deaths in 10,000.’ Principles guiding decision-making can be completely pragmatic, as in ‘we don’t know exactly where the restaurant is, but we know it’s on the water, so let’s just follow the streets along the shoreline.’ Pragmatic principles and moral principles need to be differentiated, as logics based on either are different in kind. It is also preferable to use the term principles rather than values, which can mean anything preferred (Rokeach, 1979). Values are then defined as moral justifications invoking good/bad, right/wrong rules, such as ‘we should not convict unless we are certain.’ Identifying values separately from principles makes a distinction between moral and other types of principled justification, which may be important to discern. Safety is a principle, under this definition, because it does not represent a moral judgment or fundamental philosophy of right or wrong but a pragmatic concern. Consequences are outcomes of a choice used to justify a decision and have been used to reason about a decision if someone said, ‘If we are wrong, think what will happen. His life will be ruined.’ A situation justification is invoked when a statement references a circumstance outside the substantive matter being addressed. For example, in a decision to vote for a particular candidate, one might discuss experience, policy preferences, or principles. But saying ‘this is the best option available’ does not have to do with the candidate per se, but with the failure of all candidates to meet expectations, and so is an example of a situation justification. Needs are either personal or civic, pragmatic or idealistic, e.g., ‘lets just decide so we can go home’ or ‘we need a community center here, not boating access.’ Speculations are hypothetical statements used to understand situations and justify judgments: ‘Maybe he just wanted to talk to
LOGICAL ARGUMENT STRUCTURES
303
her’ or ‘I don’t think he really meant to hurt her,’ in a jury discussion on motive. Finally, experiential judgments invoke personal experience as a justification for a conclusion, such as ‘I was in a robbery once, and I couldn’t identify correctly.’ These eight types of reasons reveal the different basis of choices on decisional topics regardless of the kind of content involved in a choice (i.e., legal matter, vacation destination, tenure) and so seem to be field-independent (Toulmin, 1958, 1988). With the typology of justifications just explained, whether a logic is based on facts or values (or consequences, etc.), or what combinations are persuasive, becomes open to discovery. Justifications are one basis of a logic’s reasonableness, and both practical and theoretical significance may obtain from being able to discern the content of justifications. Whether some types of justifications are more dominant in particular fields could be an interesting area for research, for example. Differences in the types of principles employed by scientists versus citizens could also be revealing. In practical reasoning, justifications are not rationalizations (Perelman, 1982) but are legitimate reasons to support accepting a belief or an action. Justifications give permissible reasons and do not attempt to exonerate an individual from responsibility for actions or decisions, as do excuses (Antaki, 1994). Perelman also makes a distinction between conclusions derived from reasoning, and conclusions that are, as he puts it, merely justified: ‘[We] need judges where reasoning does not end in a conclusion, but justifies a conclusion’ (1982, p. 144). The dual meaning of ‘justify’ is slightly confusing, but Perelman appears to recognize a difference between logically developed arguments for a choice (i.e., choices that logically follow from certain facts or reasons) versus those that are merely rationalized. The difference seems to be that in some cases, decisions are a result of reasoned consideration, while in others, justifications are merely top-dressing, applied to make an action look as if it were reasoned. Decisional topic propositions It is critical, in understanding a decision logic, to understand the overall decisional topic structure, such as books and teaching in tenure, together with the kinds of reasons that dominate decision-making. Reasons or justifications are the lower-tier ingredient in an argument’s logic, as different choices can be based on facts, principles, consequences, or experiential judgment. But as has already been discussed, it is the propositional macrostructure (PS2) (Foster, 1986; van Dijk, 1980) that provides coherence to a discourse and hence to an emergent decision logic. How this part of an interactive logic is identified thus needs to be addressed. Macropropositions propose, ‘This argument is about X,’ (PS1) and propose a view of X. The global propositions made by ‘views’ would form
304
JANE MACOUBRIE
