ISSA Proceedings 1998 – Semantic Shifts In Argumentative Processes: A Step Beyond The ‘Fallacy Of Equivocation’

In naturally occuring argumentation, words which play a crucial role in the argument often acquire different meanings on subsequent occasions of use. Traditionally, such semantic shifts have been dealt with by the “fallacy of equivocation”. In my paper, I would like to show that there is considerably more to semantic shifts during arguments than their potentially being fallacious. Based on an analysis of a debate on environmental policy, I will argue that shifts in meaning are produced by a principle I call ‘local semantic elaboration’. I will go on to show that semantic shifts in the meaning of a word, the position advocated by a party, and the questions that the parties raise during an argumentative process are neatly tailored to one another, but can be incommensurable to the opponent’s views. Semantic shifts thus may have a dissociative impact on a critical discussion. By linking the structure of argumentation to its pragmatics, however, it may be revealed that there are two practices that account for a higher order of coherence of the debate. The first practice is a general preference for disagreeing with the opponent, the second practice is the interpretation of local speech acts in terms of an overall ideological stance that is attributed to the speaker. Because of these practices, parties do not criticize divergent semantic conceptions as disruptive, but they treat them as characteristic and sometimes even metonymic reflections of the parties’ positions.

1. The fallacy of equivocation
Starting with Aristotle’s fallacies dependent on language (Aristotle 1955: 165 b 23ff.), the impact of shifts in the meaning of words on the validity of arguments has been a standard topic in the study of fallacies (as a review, see Walton 1996). Traditionally, such shifts have been dealt with by the ‘fallacy of equivocation’. We can say that a fallacy of equivocation occurs, if the same expression is used or presupposed in different senses in one single argument, and if the argument is invalid because of this multiplicity of senses. Moreover, in order to be a fallacy, the argument must appear to be valid at a first glance, or, at least, it has to be presented as a valid argument by a party in a critical discussion. Equivocation can be produced by different kinds of semantic shifts, for example, switching from literal to metaphorical meaning, using homonyms, confounding a type-reading and a token-reading, using the same relative term with respect to different standards (see Powers 1995, Walton 1996).
Like many others, Woods and Walton (1989) analyze equivocation as a fallacy in which several arguments are put forward instead of one. If the ambiguous term occurs twice, then there is at least one argument in which the ambiguous term is interpreted in an univocal way, and there is at least one other argument in which it is interpreted differently. Each of these arguments is invalid: The first argument is invalid, because in one of its assertions, the ambiguous term must be disambiguated in an implausible way to yield a deductively valid argument; the second argument is unsound, because it is deductively invalid. So, analytically, the fallacy of equivocation can be viewed as a conflation of several arguments. In practice, however, this ‘several arguments’ view seems to be very implausible. Woods and Walton posit that people reduce the cognitive dissonance that resulted from being faced with invalidating readings of the argument by conflating them into one that is seemingly acceptable. This “psychological explanation” for the “contextual shift”, that allows for two different readings of the equivocal term to occur in one argument (see Woods & Walton 1989: 198ff.), is not convincing. First, there is no reason why a person should generally be disposed to accept the argument in order to reduce cognitive dissonance – why doesn’t she simply reject it, if she discovers the fallacy? Secondly, most textbook examples of equivocation are puns or trivial jokes. Their humourous effect is founded on the incongruence between the plausible, default reading of the potentially equivocal expression on its first occasion of use and the divergent disambiguation it has to receive on its second occasion, if it is to make sense (Attardo 1994). That is, people just do not develop alternative readings, which they afterwards conflate, but they restrict themselves to contextually plausible readings.[i] It seems then that it is not a conflation of several arguments that leads to the acceptance of an equivocation. I suggest that it is simply the identity of the form of an expression that can be misleading, because it can erroneously suggest the identity of meanings, as long as there is no definite semantic evidence which points to the contrary. This view is in line with the observation that gross equivocations -for instance those that rest on homonyms which share no contextually relevant semantic features (like “bank”)- are easily discovered, while in the case of subtler equivocations, people often “feel” that there’s something fishy about the argument without being able to locate the trouble precisely.

