Voyant is a tool designed for textual analysis and was developed by Stéfan Sinclair and Geoffrey Rockwell. The statistical analysis of texts makes the analysis of large corpuses possible and allows the study of words that are too frequent to be studied one by one (Burrow 2004). Voyant supports its users with the analysis and offers different visualizations to display the results. It allows users to upload text files either individually or as a corpus and to display word frequencies and relations in word clouds, word lists, graphs, and other visualizations. Even though the tool offers many exciting options, the most interesting functions for me were the visualization of the results as a Cirrus, a word cloud visualizing the word frequencies in a document or corpus, and the word list 'Words in the Entire Corpus'. This allowed me to analyze word frequencies for the entire corpus, which consisted of the title cards of all six movies, and to determine what the most frequent words are in terms of talking about Indians. It is useful to keep in mind that since The Vanishing American and The Last of the Mohicans are the longest films and therefore have the most title cards, they are over-represented in the results. The textual analysis accomplished with Voyant is added to with the use of Google Ngram Viewer, a free online tool forming part of Google Books. It provided the data necessary for a comparative analysis that establishes differences between word choice in print media and films betwen 1910 and 1925. Google Ngram Viewer derives its corpus from Google Books and displays the percentage of the entered search terms in comparison to all words in English-language publications currently available in the Google Books database over a period of time that can be specified by the user.

The Categories 

I chose the categories based on both the secondary literature on the topic of Native American portrayals in the silent era and the necessity of limiting the number to six, considering that two of the eight categories needed to be 'title cards' and 'other' respectively, 'other' denoting everything that is not Indian in the films. Even though the strict categorization of visual content in only six categories is only approximate and cannot be taken as definite, it allows for a rough quantification of the time on screen for different Indian types and stereotypes. The category 'chief' denotes both a visual category that usually entails the wearing of feather-bonnets as well as a character that functions as a leader. He is separate from the 'warrior' in the sense that the warrior might have a following but is not a leader with authority. Like the chief, the warrior is usually a positive character and often appears as a suitor to the Indian maid or a white woman. The Indian maid appears as either the love interest of an Indian warrior or a white hero, the latter being the more frequent case. In this function, she chooses a white man over an Indian suitor and often helps the whites against her own or another tribe. Very rarely does she appear as a mother and wife. Because miscegenation was frowned upon, oftentimes she dies at the end of the movie (Marubbio 2006, 26-9). I used category 'group' to characterize Indian groups in which none of the characters stand out. Whenever Indian groups consisted only of warriors, I used that category instead. The villain is characterized by his hostility towards the whites and his degraded, bloodthirsty and savage character. I used this category both for individuals and hostile groups. The category 'family' was born of the necessity to denote scenes that showed Indian life outside the most frequent categories of chief and warrior. I used it to measure the time on screen of kids, entire families, and situations in which the depicted Indians were engaged in activities other than warfare, councils, and flirting. This category is especially important because the depiction of kids, joking individuals, laughter, and families humanizes Indians on screen. It disappeared almost completely in the Western of the following decades and, thus, establishes the difference of the silent era the best. Although by no means absolute or all-encompassing, together these categories allowed me to evaluate what type of images was most prevalent on the silent screen. Even though it is impossible to guess what contemporary audiences might have found most memorable or impressive about the Indians, the screen time allows for an educated guess about which image might have been dominant in their mind. The access to my results is organized by films and by themes. The themes organize my findings for the single categories in a topical approach.