PS2, and identifying them systematically in interaction is the concern of this section of the essay. In interactive decision-making, the valence, or positive/negative position (+/–) of justification statements helps in the identification of macro, global propositions. In other words, justifications or reasons offered inherently propose valenced macropropositions. PS2 (propositional structure 2) thus can be located by identifying global topics and justifications related to them and by noting the valence of justifications. For example, ‘I was in a robbery once and I couldn’t identify correctly’ (experiential judgment justification) argues against the credibility of an eyewitness’ testimony. Likewise, saying ‘there should be zero risk to public health’ (principle) or ‘twenty years of studies show no one got sick’ (fact) support opposite propositions on a concern about public health risk, either no risk or too great a risk. The valence of justifications provides a simple means to identify the overarching propositional structure (PS2), the propositions whose resolution leads to a particular decision. Locating logical argument structure components in interaction would proceed in four steps: 1) noting the local topics, 2) locating the global topics to which they relate, 3) identifying justifications in relation to each global topic, and 4) identifying the valence of justifications, as inherently making a ‘yes’ or ‘no’ proposition on each global topic. For discovering decisional topics, the analytic unit (part of discourse that receives a topic label) has first been the local topic of a sentence, although global topics are then derived from the comprehensive local topic list. That is, to discover the global, decisional topics, first all local topics are identified. Semantic differences over local topic labels are then eliminated. For example, in a jury deliberating on a drug possession case, two people might say ‘it was her name on the baggage tags’ and ‘aren’t they legally hers if she puts her name on there?’ One person might call the local topic of the second example, the tags. Another might say the local topic is legal ownership. People are very poor at agreeing on local topic labels (Cegala et al., 1989; Planalp and Tracy, 1980; Tracy, 1982), probably because the topical hierarchy lets people easily confuse the local with the global issues. Decisional topics are global topics, so as the foregoing shows, any semantic differences in local topic labels are unimportant as long as there is agreement about where each fits in a global topic array, the framework of the larger argument structure. After all local topics have been agreed upon, the author and assistants involved in a study have then independently specified the global topics under which they cohere, and tested agreement or reliability. In the author’s studies of juries and public participation groups, global topic reliability among different members of a research team has been consistently been stable at 0.74 and 0.75 or above (using Scott’s pi intercoder reliability statistic), without any effort to reconcile differences. With reconciling discussion, agreement has risen to from 0.84 to 0.94; 75 is the minimum level generally accepted by statisticians and social scientists. (Intercoder
LOGICAL ARGUMENT STRUCTURES
305
reliability studies are used in social science research to help verify a study’s internal validity and replicability.) The unit of analysis for identifying justifications, in the author’s research to date, has been a sentence or one-half of a compound sentence; the latter accommodates the fact that multiple justifications may be used in one sentence. Each sentence that provides a justification thus receives a code identifying it as a material fact, consequence, etc. Logical argument structures for decision: relating global propositions So far, I have proposed that identifying decisional topics, different justification types, and proposition valence will be helpful in locating an overarching logical argument structure. The last component of decision-making argument that needs to be understood, then, is how to map the logical connections between decisional topic propositions. The term logic intrinsically implies that such connections exist, since relationships between ideas or concerns constitute a logic. For purposes of describing practical reasoning, three kinds of linkages between decisional topics seem essential to recognize. The first two, convergent and dependent relationships, are well known. A third new type of logical relationship, a filtered relationship, will be proposed shortly. In a convergent argument, the choices on decisional topics are independent of each other, but each converges on the same conclusion (Walton, 1996). A convergence of reasons, and especially consilience (obtaining the same results by different methods) may be the ‘most solid foundation of inductive reasoning’ (Perelman, 1982, p. 142). In one of the author’s jury studies, juries considered whether a defendant was correctly identified, had a motive or opportunity, or whether there was any reasonable justification for the crime. The decision arrived would converge on a choice, because each of these decisional topic propositions is logically independent of each other. Determining a motive for murder does not depend on determining opportunity, or determining justification, and vice versa, so these choices are logically independent. They converge on a final decision; no choice follows from any other. Dependent arguments (Walton, 1996), on the other hand, are those in which one choice precedes and then directly influences others: later choices are logically dependent on the first. For example, in another of the author’s studies of interactive logics, two competing arguments were found and were both dependent arguments. This study was of a public citizen group’s collectively arrived at decision of how to clean up a toxic waste site. Within the group, the majority and minority arguments that emerged through the discussion began with consideration of the site’s risk to public health. The two groups reached different decisions because they reached different conclusions on the extent of the risk, and all subsequent considerations were dependent on that one global proposition.
