INTRODUCTION 1/ In the past, I have voted Republican. 2/ But this November, I'm voting for Doug Wilder. 3/ Marshall Coleman wants to take away the right of a woman to choose, 4/ even for poor women who are the victim[s] of rape and incest. --Commercial for Virginia Democrat Doug Wilder Taken from a television commercial for Virginia Democrat L. Douglas Wilder, the excerpt above can be seen as a statement as to why the speaker plans to vote for Wilder and an attempt to persuade the viewer to do so as well. Note that understanding the excerpt in this way requires a good deal of knowledge on the part of the viewer. This includes a working knowledge of the English language -- the meanings of the words and the grammar to combine them -- and a certain amount of cultural knowledge. For example, the argument presupposes that people vote according to party loyalty and appeals to the idea that voters may cross party lines on the basis of the candidates' positions on the abortion issue. Moreover, it seems to rely to a large degree on the case of rape and incest victims being somewhat pivotal in arguments about the morality of abortion. Moral arguments of this kind can be seen as paradigmatic of the use of language to convey information, shape attitudes, and to exert some degree of control over the actions of others. An investigation of how interacting agents utilize cognitive and linguistic resources to understand and formulate arguments about the morality of abortion, the current approach is an exploration of the role of cultural knowledge, rhetorical strategies, and pragmatic aspects of language use in Americans' arguments concerning abortion. The section below introduces the notion of a pragmatic scale (Fauconnier, 1975) and explicates how the truth-functional properties of statements interpreted on a pragmatic scale can be used to further rhetorical goals. In particular, I examine the use of linked pragmatic scales in a technique known as the double hierarchy argument. In the third section, the notion of cultural models is introduced as a means of capturing background assumptions which speakers rely on in the establishment of linked pragmatic scales. Finally, the section on reasoning with cultural knowledge contains analyses of excerpts from informants' discussion of the acceptability of abortion in the case of a pregnant rape victim. Informants' arguments are examined in terms of their reliance on cultural models outlined in earlier sections and their use of pragmatic scales to further their argumentative ends. Overall, the analysis demonstrates how pragmatic scales and cultural models support rhetorical interaction concerning the morality of abortion. PRAGMATIC SCALES AND THE DOUBLE HIERARCHY ARGUMENT Because the inferential possibilities of our utterances are so closely tied to conventions for use and interpretation of language, rhetorical efforts succeed best when they exploit culturally relevant semantic and pragmatic aspects of everyday language understanding. For example, a variety of arguments depend upon the speaker's ability to exploit the properties of *pragmatic scales*. A pragmatic scale consists of objects or scenarios ordered along relevant semantic dimensions. Once ordered, statements which concern one member of the scale entail propositions about other members of the scale. Consider the sentence: Michael Jordan makes the hardest shots. To say that Jordan makes the hardest shots is also to implicate that he makes the less difficult ones as well. This implicature falls naturally out of interpretation on a pragmatic scale. To interpret this sentence one can construct a pragmatic scale of shots ordered by their difficulty: --- "the easiest" | --- x1 | --- "the hardest" Because interpretation on a pragmatic scale involves the propagation of truth values, to assert something about an object entails assertions about objects which fall above it on the scale; to deny something about an object entails denials about objects which fall below it on the scale. Thus asserting that Jordan made the x1 shot entails his making all of the shots which occur above it on the scale ("the easiest"). By contraposition, denying that Jordan made the x1 shot entails his not making the shots which occur below x1 on the scale ("the hardest"). The inferential properties of statements interpreted on a pragmatic scale lend themselves quite naturally to a variety of argumentative strategies. The most obvious of these is argument *a fortiori*. For example, if one wanted to argue that Michael Jordan could make a particular shot -- say x1 -- one could do so by making the statement discussed above: Michael Jordan can make (even) the hardest shot. Because this statement admits an interpretation on the pragmatic scale described above it generates the inference that Jordan can make x1. That is: Jordan can make the hardest shot; so, *a fortiori*, he can make x1. Pragmatic scales can also be used in the rhetorical device known as the *double hierarchy argument* described by Perelman and Olbrechts-Tycha (1969). In a double hierarchy argument a speaker uses the ordering principle of a well-established pragmatic scale to create a second pragmatic scale whose points are ordered in the same way. Once the two scales are linked, inferences that hold on the first pragmatic scale are assumed to hold on the newly established scale. Linking two scales in this manner affords the possibility of moving from inferences generated on the least controversial scale to parallel inferences on the more controversial linked scale. Examples of the double hierarchy argument are contained in the section on reasoning with cultural knowledge. Both the possibility of ordering a pragmatic scale and of establishing links between scales relies on the existence of a set of shared beliefs within a community of speakers. These beliefs serve both as a precondition for pragmatic mechanisms and as the essential elements in an argument. But what is included in this set of beliefs and how are they organized? In