Measured Unrest In The Poetry Of The Black Arts Movement

Ethan Reed (, University of Virginia, United States of America


“Anger is loaded with information and energy,” says Audre Lorde in a 1981 speech on its political uses—but the nature of this affective information, sparked by a given political present, becomes highly vexed when articulated through literary objects (Lorde, 1997: 280). On the one hand, the cool detachment of aesthetic mediation keeps the politics of experimental works from being seen as mere propaganda, but runs the risk of appearing elitist or self-indulgent. On the other hand, the red-hot political outrage of a protest poem by Amiri Baraka or Sonia Sanchez grounds itself in the present, but may be attacked for subordinating aesthetic sophistication to political agendas.

Building on recent scholarship (like the work of Lauren Berlant and Sianne Ngai) suggesting that feeling gives structure to cultural formations, my research investigates the provocation and articulation of emotions like frustration, anger, and discontentment within recent US literary history as they relate to systemic injustice. An agitprop play that ends with shouts for workers to unite in class revolution; a poetic broadside that vents frustrations against white supremacy in America; a novel that indulges in a revenge fantasy against America’s colonial history. Unlike plays, poems, or novels that seem to obscure, submerge, or confound their own political dimensions, these works wear their hearts on their sleeves: they are frustrated, fed up with how things are, and unafraid to speak truth to power in a direct, seemingly “un-literary” way.

“Measured Unrest in the Poetry of the Black Arts Movement” offers a proof-of-concept for performing sentiment analysis on some of the most politically and affectively charged poetry of the 20 th century in America, that of the Black Arts Movement of the 1960s and 1970s. The BAM first took shape at the height of the Black Power Movement with the foundation of the Revolutionary Theatre by Amiri Baraka in 1965. As Larry Neal—one of BAM’s principal theorists—says in a 1969 manifesto, the “Black Arts movement seeks to link, in a highly conscious manner, art and politics” toward “the liberation of Black people” (Neal, 1969: 54). Moreover, what Neal calls the movement’s “black esthetic” is famous for its affective dimensions, often exploring the limits and political uses of anger, frustration, and militant poetic rage. But while BAM writers sought to link art and politics through explicitly racial terms, many—though by no means all—were marked by a failure to attend to the intersections of gender with racial injustice.

In this project I ask two questions in particular: first, how are the feelings associated with injustice in this corpus coded in terms of race and gender? And second, what can natural language processing techniques like sentiment analysis show us about the relations between different dimensions of poetry—like affect and gender—given that poetry is highly figurative and notoriously difficult to quantify in terms of sentiment or opinion?


In addressing both these questions, this project uses a small corpus of poetry—currently 26 books—from prominent BAM authors. I employ both close reading as well as machine reading techniques, combining the powerful scale of sentiment analysis with the granularity of traditional literary analysis in an effort to explore the intersections of feeling, gender, race, and injustice in the radical poetry of this period. My goal in this project is not to develop a sentiment classifier that works on experimental poetry in English. Rather, it is to see what existing classifiers can show us about a specific corpus of poetry.

In this sense, I use pre-existing sentiment classifiers like VADER and Pattern (via TextBlob) to perform a kind of exploratory computational analysis on my corpus (Hutto and Gilbert, 2014; De Smedt, and Daelemans, 2012). Rather than use these tools to make general claims about this incredibly diverse body of poetry, I test, experiment, and make targeted use of sentiment analysis techniques to pursue research questions already present in existing scholarly conversations—for example, how poets might tie heightened affects to an explicitly political quest for racial justice in America. The insights I draw from my computational analyses, then, go hand in hand with more traditional literary practices. Moreover, my methodology aims to acknowledge the fact this poetry was written in the shadow of government surveillance programs, active FBI counterintelligence operations, and a larger culture fearful of radical thought. Because of this, my project explores the fraught methodological implications of using distanced, potentially decontextualizing computational text analysis techniques to think through BAM poetry, and how these methods might best be used to pursue questions, problems, and lines of inquiry centered around black thought and experience.

