Voyant is an online textual analysis tool that allows users to discover patterns in texts or collections of texts. Some of the patterns Voyant can spotlight include: how frequently a word occurs in a single document or throughout a corpus, words that appear most frequently, differences in word frequency between documents, and a list of a document’s most distinctive words (compared to the rest of the corpus).
This tool is an excellent resource for researchers and users of a variety of backgrounds. Upon first use, individuals should begin by exploring the 5 different Voyant views that are displayed once a text is fed into the system (Cirrus, Reader, Document Terms, Summary, and Contexts). It is probably a good idea to become familiar with the system by first using a low-stakes corpus. Once the text is in the system, users should begin their exploration with Cirrus (word cloud). In order to represent more meaningful patterns, users can input a list of “stopwords” that occur frequently throughout the document but don’t provide significant insight (examples: the, a, it) so Voyant will filter these words out and exclude them from the results.
After this, users can begin to explore patterns on a deeper level. The Trends function can display patterns of a word or words over the entire corpus or a single document. The Contexts view is especially useful in providing more information about the patterns presented in Trends and Cirrus. For example, if the word “good” appears in the top 10 most frequent words, a researcher may initially hypothesize that the document overall contains positive connotations. To follow up on this theory, the researcher can use the Context function to view every incident of the word “good” in the corpus or document, along with the words on either side of it.
If most of the incidents of the word “good” appear in contexts that appear positive (“we had a…good…time at the party”), this lends support to the initial hypothesis. But if the contexts appear more negative (“he gave me a…good…whipping after that”), this allows the researcher to reform their thoughts on the significance of the word “good” in the document.