the following section the notion of cultural models (Quinn and Holland, 1987) is introduced as the way in which the shared beliefs of a community are organized. CULTURAL MODELS Derived from the notions of scripts (Schank and Abelson, 1975), frames (Minsky, 1975), and schemas (Rumelhart, 1980), cultural models are defined as representational structures which are shared by members of a given culture. Further, they are hierarchically structured so that elements can often be expanded into their own models (Casson, 1983, cited in D'Andrade, 1987). Because of their efficiently structured organization and their status as shared representational structures, cultural models figure widely in both language and thought processes. In one sense cultural models are an old product wrapped up in a new package. The definition of cultural models retains essential elements of scripts (Schank and Abelson, 1975), frames (Minsky, 1975), and schemas (Rumelhart, 1980), while emphasizing their intersubjective nature and cultural origins. Although Schank and Abelson have never addressed the cultural origins of scripts, Quinn and Holland (1987) argue convincingly that scripts are cultural knowledge. The lack of variation among culture members' description of scripts shows their cultural origins. For if knowledge of scripts were based on personal experience, one would expect more diversity than is evidenced by informants. Another sort of cultural models is the idealized cognitive model (ICM) as proposed by Lakoff (1987). Similar to a script, an ICM consists of a standardized sequence of events in a pared-down world. Although ICMs are widely shared among culture members, they need not correspond to anything concrete in the external world. An example can be found in Sweetser's (1987) discussion of what she calls the simplified speech act world. In this world people speak in order to communicate information which might be helpful to one another; their beliefs are adequately justified (and, as a result, are true); finally, people say what they believe. Obviously, the simplified speech act world is an idealized construct. People do not always say what they believe, nor do they always speak with the intention of providing helpful information. Nonetheless, the simplified speech act world is an efficient representation for reasoners to use in the definition of speech acts that do NOT fit into the model. Sweetser (1987) provides a unified account of the semantics of the word lies and some related phrases (e.g. social lies, white lies, and mistakes) by showing how speakers might employ the speech act model to note deviations from its component parts. Moreover, one can see from its resemblance to Grice's (1975) maxims, that the model is used implicitly by speakers to understand the pragmatic implications of each others statements. While speakers use ICMs in their language and reasoning, they do not explicitly believe the tenets of these models. Although speakers are well aware of the idealized character of these models, they are often not aware of the extent to which they rely on them in language and reasoning. Nonetheless, speakers' awareness of the idealized character of some of the models they use frequently causes disputes over the extent to which the inferences generated by these models will be accepted. A further example of a cultural model is provided by D'Andrade (1987) in his exploration of the American folk model of the mind. Describing the American common sense view of the mind, D'Andrade details what sorts of states and processes occur in the mind (e.g. beliefs, desires, perceptions, intentions, resolutions, and emotions); the syntax of the relevant terms; the relationships that can exist between internal states and external actions; and also, the relationships that can exist among the internal states themselves. Cultural Models of Action, Responsibility, and Punishment The idealized model of action, for example, concerns an agent with an intention, who performs an action which has consequences. The belief that action A will produce consequence C and the desire to produce consequence C constitute the intention to produce consequence C. On the model, intentions cause actions and actions cause their intended consequences. This model is what underlies Americans' (and D'Andrade suggests the model holds for many Europeans as well) interpretation of their own and each other's actions. Our default assumption is that all human acts are caused by intentions. If we observe a person performing an action (e.g. opening a door) we assume that the act is caused by the person's intention (the intention to open the door) . We don't mention intentions in our description of actions because it is implicit to the very notion of action that it be mediated by an intention. It is not adequate to explain an action by pointing to the intention which caused it. Because all actions are assumed to stem from intentions this information is seen as trivial. However, it is considered relevant to specify the cause of the intentions. Because intentions can be caused by desires, wishes, needs, beliefs, and combinations of these, one might explain another person's actions by recourse to the desire which gave rise to the intention, which, in turn, gave rise to the action. Alternatively, one might simply make reference to the external cause which gave rise to the desire (wish, need, etc.) which, in turn, gave rise to the intention and, consequently, to the action. The concept of responsibility is based on this idealized model of action. An agent is responsible for her actions because they are caused by her intentions. Similarly, an agent is responsible for the consequences of her actions, first, because of the causal link between the agent and the action on the idealized model of action; and, second, because actions and consequences are causally linked. The attribution of praise, blame, and punishment are based on the concept of responsibility and consequently involve a similar use of this idealized model