So, why have I deployed these reflections on the interpretive structure of equivocation? In typical cases of equivocation, there are two or more instances of the problematic expression. Mostly, none of them is ambiguous in context, that is, there is only one plausible disambiguation for each instance of use, but these disambiguations are different. This difference in turn results from a potential ambiguity of the lexical item out of context. Walton (1996: 21ff.) seems to acknowledge this point, as he draws a distinction between potential, lexical and pragmatic ambiguity in use. But he is wrong, if he says that pragmatic ambiguity was the interesting case, because most equivocations do precisely not arise from pragmatic ambiguities (though this might also be the case), but from the exploitation of lexical ambiguities. I now will focus on those candidates for equivocation in which expressions are not ambiguously used at the moment of their use, and I will term them ‘semantic shifts’: The meaning that is attached to an expression changes from a first instance of use to a next one.

2. The empirical case: the keyword “freedom” in a discussion on environmental policy
My inquiry into semantic shifts in natural argumentation is based on so-called ‘keywords’ (Nothdurft 1996). Keywords are expressions that obtain a crucial status concerning the topics discussed and the positions unfolded over the course of a discussion. Because of their importance for the argumentative process, and since they are used repeatedly, they are especially apt to a study of semantic alterations over the course of an argumentative process. My examples are taken from a study on public debates about environmental policy. I analyzed six videotaped discussions that were subsequently transcribed. The analysis was carried out in a conversation analytic mode (Deppermann 1999; Heritage 1995). Here I will focus on one exemplary case. It comes from a staged discussion titled “ethical questions concerning waste”. A theologian and a representative of a producer of packages argue about the changes of consumption habits that were necessary for ecological reasons, and how these changes were to be brought about. During this discussion, “freedom” emerges as a keyword. While Meyer, the industrial representative, holds that there was no legitimate way to restrict the consumers’ freedom to decide for themselves what to buy, his opponent Weiss, the theologian, insists that consumption needed to be limited for ecological and for psychological reasons. Before we turn to the analysis of specific semantic and argumentative properties, I give you a typical sequence of segments, in which“freedom” becomes crucial for the argument, and I will explicate in short the main semantic, evaluative and argumentative characteristics concerning the use of “freedom” in each segment. These segments are not adjacent parts of dialogue, but they are subsequent instances in which “freedom” is talked about, and the participants relate the segments to one another.

(1) Meyer had already asserted that there were no legitimate grounds for restricting the consumers’ freedom to decide which needs they would like to satisfy. Weiss replied that, for instance, a reduction of mobility was not a loss of freedom, but might increase the quality of life. Now Meyer insists on his position: „But who is to define the quality of life? I believe that we are all wholly individual beings, and, with my expression ‘being man’, I find it very very important that I’m not anyhow forced by any social group or by the state to live in a certain way. Like I had to sit at home every day of the week and read a book. I simply defend myself against this absolute either-or. I like to reconcile both: I like to get to know new cities and new countries and stuff and that’s what I perceive as a piece of freedom.“

Meyer takes the position of liberalistic individualism by emphatically explicating his conception of freedom. He defines it by the absence of any constraint or prescription, he explicitly includes mobility – his example is travelling- in the extension of freedom, and gives it an unquestionably high value. Meyer argues that the irreducibility of freedom was derived by the fact of individuality, because individual differences between people made any claim to general rules illegitimate. In his perspective, quality of life then is not superior to freedom, but freedom is itself the precondition for defining quality of life.

(2) Little later, a discussant from the audience takes up the issue of restricting freedom; he addresses Meyer: “I think you still owe us an answer to the question: how far should our freedom reach? Because there is the freedom to live at the expense of others, to consume at the expense of others. Now we have still learnt: freedom – my own stops where the freedom of the other begins, and if I don’t grant others to live as I do, then I cannot go on living that way, at least not in the long run. And that’s why we have to start to live in a different way.“

In his contribution, the discussant defines freedom not as an irrelational, individual affair, but sees it as a reciprocal, social matter. He values freedom negatively, as he points to harmful consequences arising from it. He claims that the current practice of freedom prevented other people from living the same way. Since he sees this as a violation of a basic moral maxim – he alludes to a famous dictum of Rosa Luxemburg-, he concludes that the way of life had to be changed, which implied that freedom had to be restricted. Interestingly, he doesn’t state this last thesis explicitly, but formulates it in terms of a question, by which he starts his argument. This kind of indirectly stating a position is a common rhetorical device in the debates I analyzed. It is also used in the segments (1), (3), and (4).