306
JANE MACOUBRIE
Based on the facts available, the majority argument asserted that risk was low. That being their assessment, they advocated a low cost and low community-burden solution. By contrast, the minority argued for the highest degree of cleanup, based on their decision that ‘yes’ there was risk to public health. Since they believed risk still existed, this minority wanted the safest clean up method possible and did not concern themselves with cost or other issues. In both cases, all other parts of each group’s argument follow from their proposition on the first issue, either yes or no on risk to public health. The solutions preferred also follow from the global topic propositions, as well. That is, it follows that if there is no risk, there is no need for a high cost or burdensome cleanup. Similarly, if there is risk, the safest solution would be preferred. Interestingly, this study also found that these two arguments had very different justification bases; the majority argument ‘no risk’ was based largely on facts, whereas the minority argument ‘too much risk’ was based primarily on principle. Filtered arguments represent a third type of argument logic, one that appears to exist alongside those that are convergent or dependent. In the proposed filtered argument, one argument element has become a barrier over which all other choices must pass. In this situation, the argument that is formed is neither dependent nor convergent, but has a distinctively different type of construction. An example will illustrate this first, followed by a diagrammatic contrast between the three argument types. The initiating example of a filtered logic comes from substantive arguments in mock juries’ logics. Thirteen mock juries had all debated the same case so that similarities and differences in their arguments could be studied. In all 13 arguments, identification of the defendant was heavily contested. In all the arguments, as well, the principle of ‘reasonable doubt’ was elevated from its normal supporting role (as a reason given); reasonable doubt, a principle, became a global decisional topic that seems to function as a filtering decision rule. Reasonable doubt acted as an evaluative criteria, then, but as global-level issue, and the argument structure was also neither convergent nor dependent. Resolution of the decision depended on the decision on reasonable doubt, that is, but it did not help form a dependent logic. That some choices are derived from others is the essence of a dependent argument. Reasonable doubt is not part of a dependent logic, because deciding identification or motive did not depend on either ‘yes’ or ‘no’ on reasonable doubt, and vice versa. Motive could be decided regardless of one’s thinking on reasonable doubt, that is, and if the argument is ‘yes’ on reasonable doubt, identification does not even have to be resolved. What has happened, then, is that the other issues in the juries’ deliberation depended on resolution of reasonable doubt, but the logics are not dependent logics as they are presently defined (Walton, 1996). Neither is the argument just described merely a convergent argument.
