Sentiment Analysis

• The NRC dictionary associates a word with the following sentiments: positive, negative, anger, anticipation, disgust, fear, joy, sadness, surprise, and trust.
• The sentiment of a body of text equals the number of words contributing to that sentiment.
• Words referring to race are biased within the sentiment libraries, so we removed them to mitigate racial bias in our analysis.

• The plot was created from the words used uniquely by each group, so the words used to evaluate sentiment for black soldiers were never used by white soldiers and vice versa. The unique words of black soldiers corresponded with more fear, disgust, anger, and sadness than those of white soldiers.
• The wordclouds below visualize the top 20 frequently used unique words by black and white soldiers, on the left and right respectively. The size of the word is proportional to the number of times it appears in the text.

• The unique words reveal what topics aren't addressed by the other group. Perhaps, the most important takeaway from these wordclouds is that the words slavery, lynch, or jail never appeared in the response of a white soldier.
• White soldiers unique words were significantly more lighthearted as they discussed things such as music and pets.

Co-Occurence Networks

• Used community detection algorithms on the co-occurence networks to identify groups of related words.
• The size of the nodes corresponds to the degree of the node ie. it is connected to more words.

Black Soldiers' Long Comment

White Soldiers' Long Comment

Topic Modeling

• A useful method to understand what the general themes are within a corpus by modeling the words which constitute a topic
• Used Biterm Topic Modeling to model the topics within each corpus of answers
• This works better on shorter texts
• It's a Gibbs Sampler that directly models the word-word co-occurrences patterns as opposed to word-document co-occurrences

Black and White Soldiers' Long Comment Gender Network

• In the above plot, we see that black soldiers often write that "woman holler rape".
• Black men often were often wrongly and publicly accused of sexual assault during this time period, and we can see that black soldiers are discussing that in their responses.
• Black soldiers are talking about interracial marriage.
• A common word sequence used by white soldiers was "white like negro".
• White soldiers are also discussing interracial marriage.

White Soldiers' Long Comment Spatial Network

• Spatial segregation of Black Americans was enforced in the American South with numerous local and state laws known as the Jim Crow laws from late 19th century to early 20th century.
• Are themes of spatial segregation prevalent in the soldiers' responses and experiences in the military during WWII?
• We created a spatial dictionary with words such as camp, messhall, etc. to see what the soldiers were saying when they were talking about these spaces.
• We see that there's a greater amount of commentary from black soldiers than white soldiers.
• They were very aware of these spaces and associated words like "mix", "together", "same", "separate", and "individual."
• Although some white soldiers were in favor of integrating outfits, there was still clear commentary from the soldiers should not be sharing the space in living arrangements or social space with black soldiers.