Argumentation

, 23:21

Framing and Editing Interpersonal Arguments

Authors

    • University of Maryland
  • Ben Warner
    • Western Illinois University
  • Dorian Young
    • Western Illinois University
Article

DOI: 10.1007/s10503-008-9107-x

Cite this article as:
Hample, D., Warner, B. & Young, D. Argumentation (2009) 23: 21. doi:10.1007/s10503-008-9107-x

Abstract

Since argument frames precede most other arguing processes, argument editing among them, one’s frames may well predict one’s preferred editorial standards. This experiment assesses people’s arguing frames, gives them arguments to edit, and tests whether the frames actually do predict editorial preferences. Modest relationships between argument frames and argument editing appear. Other connections among frames, editing, and additional individual differences variables are more substantial. Particularly notable are the informative influences of psychological reactance. A new theoretical contribution is offered, connecting argument frame research to Erving Goffman’s frame analysis.

Keywords

Argument framesArgument editingArgumentativenessFrame analysisGenderGoffmanReactanceVerbal aggressiveness

1 Introduction

People may engage in interpersonal arguments with foresight or may simply be caught up in a spontaneous exchange of reasons. Once an argument is joined we bring to bear not only our content knowledge, topical attitudes, and habitual argument patterns but also our preconceptions about what we are doing and our predispositions about how to do it. This study is centrally about those last issues: the preconceptions, the predispositions, and the means of arguing. We are interested in discovering new relationships among these features of face-to-face arguing.

1.1 Frames

Goffman (1974) has provided us with an elaborate system of what he calls frame analysis. Some of his central ideas are frame, natural strip, and keying. His use of “frame” is taken from Bateson’s (1987/1955) classic observations of monkeys’ play-fights at a zoo. Bateson noticed that all the specific actions the animals took were fighting behaviors. But somehow the monkeys demonstrably understood that they were playing and not fighting. In other words, the animals imposed a play frame on fighting behaviors and so were really playing. The analogue to a picture frame occurs in both Bateson’s and Goffman’s work. By placing a particular frame around some central painting or object, it is set off from its surround and given a particular meaning. So a framed ribbon is more special than a ribbon lying on a table. Similarly a social behavior can have its meaning changed by a new framing.

Goffman calls the base actions a natural strip of behavior. Such a strip (e.g., eating a meal) can be directly experienced and understood or can be (re)framed in a particular way (e.g., as a demonstration of gourmet insight). The reframing is accomplished by means of keying. Goffman seems to have a musical analogy in mind here, such that the same melody can be put in another key where it is at once recognizable as and different than its base form or natural strip. A natural strip of behavior may be rekeyed and so put into a different frame.

The application of these ideas to interpersonal arguing is as follows. The natural strip of behavior under study is the face-to-face exchange of reasons, and this is done in order to resolve some disagreement or get one’s way. Sometimes such a strip can be reframed so that it becomes something else entirely (just as the monkeys’ actions genuinely became play). As we will see momentarily, arguing strips can be rekeyed so that they are really play or identity projections or dominance displays, and any of these can be fierce or friendly. Arguing strips could also be rekeyed into a movie or classroom frame where they become character developments or examples even further removed from any status as actual, personally consequential arguments. Play can be done by arguing just as it can be done by fighting. Each rekeying leads to what Goffman calls a lamination, a different layer of meaning that is ordinarily opaque to any beneath it. So arguing can be rekeyed into play, which can be rekeyed into dominance attempts, which can be manipulated into some sort of confidence swindle or practical joke. Many laminations are possible in practice, though the present analysis keeps things simple.

For Goffman the frames are more or less objective features of the interactive world—that is, they are really “out there” and assist in the organization of social life. He discusses in passing the likelihood that participants have some self-consciousness about what lamination they are in and what sort of keying accomplished it (particularly when he discusses swindles, practical jokes, and the like), but this is not a major preoccupation of his work. It is a major preoccupation here. We want to learn what framing possibilities people can see (that is, what keys they know about) and how they orient to various lamination possibilities. These matters are approached by direct study of arguing frames. While we understand that these frames are “out there” in the way Goffman assumes, here we are mainly interested in people’s perception and anticipation of these keying possibilities.

