ISSA Proceedings 1998 – The Narrative As An Argument Component
A narrative is an account typically consisting of a temporal sequence of events that is focused upon characters, their actions, and the outcomes of such actions. In recent decades the narrative has been the object of much analysis, study, and debate. Psychological research on narratives has involved the study of story grammars, syntactic-like structures that describe the generic elements of narratives (e.g., Stein & Glenn, 1979). Other psychological research of narratives has included the study of causal structure (e.g., Trabasso, van den Broek, & Suh, 1989), and inference generation (e.g., Graesser, Singer, & Trabasso, 1994). Narratives also have received considerable attention in relation to their role and importance in the study of history (e.g., White, 1987).
Narratives have also been examined with respect to the purposes they serve. According to Focault (1969, 1972), narrative is used by those in power as a means of maintaining power while the alternative narratives of those out of power are suppressed by those in power. Narrative is also used to delineate official and unofficial history (Wertsch & Rozin, 1998). In the Soviet Union the official history was a Marxian account of the 1917 Revolution and post-Revolution period. Unofficial history, however, embraced a narrative that was historically Russian, extending farther into the post than the 1917 Revolution. Similarly, Epstein (1996) has shown that European American eleventh graders provide a narrative of U.S. history that follows the traditional colonization, French and Indian War, Revolutionary War, Civil War, and into the late nineteenth and twentieth centuries format, while Afro-American students provide a narrative emphasizing racial inequality. Narratives held thus relate to belief and experience, and indeed, the historian Mink (1987) has indicated that narratives provide information about the past, and the background of the narrator needs to be taken into account to understand the narrative. Narratives also have been viewed as deceptive, as White (1987) has stated, “narrative discourse …. endows events with illusory coherence” (p. ix). In any event, the narrative is used to provide continuity to a series of linear events and is the subject of this paper, a topic, incidentally, which is not new.
Narrative and Argument
The present paper is concerned with narrative as argument. Relating narrative to argument is not new, as Aristotle spoke of it as one of two types of argument within rhetoric, the other being the enthymeme. Probably the two most obvious contexts for the use of narrative as argument are those of history and of law. The study discussed here is in the jurisprudence context, primarily because of the likely greater difficulty in conducting the equivalent experiment in the context of history. Consider the statement “Capital punishment should be abolished because it is cruel and inhumane treatment.” In the Toulmin (1958) model, “Capital punishment should be abolished” is the claim and “because it is cruel and inhumane punishment” is the datum or grounds.
Let us assume that we maintain some claim such as “Capital punishment should be abolished” but to support this claim we do not provide the a supportive reason in the usual sense but we provide the statement “Because of the following story,” and then proceed to tell a narrative which has the point of showing that capital punishment be abolished. In this case the support is a narrative. This use of narrative, incidentally, is quite close to what Deanna Kuhn (1991) described in her book on argumentation as
Let us now imagine that we are in a courtroom and a prosecuting attorney makes the statement “This person, the defendant, is guilty,” and then supports this claim by providing a narrative describing what happened leading to the crime, the defendant’s presumed role in it, and how and why the defendant committed the crime.
A narrative supporting the attorney’s claim of the defendant’s guilt such as that just described is likely to have two components. One is the so-called “facts” of the case. This category consists of the statements of witnesses and exhibits of the case, which essentially constitute a list of information. The second component is the narrative, the story or account that the prosecuting attorney weaves and develops that has the goal of supporting the claim of the defendant’s guilt. The two components then are the “facts” of the case and the narrative, which integrates the “facts” into a story. The use of the narrative to support a claim and the two-component distinction just made leads to the possibility that the narrative can play a role in the judgment of the jury. It may be that a good narrative, with the “facts” included, will be more likely to produce a “guilty” judgment than a poor narrative, even with the same “facts” included.