of the mind. One receives praise, blame, and/or punishment as a consequence of responsibility for action. If the action is judged as having good consequences the actor responsible receives praise; if the action is judged as having bad consequences, the actor who is responsible receives blame and, possibly, punishment. The cultural model of just punishment is generated when the action frame is instantiated by an agent in such a way as to yield the negative evaluation of others. In such a case, the model dictates that the agent responsible requires punishment. The crucial aspect of this model is that the recipient of punishment be coextensive with the agent responsible for the action and consequence, otherwise the cultural model of unjust punishment applies. Note that there is no a priori reason why an agent should be responsible for the consequences of his actions. The attribution of responsibility to the agent results from the use of the idealized models of mind and action. Agents are accountable for the consequences of their actions because, on the model, acts committed are causally linked to the consequences by the knowledge of the outcome and the desire for that outcome. The attribution of responsibility, and, by extension, the attribution of praise and blame, becomes problematic when the assumptions of the idealized model are violated. The main case of this involves actions with unintended consequences. In such a case, the normal method of assessment of responsibility is inappropriate because the identification on which it rests (that between intentions and consequences) does not obtain. One way of getting around this is to assess responsibility on the basis of the consequences the action was intended to produce. This is determined by exploring the agent's beliefs about the consequences of her action and drawing inferences about the desires which produced the intention. Note that the attribution of responsibility in cases which deviate from the default model of action is still based on the use of the model linking intentions to actions and actions to consequences. In the case of unintended consequences, for example, the existence of a dissociation of the customary link between intentions and consequences does not result in the rejection of the idealized model of mind. Rather it results in a shift of focus from the action's consequences to the agent's intentions. Because, in the default case, the assessment at the level of consequences is only an efficient means of accessing the agent's intentions, we can see this shift as the use of different components of the model in the same way. REASONING WITH CULTURAL KNOWLEDGE: The Rape Exception In the following section we look at how speakers use cultural models in their reasoning as they attempt to persuade one another about the morality or immorality of abortion. Arguments both for and against the rape exception are analyzed to reveal what models people use in their reasoning and how they manipulate those models to serve their argumentative ends. Arguments for the Rape Exception Asked to justify abortion in the case of rape, informants point to two sorts of things. The first is related to the undesirable effect that pregnancy and childrearing will have on the rape victim. These include points such as: o the mother will not want a child which is the result of rape; o the child will remind the mother of the rape; o bringing up the child will be painful for the mother. Moreover, the trauma caused by pregnancy and childrearing in such a case is linked to undesirable ramifications for the quality of the child's life. That is: o the mother will not do a good job of bringing up such a child; o both the child and the mother will suffer from such an arrangement. The second type of thing informants point to has to do with the unintended nature of the pregnancy caused by rape. An example of the latter sort of response can be found in the following excerpt from an interview with a young woman in her early twenties. Asked whether or not it would be wrong for a rape victim to obtain an abortion, P responds: (1a) 1/ S: OK, how about um J is 17 years old and is pregnant as result of a rape. 2/ Would it be wrong for her to have an abortion? 3/ P: No. . . . Because it wasn't her choice. 4/ It was something she couldn't prevent. 5/ And there's no use traumatizing her with pregnancy 6/ and um being young and being pregnant. 7/ Having this child um, 8/ I mean I would have a very hard time. 9/ I would probably have an abortion if I were raped. 10/ . . . No matter what the age I would support abortion. In order for us to see the claims made by P as a justification for the rape exception they must be seen as functioning against the background of a set of taken-for-granted beliefs which provide an explanatory framework for P's specific claims (Hutchins, 1980). This framework includes a number of cultural models and conventions for negotiating the inferences which the models can be used to generate. Thus construed, P's arguments can be seen as denying the applicability of the ICM of action and, consequently, the applicability of the ICM of responsibility. In response to the interviewer's question, P denies that it would be wrong for a rape victim to obtain an abortion, and supports her claim with the assertions in lines 3 and 4. These assertions consist first, of the denial that the rape victim chose to get pregnant and, second, the denial that she could have prevented pregnancy. P has thus invoked the model of action discussed earlier, in which an action is the result of an agent's intentions. By denying the rape victim's intention to get pregnant, P's move in lines 3 and 4 is to deny that the sex which occurs in rape constitutes an action on the part of the woman. Note that this move rests on the existence of a cultural model of rape in which sex occurs entirely as the result of the man's intentions and contrary to those of the woman. P's denial of the applicability of the ICM of responsibility suggests that it is