Comparing Tools of Text Analysis


When working in the field of Digital Humanities, there are three computer based tools that can be indispensable for assisting with vast quantities of textual analysis, for digitized data sets that would be too time consuming and cumbersome to read through individually. Voyant Tools, Kepler.gl, and Palladio are all web based, easy to use, interactive and free. These digital tools have the capability to visually show patterns and trends that would otherwise be difficult for users to detect without the help of computers. All three were used in studying the Works Progress Administration’s Slave Narrative Interviews that took place in the 1930s. With over 2,000 interviews conducted, the text based tools help users detect certain trends or patterns through visualizations. 

Voyant Tools takes uploaded sets of digitized textual data to create a variety of visual charts and graphs that help users with quantitative text analysis, based on the words used in the data set. For the WPA Slave Narratives, Voyant Tools can create a word cloud to show which words were used most frequently, while the filters feature allows the user to eliminate unnecessary words. The other features include graphs of most frequently used words, a summary of the number of words in each document, and where particular words can be found within the documents. Voyant Tool’s text analysis can show if certain words were used more frequently based on geographic region. However, Voyant Tools does not have the capability to create maps, in order for users to visually see where the interviews took place, or where the former enslaved lived. 

Kepler.gl, developed by the transportation company Uber, is a geospatial data visualization tool that uses latitude and longitude coordinates to create maps that can show visible patterns in a variety of ways. Users can choose points, arcs, clusters, and more to show the relationship between where interviews took place, and in which geographic regions or cities did the former enslaved live and work. Another helpful feature of Kepler.gl is the timeline graph that shows when the majority of interviews took place. There are also filters where users can narrow down those who were interviewed to show whether they worked in a house or fields. 

Palladio shares similarities with Kepler.gl, in that they both have mapping capabilities. Where they differ is that Palladio can visually show the network links between people, places and topics or ideas. Neither Kepler.gl or Palladio offer any text based analysis of the words in the body of interviews. Palladio’s strength lies in its visual webs of relationship connections. Users can filter webs based on only one gender, or ages of those interviewed, or by who conducted the interviews. Users can look for whether there were regional differences in topics more commonly mentioned, or if there were differences of topics in the interviews based on gender of the interviewee. 

Because each of the text analysis tools has certain features the others are lacking, if used together they can be most helpful to users in order to gain a richer picture of any large data set comprised of texts. Each of the text analysis tools has ways of presenting large quantities of digitized data and creating visualizations that reveal otherwise unseen patterns and trends through charts, graphs or maps. 


Leave a Reply

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