(3) Meyer doesn’t respond to the claim of the discussant and instead opposes to Weiss’ thesis that the production of unnecessary goods had to be stopped: “There’s a bottle of beer on the table. I don’t drink beer, so in my opinion it’s superfluous. But I like other goods very much. And there are people, perhaps you, who would say that’s totally superfluous. So, who defines it in the last resort? Again, that’s the aspect of freedom.” Meyer repeats his conception of freedom we know already: its essential semantic aspect of the individual definition of preferences of consumption, its positive valuation and the argumentation that it was irreducible.

(4) Weiss now directly attacks Meyer’s position: “Those market-mechanisms of supply and demand are not decisions of freedom that I can make by myself. If a system once is established, I cannot elude it. The average worker must buy at Trashy’s [name of food store], he’s got no choice, but to buy these one-way packages. The question must be put another way. It’s not, whether I take the freedom to buy my things somewhere else, but, how must I organize economy, how must I organize man’s dealings with the resources. It’s not necessarily this ‘I must have my freedom’, but, perhaps, the deeper question is, if this devouring of products, this mentality of a suckling, if this really makes people happy.“

Weiss first introduces “market-mechanisms” as an antonym to “freedom” and denies the existence of the consumers’ freedom. This is a contradiction to her earlier statements, when she criticized and devaluated the consumers’ freedom and thereby presupposed its existence. Later on, she seems to suggest that Meyer (like many others) had fallen prey to an ideological self-deception: What he takes for his freedom was really the “mentality of a suckling”, which means – as she specifies later – that the consumer psychologically also is not free, but depends on consumption like a drug-addict. By her first argument, Weiss denies that it made sense to argue about how the individual might consume more ecologically. The second argument subordinates the question of the consumers’ freedom to the question of happiness.

(5) Weiss continues this line of reasoning up to a point where she inverts Meyer’s conception of freedom: “So my question actually is: How much freedom or time or creativity or occasions of communication am I deprived of by, for instance, the consumption, the acquisition of certain things?“ Freedom now is equalled with other immaterial goods, that means, with her conception of happiness; its extension not only doesn’t include consumption, but consumption is seen as the rival of freedom.

3. Properties of semantic shifts: Local semantic elaboration and processual reinterpretation There is an enormous variety of semantic aspects of “freedom” that are deployed in the segments presented. We find different conceptions of
– extension and exemplification (freedom includes (not) consumption, (not) mobility),
– implication (freedom implies travelling, happiness implies freedom),
– co-hyponymy or partial synonymy (quality of life, time, creativity, occasion to communicate),
– antonymy (to be forced to live in a certain way, market mechanisms, mentality of a suckling),
– perspective (individual/self vs. social/others; prerequisites and consequences of freedom),
– factuality (freedom exists, doesn’t exist, “freedom” is an ideological deception),
– deontic meaning (freedom needs to be restricted, must not be restricted),
– valuation (positive, negative, subordinated to other values),
– and different semantic modes (use, citation).

Clearly, these conceptions don’t sum up to a homogenous semantics of freedom, but they are continuously reworked from segment to segment. We get several kinds of semantic shifts in the meaning of “freedom” between and sometimes even within segments, as, for instance, narrowing, widening and negation of extension, oppositive valuation, rejection, addition or exchange of semantic aspects and structural relations.

