Capital murder cases are composed of two phases: the guilt phase and the sentencing phase. During the guilt phase, the defense attorney and the prosecutor present their sides of the case, and the jury determines whether or not the defendant is guilty. During this initial phase, a prosecutor’s goal is to provide sufficient evidence to convince the jury to find the defendant guilty. If the defendant is found guilty, the trial moves to the sentencing phase. During this phase, a prosecutor’s goal is to provide sufficient evidence to lead the jury to impose the requested sentence (Barkan and Bryjak 2014). The prosecutor accomplishes these goals through interactions with the jury, specifically through carefully chosen language.

In capital murder cases where the defendant has already been found guilty, during the sentencing phase, the prosecutor often asks the jury to impose the death penalty. Even though jurors in these cases will have already been “death penalty qualified,” meaning that they have asserted that they will be able to impose the death penalty if appropriate, this can still be a very difficult decision for most jurors, who may have significant moral or ethical reservations about actually sentencing someone to death. As an extension to social cognitive theory, Bandura (1989; 1999) described several cognitive mechanisms that he proposed were used by individuals to rationalize or distance themselves from the moral implications of actions such as sentencing someone to death or actually administering an execution. Haney (1997) talked about similar mechanisms and suggested that capital punishment, as a system, would not be feasible if people did not employ cognitive mechanisms to distance themselves from the consequences of these decisions. Haney suggested that prosecutors may influence jurors to use these types of mechanisms through the strategy of carefully wording their statements to jurors during the sentencing phase, thus focusing on information that would make it easier for jurors to impose a death sentence.

Previous research has found that language can affect decision-making (e.g., Tversky and Kahneman 1981), but only a few studies have investigated the effect of language on jury decisions in the courtroom setting. Schmid and Fielder (1998) explored the use of subtle language strategies in attorney’s closing statements in simulated court cases. They videotaped simulated trials conducted by student lawyers and coded the language used. The videos were then shown to mock jurors who were instructed to determine a verdict and a sentence for guilty defendants. The results suggested that the language strategies used by the lawyers did affect the jurors’ attributions of blame and guilt. A second study conducted by Schmid and Fiedler (1998) investigated prosecutorial language strategies in a real-life scenario, using transcripts from the Nuremberg trials. Prosecutors made references to the defendants’ in-group (i.e., the Nazi party) and undesirable characteristics of the defendants.

Other investigations of juror decision-making have focused on decision-making in mock juries or post-trial interviews with actual jurors to investigate the causal determinants of a juror’s decision (Kerr and Bray 2007; Sandys et al 2009). Both methodologies tend to agree: what mock jurors say that they focused on is similar to what actual jurors (after the fact) say that they focused on, which are often the types of cognitive strategies suggested by authors such as Bandura (19891999) and Haney (1997).

Haney (1997) identified five psychological mechanisms, which he called “mechanisms of moral disengagement,” that may equip capital jurors to overcome their inhibitions about imposing the death penalty. Through the use of these mechanisms, jurors may be able to reduce their cognitive dissonance and distress over the decision to sentence a person to death and may be able to believe that they made a correct and ethical decision. These mechanisms are similar to the work on moral reasoning by Bandura (19891999) and include dehumanization, viewing the defendant as defective/different/deviant, using the death penalty as a form of self-defense, minimization of personal consequences of the decision, and instructional authorization of the death penalty.

Dehumanization of the defendant involves portraying him/her as less than human, perhaps more akin to animals or even monsters. If the prosecutor uses language in describing the defendant that depicts him/her as lacking in human qualities, perhaps by describing the heinous acts that led to the trial, the jurors may be more able to justify a death sentence. Conversely, defense lawyers should use the opposite strategy of attempting to humanize the defendant for the jury, by providing information on mitigating circumstances that might allow the jury to identify and empathize with the defendant as a fellow human being.

The second mechanism is related to dehumanization but is less extreme. It may be an expression of our species’ natural tendency to categorize people in terms of their similarity to ourselves. It may not be necessary to completely dehumanize a defendant to allow us to decide to sentence them to death; it may be sufficient to simply see them as possessing qualities that make them fundamentally different from ourselves, or defective in some way. Prosecutorial language that highlights or exaggerates any such existing or presumed differences may facilitate the jury’s decision to impose the death penalty.