LOGICAL ARGUMENT STRUCTURES
307
Resolution of reasonable doubt might alone decide the verdict and does not converge on the verdict with the other propositions. Instead, the other issues have to pass the test of reasonable doubt: Correct identification of the defendant needs to be beyond a reasonable doubt, and saying ‘yes, had a motive’ needs to be beyond reasonable doubt. Perhaps because the other propositions are highly equivocal, the propositions cannot converge on a choice; reasonable doubt then provides resolution. By rising to the global topic level, reasonable doubt creates a principled barrier against which the other issues are evaluated. The logic just described is not convergent and not dependent, then. Instead, it places reasonable doubt in the role of a screen or threshold, a final hurdle in the global proposition chain over which all other considerations must pass. To illustrate this idea further, it may help to reduce the three argument types to simple logic flows: Dependent: From A, B and C, then X Convergent: If A, and B and C, then X Filtered: If A and B, determined by C, then X
The idea of a filtered argument acknowledges a principle that has become a global-level decision rule, and acknowledges the unique structure of such an argument. All the other choices, it thus seems, were ultimately subjected to the test of reasonable doubt: This is the screening concern, the threshold that must be passed for a guilty verdict to be achieved. I propose, then, that filtered arguments are a distinct type of argument structure, like convergent and dependent arguments. Diagramming decision logic arguments The sort of logical structure of argument (decision logic) that has been discussed in this essay is formed by 1) decisional topics, 2) the reasons that support choices, 3) propositions on decisional topics (justification valence), and 4) their logical connections in convergent, dependent, and filtered argument structures. Describing logical argument structures in this way focuses on overarching logics, macro logics, rather than the logic of individual, statement-level arguments or conclusions. To model this kind of logical argument structure, we might use three types of graphic means for various purposes. For macro-scale argument, graphic diagrams can only be illustrative, as attempting to reproduce the entirety of complex argument graphically would be a massive undertaking and probably also impossible to understand. The simplest kind of graphic model would be like the one shown in Figure 3. This type of model shows only demonstrates the flow of global topics, and in this case, illustrates how some issues might be filtered through a higher-level topic/principle. This form of representation of a macro-argument logic shows the choice
308
JANE MACOUBRIE
Figure 3. Topical diagram of a filtered argument.
pattern that leads to a decision, but does not provide much clarity about the types of justifications that created each of those choices, which contribute significantly to understanding a logic. If the types of justifications used are necessary to understanding an argument fully, of understanding the genesis of a choice, a second type of graphical diagramming method is thus proposed (Figure 4). This second type of diagram is also relatively simplistic, but adds the dominant justification types to the proposition flow, which as the arguments below demonstrate, can explain a great deal about the ultimate outcomes of arguments. In Figure 4 there are two logics: the global propositions and dominant justification types used by the majority (winning) and minority community factions debating, in this argument, a toxic waste site’s remediation. The upper structure of the diagram shows the decisional topics, and the lower portion shows the proposition-related reasons or justifications. Both diagrams are arranged to illustrate the logical relationships linking the propositions, which are both dependent arguments. This second type of graphic representation of an argumentative case or decision logic reveals more of the essential features that explain the decisions: global propositions, distinctive justification types, and proposition flow or connections.
CONCLUSIONS
This essay has described a way to resolve a number of problems. First, as Jacobs (2000) noted, how an argument’s messages function in informational assemblies is under-theorized. We have focused on component pieces of argument because a systematic way to assemble these pieces has been missing. Yet as demonstrated in this essay, some basic principles of discourse can be integrated to create a means of systematically describing (and opening to evaluation) argument logics in discourse. The argument structure described here uses concepts from linguistics, communication theory,
LOGICAL ARGUMENT STRUCTURES
309
Figure 4. Qualitative Model 2.
and argument studies in new ways and extends them to explain the genesis of interactive decisions, argument-based content logics. The type of logical argument structure proposed in this essay also seems to have several strengths. First, it is responsive to the substantive content found in particular arguments and draws its logic from that content and its relationships, as Perelman and Olbrechts-Tyteca (1969) argued must be the case. That is, the logical structure explained in this way is derived from the content of a particular argument, but is found via predictable types of argument components and connections. That logics may vary in their structure is not a negative, but a necessity; the logical structure described here is also flexible enough to describe all types of arguments and very probably is field-independent, although the types of principles in use may vary (legal versus pragmatic, for example). Second, the logical structure discussed here handles complexity easily and models interactive logics successfully. In decision-making research, how decisions are reached in groups is often of interest, so that a means to analyze interactive decision-making could be quite useful. Argument as informational assemblies are under-theorized as they relate to group decision-making, where complex decision-making is the rule rather than the exception. Perhaps most importantly, the two studies used as illustrations here support that the proposed substantive structures can reveal logics used in decision-making, or the interactive construction of an argument that logically leads to a certain decision outcome. In each case studied (13 juries and 3 public participation groups), the argument structures discovered via the techniques described here did logically lead to the decisions made. That
310
JANE MACOUBRIE
is, the author believes a reasonable person using the same decisional propositions, linked in the same ways and underpinned by the same types of justifications, would reach the same conclusions. There is a logic of decision-making in these argument structures, in other words, leaving aside the issue of whether one would evaluate the logics as correct or the best possible. Arguments for action, after all, are best understood as adequate (or not) in comparison to the other arguments that could be made. Finally, the original contribution of this type of argument structure is threefold. It provides a means to analyze logics in interaction, operationalizes practical reasoning theory’s proposal that practical reason’s logic is formed by content liaisons, and proposes recognition of a new type of argument logic, a filtered argument.