1.2 Arguing Frames

The preconceptions we study are the argument frames described by Hample (2003). These so-called frames summarize some of people’s leading expectations about the activity of arguing. When two people argue, what do they think they are doing? Do they think they are seeking personal gain? Do they think they should express themselves diplomatically or blurt the first thing that comes to mind? Such questions point us toward discovery of what laminations people position themselves to notice and toward a description of how (e.g., cooperatively or competitively) they orient to each keying. Hample (2003, 2005a) theorizes that there are three general categories of frames. These are in ascending order of sophistication, and perhaps also in order of developmental acquisition.

The first category is called primary frames, and these most immediately connect to Goffman’s conceptions. One’s primary frames are focused on self and on one’s life goals in the moment. A person might make an argument to avoid an errand or to encourage a loan, and might notice only that sort of utilitarian purpose for arguing. Although this practical motivation for arguing is the one most often noticed in argumentation literature, people may have other impulses. A primary frame is defined as being in play when self’s desires are the foci. Hample specifies four such frames: utility (using an argument to one’s advantage), dominance (arguing to display power over the other), identity (arguing to display some feature of self), and play (arguing for entertainment). These four were suggested because of their frequency or theoretical interest, and others may yet be added to the list.

The second group of argument frames involves whether and how one connects with the other arguer. In retrospect, they are not quite frames in Goffman’s sense but are closer to keys (Goffman 1974, pp. 43–45; Hymes 1972, p. 62; nonetheless Hample’s terminology will be retained here). They display how one goes about participating in any of the primary frames (and by being keys, hint at the possibilities for rekeying or transforming to a different frame). This group of frames is oriented to the fact that in an interactive argument, another active person is present and brings another set of motives and plans to the episode. However, people do not always acknowledge the other person in what we might loosely call a genuine way. Sometimes the other might as well be inanimate, and is apparently seen only as a foil, a means or obstacle to achieving one’s goals. So the first theoretical issue in this second kind of frame is whether or not the arguer even arrives at this stage. Thus, we examine blurting (speaking without planning) because blurts come simply out of cognition without alteration, and perhaps without any adaptation to the other arguer’s personal reality. Blurters never make use of the second sort of frame because they do not connect own goals to other’s. Other arguers do. For people who do make a conscious or unconscious effort to conjoin own impulses with other’s needs and rights, a key question is whether the attempted connection (the key) is cooperative or competitive. Both require genuine notice of the other person. However, cooperation is regarded as displaying a more sophisticated understanding of what people do when they argue. (This is not to say that people should always cooperate when arguing; it merely acknowledges that competition is the more basic urge, and that some people never notice the possibility of cooperating while disagreeing). The degree to which one expects that arguments are civil is a measure that straddles the second and third set of frames. Seeing arguments as uncivil and brutish is partly connected to whether one frames arguing as permitting polite cooperative interaction. But civility is a key part of an advanced third frame as well.

The third argument frame is the one that requires reflective consideration, a thoughtful theory of arguing. It raises the possibility of whether a person understands the possible frames and available keys, or is “contained” within a less reflective understanding (Goffman 1974). Arguers may intellectualize the activity of arguing. If they do so well, they achieve or approach the views about arguing that are held by argumentation professionals, particularly scholars. The conceptualization and operationalization of the third frame derive from the frank bias that scholars are correct about the nature of argument and that ordinary actors are quite often wrong (Hample 2003, 2005a). Many people believe that arguments are inherently nasty, irrational, hurtful, pointless, damaging, and potentially violent (Benoit 1982; Hample and Benoit 1999; Martin and Scheerhorn 1985). While these views need to be studied and acknowledged, as scholars we are entitled to say that they are also wrong. The key variable that operationalizes this third frame is therefore called professional contrast, and the items ask respondents to agree with either the professional view or the cruder understanding of many ordinary actors.

So these are Hample’s three classes of argument frames. Translated into Goffman’s terminology, we might say that the primary frames are Goffman’s frames as well, the second set of frames are more like Goffman’s keys, and the third frame points toward what Goffman called containment. Although empirical research on argument frames is only starting, the basic idea is that the frames contain people’s expectations and immediate orientations to arguing. These should influence people’s emotional reactions to arguing. Given the importance of arguing to daily activity, these emotional reactions may be important to the subjective quality of one’s life. Frames should also affect what one anticipates is going to happen in an argument, thus proposing opening argumentative moves and suggesting the chance of self-fulfilling prophecies.

