A central question facing scholars and policymakers is how to anticipate the behavior of rogue political actors, including authoritarian leaders and non-state actors like rebel groups and terrorists. By analyzing the linguistic patterns of leaders and other key political actors, new data and metrics can be collected that assist policymakers to build improved methods for anticipating potential areas of unrest, instability, and conflict. Our goal is to determine if political actors, such as leaders, use particular language or styles of speaking before performing acts of violence and acting on threats. Distinguishing between bluffs and threats is one key issue that we believe linguistic analysis can help address. Actors in global politics, whether leaders of countries, leaders of rebel movements, or leaders of terrorist networks, have the opportunity to make speeches demanding policy concessions from other actors. We describe these demands as either credible threats, or bluffs. As a part of the bargaining process, under some conditions, actors will be required to follow through on their threats, whereas other circumstances allow actors to use bluffing strategies to obtain policy concessions (Powell, 2002; Putnam, 1988). This is particularly problematic when dealing with rogue states, which may bluff repeatedly before carrying out a threat. Distinguishing between the two conditions is difficult by intention: actors want their bluffs to sound credible. These speeches include bluffs made by North Korean leader Kim Jong Un (n=9), and credible threats – those made and followed through on – from a sample of world leaders (n=21). The bluffs and threats are statistically significantly different across several linguistic dimensions.
Preliminary analyses indicate that credible threats tended to use more negative language and far more passive voice than other speeches. The increased use of passive voice for serious threats may indicate that leaders either use more bluster for empty threats (speak loudly and carry a small stick), hedge their bets (leave an “out” open), or actively downplay the severity of their threat when they actually plan to attack, protecting their private information (deception). Credible threats also reference the past and future more often than do bluffs. Bluffing strategies tend to be more verbose than the more concise credible threats (“words per sentence”). This could indicate that irresolute leaders who do not intend to strike militarily tend to bloviate more than resolute leaders who are prepared to use force. We may also be able to extrapolate from this measure who the intended audience is. It is widely understood that the series of bluffs levied by Kim Jong Un were less directed at initiating conflict with either South Korea or the United States, but were more indicative internal posturing for his domestic audience with whom he was trying to establish credibility. As a captive audience under a repressive authoritarian regime, they have no choice but to listen to their leader’s oratory. Bluffs and credible threats both tend to use informal language, but credible threats are much less formal than bluffs. Credible threats use more in-group language (we, us) than do bluffs. This may indicate that the speaker has overcome the collective action problem inherent in mobilizing for conflict, as collective language can signify the presence of a coherent, coordinated group a clear in-group with a definite agenda (Olson, 1965).