Skip to content

Reconciliation in Chaos: Tracing word distributions and sentiment in the monologues of “Palace of the End”

By: Nathan Drezner

Judith Thompson’s 2007 docudrama play, Palace of the End, presents a series of characters entrapped in their own histories, tangled up in pasts that prove torturous to their everyday selves. Each speaker is faced with the burden of themselves, coping with the consequences of both their own actions and the actions of others. The monologues presented in Palace provide these individuals with a voice—a direct line of communication between their personal conflict and the outside world. The play consists of three monologues spoken by narrators connected to either the Iraq War or the Persian Gulf War; two of these narrators died as a result of the conflicts and, through their monologues, look back at their lives after their deaths. An unnamed soldier inspired by the true story of Lynndie England struggles to define and come to terms with her abuse of the prisoners at Abu Ghraib in “My Pyramids”; David Kelly seeks solace after his suicide over the actions he undertook while investigating the production of biological weapons in Iraq in “Harrowdown Hill”; Nehrjas tries to find peace from the constant torture she and her family face and her subsequent death in the Gulf War in “Instruments of Yearning.” In their monologues, the characters work to reconcile their histories after their involvement with American operations in the Middle East—whether that be reconciliation via ostracization, death, or both.

To better understand how language defines the ways these characters cope with the effects of their pasts, I have conducted a computer-aided analysis—a sort of “digital dramaturgy”—of the language in each monologue by tracing their word counts, or the frequencies of word usage within them, and sentiments of their dialogue to find patterns of emotion expressed by each character. The data obtained through this analysis yields surprising results and reveals thematic connections within the play. In particular, each character’s perception of themselves and of those around them guides their method to finding peace as they demonstrate a significant growth of trust over time through their choices of language. Each monologist finds reconciliation for themselves as they reflect on their tortuous pasts, realizing the consequences of their actions and stories as they work their way through their respective personal histories. 1 These findings elucidate important textual questions, such as whether a character’s sense of reconciliation can counteract their problematic history.

The first method of analysis categorizes and visualizes the word count of each monologue, connecting each character’s expressions of empathy and selflessness to their relationships with others.2 In tracing the word counts, I worked my way backwards through the text, first identifying repetition in “Instruments of Yearning.” This monologue’s visualized word cloud contains a significant amount of words relating to “the Other,” an individual or a group defined by their differences to the Self. The most prominent appearances of the Other in “Instruments of Yearning” are “son,” “people,” and “woman,” which appear 26, 12, and 11 times respectively.3 Several other such words, including “children,” “mother,” and “child” also appear. This diction suggests that Nehrjas has a strong connection to others—of course, her story is one of family, as it relays her abuse and torture alongside her son. Nevertheless, other vocabulary in this monologue correlates to a strong expression of humanity, perhaps ironically, considering her own assessment that “an infidel has no friends” (Thompson, 31). Nehrjas feels isolated, yet she is also intensely connected to those around her—her husband, her son, her people. She is surrounded by others.

By contrast, the diction of David’s monologue shows a character bound tightly to a smaller set of others, compared to the immense breadth of connection expressed by Nehrjas. Words relating to the Other that appear most frequently include “family,” which occurs nine times, “people,” which appears eight times, and “child,” which appears seven times. References to family, however, aren’t related to David’s own relatives. Instead, the references are connected to the story of Jalal, a man whose family was brutally raped and murdered by American soldiers. David’s references to the Other indicate that he is only connected with a small set of close friends, showing a man who is far more alone than Nehrjas. David finds solace by reflecting on his connection with another man’s family rather than his own. His fears related to his own family—specifically that his daughter will find him in the woods—are distanced from his concerns with other families. David’s connection to the Other is further removed than that of Nehrjas’.