1.3 Editing

Of the various processes involved in generating arguments, we have selected argument editing for special attention (Hample and Dallinger 1990). Editing is theorized to intervene between the initial private production of a message (the impulse that spontaneously generates content in service of a primary goal) and the final public production (Hample 2006). Should that initial form strike the arguer as somehow inappropriate or unwise, it may be edited so that the public statement is better suited to the arguer’s primary and secondary goals (the latter often have to do with facework or relational maintenance). People vary in the degree to which they bother to edit at all, and those who do commonly edit may still differ among themselves as to the standards featured in the reshaping of utterances.

Prior research has determined that editorial standards fall into two or three classes (Hample and Dallinger 1992). Most of the work has been done by providing respondents with lists of arguments that they could make or reject for specified reasons. The endorsement choice means that the arguer would be willing to offer that message, and no editing is needed. The first class of editorial standards has to do with effectiveness. A potential message can be rejected because it is not expected to work, and this choice most obviously represents the larger class called effectiveness. Alternatively, an argument might be suppressed because it is too negative to use, and this is the second choice in the effectiveness group. The second class is called person-centered. Here, an argument might be rejected out of concern that it would harm self (perhaps violating one’s self identity or conscience). It might be suppressed because it is seen as possibly harming the other (perhaps the other’s identity or feelings). The last choice in this group of standards is harm to relationship, which respondents sometimes do not clearly distinguish from harming the other. The third class, which is often statistically joined to the first, is called discourse competence. Here, possible arguments are suppressed because they are false or irrelevant, in the view of either self or other. Finally, the instrumentation permits a residual choice, which involves rejecting a possible argument for reasons not otherwise listed.

These editorial standards also represent the goals an arguer has in play while arguing. Editors tend to take on one of two styles. Some people edit primarily on grounds of effectiveness, and represent themselves as willing to say nearly anything that will work argumentatively. Others are person-centered editors. They won’t say things that have negative identity or relational repercussions, and may very well swallow potentially effective messages.

Argument frames have not yet been empirically connected to editorial behavior. This study will explore the possible associations. Both research traditions derive theoretically from concern for arguers’ goals, their understandings of what is wanted. We suppose that people with different frames may have different editorial inclinations. Several personality variables have been found to affect both frames and editorial activity, and this also encourages us in the expectation that we will discover direct relationships between them.

1.4 Other Individual Differences Variables

Both frames and editorial styles are individual differences variables, but others are relevant to interpersonal arguing as well. Here, we have chosen several which we think are of special importance, and which themselves may be interconnected. They all seem conceptually to bear on one’s combativeness which at the very least is an important possible key for arguing.

Argumentativeness (Infante and Rancer 1982) and verbal aggressiveness (Infante and Wigley 1986) are perhaps the most important personality traits to examine here. These two sorts of aggressiveness refer respectively to constructive efforts to engage the other’s position (argumentativeness) and to destructive impulses to attack the other’s feelings, identity, and person (verbal aggression). Both variables have been found to affect people’s beliefs about arguing (Rancer and Avtgis 2006). Argumentativeness, for instance, is connected to one’s beliefs about hostility, dominance, and self image, and verbal aggression levels predict the degree to which one believes personal attacks are hurtful. These findings immediately suggest connections to argument frames and editing styles. Some empirical work on these points will be mentioned momentarily.

To enlarge this portrait of arguers’ predispositions, we also examine their gender orientation (Bem 1974). Estimates of people’s masculine and feminine trait levels help us to understand the degree to which they will be nurturant or combative, for instance. Males and females (biological sex, not psychological gender) differ on argumentativeness and verbal aggressiveness (Rancer and Avtgis 2006), a finding that roughly suggests that masculines and feminines might show parallel results. Both sex and gender have shown weak connections to editorial patterns (Hample 2005a). Gender has a more marked association with argument frames (Hample 2005b).