The third mechanism of moral disengagement that Haney proposed is the tendency to view imposition of the death penalty as vicarious self-defense against a demonstrably dangerous defendant. The trial process itself exposes the often horrifying details of the murder and may lead jurors to fear the defendant’s capacity for violence. It is an accepted tenet of our legal system that an individual is allowed and expected to engage in self-defense behaviors when threatened by a violent individual. Jurors may be able to rationalize a death sentence as a way of protecting themselves and the community from seemingly inevitable future violence perpetrated by a dangerous defendant. Jurors may even reason that defendants sentenced to life in prison without the possibility of parole will still have the opportunity to commit violence against others, and they may not view life imprisonment as an appropriate option (Sorenson and Pilgrim 2000). Prosecutorial language that emphasizes the defendant’s potential for future dangerousness may activate this mechanism. There is evidence showing that jurors are not particularly accurate when it comes to predicting future dangerousness, but they are still heavily influenced by future dangerousness arguments by prosecutors in states that allow such arguments (Marquart et al. 1989).

A fourth potential mechanism involves minimizing the personal consequences of imposing the death penalty. In other words, jurors may be more likely to impose this punishment when the consequences to themselves seem insignificant or distant. There are several ways to distance oneself from the consequences of the decision. At the outset, they share with the other jurors the responsibility for the decision to impose the death penalty. This in and of itself diffuses the responsibility. The fact that jurors do not need to carry out the execution themselves also provides a measure of distance; the sentence is carried out by others, sometime in the future. Many jurors may even believe that the death sentence will never be carried out because of numerous appeals or other delays, thus mentally shifting the ultimate responsibility for the decision to appellate judges. All of these possibilities can result in a diffusion of the individual juror’s responsibility for the decision.

The final mechanism described by Haney (1997) is instructional authorization for capital violence, or a displacement of responsibility, which occurs when jurors might feel relieved of responsibility for a death penalty decision because they are following what they perceive to be official authorization or even an expectation to do so. This is illustrated by the classic Milgram obedience studies (see Milgram 1965), in which participants delivered dangerous electric shocks to other participants because they had been ordered to do so by the researcher, an authority figure in that situation. In the capital murder trial, the initial process of “death qualifying” each juror could have the effect of conveying to the jurors that the legal system expects them or even requires them to impose the death penalty (Barkan and Bryjack 2014). Judicial or prosecutorial statements about following the law may have the same effect.

Osofsky et al. (2005) investigated whether the use of moral disengagement mechanisms enabled prison personnel to carry out the death penalty. One might surmise that actually carrying out an execution might be much harder psychologically than casting a vote to impose the death penalty. The authors demonstrated that the use of moral disengagement mechanisms varied among prison personnel, depending on their level of involvement in the execution process. The execution team (personnel who carry out the execution) exhibited the highest level of justifications, dehumanization of the inmate, and disavowal of personal responsibility, compared with support teams (personnel who provide supportive services to the inmate and his/her family), and prison guards who had no involvement in the execution process. The mechanisms used by the execution staff overlapped with those identified by Haney (1997) and Bandura (1999).

To our knowledge, no prior research has examined whether prosecutorial language used in actual capital murder cases influences the sentencing decisions of jurors. The purpose of the present study is to determine whether the extent to which prosecutors’ use of specific language strategies in their closing statements might encourage moral disengagement by jurors and thus might influence their sentencing decisions. The types of language that were examined were similar to and based on mechanisms discussed by Haney (1997), Bandura (1999), and Osofsky et al. (2005). We looked for prosecutorial language that dehumanized the defendant, portrayed the defendant as dangerous, invoked moral justification, and provided displacement of responsibility on the part of the jurors. We predicted that prosecutorial use of these types of language would be associated with increased imposition of the death penalty versus life without parole. We also looked for the presence of any empathetic language toward the victim of the defendant, which might lead jurors to feel sorry for the victim or for the family or friends of the victim. Given that Shelton and Rogers (1981) have demonstrated that empathy can lead to attitude change, we predicted that empathetic language toward the victim by prosecutors might also increase a jury’s tendency to impose the death penalty.