A question then raised by this analysis is how may narrative quality be defined? Fortunately, Leinhardt, Stainton, Virji, and Odoroff (1994) asked a number of historians to indicate what they thought to be the qualities of a good narrative. Five attributes were noted, namely, coherence, causality, chronology, completion, and colligation or, more or less, contextualization. Coherence refers to the narrative having a coherent whole. Causation to the narrative’s need to show causal relations. Chronology is that the events of the narrative follow in a chronological order. Completion refers to whether the historians used all available information, and colligation to the narrative occurring in the appropriate historical context. Pennington and Hastie (1993), in their work on jury decision making, also considered narrativity, and emphasized the importance of coherence, coverage (similar to completeness), uniqueness (the most appropriate narrative), and (being psychologists) they included the goodness-of-fit of the narrative.
The rationale of the study was as follows. The first hypothesis was that if a hypothetical prosecuting attorney states the defendant is guilty and provides a narrative in support of this claim, ratings of the quality of the narrative are a function of the extent to which the narrative maintains the criteria of good narrativity. If, for example, a narrative is made less coherent, the quality of the narrative will be rated lower than the original narrative, before it was made less coherent. The second hypothesis is that if a narrative is degraded, the ratings of the defendant’s guilt are lower than guilty ratings provided for the original narrative. In other words, with a narrative having less coherence than the standard, both ratings of narrative quality and ratings of guilt would be lower than found for the standard narrative. The reason guilty ratings are likely to be lower is that a poor narrative presumably acts to hurt the prosecuting attorney’s case. In the experiment conducted there were four narrative conditions. One was a standard narrative. One contained the identical sentences as the standard narrative but the sentence order was changed. This version maintained local coherence. This was called the coherence/chronology condition because it decreased the narrative’s coherence and the chronological order. A third condition, the causal condition, decreased the causality stated in the standard condition. The fourth condition, the completion condition, deleted some of the information in the standard narrative but did not delete any of the “facts.” It should be especially noted that in all four conditions the “facts” of the case were included, thus making the design one of holding the “facts” constant and varying the narrative, modifying the standard narrative in three conditions to lower its quality according to the previously mentioned criteria.
Four texts were employed, each being a murder case. Each text had four versions, each version of each text corresponding to the four types of narratives. Participants were 64 college students, with 16 serving in each row of a greco-latin square, that is, each participant read each of the four texts once, also serving one time in each of the four narrative conditions.
The baseline or standard condition for one of the texts, “The Car Accident” follows. Participants were told that they were to consider the text to be the prosecuting attorney’s summary statement.
The victim, Roger Wilson, had dropped off his co-worker, Susan Walker, at her home. He then was driving on Crawford Street in order to get to the freeway. As he was driving, a six year old girl, Marjorie Moran, ran out from behind a parked car. Before Roger could stop, his right fender hit her and she fell to the ground. He quickly got out of his car to check on her and found that she was not seriously injured. Despite this fact, a number of neighborhood teenagers, who were standing nearby, began to push him around, saying things such as “Don’t you know how to drive?” Then someone from the crowd took a baseball bat and hit Roger in the head, killing him. This action was seen by a resident living across the street from the altercation, but he was unable to identify who had used the bat. When the police got to the scene of the crime, they took statements form several witnesses, and looked for the bat. In a few minutes, they found a baseball bat in the back seat of a car that was parked nearby. The car belonged to Matthew Moran, the girl’s older brother. Matthew Moran had been among the crowd that attacked Roger Wilson. He was very protective of his younger sister, and sometimes got into fights with people he determined were trying to hurt her. Analyses later revealed that the victim’s blood and hair were on the baseball bat. This evidence indicated that Matthew Moran’s bat must have been the bat used to hit Roger Wilson. Furthermore, Matthew’s were the only fingerprints found on the bat.
