Source: Johnston. 2013
Steven Universe is the story of a 12 year-old boy as he learns to control his mother's powers while trying to live up to her legacy. Steven lives in a beach house with his three alien caretakers, Garnet, Amethyst, and Pearl, struggling with ordinary slice-of-life problems, solving cosmic power struggles, and coping with his own existence.
During its run from 2013-2020, Cartoon Network aired five seasons, a movie, and a limited epilogue series.
Steven Universe is full of plot twists that cause major shifts in the status quo. Throughout the series, characters undergo massive struggles, triumphs, and regressions. Most characters that are introduced as villains are redeemed, and some that start as heroes slowly become villains. Rose Quartz is introduced as an infallible martyr who sacrificed herself for the joys of life on Earth. By the end of the series, however, she is revealed to have been the villainous Pink Diamond who had committed an innumerable amount of atrocities throughout her life. As the show released, fan-writers had interesting reactions to the twists along the way; many even predicted Rose's alternate identity years before it was revealed. Using tags for fluff and angst, diction, and fan interaction on the ways the fandom reacted to Pink Diamond's arc, one can learn when plot twists occur and how the fandom perceives them.
For this project, I scraped the entire Steven Universe fandom on Archive of Our Own on November 9, 2022. After excluding non-English fan-fiction and crossovers, 17,475 works remained. I converted this to individual text files so that it would be compatible with DocuScope, and I created individual spreadsheets of the metadata so that it would be compatible with Tableau. Further, I compiled a list of all major events that happened in Steven Universe and the release dates of major episodes and events throughout the series pertaining to Rose Quartz or Pink Diamond. The following are the methods I used to acquire data for each section:
Using Excel, I added two columns to the metadata file: angst and fluff. Next, I filtered all the additional tags by whether they contained the phrase "angst", and I replaced those rows with 1 using find and replace, and I replaced all the remaining rows with 0. I repeated this process for the phrase "fluff." By loading these changes into Tableau, I modeled both the average angst and fluff of stories containing Rose Quartz or Pink Diamond. I created two types of plots for four time periods. One displays the average fluff values in orange and the average angst values in blue over time; the other displays the amount of works released over time, and each segment is colored blue or orange depending whether there are more angst or fluff tagged stories during that time period.
I used the corpus analysis feature in DocuScope, to input all the text files generated from my initial scrape. This generated several cluster and dimension files containing the distributions of each DocuScope category. Next, I reformatted the filename column in the normalized cluster file to not have the ".txt" extension. I added the normalized cluster file and metadata file to Tableau and inner-joined with respect to filename and work ID. This allowed me to graph number of works over time and average positive and negative scores over time filtered by Rose Quartz and Pink Diamond.
Using Excel, I added one more column in the same way as in "Fluff and Angst Tags" above based on the tag "Rose Is Pink Diamond Theory" where rows containing this tag are 1 while all others are 0. After loading these changes into Tableau, I modeled average comments and average kudos on the same axis, and I modeled average hits on another plot. I also set the color to change from blue to yellow to red depending on how many works use this tag and limited the publish date to before Pink Diamond's reveal (May 8, 2018). Like before, I filtered the results to only feature works containing Rose Quartz and Pink Diamond.