Method

Materials

Transcripts from 25 actual capital cases were selected (13 received a sentence of life without the possibility of parole and 12 received the death penalty). All cases involved male defendants to control for potential differences in sentencing of males and females. Only cases from states that allow future dangerousness arguments were included. The age of the defendant ranged from 20 to 66 years old (M = 37.36 years; SD = 14.46 years). The number and age of the victims were similar across most cases; only two cases involved non-adult victims (one age 2, and one age 15). Thirteen of the defendants were White, seven were Black, and one was Hispanic. Four cases did not specify the race of the defendant. See Table 1 for the demographic details per case.

Table 1 Percentages of control variables

Transcript Coding

A coding system designed by the authors was used to identify the presence of four types of prosecutorial language that might be designed to invoke moral disengagement mechanisms in the jurors: dehumanization, future dangerousness, moral justification, and displacement of responsibility. Additionally, language that may increase empathy toward the victim was also coded. More detailed descriptions of the definitions of the variables as used in this study and the coding instructions are contained in Appendix. During coding, it was determined that the diffusion of responsibility category should be omitted due to the difficulty of determining whether the prosecutors’ use of the word “you” referred to the individual juror, or the jury collectively.

All identifying information was redacted from the transcripts before they were coded. The actual sentencing decision of the jury was available for each of the cases, but was not known by the coders. Only the closing statements from the prosecution were coded; all of the transcripts were independently coded by two of the authors. Any coding discrepancies were resolved through discussion. Each instance of the presence of one of the variables during the closing statement was tallied.

Results

Inter-rater Reliability

To determine inter-rater reliability, the percentage of agreements by the two coders for each variable, for each transcript, was calculated. The percentage of agreements for the four variables was averaged for each transcript, creating the overall inter-rater reliability for that transcript. Finally, the overall averages for all of the transcripts were averaged together to establish the overall inter-rater reliability, which was 99%.

Language Strategy Analysis

The first goal of this experiment was to determine if four moral disengagement variables ((1) Dehumanizing language, (2) Future dangerousness, (3) Moral justification, and (4) Displacement of responsibility) were correlated with verdict (i.e., receiving the death penalty or life in prison) in 25 capital murder cases. A point-biserial correlational analyses (see Table 2) between the four moral disengagement variables and verdict were conducted. In this case, a positive correlation would indicate that increased instances of moral disengagement language are associated with the death penalty verdict. A negative correlation would indicate that fewer instances of moral disengagement language are associated with the life in prison verdict. The analysis showed that only future dangerousness shared a significant positive relationship with verdict (r = 0.40, p = 0.045). This means that the increased use of future dangerousness language increases the likelihood that the defendant will receive the death penalty. Dehumanizing language (r = 0.34), moral justification language (r = 0.02), and displacement of responsibility language (r =  − 0.13) were not significantly correlated to verdict (all ps > 0.05), showing that these language types did not relate to the likelihood of receiving the death penalty.

Table 2 Correlations between five language strategies and sentencing outcome

The researchers had some concerns that variables other than the four moral disengagement variables could be impacting the aforementioned correlations; thus, four control variables that could impact verdict were identified: (1) race of defendant, (2) number of victims in crime, (3) whether or not the crime involved a child (i.e., under 18 years of age), and (4) whether or not an officer was killed in the line of duty during the crime. A partial correlation on the four moral disengagement variables and verdict was conducted, controlling for the four control variables (see Table 2). The results of the partial correlation were similar to the aforementioned point-biserial correlation analysis assessing the relationship between the moral disengagement variables and verdict. Again, the number of future dangerousness language uses had a significant positive relationship with verdict (r = 0.52, p = 0.017), such that as this language type increased in the closing argument, so did the likelihood of receiving the death penalty. However, dehumanizing language (r = 0.22), moral justification language (r = 0.03), and displacement of responsibility language (r =  − 0.12) were not significantly related to verdict (all ps > 0.05). This again shows that these three language types did not relate to the likelihood of receiving the death penalty.

The next goal of this experiment was to determine if the four moral disengagement variables could predict verdict. A binomial logistic regression was conducted with verdict as the criterion variable and number of times each of the following language types was used in the prosecutor’s closing argument during the sentencing phase: dehumanizing language, future dangerousness, moral justification language, and displacement of responsibility language. The results showed that the model was not statistically significant χ2(4) = 7.13, p = 0.129 and could only explain 33% (Nagelkerke R2) of the variability in jury verdict, with 72% of cases correctly classified. This indicates that there may be some other variables that impact the verdict that the jury delivers other than the four moral disengagement variables.