Finally, in this study we include psychological reactance (Brehm 1966), an under-appreciated variable (Burgoon et al. 2002). Reactance refers to one’s impulse to resist and counter when pressured. Levels of reactance are implicated in various personality disorders (Seibel and Dowd 2001), and are connected to agreeableness, openness, and extraversion in ordinary people (Seemann et al. 2005). Reactance should help us to understand when people push back or concede gracefully when they are pressed during an argument. People differing in reactance may well have different arguing frames and editorial orientations.

2 Specific Connections and Predictions

Prior work has, in some cases, directly associated some of these measures with one another. The major exceptions are that frames and editing have not been studied together, and reactance has not been introduced into any of these research programs. Reviewing previous findings may sharpen our expectations about what those missing relationships might be.

Gender orientation has been related to the frames in two studies. Hample and Dallinger (2002) use a preliminary version of the frames instrument. They report that results from the professional contrast scale (the third frame, the most reflective) show that people with a feminine gender orientation have higher (more advanced) scores. Androgenous respondents—those with high scores on both masculinity and femininity—regard arguments as more civil than do people with lower gender scores, although this is a weak effect. Hample and Dallinger also use a scale called winning, which does not survive into later versions of the frames instrument, but has some connection to the competitive and utility frames. The view that winning is the main desiderata in arguing is most typical of masculines and negatively associated with feminine orientations. Hample (2005b), using better frames scales, replicates and extends these findings. Although utility is not successfully operationalized in that study, the other first frame measures (identity, dominance, and play) are positively associated with masculinity, and negatively correlated to femininity. Cooperation (operationalized as the opposite of competition) is positively associated with high feminine orientations, partly replicating the earlier finding about winning. Femininity is again positively associated with advanced scores on the professional contrast scales.

Hample (2005b) also presents evidence about the associations between argument frames and the argumentativeness and verbal aggressiveness instruments. High argumentatives are unusually likely to see play as a legitimate goal for arguing. People who are very verbally aggressive see domination as an appropriate reason to argue, and are noticeably more competitive in their orientation. They also get low scores on the professional contrast scale.

These same measures have been used in the argument editing research program (the current summary is Hample 2005a). Arguers high in masculinity are unusually likely to edit for effectiveness, but are noticeably disinclined to worry about harm to self, harm to other, or harm to relationship, when deciding which arguments to express and which to suppress. Femininity scores do not predict editorial choices. Argumentativeness has two subscales, argument approach and argument avoid. Those with high approach scores endorse more arguments, meaning that they edit less overall. On the other hand, people with avoidant predispositions are very concerned with harm to other when deciding what arguments to make. Verbally aggressive arguers endorse more possible arguments and are relatively unconcerned with harm to other.

The current study permits several of these results to be replicated, but its main purpose is to connect frames and editing. The data record affords some expectations about what such connections might be.

Masculine respondents are oriented to the identity, dominance, and play frames, and feminines are cooperative and resemble scholars in their understanding of arguing. At the same time, masculines edit for effectiveness but not for person-centered reasons. So we expect that the first frames (identity, dominance, play, and utility) should be the province of those who are effectiveness oriented in their editing. At the same time, we suppose that those who edit for person-centered reasons will have higher scores for cooperation and professional contrast.

High argumentatives are playful, edit less, and are less worried about hurting the other’s feelings with their arguments. This suggests that people with high play orientations will edit less, and will prefer effectiveness standards to person-centered ones. These expectations are consistent with those flowing from the gender work.

Verbally aggressive people orient to domination, have competitive orientations to arguing, and generate low scores on the professional contrast scale. These same people edit less and display less interest in protecting the other from harm. So we suppose that domination will be directly associated with endorsement and negatively associated with use of the harm to other standard. The opposite relations should hold for highly cooperative people and those with high professional contrast scores.

Other relationships between framing and editing are also possible, and we will test for them as well.

3 Method

Two hundred and five undergraduates at our institution served as respondents. 51% were male and 49% were women. Their average age was 19.9 years. Students received minor extra credit in their courses for participating. Data were collected in two separate sessions. The study was approved by our Institutional Research Board.