Matthew Moran claimed that his fingerprints were on the bat because he had used it earlier in the day to play baseball, but playing baseball could not have placed the victim’s blood and hair on the bat. The evidence indicates that Matthew Moran’s bat must have been used to hit Roger Wilson, and since there were no fingerprints on the bat besides those of Matthew Moran, he must have been the person who hit Roger Wilson with that bat. Matthew Moran, who had the motive, the means, and the opportunity, is guilty of killing Roger Wilson.
In the causation condition the following changes were made. (Text prior to arrows was in the standard text and changed to the material found after the arrows.)
– He quickly got out of his car to check on her and found that she was not seriously injured. -> He quickly stepped out to check on his car and found that it was not damaged.
– Despite this fact, a number of neighborhood…. -> A number of neighborhood….
– This evidence indicated that Matthew Moran’s bat … -> Matthew Moran’s bat….
– Matthew Moran’s bat must have been the bat used to hit Roger Wilson… -> Matthew Moran’s bat must have come into contact in some way with Roger Wilson….
– Matthew’s were the only fingerprints found on the bat… -> Matthew’s were the only fingerprints found on the bat, indicating that he had touched it and, that no one else could have touched it, unless they were wearing gloves…
– … had used it earlier in the day to play baseball, but playing baseball could not have placed the victim’s blood and hair on the bat -> … had used it earlier in the day to play baseball.
– bat must have been used… must have been the person… -> bat was probably used… he was probably the person…
In the incomplete condition, the following deletions were made:
– his co-worker
– He quickly got out of his car to check on her and
– but he was unable to identity who had used the bat.
– When the police got to the scene of the crime, they took statements from several witnesses, and looked for the bat.
– In a few minutes, they found a baseball bat in the back seat of a car that was parked nearby.
– The car belonged to Matthew Moran, the girl’s older brother.
– Matthew Moran claimed that his fingerprints were on the bat because he had used it earlier in the day to play baseball.
The order of sentences in the coherence/chronology condition, of the sentence in the standard narrative, were: 7, second half of 17, 4, 3, 1, 5, 11, 13, 15, 14, first half of 17, 9, 10, 12, 16, 8, 6, 18, 19.
Participants, after reading each narrative, provided 1-10 ratings for each of five questions and then subsequently answered these questions. The five rating scale questions were: “Do you think the accused is guilty?” “How confident are you in your decision?” “Please rate the overall quality of the summary statement.” “How convincing or persuasive was the statement?” “How good an argument did the lawyer make for the case?” “The three open-ended questions were ”What was good about the argument?” “What was missing?” “How could the statement be improved?”
The results indicated that the mean guilty rating (1=definitely not guilty, 10=definitely guilty) was 7.5, 7.5, and 7.4 for three of the texts. The fourth text, however, provided both a considerably different mean of guilt ratings and a quite different distribution of ratings. Only the three consistent texts were therefore used in the analyses. The mean guilty rating for the baseline condition was 8.0 and for the completeness condition was 7.9. However, for the causation condition the mean guilty rating was 7.0 and for the coherence/chronology condition was 6.9, the latter two means being statistically significantly lower from the first two. The confidence rating means were 8.0, 7.4, 7.6, and 7.8 for the four respective conditions, as listed in the order of the preceding sentence. The only significant difference was that the baseline condition yielded more confidence than the causation condition.
The three ratings of narrative quality yielded highly similar results. The means for the respective baseline, causation, completeness, and coherence/chronology conditions for overall quality of the narrative were 7.8, 6.9, 7.8, and 4.9; for convincingness, 7.6, 6.9, 7.9, and 5.3; for the argument stated 7.6, 6.9, 7.9, and 5.1. For all three narrative measures, the standard condition yielded significantly higher narrativity ratings than the causality and the coherence/chronology conditions, but not the completeness ratings.
The data show both hypotheses to be supported for the causality and coherence/chronology condition. Specifically, modifying either the causal or the coherence/chronology narrative structure produced lower judgments than the standard condition for narrativity and for the guilt ratings. With respect to the completeness condition, the deletion of information that did not involve the “facts” of the case likely produced little description in the participants’ consideration of the narrative.