Finally, the researchers wanted to explore the relationship between how empathy toward the victim is related to jury verdict. Prosecutors will often use empathizing language in regard to the victim, which may garner empathy from the jury. We cannot state with any certainty that the jury does feel empathy when hearing this language, only that garnering empathy seems to be the prosecutors’ intent. A correlation between the number of empathy toward the victim language uses and verdict showed a significant positive correlation (r = 0.40, p < 0.05). This suggests that the more the prosecutor attempts to garner empathy for the victim, the more often the jury gives the death penalty sentence. A binomial logistic regression also showed that empathy toward the victim language is a significant predictor of verdict, χ2(1) = 5.55, p = 0.018, that can explain 27% of the variability in jury verdict (Nagelkerke R2), with 76% of cases correctly classified. Although empathy toward victim language does seem to have some impact on the jury ruling, it is unclear what emotions this garners (e.g., it could be anger toward the defendant); thus, this result should be interpreted with caution. Future studies should attempt to not only determine what impact this language has on a juror’s verdict decision, but also what emotion it elicits from jurors.

Discussion

Language has previously been found to significantly impact decision-making (Haney 1997; Schmid and Fielder 1998). During trials, both defense and prosecution attorneys use language to persuade jurors to believe their arguments. Determining what types of language are the most persuasive would allow attorneys to produce the most effective arguments. The current study analyzed prosecutorial closing arguments in capital murder cases. Not only does the current study confirm that language can influence decision-making, but it also sheds light on how everyday citizens (i.e., jurors) might be able to make the decision to sentence another human being to death. Some of the moral disengagement strategies previously identified by other authors (Haney 1997; Bandura 1999) were not correlated with a death penalty sentence. However, language that emphasized future dangerousness and empathetic language toward the victim were significantly correlated with a death penalty sentence, as opposed to life in prison without the possibility of parole.

Previous studies (Marquart et al. 1989) found that jurors do not accurately predict a defendant’s future dangerousness, yet the current study found that assertions about future dangerousness significantly impact the jury’s willingness to sentence a defendant to death. Because arguments about future dangerousness do not necessarily correspond to actual future dangerousness, many states have banned these types of arguments from the courtroom; mentioning future dangerousness in these states is grounds for a mistrial (Marquart et al. 1989). The current study shows that in states that allow these arguments, assertions about future dangerousness do impact the juries’ decision to sentence a defendant to the death penalty.

The current study is unique in two ways. Firstly, to our knowledge, it is the only study that has analyzed language strategies used by prosecutors in real capital murder cases rather than simulated cases. Secondly, this study incorporated empathy for the victim as a linguistic variable that may influence jurors’ decisions. To our knowledge, empathy has not been previously studied in research investigating moral disengagement and juror decisions. The results of this study indicate that language that may evoke empathy for the victim may be an important means by which jurors rationalize sentencing a defendant to death. This should be a variable of interest in future studies on moral disengagement in juror decision-making.

While this study has contributed to moral disengagement research, it is not without its limitations. A significant limitation is the small number of cases (25) that we were able to include. It was also challenging to control for all possible variables. In the current study, we were able to hold constant the gender of the defendant (male) and the type of state in which the trial took place (a state that allowed future dangerousness arguments); however, controlling for other possibly mitigating variables was difficult. Some variables that were specified in some transcripts were not specified in others (e.g., race of defendant and/or victims). Two of the cases involved non-adult victims. When those cases were omitted from the analyses, the variables future dangerousness and empathy for the victim were only marginally significant. This could mean that these variables are especially influential for non-adult victims; however, one of the cases, which involved a 2-year-old child, resulted in a sentence of life without the possibility of parole, so that change in significance can simply be due to a loss of power.

Future studies could account for the prosecutor’s attitudes toward the death penalty to see how it impacts sentencing outcomes because they could subtly lean toward a sentencing outcome in their arguments by employing (or not employing) moral disengagement strategies. The authors obtained anecdotal information from criminal attorneys to the effect that prosecutors are typically not explicitly trained in the use of language strategies that might evoke moral disengagement in capital juries, although they are trained in argumentative tactics, such as adjusting one’s tone of voice. The current study adds to the existing body of knowledge regarding moral disengagement strategies and jury decision-making by providing further evidence for the influential nature of future dangerousness arguments and by introducing a new moral disengagement strategy (empathy for the victim) that may influence jury sentencing decisions.