The already vibrant conversations on sentiment analysis and natural language processing more generally have been illuminating in forming these thoughts and questions. The discussion between Matthew Jockers and Annie Swafford on the  Syuzhet package and “archetypal plot shapes” has helped me not only to consider the current possibilities and limitations of sentiment analysis as applied to literary corpora, but also to think through the kinds of results we expect from digital projects and how we verify those results as an academic community (Swafford, 2015; Jockers, 2015). With regards to poetry and NLP more specifically, Lisa Rhody’s topic modeling of highly figurative ekphrastic poetry is a great model for how unexpected failures in textual analysis can also be productive, prompting us towards new questions as well as new understandings of familiar methods like close reading (Rhody, 2012).


I have implemented NLP techniques with NLTK and TextBlob, a text-processing Python library, on my collection of 26 books of poetry. I have also used two sentiment classifiers—Pattern (via TextBlob) and VADER—to evaluate my corpus for sentiment and interpret my results. While this work is ongoing, so far my work comprises explorations and experiments in the smaller-scale uses of sentiment analysis in the study of poetry and affect.

For example, Pattern considers Quincy Troupe’s “Come Sing a Song”—from his 1972 collection Embryo Poems, 1967-1971—to be the most negative poem in my entire corpus. In a corpus of poetry containing direct attacks, extreme invective, and explicit takedowns of individuals, groups, and institutions, I did not find this poem to contain an exceptional amount of negative sentiment. On the contrary, I found “Come Sing a Song” to be positive and celebratory with regards to black life and black artistic expression. VADER, meanwhile, considers Nikki Giovanni’s “The True Import of the Present Dialogue, Black vs. Negro”—from her 1968 Black Feeling, Black Talk—to have the most negative sentiment in the corpus. These results are very much in keeping with other human readers of this poem: critics consider it to be one of the most significant and famous examples of a certain type of angry, militant, even aggressive poem. Where Pattern and I disagree strongly over the feel of Troupe’s “Come Sing a Song,” critics and VADER seem to agree that Giovanni’s “The True Import” has, on the surface, an exceptional amount of negative sentiment compared with its contemporaries.

Among other things, my project analyzes discrepancies and correspondences such as those described above. Already, my findings have revealed an interpretive disjoint between the denotative affective impact of words—what might be called their surface sentiment—and their more nuanced affective import as shaped by poetic, literary, social, and political contexts. A sentiment classifier like VADER, for example, highlights the intensity of negative sentiment in a poem according to the words and phrases it contains without the literary and historical context of their use. This kind of surface reading, attuned specifically to words’ immediate affective impact, anticipates the space between a surface anger that can spark feelings regardless of context and a poetic form that, in the case of Giovanni’s “The True Import,” leverages negative sentiment to address meaningful social issues in a productive, ultimately positive way. By investigating these poems through conventional literary methods (i.e., historical contextualization, close reading, consideration of relevant scholarship) and computational methods (in this case Pattern and VADER), while also investigating the histories, intended use contexts, and potential biases of the chosen computational methods, this project provides an opportunity to examine what it is, exactly, that provides a book, poem, or poetic line with its emotional charge.

Appendix A

  1. De Smedt, T. and Daelemans, W. (2012). “Pattern for Python.” Journal of Machine Learning Research 13: 2063–67.
  2. Hutto, C. J. and Gilbert, E. (2014). “VADER: A Parsimonious Rule-Based Model for Sentiment Analysis of Social Media Text.” Eighth International Conference on Weblogs and Social Media. Ann Arbor, MI, June 2014.
  3. Jockers, M. (2015). “Revealing Sentiment and Plot Arcs with the Syuzhet Package,” February 2, 2015. (accessed 27 February 2018).
  4. Lorde, A. (1997). “The Uses of Anger.” Women’s Studies Quarterly 25, no. 1/2: 278–85.
  5. Neal, L. (1969). “Any Day Now: Black Art and Black Liberation.” Ebony 24, no. 10: 54–62.
  6. Rhody, L. (2012). “Topic Modeling and Figurative Language.” Journal of Digital Humanities 2, no. 1. (accessed 27 February 2018).
  7. Swafford, A. (2015). “Why Syuzhet Doesn’t Work and How We Know,” March 30, 2015. (accessed 27 February 2018).