Perhaps, you would question, if really all of these shifts concern semantic matters. So, is valuation part of semantics, or does it rather relate to a state of affairs? Or is there really a shift in meaning involved, if you point to the harmful consequences of freedom for others, instead of focussing on the benefits for the individual? The answer to such questions depends on your semantic theory, especially on what you consider as the scope of semantics. But beyond differences in theoretical outlook, it seems to be impossible to draw a clear-cut distinction between the semantic properties of words-in-context and the assertions that are made about certain states of affairs that are designated by these words. This becomes especially clear in the case of opposing valuations. The positive or negative value of “freedom” is not attributed to a state of affairs or a semantic conception that is expressed independently of the valuation. On the contrary, it is by expanding different semantics of “freedom” that valuations are made. Consider segments (1) and (2): Meyer’s view of individual choice implies a positive valuation of “freedom”, a negative valuation is implied by the discussant, who conceives of “freedom” as a social threat to others. It is highly improbable that they talk about the same referents of “freedom”, and it is for sure that they don’t mean the same intension of “freedom”.

As the instances of “freedom” show, speakers actively shape the meaning of words with respect to their context of use. They do this by practices of what I would call ‘local semantic elaboration’: by explicating and exemplifying the semantics of a word, by contrasting it with other words or by establishing relations of class-inclusion, implication or synonymy. Context-dependency doesn’t only relate to such clearly pragmatic dimensions of semantics as reference or deontic meaning, it also affects dimensions that are commonly held as lexically determined, such as denotation or position within lexical fields.[ii] These contextual constructions of meaning are not merely discursive realizations of lexical relations that would hold independently of actual use. Rather, lexical relations are selectively constructed and portrayed as relevant for the specific context of use. These semantic constructions are ‘local’, because they are intrinsically context-bound; the speaker might consider them as irrelevant or even wrong, when he uses the word “freedom” for the next time. As the examples of the antonyms “market mechanisms” and “mentality of a suckling” demonstrate, these lexical relations can not simply be viewed as actualizations of a pregiven lexical structure, but they are created with respect to the specific contextual matters at hand.
In most cases, these local semantic elaborations do not result in gross equivocations or even contradictions. Rather, most of them constitute different specifications of a very abstract and vague basic meaning. In the above segments, a definition of “freedom” as “to be allowed to do whatever one wants to do” would work for most, though not all instances.[iii] But this is clearly not a definition that covers all semantic aspects of “freedom” that are relevant in each individual instance of use. Indeed, it is often very hard to decide, if the semantics of any two instances of “freedom” are sufficiently similar for considering them as relevantly concerning the same matter or if they are relevantly different.[iv] The simple distinction between “same meaning” and “different meaning” is quite pointless, because there is always some semantic aspect that is subject to change.
The complexity of the semantics of words-in-context is further complicated by the fact that meaning is not invariably fixed by the end of an utterance. Speakers may add or correct certain aspects, they may give further specifications and clarifications. In addition, the activities of other speakers can affect the meaning of the words that a speaker has used. Consider, for instance, segment (2). By claiming that unrestricted individual freedom was a danger to the freedom of others, the discussant contests an aspect of Meyer’s conception of freedom that remained implicit in segment (1), namely, that “freedom” in Meyer’s sense was available to everyone. Meyer didn’t state this availability, but it can be attributed to his semantics of “freedom” as long as he doesn’t exclude this aspect explicitly. Semantic activities of one speaker thus can lead to emergent reinterpretations of the semantics of words that another speaker has used. So we really are faced with semantic processes in which interpretations are locally made and continually reworked. Because of this local semantic elaboration and processual reinterpretation, semantic shifts in argumentative processes almost inevitably occur.
Most theorists of argumentation still at least tacitly seem to cling to a conception of logical semantics. This might also be the main reason for the fact that they conceive of semantic shifts nearly exclusively as potential sources of fallacies. My short analysis on the semantics of natural language in everyday dialogue suggests that we need a more complex, more interpretive and more contextually sensitive conception of semantics. Especially the aspects of active constitution of meaning in context, of processuality and of multiplicity of the dimensions of meaning have to be considered more seriously. They must be viewed as basic features of semantics and not primarily as flaws.