ACKNOWLEDGMENTS
The author wishes to acknowledge the generous contributions of Valerie Manusov and Barbara Warnick, who mentored the dissertation from which this article derives, as well as very useful critiques of this article by Valerie Manusov, Barbara Warnick, Ken Zagacki, Vicki Gallagher, and the anonymous reviewers. NOTES 1
The principles of practical reasoning relied on here are not the best-known concepts of Perelman and Olbrechts-Tyteca’s work (Perelman, 1979, 1980, 1982; Perelman and Olbrechts-Tyteca, 1969). The present essay directs attention to their thinking about overarching logics, rather than to the logic of discrete conclusions or inferences (e.g., Warnick and Kline, 1992). This article also does not make use of their most-known concept, that of the universal audience (Fisher, 1982; Golden and Pilotta, 1986). As well, although their idea of justifications is used here, justifications in this view are not functional speech acts, as they are often utilized in group argument research (e.g., Canary et al., 1995; Meyers and Brashers, 1988). 2 The author has completed two studies of interactive argument logics and a fourth is presently in progress. The first study (1996) pilot-tested the analytical scheme explained here and examined the logics used by a community group deciding on the appropriate way to remediate a toxic waste site. Interviews were the source of argument data. In the second study (1998), 13 mock juries’ deliberation was studied; transcripts of the deliberation were the data source. Actual U.S. juries cannot be studied without state Supreme Court permission, and mock juries are valid for study when certain conditions are met. Because of the quantity of data from these juries, log linear modeling was used to describe the relationships between particular justification types and their respective decisional topics, as well as to provide a model of the decisional topics themselves. Log linear modeling is a type of non-additive chi square, a statistical technique for locating beyond-chance relationships between categorical variables.
LOGICAL ARGUMENT STRUCTURES
311
REFERENCES Antaki, C.: 1994, Explaining and Arguing: The Social Organization of Accounts, Sage Publications, Thousand Oaks. Barth, E. M. and E. C. W. Krabbe: 1982, From Axiom to Dialogue: A Philosophical Study of Logics and Argumentation, Walter de Gruyter, Berlin/New York. Bales, R. S. F.: 1951, ‘Phases in Group Problem Solving’, Journal of Abnormal and Social Psychology 46, 485–495. Burke, K.: 1966, Language as Symbolic Interaction: Essays on Life, Literature, and Method, University of California Press, Berkeley. Canary, D. J., J. E. Brossman, B. G. Brossman and H. J. Weger: 1995, ‘Toward a Theory of Minimally Rational Argument: Analysis of Episode-specific Effects or Argument Structures’, Communication Monographs 62, 183–212. Cappella, J. N.: 1994, ‘The Management of Conversational Interaction in Adults and Infants’, in M. L. Knapp and G. R. Miller (eds.), Handbook of Interpersonal Communication, Sage Publications, Thousand Oaks, pp. 380–418. Cegala, D., C. L. Bayer, J. C. B. Teboul, M. Dewhurst and A. Sears: 1992, ‘A Study of Topic of Conversation as an Assessment of Intersubjectivity’, unpublished manuscript, The Ohio State University, Columbus. Cegala, D., M. G. Dewhurst, G. Galanes, J. M. Burggraf, J. Thorpe, J. Keyton