The soldier finds herself removed from the Other to the greatest degree. In her monologue, words relating to the Other include “man,” “girl,” and “Charley.” Both “man” and “girl” are used detachedly, the latter referring to the soldier herself. Meanwhile, “Charley” refers to the man who coerced the soldier into committing war crimes—including the titular human pyramids that prisoners were forced to form in Abu Ghraib. The soldier’s lack of empathy separates her from other speakers in the play, who forge connections within a community. Nonetheless, a transition between each of the three monologues, guided by a connection between each character and the Other, shows how the play defines empathy between characters. The language of each character’s monologue reveals this difference: the soldier finds difficulty empathizing with a local community, David finds solace in the families of others, and Nehrjas finds consolation through her own nuclear family and friends.

As a secondary mode of analysis, I focused on emotional transitions between monologues using the Saif Mohammad and Peter Turney database.4 Several trends quickly became apparent. First, an emotional transition over the course of the play leads to an increase in words related to anticipation, trust, and surprise, while words related to anger, fear, and sadness increase throughout the play then dissipate in Nehrjas’ monologue. These changes in sentiment indicate developments over the course of the narrative. The number of words related to trust notably increases, which shows that as the play progresses, each character’s belief in the veracity of their own narrative also grows. The soldier has the smallest percentage of words related to trust in her story, while David has a higher density of words related to trust, and Nehrjas has the most words related to trust relative to other vocabulary in her monologue. The increased usage of vocabulary associated with trust indicates that each character begins to believe more of what is happening to them and better understands their own situation. As a method of reconciling history, this shift towards belief in oneself is quite important, demonstrating that even as they retell their histories, characters must trust the language they use to define their own experiences. By the conclusion of the play, there are far more words related to trust, concluding the arc of the three monologues with more confidence in each related history.

Over the course of the narrative, the overall relationship between positive and negative words also shifts. In “My Pyramids,” the number of positive words spoken exceeds the negative words, while “Harrowdown” presents an equal portion of positive and negative words, and in “Instruments of Yearning” positive language is more prevalent than negative. The soldier, who perpetuates the violence, uses far more negative language than any of the other characters. David, who caused more violence when he lied about the existence of weapons of mass destruction in Iraq, but sought to remedy his actions by leaking the lies to the press, shows conflict in his sentiment: the ratio of positive to negative words in his monologue is equal. Nehrjas, a victim of the Persian Gulf War, is characterized by her use of more positive language than negative language. These findings suggest a relationship between a character’s sense of positivity and negativity and their role as perpetrator or victim. The language used in the narratives of the victim is, in a certain twist, more optimistic than that of the perpetrator, as it is characterized by more positive diction. David, caught between the role of victim and perpetrator, finds himself unable to escape either role. Instead, he is trapped in a state of conflict wherein he has difficulty reconciling his past and present.

The third study I conducted traces the patterns of positive and negative emotion over the course of each narrative.5 The findings from this study represent a specific analysis of the overall sentiment of each monologue and the play as a whole, tracing how each character’s interpersonal relationships evolve over the course of each monologue. The results highlight specific moments in the text, especially when sentiments of language dictate the narrative. In particular, these results point to transition points and climaxes. An extreme sense of negativity characterizes “My Pyramids,” as an overwhelming trend of negative words appear in each paragraph, made apparent in both the Bing and Finn Årup results. The graphs point to the opening paragraph, where the soldier reads responses to her actions: “Drown the slut in acid, she should be hogtied, damn she’s ugly I’d put my wang in her ass she is the ugliest female I have ever seen drown that bitch in acid,” she reads (Thompson, 6). Despite the soldier’s separation from significant others—friends, family, or other connections—she is desperately attached to the language the world uses to describe her. The soldier is alone save for a constant barrage of online criticism, tearing apart her empathic tendencies and ostracizing her from those around her—and from the world at large.