4. Semantic shifts in the argumentative process: Reciprocal constitution of semantics, question of debate and position
How are these semantic properties linked to argumentation? First of all, semantic shifts are closely tied to alterations of the question of the debate.[v] A very obvious case is segment (4): Weiss first deals with the economical question, how consumption might be arranged in a way that is ecologically favourable; she then unmarkedly turns to the psychological question, in which relation the consumers’ freedom stands to happiness. Alterations of questions are still more common between subsequent contributions of different parties. So we find alterations of the question between segments (1) and (2), between (2) and (3), and partially between (3) and (4). Take, for example, segments (1) and (2). In segment (1), Meyer talks about the question “who is to define quality of life?“; his position is that everyone had the right to decide on his own about his way of life; this position rests on the semantics of “freedom” as an irreducible individual right. In segment (2), the discussant talks about the question “how far should freedom reach?“; his position is that freedom was to be restricted; this position rests on the semantics of “individual freedom” as a limitation to the freedom of others. We see that alterations of the question of the argument are in line not only with the semantics of the keyword “freedom”, but also with the position advocated by the speaker. In other words: There is a reflexive relationship between the question of the argument, the position taken und the semantics of crucial words. This reflexive relationship consists in a self-referential and reciprocal constitution of the three elements question, position and semantics, which bolster and stabilize one another. Semantic shifts thus can gain an important role for the elaboration of positions. A major part of the confrontation between the parties is realized by deploying different semantics of “freedom”. Although these semantic shifts can cause dissociations[vi] of the argumentative process, they are vital to the unfolding of the parties’ positions and therefore also for their communication.

Let me go a little bit further into this, because it especially matters to the relevance of the fallacy of equivocation for dialogic argumentation. First, we have to keep in mind that the fallacy of equivocation is only in case, if a semantic shift in the meaning of a word affects assertions that are tied together in one argument[vii], and that means also: they have to be framed as relating to the same question. An equivocation that meets this criterium can be found in segment (4). Weiss claims to refute Meyer’s assertion that the consumer should be free to decide which sort of product he wants to buy. She objects that the consumer couldn’t avoid buying goods which are wrapped up in one-way packages, and that the consumer thus was not able to decide freely. This alleged refutation rests on an equivocation; more specifically it is a fallacy “secundum quid” that consists in the neglect of relevant semantic qualifications: While Meyer spoke of “freedom” in terms of subjective preferences for certain products, Weiss speaks of “freedom” in terms of the choice of ecologically favourable products.
However, most of the semantic shifts that can be observed in our examples of “freedom” do not lead straightforward to fallacies of equivocation. There are at least three other argumentative moves that are accomplished by shifts in the semantics of “freedom”. The first move is to argue about the right definition of “freedom”. For instance, in segment (1), Meyer explicitly defines “freedom” in terms of travelling, whereas Weiss had claimed that mobility wasn’t part of freedom. The second argumentative move is to downgrade the relevance of the opponent’s position and the question he deals with by semantic shifts. In segment (2), the relevance of Meyer’s claim to individual freedom is downgraded by focusing on the detrimental aspects of freedom. By downgrading relevance, the validity of the opponent’s position and his semantics of the keyword are not really rejected, but they are either ignored or treated as less relevant in relation to some higher-order concern and become superseded by an alternative conception that is presented as being more relevant. By downgrading relevance, parties to an argument leave open, if they share an opponent’s assertions. They manage to maintain opposition, even if they actually share the opponent’s views, and they refuse consent which could be exploited by him. A third argumentative move that rests on semantic shifts is made by refuting positions which have not, at least not exactly in this way been taken by the opponent. In segment (2), for instance, the discussant refutes the position that there was generally no limit to individual freedom, even if it does harm to others. The refuted position is framed as if it had been taken by Meyer, though Meyer had not talked about potentially detrimental aspects of freedom. The refutation thus is a valid argument in itself, but it rests on a semantic shift. Again, self-reference is at work here: Speakers build arguments that are framed as refutations of the position of others, while the refuted position is not the opponent’s original position, but rather a more or less altered representation of it.