In the first session, students completed the 63 item frames instrument, the 28 item Therapeutic Reactance Scale (Dowd et al. 1991), and the 60 item Bem Sex Role Inventory (Bem 1974). Reliabilities and other descriptive statistics for these and the other measures in the study are in Table 1. Of special note are the results for the utility and blurting scales from the frames instrument. Prior work has consistently had difficulty with these measures. Here, for the first time, we factor analyzed these subscales, and found evidence that the scales are multidimensional. While the reliabilities are somewhat problematic here, we can now see how to improve the measurements in the future. With the additional exception of the cooperation/competition subscale, reliabilities are acceptable.
Table 1

Reliabilities and descriptive statistics

Scale

Cronbach’s alpha

Mean

Standard deviation

N

No. of items

Utility (constructive resolution)

.62

18.5

2.67

202

5

Utility (self interest)

.58

8.7

2.09

201

3

Identity

.76

27.9

4.82

202

8

Dominance

.75

16.0

4.19

202

6

Play

.81

10.9

3.82

202

4

Blurting (take care of business)

.52

5.0

1.41

202

2

Blurting (say what’s on my mind)

.59

9.9

2.12

202

3

Blurting (argue w/out thinking)

.57

5.7

1.68

202

2

Cooperation

.59

18.1

2.80

202

5

Civility

.78

33.5

5.21

201

10

Professional contrast

.81

24.4

5.00

201

7

Verbal aggressiveness

.80

34.2

9.38

197

20

VA (prosocial)

.71

17.6

5.13

198

10

VA (antisocial)

.81

16.7

6.23

201

10

Argument approach

.83

22.2

6.80

202

10

Argument avoid

.80

18.9

6.73

199

10

Masculinity

.83

73.4

9.33

197

20

Femininity

.82

71.9

9.09

201

20

Therapeutic reactance

.75

84.4

9.60

196

28

In session two, normally a week or less afterwards, respondents filled out the argumentativeness (Infante and Rancer 1982) and verbal aggressiveness (Infante and Wigley 1986) instruments after completing the editing tasks. Both of these scales are now thought to have subscales (Levine et al. 2004; Rancer and Avtgis 2006), and results for those appear in Table 1.

The editing task involves some minor departures from earlier work. Four situations were used. Following Johnson’s (2002) recommendations, two of these were public issues (getting your roommate to recycle, and getting your father to attend a political rally) and two were personal ones (getting your friend to give you a ticket to a football game, and borrowing your friend’s car). Each participant responded to all four situations, although they were presented in four different orders. Each one paragraph situation was followed by 20 possible arguments. These were intended to afford substantial variability in possible responses, but were not generated according to a standard typology, as had been done in earlier work. Once the 20 arguments were written for the first situation, this list was used as a template for the other three situations, yielding rough equality for matters such as non-topic content, prosocial versus antisocial, direct versus indirect, and so forth, across the situations. For each possible argument, students make one of nine responses. The first choice is endorsement. The other eight are reasons for suppression: ineffectiveness, too negative to use, it would hurt self, it would hurt the other, it would damage our relationship, it is false, it is irrelevant, or some other reason for rejection. This is the usual list of available responses. Figure 1 displays one set of experimental materials.
https://static-content.springer.com/image/art%3A10.1007%2Fs10503-008-9107-x/MediaObjects/10503_2008_9107_Fig1_HTML.gif
Fig. 1

An experimental situation, message set, and list of editorial choices

This design permits an assessment of the internal reliability of the editing instrumentation. By cumulating all the endorsements for situations one, two, three, and four, all the rejections for ineffectiveness for situations one through four, and so forth, we constructed a data set with four measures for each editorial code. Cronbach’s alphas are as follows: endorsement, .83; ineffectiveness, .87; too negative to use, .90; harm to self, .73; harm to other, .84; harm to relationship, .63; false, .66; irrelevant, .78; and residual, .83. These are conservative estimates of reliability, since they ignore the possibility that different situations and argument lists might call out different editorial behavior. This information is new to the editorial research program.

4 Results

Although the key issues here concern the relationships among the frames and editorial variables, some preliminary results are of substantial interest.