and L. Makay: 1989, ‘A Study of Participants’ Judgments of Topic Change During Conversation: Global Versus Local Definitions’, Communication Reports 2, 62–71. Davis, J. (ed.): 1981, Stasis Theory, Harcourt Brace, Orlando. Dijk, T. A. van: 1980, Macrostructures: An Interdisciplinary Study of Global Structures in Discourse, Interaction and Cognition, Lawrence Erlbaum, Hillsdale, NJ. Ervin-Tripp, S.: 1964, ‘An Analysis of the Interaction of Language, Topic and Listener’, American Anthropologist 66, 86–102. Eemeren, F. H., van and R. Grootendorst: 1984, Speech Acts in Argumentative Discourse: A Theoretical Model for the Analysis of Discussions Directed Towards Solving Conflicts of Opinion, Foris Publication, Dordrecht/Cinnaminson, PDA 1. Fisher, W.: 1986, ‘Judging the Quality of Audiences and Narrative Rationality’, in Golden and J. Pilotta (eds.), Practical Reasoning in Human Affairs, Reidel Publishing, Boston. Foster, S.: 1986, ‘Learning Discourse Topic Management in the Preschool Years’, Journal of Child Learning 13, 213–250. Foster, S., and S. Sabsay.: 1982, ‘What’s a Topic?’ Unpublished manuscript, University of California, Berkeley. Golden, J. and J. Pilotta: 1986, Practical Reasoning in Human Affairs, D. Riedel Publishing Company, Boston. Gouran, D. S.: 1983, ‘Communicative Influences on Inferential Judgments in Decisionmaking Groups: A Descriptive Approach’, in D. Zarefsky, M. O. Sillars and F. Rhodes (eds.), Argument in Transition: Proceedings of the third SCA/AFA Conference on Argumentation, Speech Communication Association, Annandale, VA, pp. 667–682. Gouran, D. S.: 1986, ‘Inferential Errors, Interaction, and Group Decision-making’, in R. Y. Hirokawa and M. S. Poole (eds.), Communication and Group Decision-making, Sage, Beverly Hills, pp. 93–112. Gouran, D. S. and R. Y. Hirokawa: 1996, ‘Functional Theory and Communication in Decision-making and Problem-solving Groups’, in R. Y. Hirokawa and M. S. Poole (eds.), Communication and Group Decision-Making, 2nd ed., Sage, Beverly Hills, pp. 55–80. Gouran, D. S., R. Y. Hirokawa, K. Julian and G. Leatham: 1993, ‘The Evolution and Current Status of the Functional Perspective on Communication in Decision-making and Problemsolving Groups’, in S. Deetz (ed.), Communication Yearbook, Sage, Beverly Hills, pp. 573–600.
312
JANE MACOUBRIE
Hirokawa, R. Y.: 1983, ‘Group Communication and Problem-solving Effectiveness II: An Investigation of Procedural Functions’, Western Journal of Speech Communication 47, 59–74. Hirokawa, R. Y.: 1985, ‘Discussion Procedures and Decision-making Performance: A Test of a Functional Perspective’, Human Communication Research 12, 203–224. Inch, E. and B. Warnick.: 1997, Critical Thinking in Communication, 3rd Ed., PrenticeHall, Needham Heights, NY. Jacobs, S.: 2000, ‘Rhetoric and Dialectic from the Standpoint of Normative Pragmatics’, Argumentation 14, 261–286. Keenan, E. O. and B. B. Schieffelin: 1976, ‘Topic as a Discourse Notion: A Study of Topic in the Conversations of Children and Adults’, in C. N. Li (ed.), Subject and Topics, Academic Press, New York, pp. 337–384. Kline, S.: 1979, ‘Toward a Contemporary Linguistic Interpretation of the Concept of Stasis’, Journal of the American Forensic Association 15, 95–103. Lorenzen, P. and K. Lorenzen: 1978, Dialogische