at least possible for someone to make a claim for its applicability, either specifically in the case of rape or to unwanted pregnancy in general. Note the similarity of the schema for the attribution of responsibility in the cultural model of pregnancy to the schema for the attribution of punishment. What, under normal circumstances, is considered the attribution of responsibility for one's actions can also be portrayed as the attribution of punishment. Whereas pregnancy is normally considered the consequence of an action (sex) for which one is responsible, the denial of the applicability of conventional ICMs of action and responsibility make such a conceptualization untenable in the case of a rape victim. P suggests that to require the rape victim to continue her pregnancy would be to punish her. To punish her for her pregnancy would be to blame her for her pregnancy. Moreover, to blame her for her pregnancy would be unfair because she had no choice in getting pregnant, and it is unfair to blame someone for something in which they have no choice. Further, the move from dispute of the legitimacy of pregnancy as punishment to support for abortion is non-trivial. In order to construe this move as coherent we must postulate the existence of a model of abortion as a means for preventing ineffective and unjust punishment. (1b) After P answers the question about abortion in the case of rape, she is asked about the case of incest. She responds as follows: [S asks about the case of incest.] 1/ P: More of a reason probably. 2/ S: More of a reason. Why is it more of a reason? 3/ P: Because then you're dealing with rape and you're dealing with incest. 4/ . . . I have a feeling it would be more traumatizing to be raped by a member 5/ of your family. P's statement that incest is more of a reason for abortion than rape is supported by a double hierarchy argument. In the double hierarchy argument a speaker introduces a relationship between terms in a disputable hierarchy supported by the relationship between terms in an accepted hierarchy. The double hierarchy argument rests on there being a sensible connection between the two hierarchies which warrants the transference of the superiority relationship. This relationship can be sequential (as in cause and effect, means and end) or coexistence (when the elements in one of the hierarchies are a property of the elements in the other hierarchy). Further, this sort of argument requires previous agreement about at least one of the hierarchies. In this argument P sets up a hierarchy of traumatic pregnancies and links it to a hierarchy of the acceptability of abortionin those cases. The pragmatic scale of pregnancies ordered by the trauma they cause the woman is linked here to a scale of pregnancies ordered by the acceptability of abortion in each case in an argument from double hierarchy. Because the hierarchy of traumatic pregnancies is relatively uncontroversial, it can be used to order the more controversial hierarchy of acceptable abortions. The link between the two hierarchies is the model of abortion as a means to the prevention of the trauma caused by unwanted pregnancy. Because inferences which hold on the initial scale are assumed to hold on the linked scale, one can imply statements interpreted on the scale of acceptable abortions simply by making statements interpreted on the scale of traumatic pregnancies. P has already established that rape is traumatic and that it creates a very strong case for abortion. In order to use the argument from double hierarchy, P shows that incest is a more traumatic cause of pregnancy than is rape. She does this by noting first a quantitative difference between the two cases: the rape victim must deal with the trauma of one sort of event (rape), while the incest victim must deal with the trauma of two sorts of events, (both rape and incest). Moreover, she follows by describing a qualitative difference between the two sorts of rapes. So, having established rape as a case which warrants the acceptability of abortion, P's arguments as to why incest is more traumatic than rape locates incest as higher on the scale of traumatic pregnancies. Because inferences which hold on one pragmatic scale in the double hierarchy argument are assumed to hold on the scale to which it is linked, locating incest higher on the trauma scale generates the inference that incest, like rape, warrants the acceptability of abortion. Argument from double hierarchy arises out of our need to compare things to one another and the existence of ordering principles. Further, this sort of argument is bolstered by our ability to establish one item into the context and then to compare it to another item interpreted on the same pragmatic scale. Thus the same cognitive capacities invoked to interpret statements on pragmatic scales are also used to understand the double hierarchy argument. Arguments Against the Rape Exception (2) Nathanson 1/ When pregnancy is confirmed, 2/ one can no longer discuss morality in the framework of 3/ rape or incest 4/ (or whether birth control was or was not used, 5/ or whether the child is 'wanted' or whatever). 