Table 2 shows the intercorrelations among the argument frames variables. As in prior work (Hample 2005b) the first-frame measures (utility, dominance, identity, and play) are positively correlated. This is sensible, since the first frame is self-centered and focused only on what the arguer seeks to gain by arguing. The utility data are new, as are the blurting results. Blurting, which indicates whether or not the arguer even participates in the second frame, is negatively associated with seeing arguments as constructive options, but positively correlated with the view that arguments have utility for getting what self wants. When blurting is justified as “taking care of business,” it is negatively associated with more advanced frames: seeing arguments as essentially cooperative, civil, and partaking of professionally-advanced characteristics. These negative correlations follow from the theoretical point that blurters do not really enter into second-frame orientations. Professional contrast scores are closely connected to civility and cooperativeness, as in earlier work.
Table 2

Correlations among frames variables

 

a

b

c

d

e

f

g

h

i

j

(a) Utility constructive

          

(b) Utility self interest

−.17

         

(c) Identity

.37

.14

        

(d) Dominance

−.00

.41

.35

       

(e) Play

.19

.18

.56

.41

      

(f) Blurt take care of business

−.30

.31

.01

.25

.07

     

(g) Blurt on my mind

.07

.14

.12

.11

.13

.01

    

(h) Blurt argue w/out thought

−.17

.09

−.05

.13

−.02

.11

.06

   

(i) Cooperative

.20

−.08

.13

−.19

−.14

−.28

.01

−.11

  

(j) Civility

.44

−.31

.26

−.23

.11

−.18

−.05

−.27

.14

 

(k) Professional contrast

.26

−.25

.02

−.28

−.05

−.17

−.09

−.23

.19

.51

Note. Statistically significant correlation are boldfaced. Sample sizes range from 202 to 195

Table 3 shows how the frames instruments are associated with the other individual differences measures. These results roughly replicate the parallel outcomes in Hample (2005b), when they are available. Masculinity is directly correlated with utility (constructive), identity, play, and blurting (say what’s on my mind). Femininity is negatively correlated with dominance, play, and blurting (say what’s on my mind), and positively with cooperativeness and professional contrast. People with high motivations to approach argument also see arguing as utilitarian (constructive), serving identity, dominance, and play functions, and being civil. Argument approachers are also less likely to report that they blurt (argue without thinking). Those highly motivated to avoid arguments have essentially the reverse relations. Respondents who have high scores on the antisocial verbal aggression items see arguments as utilitarian (serving one’s self interest), and fulfilling identity, dominance, and play functions. These people report that blurting is good because people should simply “take care of business.” They have low scores for cooperativeness, civility, and professional contrast. The reactance data are not precedented. Highly reactive people see arguments as useful (serving self interest), fulfilling identity, dominance, and play functions, and being competitive. They also blurt because they prefer to “say what’s on my mind.”
Table 3

Correlations between frames and other non-editing variables

 

Reactance

VA prosocial

VA antisocial

VA total

Argument approach

Argument avoid

Masculinity

Femininity

(a) Utility constructive

.07

−.01

−.07

−.05

.24

.34

.14

.06

(b) Utility self interest

.14

.24

.28

.33

.07

.07

−.06

−.13

(c) Identity

.40

.12

.26

.25

.47

.35

.25

−.08

(d) Dominance

.38

.36

.49

.54

.35

−.14

.08

.32

(e) Play

.52

.27

.29

.35

.52

.44

.25

.20

(f) Blurt take care of business

.09

.19

.25

.28

.06

.09

.08

−.12

(g) Blurt on my mind

.25

.17

.08

.15

.10

.19

.24

.14

(h) Blurt argue w/out thought

−.09

−.07

.10

.11

.21

.15

−.12

−.07

(i) Cooperative

.23

.39

.24

.38

−.12

.13

−.12

.31

(j) Civility

.03

.12

.19

.20

.14

.21

−.01

.12

(k) Professional contrast

−.05

.18

.18

.23

.07

−.10

−.01

.22

Note. Statistically significant correlation are boldfaced. Sample sizes range from 202 to 195