Logic (Dialogue Logic), Wissenschaftliche Buchgesellschaft, Darmstadt. McLaughlin, M.: 1984, Conversation: How Talk is Organized, Sage, Beverly Hills, CA. Meyers, R. A. and D. E. Brashers: 1998, ‘Argument in Group Decision-making: Explicating a Process Model and Investigating the Argument-outcome Link’, Communication Monographs 65, 261–281. Perelman, C.: 1980, Justice, Law, and Argument: Essays on Moral and Legal Reasoning, Reidel Publishing, Boston, MA. Perelman, C.: 1979, The New Rhetoric and the Humanities: Essays on Rhetoric and its Application, Reidel Publishing, Boston, MA. Perelman, C., 1982: The Realm of Rhetoric, University of Notre Dame Press, Notre Dame, IN. Perelman, C. and L. Olbrechts-Tyteca: 1969, The New Rhetoric: A Treatise on Argumentation, University of Notre Dame Press, Notre Dame, IN. Phillips, G. M.: 1973, Communication and the Small Group, Bobbs-Merrill, New York. Planalp, S. and K. Tracy: 1980, ‘Not to Change the Topic But . . . : A Cognitive Approach to the Management of Conversation’, in D. Nimmo (ed.), Communication Yearbook 4, Transaction Books, New Brunswick, NJ, pp. 237–258. Reichmann, R.: 1978, ‘Conversational Coherency’, Cognitive Science 2, 283–327. Reinhart, T.: 1981, ‘Pragmatics and Linguistics: An Analysis of Sentence Topics’, Philosophica 27, 53–94. Rieke, R. and M. Sillars: 1997, Argumentation and Critical Decisionmaking, 4th ed., Longman, New York. Rokeach, M.: 1979, Understanding Human Values, Free Press/Macmillan, New York. Ryle, G.: 1976, The Concept of Mind, 5th ed. (1st edition 1949), Penguin Books, Harmondsworth, England. Scheidel, T. M. and L. Crowell: 1964, ‘Idea Development in Small Discussion Groups’, Quarterly Journal of Speech 50, 140–145. Simon, H.: 1976, Administrative Behavior: A Study of Decision-Making Processes in Administrative Organizations, Free Press, New York. Sykes, R. E.: 1990, ‘Imagining What We Might Study If We Really Studied Small Groups from a Speech Perspective’, Communication Studies 41, 200–211. Toulmin, S. E.: 1958, The Uses of Argument, Cambridge University Press, London. Toulmin, S. E.: 1988, The Uses of Argument, 9th ed., Cambridge University Press, London. Tracy, K.: 1983, ‘The Issue-Event Distinction: A Rule of Conversation and its Scope Condition’, Human Communication Research 9, 320–334. Tracy, K.: 1982, ‘On Getting the Point: Distinguishing “Issues” from “Events”, An Aspect of Conversational Coherence’, in M. Burgoon (ed.), Communication Yearbook, Transaction Books, New Brunswick, NJ, 280–301.
LOGICAL ARGUMENT STRUCTURES
313
Van Lear, C. A. and E. A. Mabry: 1999, ‘Testing Contrasting Interaction Models for Discriminating between Consensual and Dissentient Decision-making Groups’, Small Group Research 30, 29–58. Walton, D.: 1996, Argument Structure: A Pragmatic Theory, University of Toronto Press, Toronto. Warnick, B. and S. Kline: 1992, ‘The New Rhetoric’s Argument Schemes: A Rhetorical View of Practical Reasoning’, Argumentation and Advocacy 29, 1–15. Wenzel, J.: 1979, ‘Jurgen Habermas and the Dialectical Perspective on Argumentation’, Journal of the American Forensic Association 16, 83–92. Wenzel, J.: 1990, ‘Three Perspectives on Argument: Rhetoric, Dialectic, and Logic’, in R. Schuetz and J. Trapp (eds.), Perspectives on Argumentation, Waveland Press, Prospect Heights, IL, pp. 9–27.