Text Analysis with Voyant

Content Warning: presence of the N-word in visuals (with the hard-R), topics surrounding chattel slavery in the United States

Voyant is a digital text-mining tool that allows for the user to analyze their submitted corpus through different visual tools. This includes (but is not limited to) word clouds, relative frequency graphs for chosen words/phrases for one or all of the documents in one’s corpus, a context tool to analyze the preceding and proceeding words in sentences containing a chosen word/phrase, a text reader to provide full texts that are being analyzed, and a summary tool that spots patterns and distinctive qualities of each document in one’s corpus. Each tool has different parameters one can set to fit their analytical needs; creating a malleable tool with a lot of different options for exploration!

Text mining can be useful if one’s source material relates to each other in some way, but the corpus’ overall length remains an obstacle for extensive/effective analysis. Some projects utilize millions of pages of text, therefore digital tools like Voyant were created to parse through raw data to create a user-friendly experience.

Voyant’s utility is in its ability to find core themes within documents or throughout the corpus. Below is an example of a word cloud generated through the “cirrus” tool; using a corpus of interviews with formerly-enslaved people in the state of Georgia:

This was useful in determining topics, themes, and the framing of the interviews. The word cloud shows themes of racial disparity, age, a sense of place through the home, plantations and the hierarchy inherent to slavery, and some instances of African-American Vernacular English (or AAVE). It sums up different aspects of the interview, which while lacking its entire context, can be a representative introduction to the content of specific documents or corpora.

The “context” tool was another feature that allows for a much better understanding of this corpus. In Maryland interviews, the “summary” tool identified the distinctive phrase “rezin” seven different times. When investigating further, it turns out that this was a local “freeman”; “Uncle ‘Rezin’ Williams.” It not only brought an individual’s story to life, but also gave me an opportunity to explore something specific to the state of Maryland through these interviews. The context tool filtered to show instances of “rezin” is pictured below:

Prior to exploring digital tools, I was not familiar with text-mining tools, or how they worked. Voyant (and I imagine other programs in the same vein) enabled me to explore texts in a whole new way. Being more of a visual learner, I appreciate when projects or texts have an interactive element that deepens one’s understanding of the topic that they are exploring. Through visual analysis driven by text, it is both a way to connect to the broader public, while succinctly conveying one’s points through more accessible means. I am grateful for tools like this, and I hope they only get utilized more in the years to come!

Leave a Reply

Your email address will not be published. Required fields are marked *