6/ The unwanted pregnancy flows biologically from the sexual act, 7/ but not morally from it. 8/ . . . even degradation, shame and emotional disruption 9/ are not the moral equivalent of life. 10/ Only life is. --Bernard Nathanson [quoted in Whose Life?] Nathanson's argument can be seen as aimed at rebutting two double hierarchy arguments often used to substantiate abortion in the case of rape, incest, and various other unwanted pregnancies. The first of these double hierarchy arguments links the moral evaluation of the cause of pregnancy to the moral evaluation of the pregnancy itself. The second argument links the trauma which the pregnancy causes the woman to the acceptability of abortion in that case. Hierarchy from Cause to Effect: Nathanson's Rebuttal The first argument which Nathanson seeks to rebut links the accepted hierarchy of the moral status of the cause of pregnancy to a hierarchy of the desirability of the pregnancy itself. The link between the two hierarchies in this argument is a liason of succession -- the relationship between a cause and its effect -- and is based on the cultural model of sex as the cause of pregnancy. Thus pregnancy which results from an illicit sexual encounter is morally contemptuous and should be eliminated, whereas pregnancy which results from the desire to provide a loving home for a child has a high value. To challenge a double hierarchy argument one can either dispute the placement of one or more terms in the accepted hierarchy, or dispute the liaison that connects the two. Nathanson chooses the latter strategy, arguing (in lines 6 and 7) that while the rapist causes the pregnancy, the morality of the rapist does not cause the morality of the pregnancy. This constitutes a challenge to the first double hierarchy argument discussed above, in which the moral evaluation of the cause of pregnancy is linked ot the moral evaluation of the pregnancy itself. Nathanson does not dispute the causal nature of the link between the two hierarchies, but the legitimacy of the transference of moral judgments from cause to effect. The moral responsibility engendered by pregnancy does not originate in the action which causes it. Rather, pregnancy itself causes responsibility. With the statement in lines 8 through 10, Nathanson invokes the technique of equalization to show the irrelevance of the hierarchy of the means of getting pregnant to the hierarchy of the acceptability of abortion. Equalization is a rhetorical technique used to transform differences of kind to differences of degree or intensity (Perelman and Olbrechts-Tycha, 1969). This technique is used to accentuate the similarity between two different kinds so that the inferences and moral consequences that obtain for one will come to obtain (to a certain degree) for the other. Specifically, Nathanson wants to equate abortion in the case of rape or incest to the case of abortion in a planned 'wanted' pregnancy. Whereas the morality of abortion is ambiguous in cases of incest and rape (and thus has to be argued for), it is unambiguously immoral in the case of a planned pregnancy. Thus, if we can come to see the two cases as equivalent on some dimension that brings the case of unwanted pregnancy closer to the case of planned pregnancy, we will be more likely to judge the ambiguous case similarly to the way we judge the unambiguous case of a planned pregnancy. Hierarchy from Trauma to Acceptability: Nathanson's Rebuttal Besides rebutting the double hierarchy from cause to effect, Nathanson's argument in this passage can also be seen as a challenge to another argument from double hierarchy similar to that invoked by P in example (1b) above. In this argument, an established hierarchy of traumatic pregnancies is linked to the acceptability of abortion. The link between these two hierarchies is a cultural model of abortion as the means to preventing the trauma caused by the pregnancy in each case. Marked by *even*, the statement in lines 8 and 9 indicates that the stated proposition (". . . shame, degradation, and emotional disruption are not the moral equivalent of life,") is more informative than some contextually established proposition interpreted on the same pragmatic scale (Kay, 1990). In order for the statement to be considered coherent we must view it as a rebuttal to the double hierarchy argument of trauma and acceptability. Because that argument relies on the model of abortion as the means to the prevention of trauma caused by continuing the pregnancy, it establishes shame, degradation, and emotional disruption as morally relevant factors which abortion will prevent. Nathanson's statement, then, relies on the existence of the contextually established double hierarchy (trauma to acceptability) which set up a pragmatic scale of morally relevant factors which abortion will prevent. However, in order for the ends to justify the means, the ends must be morally equivalent to the means. This requires a comparison of the moral factors which are involved. However, with the statement in line 10 Nathanson argues that such a comparison is ludicrous. Equalization is used here by Nathanson in a way that it is often used: to invoke a deontic principle which precludes a strict utilitarian interpretation of the situation (Perelman and Olbrechts-Tycha, 1969). The possibility of ordering pregnancies by the amount of trauma they cause the woman sets us up for the idea that abortion can be more or less acceptable. To Nathanson, this idea is simply untenable. Life is presented as something which cannot be placed on this particular pragmatic scale. Nathanson's argument, besides rebutting the double hierarchy from trauma to acceptability, serves to erect yet another double hierarchy. In Nathanson's double hierarhcy, a hierarchy of the moral importance of the effects of continuing pregnancy are linked to the acceptability of abortion. Linking the two hierarchies is the model that the effect which is most important should determine the acceptability of abortion. By inviting the inference that the moral importance of continuing life makes abortion unacceptable, it entails the unacceptability of abortion regardless of the effects of continuing pregnancy. Note that in the establishment of this double hierarchy, Nathanson makes use of the model of abortion as a means to preventing an undesirable end. However, by introducing the deontic concept of life he also makes use of a model of abortion in which the procedure, as the termination of the life of the fetus, is an illegitimate end in itself. (3) Ron Reagan Show The excerpt below concerns a discussion between Ron Reagan, a talk show host, and Joseph Scheidler, the executive director of Pro-Life Action, with an interjection from Randall Terry, director of the antiabortion activist group Operation Rescue. Comments must be understood in terms of their relationship to the speakers' various positions on the issue and their goals in the conversational exchange. As the host of the talk show, Reagan's primary goal is to provoke interesting discussion. This involves posing challenging questions to guests and attempting to elicit controversial statements from them. However, he is also obligated to satisfy time constraints such as making time for commercial advertising and controlling turn-taking to allow all guests to speak. On the other hand, Scheidler's presence on the show is as a representative of a pro-life organization. His primary goal is to inform the audience of his position and to persuade them of the morality of that position. This obligates him to argue convincingly for his own position and to belittle the arguments of his opponents. 