The results of the present study indicate that under particular circumstances, the narrative may be considered as a component of argument, a statement that supports a claim. Furthermore, the results indicate that if the narrative is of relatively low quality, as determined either by the causality it states or by the lack of coherence and chronology, the persuasiveness of the argument will suffer. Another interpretation of the results, although not mutually exclusive, is that the presenting attorney may have lost his ethos, that is, by presenting a relatively poor narrative, professional respect for lives as are authority may have diminished. The present data do not, however, provide evidence regarding this notion.
Possibly the most interesting result involves the causation condition. Why does making some statements probabilistic, statements that do not involve the critical events, produce lower narrativity and guilt ratings? One possible explanation is that the probabilistic wording generalizes to the entire paragraph, giving the participant a sense of relative uncertainty for all paragraph events.
Performance in the coherence/chronology condition suggests that individuals are quite sensitive to the need for coherence and chronology in the narrative. In reply to an open-ended question, there were 23 comments that the text “made little sense,” “jumped around,” or were “mixed up,” as compared to such statements in the other narrative conditions.
There are a number of questions raised by the present findings, such as how would the guilt judgments be related to narrative judgments when both the presenting and defense attorney cases are presented as alternative narratives. More broadly, there is the question of how beliefs about the structure of a narrative play a role in guilt ratings and whether it is possible that an excellent narrative could be constructed with few facts that would provide a relatively high guilty rating. In other words, could under appropriate conditions, narrativity dominate the factual evidence.
In conclusion, the study indicates that narrative, when used as support for a claim, may be judged for its quality and that judgment is related to the convincingness of the argument presented. Finally, it needs to be mentioned that in the case of the enthymeme, the two primary criteria of support are that the reason is acceptable and that the reason provides support for the claim. The present results suggest the narrative quality influences the acceptability of the reason, and with less acceptability, less support may be provided and the proposed strength of the argument is diminished.
Epstein, T. L. (1996). Urban adolescents’ historical understanding: Differences in Black and White. Theory and Research in Social Education, 26, 399-402.
Foucault, M. (1969, 1972). The archeology of knowledge. Trans. by A. Sheridan Smith. New York: Pantheon.
Graesser, A. C., Singer, M., & Trabasso, T. (1994). A constructionist theory of inference generation during narrative text comprehension. Psychological Review, 101, 371-395.
Kuhn, D. (1991). The skills of argument. Cambridge: Cambridge University Press.
Leinhardt, G., Stainton, C., Virji, S. M., & Odoroff, E. (1994). Learning to reason in history: Mindlessness to mindfulness. In: M. Carretero and J. F. Voss (Eds.), Cognitive and Instructional Processes in History and the Social Sciences. Hillsdale, NJ: Erlbaum. Pp. 131-158.
Mink, L. O. (1987). Historical Understanding. Ithaca, NY: Cornell University Press.
Pennington, N., & Hastie, R (1993). The story model for juror decision making. In: R. Hastie (Ed.), Inside the Juror. The Psychology of Juror Decision Making. Cambridge: Cambridge University Press. Pp. 192-221.
Stein, N. L., & Glenn, C. G. (1979). An analysis of story comprehension in elementary school children. In: R. O. Freedle (Ed.), New directions in discourse processing. Vol. 2, pp. 53-120. Norwood, NJ: Ablex.
Toulmin, S. E. (1958). The uses of argument. (2nd ed.). Cambridge, England: Cambridge University Press.
Trabasso, T., van den Broek, P., & Suh, S. (1989). Logical necessity and transitivity of causal relations in stories. Discourse Processes, 12, 1-25.
Wertsch, J. V., & Rozin, M. (in press). The Russian Revolution: Official and unofficial accounts. In: J. F. Voss and M. Carretero (Eds.), Learning and Understanding in History. London: Woburn.
White, H. (1987). The Content of the Form. Baltimore: Johns Hopkins University Press.