David’s monologue in “Harrowdown Hill” tells a different story—one of incredible conflict. Just as David’s uses of positive- and negative-associated language parallel each other, extremely negative dialogue compliments moments of extreme positivity in his monologue. These situations take two forms: first as a transitional moment, second as a moment of pure conflict. Early on, as David describes his transition into suicidal ideation, his language parallels his internal conflict. David explains how he “was fine. [He] was strong. Very excited to go back to [his] beloved Iraq” (Thompson, 19). Words like “beloved,” “fine,” strong,” and others signify the positive sentiment of this passage. David’s tone, however, quickly shifts when he describes his methods of suicide, detailing the horrible contradiction he faces between a life of service for his family and country and the consequences his actions in Iraq. This contradictory language creates a sense of displacement that continues throughout the monologue, with David later describing his appreciation of Jalal’s family, and the subsequent rape and murder its members face. Words like “beautiful,” “delicious,” “fine,” and “loved,” appear twice, indicating a sense of positivity, while language including the words “crude,” “cry,” “intimidating,” and “raped” contrast David’s sense of love. The semantic contradictions in David’s account of his history demonstrate the conflict he endures in the moments leading up to and during his suicide. Just as his actions counteract each other—such as when he lies and subsequently leaks information about biological weapons in Iraq—so too does his language. David’s monologue clarifies the context of his suicide: through his contrasting diction, David indirectly expresses the difficulty he experiences in reconciling his actions.

“Instruments of Yearning” positions Nehrjas as the most resilient narrator in the play, and her language demonstrates a level and clarity of emotion absent from both David and the soldier’s perspectives. Her optimism is devastating, yet it keeps her steady. The events that unfold throughout her life are similarly devastating as David’s, yet her optimism is unrestrained. Just before describing her torture, she thinks “of an amusing story,” telling of an interaction with an American ambassador; here, her positive approach to politics acts as a prologue to the horrors of her torture (Thompson, 34). Unlike the others, Nehrjas’ positivity remains. Her monologue fluctuates from the negativity of the descriptions of her history (paragraphs 65-86) into a more positive outlook, trading back and forth between an optimism for her past and future and the contradiction that the truth of her life and torture holds.

At the conclusion of her monologue, Nehrjas’ emotional intensity declines. This trend parallels models of the other monologues—in all of their conclusions, the frequency of words expressing strong emotions declines. This trend indicates that the characters have, at the end, achieved a sense of solace. The final stage direction symbolizes that all three characters achieve a sense of emotional security: “The three performers stand, somehow communicate with each other, and walk off” (Thompson, 40). This sense of reconciliation—stemming from a decline in emotional upheaval—suggests that the characters are able to move forward from the horror of their histories. This reconciliation through emotional displacement occurs differently for each orator. For the soldier, an upward trend represents her reconciliation, moving from a highly negative opening towards a conclusion of balanced emotion. David’s moments of conflict increase, peak, and finally decrease toward a balanced sense of emotion. Nehjras’s emotional output declines over the course of her monologue, reaching a similar point of emotional neutrality by its conclusion. The monologues all show closure through neutrality, and demonstrate that an exhaustion of emotional output signifies reconciliation for each character.

The evolution towards reconciliation raises the question of whether each character counteracts their history by finding reconciliation with it. The soldier’s actions at war outweigh her ability to achieve a sense of closure and reconcile what has happened; even if she reconciles with her past, her actions remain unchanged, and the torture she conducted continues to inflict its damage. For both David and Nehrjas, reconciliation does not change their deaths. Finding reconciliation with their own actions and history does not mitigate the violence they have enacted or endured, although they can let go of the burden of these actions. The internalization of both characters’ personal traumas represents a different sort of conclusion, suggesting that the purpose of each monologue is not to solve the problems of history, but instead to personalize it, inflecting the language with the sentiment of each individual rather than the voices of outsiders. Personal reconciliation takes precedence over the resolution of one’s history, suggesting that each character’s ability to reconcile what has happened is the best solution they can undertake in the light of what they have done and what has happened to them. Reconciliation, then, doesn’t matter for history, but it matters for the person. As the play focuses on realizing the individual hidden behind a headline, this reconciliation is hugely important in identifying how these histories were allowed to occur in the first place.