1/ Reagan:OK. Joseph, if one of your daughters is raped and becomes pregnant, 2/ should [she] be able to have an abortion? 3/ Scheidler: I become a grandfather then. 4/ Reagan: If, if, god forbid, and I'm not suggesting this is a possibility, 5/ you committed incest with your daughter and she became pregant-- 6/ Terry: The professionalism of this show is just going down the crapper. 7/ Reagan: No, it happens. 8/ Incest is a reality and girls become pregnant because their fathers impregnate them. 9/ Scheidler: Ron, that is the most unnatural thing I can think of. 10/ Reagan: I agree but the, 11/ should the result of that unnatural act be expunged in any way? 12/ Scheidler: It is so repugnant, but to answer your question, 13/ I would never kill a child because of the sin of the father. 14/ I just would not do it. 15/ Reagan: That's all I wanted to know. 16/ Got to go to a commercial. Be right back. 17/ commercial break 18/ Reagan: OK, we're back with our discussion. 19/ Is there any circumstance, Joseph, in which it is proper to abort a fetus, 20/ severe birth defects for instance. 21/ Scheidler: No. . . . (from The Ron Reagan Show, "Abortion: The Hidden Agenda"; October 8, 1991) In this excerpt, Reagan first asks Scheidler if abortion is acceptable in the case of rape, then follows with the case of incest. Following a commercial break Reagan asks Scheidler if there is any circumstance in which abortion is permissible and offers birth defects as a possible answer. As we have seen above, the cases of rape and incest are more ambiguous as to the acceptability than, say, abortion in a case where no rape has occurred. People who are otherwise opposed to abortion are most likely to admit the acceptability of abortion in the case of rape or incest than any other. Thus, Reagan's question is relevant in spite of the fact that we know that Scheidler is pro-life and so opposes abortion. Reagan's first question is pertinent, not simply because it reveals Scheidler's position on abortion in the case of a rape victim, but because of how it can be interpreted in the context of a pragmatic scale of reasons for abortion. Because rape figures high as a reason for abortion, answering 'no' to Reagan's question entails a 'no' to the same question in the majority of other cases. Because of what a 'no' entails, Scheidler's answer will most certainly offend audience members who think abortion is generally acceptable. Moreover, it will also offend those people who oppose abortion generally, but find it acceptable in the case of a rape victim. Reagan's question can be seen as accentuating for the audience the extremity of Scheidler's position. Further, the question can also be seen as baiting Scheidler to reveal the extent of his beliefs about the immoral nature of abortion. Note also that Reagan does not ask about rape in the abstract, but in the case of the rape of Scheidler's own daughter. This appeals to the empathic basis of the rape exception (as seen in P's discussion) and supports the idea that Reagan's question is aimed at enticing Scheidler to admit to the rape exception. Scheidler's response, although it does not constitute a direct answer to the question, implies a negative answer to Reagan's question. However, because it requires the listener to invoke a number of cultural models it does not admit of straightforward interpretation on the pragmatic scale. First, we know that one becomes a grandfather when his son or daughter has a child. However, Reagan has not brought up a scenario in which Scheidler's daughter has a child, but one in which she is pregnant. In order for Scheidler's statement to make sense, he must believe that pregnancy is equivalent to having a child. This belief is also consonant with his conviction that it is immoral to terminate a pregnancy. Moreover, it is wrong to kill one's grandchild. So, if in this case, abortion is killing one's grandchild, the daughter should not be able to abort. Fetus = Child Abortion kills fetus. therefore, Abortion kills child. Fetus = Child Scheidler's Daughter's Child = Scheidler's Grandchild Scheidler's Daughter's Fetus = Scheidler's Daughter's Child Scheidler's Daughter's Fetus = Scheidler's Grandchild If Reagan's question, by personalizing the situation for Scheidler, was intended to appeal to Scheidler's sense of compassion for his daughter, Scheidler's response employs a parallel strategy of showing the familial relationship which he sees as existing between himself and the fetus. Scheidler's response both highlights the existence of the fetus and implies that it should somehow factor into the equation. Reagan's strategy of personalizing the scenario of the rape victim backfires when he extends it to the incest victim. Whereas the strategy in the first case was intended to make S empathize with the plight of the rape victim by suggesting that his own daughter might be in such a position; implying that Scheidler's own daughter might be the victim of incest serves only to offend Scheidler because (if only hypothetically) he is assigned the role of the incestual rapist. Scheidler's response attempts to dismiss the question on the grounds that it is far-fetched. However, Reagan substantiates those grounds by arguing that the possibility of pregnancy as the result of