Voyeur

Hermeneuti.ca is a collaborative project by Stéfan Sinclair & Geoffrey Rockwell to think through some foundations of contemporary text analysis, including issues related to the electronic texts used, the tools and methodologies available, and the various forms that can take the expression of results from text analysis.

Your task is to learn a bit more about and play around with Voyeur. How could you use this? How could the tool/instructions be improved?

3 Responses to “Voyeur”

  1. Carrie Tallichet Says:

    As a cultural historian, I see Voyeur as useful in analyzing published works, newspapers, etc. to gauge public opinion at a specific time. It could help contextualize an event, person or single publication within the environment of greater cultural trends. The graphs and accessible interface facilitates comparison of frequently used words and would allow me to determine whether the relationship between two ideas is worth further examination. In short, I see Voyeur as a useful tool in both the initial stages of research and in reconstructing cultural framework by isolating most frequently discussed topics.

    From what I gather, Voyeur does not operate as a plug-in, which may be a drawback. While Voyeur allows you to save the results, can you still manipulate the data as though you were on the Voyeur site? That is, after initially running analysis, I would like to be able to return to my results and add to/manipulate them easily. I was unclear if Voyeur allows you to do this.

  2. Eron Says:

    Voyeur offers some interesting possibilities for discerning rhetorical patterns in texts, as the creators demonstrate in their analysis of the speeches on race by Obama and Wright. It may be especially useful for analyzing a large corpus that would otherwise demand a lot of time from the reader; however, as a first-time user, I had some difficulties and found some limitations with the program.

    The creators list some of the limitations of the beta version: lack of advanced linguistic processing, proximity of searching terms, etc. It would be nice if the program could pull up full sentences (even full paragraphs) in which specific words are located to get a better sense of the linguistic context. Otherwise, how are we to know whether two collocates (adjacent words) are being compared or contrasted, equated or negated by the author?

    An even more basic problem I had–unless I’m missing something–was identifying which document or web page contained which words. This may not be a problem if the user is importing a bunch of essays with distinctive titles, but what if she is importing a bunch of URLs from the same website and they all have the same name? Perhaps some sort of numbering system could remedy this problem.

    I would also like to see an option for completely removing common words like “a”, “the”, and “and” from consideration. Such words can be removed in the main words section, but they still appear in the collocates list. Finally, the program crashed on me a couple of times when I was using it, and this should be remedied before the 1.0 version is released. This tool has a lot of potential, and I look forward to seeing some improvements in the near future.

  3. Carolyn M. Says:

    I’m not a fan of this tool at all for textual analysis. Maybe if a historian is looking to give bare statistical data, like how often a name or event is mentioned, it would be helpful- but how often do historians bother (or need to bother) giving such exact data? In the few examples I played around with, nothing was revealed that surprised or even interested me. I uploaded my senior thesis, which was on WWII propaganda, as an example. Besides the, a, and, etc., the most common words were, shockingly: world, war, II, propaganda, American. Of course I know that the results are limited the shorter the document you seek to analyze, but I just don’t see how the information you can get from a tool like this would be useful at all.

Enhancing Historical Research With Text-Mining and Analysis Tools