The use of Voyant Tools to analyze bodies of text is an amazing development in digital history, but it may actually ask more questions than it answers. These tools quantify and analyze the incidence of words, relationships of phrases and give some idea of the construction of the documents being examined. They can produce some interesting visual representation of the use of words, and allow comparison of different documents, such as the Cirrus pattern shown above. This pattern is derived from this article.
Such tools do augment the traditional reading skills that historians bring to bear, in that they point out the use of words that wouldn’t be easily apparent otherwise. To the present time, however, understanding of the meaning, the underlying sub-text, or the psychological effect of a document is not possible in a software tool. For instance, the tool can identify that a word has occurred more than a dozen times in a document, but it cannot explain why, or what that means.
Publicly available, free Voyant Tools are somewhat frustrating to use, because although one can load multiple documents into it, there doesn’t seem to be a way to save them. This means one must continually re-load the corpus if one leaves the tool for any reason. There also does not seem to be a way of automatically comparing the two documents, although you can manually compare the ratio of words to types, average number of words in sentences, and the recurring rate of use of various words.
But what does this mean? The tools will be much more useful to those who have training in the psychology of word use or historical information on how certain words resonate with certain nations, and how juxtaposition of phrases indicates a hidden agenda or sub-text. This kind of training will have to include how words might have been used in the past, and of course documents that have been translated from other languages than English will have to be analyzed differently too.
These tools also can’t analyze the intent of the writer. If a high emphasis is placed on one word or phrase, we cannot know why the writer chose to use that word or phrase so often. Of course, a historian reading the same word or phrase often in a document can readily understand that it is important to the subject. Was this the writer’s choice, or was it by accident? It would be helpful to be able to analyze words that are used less often, too, and compare usage in many related documents. Will we gain information from the use of Voyant? Absolutely we will. Will the information be relevant? That is uncertain.