Table 4 cumulates these last results into a canonical analysis of the relations among the frames variables, considered as a group, and the other individual differences measures, considered as another group. The relationships are very substantial, as indicated by the size of the squared canonical correlations, the amounts of variance in the groups accounted for by the latent variables, and by the consistency with which each individual variable is predicted by the variables in the other group. Root 1 shows mainly that first-frame issues (identity, dominance, and play) are directly associated with reactance, antisocial verbal aggressiveness, and argument avoidance. Interpreting further roots is delicate because the first root has already extracted considerable variance. However, root 2 seems to show that utility (self interest), dominance, and blurting (taking care of business and arguing without thinking) are all directly related to verbal aggressiveness (the antisocial subscale), and negatively with verbal aggressiveness (prosocial) and femininity. Utility (constructive), cooperativeness, civility, and professional contrast scores, a generally advanced set of frames, are positively associated with high scores on the prosocial measure of verbal aggressiveness and femininity. Root 3 does not show much more than that masculines are competitive. It should be emphasized, as the final column of the table shows, that every one of the frames is substantially predicted by the other individual differences variables. This suggests that we are working with a nice cluster of argument-relevant personality indices.
Table 4

Canonical correlation between frames and other individual differences variables

 

Root 1

Root 2

Root 3

R2 from other root

Utility constructive

−.23

−.49

−.05

.15***

Utility self interest

−.21

.55

−.39

.17***

Identity

−.66

−.27

−.36

.31***

Dominance

−.60

.59

−.33

.35***

Play

−.82

−.16

−.20

.41***

Blurt take care of business

−.17

.47

.05

.10*

Blurt on my mind

−.38

−.02

.37

.12**

Blurt argue w/out thought

.18

.41

.04

.09*

Cooperative

.33

−.49

−.50

.19***

Civility

−.06

−.56

−.29

.14***

Professional contrast

.04

−.50

−.34

.11**

    Variance explained

17.1%

19.7%

9.3%

 

Reactance

−.85

.03

−.03

.44***

VA prosocial

.51

−.56

−.21

.30***

VA antisocial

−.55

.60

−.23

.33***

Argument approach

−.79

−.17

−.34

.40***

Argument avoid

.67

.38

−.16

.33***

Masculinity

−.47

−.24

.60

.22***

Femininity

.45

−.50

−.26

.23***

    Variance explained

39.6%

16.3%

9.4%

 

RC

.76

.59

.38

 

Note. Pillais approx. F (77, 1169) = 1.29, p < .001. df for the regressions in which the frames variables are dependent = (7, 171). df for the regressions in which the other individual differences variables are dependent = (11, 167). * p < .05; ** p < .01; *** p < .001

Tables 5 and 6 show the associations between the frames and editing data. Table 5 displays a scattering of significant correlations, not many more than would be expected given the significance level and number of results (8 significant out of 99 calculated). Table 6 shows the overall canonical correlation between the two variable sets. While the RC is substantial, neither latent variable accounts for much of the variance in its component variables. Nor are many of the variables predictable from those contributing to the other latent variable. However, the significant root shows that endorsements are especially common among people who see arguments largely in terms of the first frame set (utility, identity, dominance, and play), who blurt, and who have unsophisticated appreciations of the cooperative, civil, and other possibilities for arguing. The opposite pattern appears for editors who make substantial use of the “too negative to use” standard. While frames and editorial orientations are significantly connected, the extent of the connections is less than we had expected. The results that do appear, however, are generally consistent with our original expectations.
Table 5

Correlations between frames and editorial standards

 

Endorse

Ineffective

Too negative

Harm self

Harm other

Harm relationship

False

Irrelevant

Residual

Utility constructive

.12

−.01

−.06

−.06

.02

−.01

−.10

−.07

.12

Utility self interest

.11

−.05

−.02

.11

−.04

−.02

.03

.07

−.08

Identity

.13

−.11

.18

−.03

−.03

.13

−.01

.10

.13

Dominance

.18

.05

.16

.11

−.11

.01

−.01

−.01

−.08

Play

.15

−.00

−.12

−.03

.17

.11

−.10

.04

.05

Blurt take care of business

.08

−.09

−.06

.15

.02

.04

.02

.02

−. 1

Blurt on my mind

.09

−.01

−.01

.11

.00

−.03

.05

.12

−.09

Blurt argu w/o thought

−.02

.01

.01

−.04

−.13

−.12

.07

.04

.07

Cooperative

−.02

−.01

.06

−.13

.13

.01

−.03

−.07

−.07

Civility

.10

.07

−.14

−.07

−.09

.07

−.06

−.06

.14

Professional contrast

.07

.01

.19

−.00

−.01

.07

.08

−.04

.07

Note. Statistically significant correlations are boldfaced. Sample sizes range from 197 to 202

