Voyant Between the Lines

This blog post is my introductory reflection after using Voyant, a text analysis tool!

1. What information did you choose to delete and keep? What stopwords did you choose and why? What was your thinking about how these choices would impact your work in Voyant and your reading of the text?

For my initial analysis using Voyant I used the preselected collection of Austen’s novels. Because of this I did not clean or filter my data before uploading it to the tool. However, by instinct I would porbably be interested in removing “non-critical information” which is to say information that does not directly inform the reader of the context of the book. This could include appendices, tables of contents ect… That being said, the question is really situational, if you wanted to study the transformation of indices in the 19th century you would want to include only that. Voyant’s filtering capabilities permit the reader a great deal of control over the work that they examine but this is essnetially a double edged sword because it means the reader must have a clear vision of their questions or at the very least the tenacity to conduct a broad anlaysis if they choose to just a explore a collection of works.

2. What three Voyant tools did you find most helpful and relevant? Why?

I liked most of tools in Voyant but three seemed to be particularly useful to me: the word cloud, the summary, and the contexts tool. Word clouds are useful for so many reasons and in numerous contexts but generating them from preselected writings is actaully very difficult without a tool like voyant. The generation of this cloud has the ability to start conversations in classrooms and also idenitfy broad themes in collected works that may have been invisible without it. In the context of austens novels, these could include silence and the role it has played in her works. The next tool I thought was interesting is the summary tool. This one would be especially useful when working with a larger corpus of text that you may not have previously been able to quantify. The summary include things like the number of words analyzed but also more interesting metrics such as the density of vocabulary, its readability, and distinctive words used in throughout the text. Lastly, I loved the contexts tool. This is perhaps one of the most powerful tools in Voyant depending on how you use it, especially for anthropology. Understanding the context in which words emmerge is a powerful analytic and may allow a researcher to understand the relationship between certain words and the context that precipitated their inclusion in the writing.

3. What struck you most about what Voyant did or didn’t reveal? (Were you expecting to see something and didn’t? Was it what you expected? Etc.)

Voyant is clearly a powerful tool when used correctly and with precision. I think that in order for it to be the most useful in DH settings the researcher must have some goal in mind when using it. While using it as an exploritory tool may be intiially useful the computational power it uses makes working with entire collected works difficult if you cannot streamline your questions and what it is you are searching for within the documents. The tool frequently froze and crashed on me even when performing basic tasks on the preloaded datasets they provide. I also expected this to be an analytical tools and instead its almost just a huge filtration tool. What I mean to say is that without an analytical question, a theoretical framework voyant will not tell you anything other than how frequently certain words or patterns occur within your data. In and of itslef this is a pwerful strategy but before being exposed to DH and this tool I thought the tool was going to extrapolate literary themes or ideas worth investigating further. It turns out it does do this, but you ahve to assign value to those patterns. it will not do the hard work of anlaysis for you, if you were looking for that, you’ve come to the wrong place.

4. Based on what you see in Voyant, how might you read the text differently in the future, that is, what else might you read or look for?

Interestingly, I am not sure that the tool has affected how I will read in the future, and cetainly not as it applies to journal manuscripts. However, it has broadened my understanding of how to analyze text, and what possible meta analysis are possible. This is becoming a burgeoning approach in anthropology and it could considerably benefit from a tool like this. Metadata and its analysis becomes computationally intensive very quickly and though there are other computational methods that could do this, the informatic knowledge required to perform those from scratch is generally beyond the scope of most researchers which makes the general accesibility of voyant very attractive.

5. Do a multi-dimensional assessment of Voyant as you experienced it. What do you see as flaws or aspects that merit caution? What are its benefits? If you were asked by someone whether they should try it what would you advise?

One of the major flaws is that its processing ability is limited especially when you being to consider anlyses of collected work or anthologies. Even power computing systems like mine had difficulty in maintaining the website operational. However, a more detailed anlaysis of less works may work to ameliorate this flaw. Another one is that the learning curve required to understand the tools is pretty steep. The ? mark tool was very helpful but i almost wish that they had a greater introduction to the tool on the landing page so that a researcher has the ability to engage with the tool before beocming overhwlemed with its ability. That being said, it ability to visualize data and patterns within the text youve selected for analysis is far and away its most redeeming quality. For new researchers in text analysis I would suggest first using one of their collected works to explore tha capabilities of the tool before they commit to their time and effort to working with it. It can be powerful but perhaps only to more advanced researchers that have a very specific idea of what they need data on and how to retrieve it from a large corpus of text through this tool.

6. How are your ideas about “close reading” and “distant reading” impacted by your experience with Voyant?

As I had mentioned earlier I dont think the tool has affected how I read but it has changed the type of questions I thought were possible to explore in text. How would scientific text analysis benefit from a tool like this? Could broader themes of ethics, methods, and protoclls be explored with something like Voyant? Perhaps… I will ammend my earlier statement to include the fact that it might have given me a greater appreciation for what is available for analysis within the literal words and not just the metpahors of what people choose to write.