Works Cited

Thompson, Judith (November 2007). Palace of the End. Playwrights Canada Press.

Photo: “The Art of Mapping,” Yacan1

Code and data available here

1 To conduct a computer-aided analysis of each monologue, I began with Voyant, an accessible tool for language analysis. Using Voyant, I jump-started a thematic inquiry by investigating “word-clouds” generated by the system. This provides a visual demonstration of the most frequently occurring words in a text, after the removal of a pre-set list of stop-words (“the”, “a”, “and”, etc.). I then wrote several programs to determine the sentiment of each monologue and the sentiment over the duration of the play. To do this, I wrote scripts in R which take input as a digitized version of each text. First, the program divides the text into smaller sections. I did this in two ways: by dividing the text by paragraph, as defined by the monologist, and then by dividing the text into clusters of words (i.e. groups of 40, 50, 60 words etc.). Next, the program tokenizes each group of words, divided by paragraph or cluster, into individual words. Stop-words were removed from the text using a predefined library. The program compares the individual words to three libraries of pre-categorized words. First, the Saif Mohammad and Peter Turney database, a lexicon of 13,900 words categorized by their association to two sentiments (positive and negative) and eight emotions (anger, anticipation, trust, disgust, fear, joy, sadness, surprise). This database calculates the overall sentiment of each text and shows the trends of specific emotions from monologue to monologue. To determine the evolution of emotional trends over the course of each monologue, I used the Bing Liu database, a dictionary of 6,700+ words, uniquely categorized as “positive” or “negative,” as well as the Finn Årup Nielsen database, a dictionary of 2,400 words, each given a score from -5 to 5 according to their positive or negative association. This methodology represents a relatively elementary form of natural language processing: taking language and processing it in a way that is comprehensible for a computer.

2 Results were obtained using Voyant Tools.

3 My method of modeling relies on a pre-set list of stop-words, compiled without the use of a computer. As a result, this means that words that could be significant—in a novel, the word “said” might be removed based on this list—could be eliminated from data analysis. The stop-words removed from my analysis are based around Zipf’s Law, dictating most used words in language and defining words that could or should be removed in data analysis. This method, however, led to inconsistencies between Voyant’s tools and my personal methods of analysis—the word “known” provided insights from the word cloud but was not included in my corpus for sentiment analysis. This could be improved through custom stop-word removal, where the user dictates which words are removed and why, avoiding text-specific removal of important language.

4 Results were obtained by labelling each word with one of the 10 sentiments registered in the Saif Mohammad and Peter Turney database.

5 Though this data reveals notable trends and points in the text, it is also worth discussing possible problems and difficulties in this data, and how the methods of analysis could be improved. The data collected relies on individual markings for words, and, as a result, can create problems of reliability. Since individual words are often not indicative of true sentiment—context plays a huge role in meaning—these forms of sentiment analysis pose the risk of leading the researcher in incorrect analytical directions. To correct for this, I’ve worked to incorporate several methods of analysis and compared them to find trends, rather than relying on a single mode to discover a single observation. To improve this, methods of topic modeling and word embedding could be applied to the text. These methods of analysis provide contextual information about the text, finding meaning via machine learning rather through subjective, human labeling. Future analyses using these methods could create more accurate clusterings of language and perhaps point to different significance in a text.

6 I conducted this analysis in two ways: first, by splitting each monologue by paragraph, and second, by splitting each monologue into sections of a certain number of words. I found the first method  far more enlightening as author-defined sectioning showed more specific points of contrast and conflict. Sectioning by word count, however, revealed similar trends towards neutral emotional output, again suggesting a trend towards emotional neutrality over the course of each text.

7 “afinn” refers to data obtained with weighted measurements, i.e. each word in the text was ranked between -5 and 5, a scale of negative to positive. “bing” refers to data obtained with unweighted measurements, i.e. each word in the text was labeled as either “positive” or “negative” and the results were tallied.