Table 6

Canonical correlation between frames and editorial standards

 

Root 1

R2 from other root

Endorse

−.53

.08

Ineffective

−.03

.06

Too negative

.67

.11*

Harm self

−.33

.07

Harm other

.37

.10**

Harm relationship

−.17

.04

False

−.13

.05

Irrelevant

−.11

.05

    Variance explained

13.0%

 

Utility constructive

−.08

.03

Utility self interest

−.33

.04

Identity

−.45

.08*

Dominance

−.63

.09*

Play

−.48

.09*

Blurt take care of business

−.34

.04

Blurt on my mind

−.27

.07**

blurt argue w/out thought

−.02

.04

Cooperative

.37

.05

Civility

−.27

.05

Professional contrast

−.34

.05

    Variance explained

13.3%

 

RC

.45*

 

Note. Pillais approx. F (88, 1496) = 1.28, p < .05. df for the regressions in which the editorial codes are dependent = (11,187). df for the regressions in which the frames are dependent = (8, 190). * < .05; ** p < .10

5 Discussion

5.1 Implications of Empirical Results

This project successfully replicates some leading results from the nascent research program dealing with arguing frames. These expectations about arguing now seem to have consistent and theoretically sensible relationships with several powerful trait variables: argumentativeness, verbal aggressiveness, and gender orientation. The new results associating the frames with psychological reactance are promising, and conceptually consistent with prior work on both topics.

In addition, some psychometric improvements in the frames instruments appear here, and solutions to persistent measurement problems now seem at hand. Ambiguity in the utility and blurting scales has been detected. One substantial implication is that one group of utility items, those suggesting that argument is a constructive way to resolve problems, is more directly connected with the more advanced elements of the second and third frames than with the other first-frame measures. Some reconceptualization of the scale is in order. In the case of blurting, respondents are making distinctions that were not originally apparent to the scales’ authors, and this may also require some rethinking on our part.

We are nonetheless encouraged by our discovery of relationships between arguing frames and editorial work. Endorsement, understood in this context as the inclination not to edit at all, is characteristic of first-frame people. As arguers advance in their appreciation of arguing—that is, as they begin to see that it is cooperative, ought to be civil, and should be a productive way of resolving conflict—they endorse fewer arguments and suppress more of them. However, our present results do not point to any particular editorial style for these advanced arguers.

5.2 Frame Analysis

Although the work on argument frames has been under way for several years it was born as a descriptive data-driven project without many theoretical anchors. Here for the first time it is connected to Goffman’s rich microsociology.

Goffman’s conception of frames corresponds directly to the primary arguing frames. Arguing for a utilitarian purpose is the natural strip of behavior, untransformed and working directly in the world. This strip can be transformed in various ways, and several of them are play, identity, and dominance (as well as others not studied in the argument frames program, such as classroom example or theatrical display).

An arguer can act within any of these frames in various ways and these can give a tone to the argument. Thus the second set of frames particularly features competition and cooperation, both traditionally recognized as important attitudes toward arguing. These are keys in which one might dispute or play or show off. In fact, these keys may well be those that transform the primary frames. For instance a competitive key may be what changes a display of identity to one of dominance and a cooperative key may convert dispute into play.

Goffman discusses the plight of people who are “contained” within a frame, such as victims of swindles or practical jokes. To be contained is to be unaware of someone else’s fabricated frame, in Goffman’s writing. Here the idea of containment refers to the third arguing frame, which points to whether a person has a crude unreflective understanding of argument or the more sophisticated perceptions that argumentation scholars propound.

The connections between the empirical work on arguing frames and Goffman’s theory deserve to be worked out in more detail. It may also be that the various frames and keys can be connected to the goals that arguers pursue. If that proves to be the case, we may be able to unify our thinking on a number of matters: arguing frames, argument goals, and argument editing.

Copyright information

© Springer Science+Business Media B.V. 2008