incest is not far-fetched. Moreover, he depersonalizes the question by moving to general terms. Scheidler's second response is that he is reluctant to answer the question because to do so would involve considering a scenario he finds repulsive. Reagan, in order to get an answer to his question, says, "I agree, but, the should the result of that unnatural act be expunged in any way?" Reagan has alluded to the double hierarchy which links the morality of the cause to the morality of the effect which is challenged by Nathanson's statements in (5). Reagan's choice of words here may be motivated by the desire to provoke Scheidler, or, alternatively because they reflect his view on the issue. Note, for example, that the fetus is referred to in a dehumanizing manner ("the result of that unnatural act") bound to offend Scheidler. Moreover, the abortion is not being construed (by Reagan) as a medical procedure which results in the loss of life, but rather as the ablation of an earlier moral wrong. Note that Reagan's and Scheidler's statements reflect differing conceptions of the ontology of the fetus. Although they share the proposition that abortion eliminates the fetus, their differing cultural models have different inferential properties and correspondingly different moral implications. Reagan fetus=result of unnatural act abortion=expunging incest Scheidler fetus = child abortion= killing child Reagan's frame presupposes that pregnancy as the result of incest never should have happened in the first place. Thus abortion has the effect of making the situation equivalent to one in which incest had never happened, and thus of expunging an unnatural act. On the other hand, Scheidler's frame presupposes that all pregnancies (including those that result from incest) represent the beginning of life. The idea of pregnancy as an event which never should have happened is thus less plausible. Instead, the pregnancy that results from rape or incest is the good outcome of a bad event. Construed in this manner, abortion to expunge an unnatural act is misdirected. Further, Scheidler's statement in line 13 (I would never kill a child because of the sin of the father) serves to instantiate the cultural model of unjust punishment diagrammed below. The father who commits incest with his daughter fulfills the role of offender; the fetus fulfills the role of non-offender; the abortion constitutes the punishment. The instantiation of this model generates the inference that the abortion, construed as unjust punishment, is immoral. Discussion of knowledge structures such as frames all too often proceeds as if these structures are devoid of affective content. Or, if affective content is acknowledged it is dismissed as irrelevant to the computational procedures in which it is employed. However, it should be clear from this analysis that the affective dimension of frames is both real and relevant. The choice of which frame/metaphor to invoke has important consequences for the moral judgments which we will make of the situation. CONCLUSIONS Analysis reveals that simple computational structures such as pragmatic scales and the cultural models which produce them can be seen to underly the more complex rhetorical interactions that occur in argumentation about the morality of abortion. Cultural models can be used to infer what happened, to make some sort of moral assessment of what happened, or for the attribution of responsibility and the assignment of praise, blame, or punishment. Moreover, the various uses of the model are not orthogonal to one another. One may use a model to simultaneously infer what happened, assess what happened, attribute responsibility, and assign praise or blame. Further, one can argue against the attribution of responsibility as determined by the use of a particular model. This can be done by appealing to a foundational model -- a model which the particular model is based on -- or by appeal to another model. The rape exception results from an appeal to a foundational model. The responsibility model which normally applies to pregnant women does not apply to the pregnant rape victim because it is possible to i) appeal to the action model upon which responsibility is based, and ii) to show that the action model does not apply. An example of recourse to another model is appeal to the victim frame or the unjust punishment frame. Oftentimes, both strategies named above can be used simultaneously to counter the judgments which result from the use of a model. The ability to interpret statements in the context of a pragmatic scale allows us to compare things to one another. Further, statements that can be interpreted on a given pragmatic scale generate inferences about other propositions which can be interpreted on the same scale. This property of pragmatic scales can be exploited by pointing to the truth of a statement low on the scale; if we then locate another statement higher on the scale we can infer the truth of the proposition which corresponds to that point. This facet of pragmatics was then shown to underly the rhetorical strategy called the double hierarchy argument. In a double hierarchy argument the speaker uses the ordering principle for one pragmatic scale to establish a second scale whose points are ordered in the same way. Once the two models are linked, the inferences that hold in the original scale are assumed to hold in the newly constructed scale. Thus the speaker can exploit the inferential possibilities of propositions interpreted on the first pragmatic scale to yield inferences in the second scale. Thus P in (1b) uses statements about the trauma which pregnancy causes the rape victim to yield inferences about the acceptability of abortion in that case. Pragmatic scales are grounded in the use of mutually available cultural models of social reality. Although we may have the abstract ability to set up an arbitrary scale, such a scale will not function in our arguments unless it resonates with the cultural models that exist in one's language