Though my analysis seems to suggest that this last kind of argumentative move was unfair or fallacious, this is not necessarily so. In order to advance the argumentation with respect to related or higher-order questions, it might be inevitable and perfectly right to draw on inferences and interpretations derived from an opponent’s utterances, to comment on its premises or to reject its consequences. A general problem of the analysis and evaluation of semantic shifts thus results from the complexity of dialogic arguments. This complexity is made up of several factors: usually, there are several associatively, hierarchically etc. interrelated questions; there are background issues and taken-for-granted conditions, values and so on that any argumentative contribution can be related to; the argumentative function of a specific speech act is often polyvalent and sometimes unclear; semantic interpretations of one segment can be changed later on; many semantic shifts do not occur within clear-cut arguments, but over the course of an accumulating argumentative process that is characterized by internal argumentative relations which are often highly complex, vague and multiply interpretable.

5. Semantic shifts and higher-order coherence: Indexical interpretation with respect to a global positional confrontation and preference for disagreement
Semantic shifts can lead to talking at cross purposes. This can easily be seen, if you look at the debate about freedom. From a first segment to a next, the question is regularly altered, so no specific question is settled, nor is it the case that different opinions to a question are equally clearly expressed by the opponents. While Weiss and the discussant almost exclusively focus on negative aspects of freedom, Meyer simply doesn’t respond to them. On the other hand, Weiss and the discussant partly deny, but also partly disregard the positive aspects of freedom that Meyer values highly.

The conceptions of freedom that the parties to the argument develop are incommensurable in many ways.[viii] Nevertheless, to complain of mere dissociation would be premature. The speakers themselves signal coherence between contributions by tying devices, such as
– reminding the opponent of an obligation that was established by his partner’s activity (segment (2): “you still owe us an answer”),
– highlighting that an argument refers to a position that had already been deployed (segment (3): Again, that’s the aspect of freedom.),
– using paraphrase and citation of the opponent’s position (segment (4): “It’s not whether I take the freedom to buy my things somewhere else”, “It’s not necessarily this ‘I must have my freedom’”),
– using parallel syntactic construction formats, in order to link two positions together (segment (4): repeated use of the format “It’s not …I …freedom…, but…”).

Moreover, the repeated use of “freedom” as a keyword is itself a device for establishing coherence: By using the same word, the participants signal that they talk about the same topic as they did before. One of the main functions of the keyword thus is to weave a thread which ties together different contributions in one topical unit. “Freedom” thus acquires a somewhat paradoxical status with respect to discursive coherence: While its semantic alterations produce incoherences, the repetition of the word-form indicates a general topical coherence.
In spite of these dissociations and, indeed, in part by these dissociations, there is a higher order of coherence. It is the coherence of a confrontation between two global positions. Meyer advocates the position of liberalistic individualism, and he focuses on the subjective use of products for the consumer; Weiss and the discussant advocate the position of universalistic dirigism, and they focus on questions of global ecological responsibility. These opposing positions are unfolded consistently over the course of the debate. It is performed rather as a global positional confrontation than as a discussion in which questions with a clearly restricted focus were talked about in a strict order. Single speech acts and arguments presented by one party are not interpreted and reacted to in isolation. Instead, they are indexically interpreted with respect to the global ideological stance that is attributed to the speaker. Since parties interpret local moves in terms of a global positional confrontation, it is thus perfectly to the point to reject an opponent’s thesis by simply downgrading its relevance or by switching to another aspect that relates to some similar point at issue. It all seems to be one argument – in the sense of having an argument-, rather than performing a series of arguments – in the sense of making arguments concerning a clear-cut question.
At times, this global orientation is articulated by the opponents themselves. Consider, for instance, segment (3): Meyer makes an argument that is supposed to prove that there was no legitimate way to decide which goods should be dismissed as superfluous. By concluding “again, that’s the aspect of freedom”, he links his argument to his general ideological stance. It is itself symbolized by the keyword “freedom”, which he has repeatedly used like a flag for his position. Weiss does the same regarding Meyer’s position: She refers to it by the ficticious citation “I must have my freedom” and thereby treats it as a whole that can be referred to globally. So, not only utterances of the opponent are interpreted in terms of his overall ideological stance, speakers also frame own arguments and assertions as contributions that indexically reflect their global standpoint. Dialogic argumentation thus is performed as an interpretive process which locally and globally gains crucial dimensions of its coherence by assumptions about higher order positions of the parties. The practice of higher-order interpretation clearly can cause difficulties to the analyst who doesn’t share or doesn’t manage to reconstruct such higher order assumptions.

