Deep into that Darkness Peering : A Computational Analysis of the Role of Depression in Edgar Allan Poe's Life and Death

Dean, Hannah and Boyd, Ryan (2020) Deep into that Darkness Peering : A Computational Analysis of the Role of Depression in Edgar Allan Poe's Life and Death. Journal of Affective Disorders, 266. pp. 482-491. ISSN 0165-0327

Text (Dean & Boyd (2020) - A computational analysis of the role of depression in Edgar Allan Poe's life and death)
Dean_Boyd_2020_A_computational_analysis_of_the_role_of_depression_in_Edgar_Allan_Poe_s_life_and_death.pdf - Accepted Version
Available under License Creative Commons Attribution-NonCommercial-NoDerivs.
Download (2MB)

Abstract

Background: To help shed light on the peculiar circumstances surrounding the death of the famed macabre and mystery writer, poet, editor, and literary critic, we explored the potential role of depression in the life and death of Edgar Allan Poe via his written language. Method: Using computerized language analysis, we analyzed works from Poe’s corpora of personal letters (N = 309), poems (N = 49), and short stories (N = 63), and investigated whether a pattern of linguistic cues consistent with depression and suicidal cognition were discernible throughout the writer’s life, particularly in his final years. Building on past work, language scores were collapsed into a composite depression metric for each text. Data from each work type was subsequently compiled and graphed into a single plot by year, with scores exceeding the 95th percentile (p <.05) considered statistically significant and treated as potential depressive episodes. Results: Significant, consistent patterns of depression were not found and do not support suicide as a cause of death. However, linguistic evidence was found suggesting the presence of several potential depressive episodes over the course of Poe’s life – these episodes were the most pronounced during years of Poe’s greatest success, as well as those following the death of his late wife. Limitations: Given the sampling method, it is not possible to establish direct causality; results should be considered informed but tentative. Conclusion: This investigation demonstrates the utility of language analysis for capturing disruptive/maladaptive emotional responses to life events.