community. Moreover, once a cultural model has been invoked to warrant the acceptance of a particular proposition, e.g. the models of rape and responsibility used to warrant the acceptability of abortion, they can be used to link pragmatic scales in a double hierarchy argument. It is as if the original use of cultural models establishes the existence of a relationship between the ordered terms, e.g. trauma caused by pregnancy and the acceptability of abortion, which can then be exported to the two scales. Note that the double hierarchy argument is not irrefutable. A dimension can be added to linked scales to counter the original inferences which they generated. Similarly, scales in an existing double hierarchy can be supplemented with another double hierarchy which counters the inferences obtained with the original. These techniques direct our attention to the cultural models which link scales and thereby serve as the warrant for the transfer of inferences from model to model. Because the models are almost always simplified models of social reality, they just as frequently admit of supplementation by other simplified models. Manipulating the scales has the effect of casting and recasting the models which link them. In contrast to the traditional picture of argument as the consideration of a series of points, the participants in the arguments examined here can be seen to be engaged in either building or avoiding pragmatic scales and appealing simultaneously to a whole host of propositions which issue from the use of cultural models. Because meaning occurs in the context of cultural models whose inferential possibilities are somewhat indeterminate, disputes can arise over the particular inferences which will be allowed to issue from the particular models employed. The dynamics of conversation consist of negotiating the extent to which the inferential possibilities of stated propositions will be allowed or disallowed. We have seen how people with different ideological alliances work with the same cultural models in different ways. Differential manipulation of the models results in the joint construction of sociocultural reality. Moreover, the differential emphasis in the relevance of different aspects of various cultural models employed by speakers in conversation serves to negotiate the salience of the different points in an argument. The negotiation which occurs in conversational debates can be seen as a microcosm of the dynamic negotiation of social reality which occurs in the larger society. REFERENCES Billig, Michael. 1987. Arguing and thinking: A rhetorical approach to socialsocial psychology. Cambridge, England: Cambridge University Press. Briggs, C.L. 1986. Learning how to ask: A sociolinguistic appraisal of the role of the interview in social science research. Cambridge, England: Cambridge University Press. D'Andrade, R. 1987. A folk model of the mind. In Cultural models in language and thought. Eds. D. Holland and N. Quinn. Fauconnier, G. 1975. Pragmatic scales and logical structure. Linguistic Inquiry. Volume IV: no. 3. Gentner, D. and D.R. Gentner. 1983. Flowing waters or teeming crowds: Mental models of electricity. In Mental models. Eds. D. Gentner and A.L. Stevens. Grice, H.P. 1975. Logic and conversation. In Syntax and semantics, Vol. 3, Speech acts. Eds. P. Cole and J. Morgan. Holland, D. and D. Skinner. 1987. Prestige and intimacy: The cultural models behind Americans' talk about gender types. In Cultural models in language and thought. Eds. D. Holland and N. Quinn. Holland, D. and N. Quinn. 1987. Cultural models in language and thought. Cambridge, England: Cambridge University Press. Hutchins, E. 1980. Culture and inference: A Trobriand case study. Cambridge, MA: Harvard University Press. Hutchins, E. 1987. Myth and experience in the Trobriand Ilsands. In Cultural models in language and thought. Eds. D. Holland and N. Quinn. Johnson, M. 1987. The body in the mind: The bodily basis of meaning, imagination, and reason. Chicago: University of Chicago Press. Kay, P. 1990. Even. Linguistics and Philosophy 13: 59-111. Lakoff, G. 1984. Classifiers as a reflection of mind: A cognitive model approach to prototype theory. Berkeley Cognitive Science Report No. 19. Berkeley: University of California Institute of Human Learning. Lakoff, G. 1987. Women, fire, and dangerous things: What categories reveal about the mind. Chicago: University of Chicago Press. Lakoff, G. and M. Johnson. 1980. Metaphors we live by. Chicago: University of Chicago Press. Lakoff, G. and Z. Kovesces. 1987. The cognitive model of anger inherent in American English. In Cultural models in language and thought. Eds. D. Holland and N. Quinn. Perelman, C. and Olbrechts-Tycha, L. The new rhetoric: A treatise on argumentation. Translated by John Wilkinson and Purcell Weaver. Notre Dame, IN: University of Notre Dame Press. Quinn, N. 1987. Convergent evidence for a cultural model of American marriage. In Cultural models in language and thought. Eds. D. Holland and N. Quinn. Quinn, N. and D. Holland. 1987. Culture and Cognition. In Cultural models in language and thought. Eds. D. Holland and N. Quinn. Quinn. N. 1992. Cultural Models of Marriage. Schank, R. and R. Abelson. 1977. Scripts, plans, goals and understanding: An inquiry into human knowledge structures. Hillsdale, N.J.: Lawrence Erlbaum Associates. Schiffrin, Deborah. 1987. Discourse markers. Cambridge, England: Cambridge University Press. Stubbs, Michael. 1983. Discourse analysis: The sociolinguistic analysis of natural language. Chicago: University of Chicago Press. Sweetser, E. 1987. The definition of lie: An examination of the folk models underlying a semantic prototype. In Cultural models in language and thought. Eds. D. Holland and N. Quinn. Sweetser, E. 1990. From etymology to pragmatics: Metaphorical and cultural aspects of semantic structure. Cambridge, England: Cambridge University Press. Sweetser, E. and G. Fauconnier. (1992, in press). Cognitive links and domains: Basic aspects of mental space theory.