Because of this practice of indexical higher-order interpretation, participants only very rarely criticize semantic switches as fallacious or as invalidating a refutation.[ix] Different semantics are interpreted and taken into account as reflecting the specifics of the parties’ positions, they are not treated as obstructions to a critical discussion. As Meyer’s leitmotif-like conclusion “again, that’s the aspect of freedom” and Weiss’ ficticious citation “I must have my freedom” show, a party’s position can metonymically be identified by a certain way the party uses a keyword, and that is, also by a certain semantics of the keyword.
Higher-order interpretation in terms of opposing global positions is closely linked to another pervasive feature of the argumentative process: a preference for disagreement (Bilmes 1991).[x] This is in sharp contrast to non-competitive, cooperative interactions, which are enacted according to a preference for agreement (Pomerantz 1984). This inversion of preference in a competitive debate is constituted by several features of discursive practice: Disagreements are formulated without hesitation, in unmitigated and even upgraded forms, while agreements are generally avoided.[xi] If they are produced at all, they are minimized, subordinated to disagreements, and formulated in mitigated ways. Together with higher-order interpretation, this general preference for disagreement itself lends a coherent structure to the debate as a global positional confrontation. Along with these two practices, the positions tend to become increasingly rigid. One case in point is the stabilization of certain argumentative patterns that are repeatedly used by the parties. Meyer, for instance, rejects any demand for an ecologically based regulation of production or consumption by a fixed argumentative pattern (see segment (3)): He points to some product or activity, talks about his own consumptive preferences regarding it, declares that other people would prefer different things, and concludes that there were no legitimate grounds on which to base any regulation.[xii]

The combination of local semantic elaboration with the practices of higher-order interpretation and preference for disagreement might also be responsible for the fact that the participants don’t seem to care about obvious contradictions that result from divergent semantics of “freedom”. For instance, Weiss once claims that consumption wasn’t a case of acting freely and would even deprive of freedom (segment (4) and (5)), while in a later phase of the discussion, she demands that the consumers’ freedom be restricted. Though this is an apparent contradiction, both conceptions converge with regard to a higher order of global positional confrontation.
Both of them result in downgrading Meyer’s issue of individual freedom in relation to her issue of global responsibility and the increase of happiness by changing the way of life. So it seems that assertions may be accepted as long as they are functionally equivalent with respect to a positional confrontation, even if they suffer from severe logical flaws.
Behind the dissociation of the argumentative process that is mainly produced by semantic shifts, there thus lies a coherent systematics of global positional confrontation. This coherence follows its own principles of higher-order interpretation and preference for disagreement. These principles have their specific functions for the evolution and negotiation of positions, for the constitution of the interactional relation of being opponents and for issues of their self-presentation in front of an audience.

6. Conclusion: A plea for a non-normative reconstruction of argumentative practices
My analysis has shown that semantic shifts are virtually inevitable in a critical discussion on complex subject matters. They can give rise to dissociation and fallacies, but they may as well contribute to the elaboration of questions and positions. Participants in a debate follow argumentative and interpretive principles that are at odds with traditional views of argumentation. By reconstructing such principles, namely, the preference for disagreement and the interpretation of local utterances with respect to an overall stance attributed to the speaker, we can reveal that phenomena like semantic shifts can be coherent, functional and often unproblematic for discussants. Argumentation analysis therefore should not prematurely condemn such processes as defective because of their dissociative impacts on argument structure. These alleged flaws rather should be seen as a starting point for a non-normative reconstruction of the practices, principles and functions that govern natural argumentative processes.[xiii] An empirical inquiry into natural argumentation should not restrict its focus to questions of argument structure, but it should take interactive, processual and functional matters into account. As my analysis shows, these aspects are not only interesting in their own right, they are also vital to an adequate understanding of the way discussants constitute and interpret argument structure itself.[xiv]

NOTES
i. Psycholinguistic experiments of the process of semantic disambiguation in natural language comprehension also show that, within a few tenth of a second, people choose the contextually appropriate reading and discard the implausible ones (Swinney 1979).
ii.Consider, for example, the denotational question, if the consumers’ right of choice is part of freedom, or the differing antonyms, co-hyponyms or partial synonyms that are related to ‘freedom’ by the speakers.
iii. For instance, this definition would produce a contradiction, if it was applied to ‘freedom’ in the context of the assertion ‘restriction of mobility is not a loss of freedom’.
iv. The relation of this problem to the fallacy of equivocation is discussed in the next section.
v. A related point was already made by Aristotle (1955: 92ff.), who points out that using ambiguous expressions amounts to asking more than one question.
vi. Dissociations are produced, if the argumentation loses its topical coherence and contributions relate to different issues (see Spranz-Fogasy & Fleischmann 1993).
vii. Remember Walton’s ‘argument requirement’ for fallacies (Walton 1996: 24ff.)!
viii. By the way, I personally think that this incommensurability and talking at cross purposes is one of the main reasons why debates of this kind so often leave the audience dissatisfied.
ix. Indeed, while a lot of contributions that include semantic shifts are rejected by opponents, the rejection is always justified by the alleged irrelevance of the fact, the question, etc., but never by a reproach with equivocation.
x. ‘Preference’ here doesn’t mean a psychological disposition of sharing or not sharing opinions, but refers to structural features of the discourse: Preferred activities are those that are performed without justification, and that are realized in a shorter, unmarked, and unmitigated form, while dispreferred activities are characterized by the opposite features.
xi. Hence, we often find no uptake of opponent’s positions that are likely to be shared.
xii. Meyer repeats this argumentative pattern six times during the discussion.
xiii. I elaborated further on this point in Deppermann (1997: 319ff.).
xiv. Consider, for instance, the argumentative criteria and resources that participants in natural argumentation themselves appeal to (Spranz-Fogasy 1999).

REFERENCES
Aristotle (1955). On sophistical refutations. London: Heinemann.
Attardo, S. (1994). Linguistic theories of humour. New York: de Gruyter.
Bilmes, J. (1991). Toward a theory of argument in conversation: The preference for disagreement. In: F. van Eemeren & R.
Grootendorst (Eds.), Proceedings of the Second International Conference on Argumentation. Vol 1A (pp.462-469). Amsterdam: SICSAT.
Deppermann, A. (1997). Glaubwürdigkeit im Konflikt. Rhetorische Techniken in Streitgesprächen. Frankfurt am Main: Lang.
Deppermann, A. (1999). Gespräche analysieren. Opladen: Leske + Budrich.
Heritage, J. (1995). Conversation analysis: methodological aspects. In: U.M. Quasthoff (Ed.), Aspects of oral communication (pp395-418). Berlin: de Gruyter.
Nothdurft, W. (1996). Schlüsselwörter. Zur rhetorischen Herstellung von Wirklichkeit. In: W. Kallmeyer (Ed.), Gesprächsrhetorik. Rhetorische Verfahren in Gesprächsprozessen (pp.351-418). Tübingen: Narr.
Pomerantz, A. (1984). Agreeing and disagreeing with assessments: Some features of preferred/dispreferred turn shapes. In: J.M. Atkinson &  J. Heritage (Eds.), Structures of social action: Studies in conversation analysis (pp.57-101). Cambridge: Cambridge University.
Powers, L.H. (1995). Equivocation. In: H.V. Hansen & R.C. Pinto (Eds.), Fallacies: classical and contemporary readings (pp.287-301). Philadelphia, PA: Pennsylvania State University.
Spranz-Fogasy, T. (1999). Interactional resources of argumentation. This volume.
Spranz-Fogasy, T. & Fleischmann, T. (1993). Types of dispute courses in family interaction. Argumentation 7, 221-235.
Swinney, D.A. (1979). Lexical access during sentence comprehension: (Re)consideration of context effects. Journal of Verbal Learning and Verbal Behavior 18, 654-660.
Walton, D.N. (1996). Fallacies arising from ambiguity. Dordrecht: Kluwer.
Woods, J. & D.N. Walton (1989). Equivocation and practical logic. In: J. Woods & D.N. Walton, Fal-lacies. Selected papers 1972-1982 (pp.195-207). Dordrecht: Foris.