In our time spent with this material, we have become aware of a few factors that may have negatively impacted the nature of our data. Firstly is an obvious point we came to early on in our research, which is the simple fact that these letters were being sent between soldiers and thier families back home. Naturally, as to keep morale on the homefront high, it stands to reason that the content in these letters will not fully reflect the headspace a soldier finds himself in while in the trenches. Second, we became aware of rampant censorship that most likely affected what could be included in these letters. It stands to reason that if a letter detailing the exact plans of a particular regiment fell into the wrong hands it would lead to a massacre. Additionally, if word of how horrible the conditions were in the trenches made it back to the homefront, it would destroy public opinion of an already unpopular war. Lastly was a self imposed problem in our research, that being the language barrier we faced while assembling our corpus. Since we did not have any native French or German speakers on our team, we were naturally limited in the number of letters we could draw on for our research. Now, we could have used translations of these works or our limited understanding of the afformentioned languages to work through a few letters from France or Germany, but we felt that would leave out the neuances of the letters that would be lost in translation. With this in mind, we were able to observe a notable change in both emotion and the wartime factors present within the letters over time. This change is not only illustrated in the inverted bar charts on our Analysis page, but also in the table below.
Year | Number of Letters | Majority Positive | Majority Neutral | Majority Negative | Majority Mixed | Dominant Emotion |
---|---|---|---|---|---|---|
unknown | 1 | 0 | 0 | 1 | 0 | Negative |
1915 | 1 | 0 | 0 | 0 | 1 | Mixed (Positive/Neutral) |
1916 | 26 | 8 | 2 | 10 | 6 | Negative |
1917 | 13 | 6 | 1 | 3 | 3 | Positive |
1918 | 6 | 4 | 2 | 0 | 0 | Positive |
1919 | 1 | 1 | 0 | 0 | 0 | Positive |
1920 | 1 | 0 | 0 | 0 | 1 | Mixed (Negative/Positive) |
Across the three years that we have substantial data for, a positive correlation between the distribution of emotive statements and the dominant wartime factors per year can be observed. Although 1916 encompasses the largest chunk of our corpus and yielded the most letters that trended negative, it should be noted that this data only represents a small sample of what exists and is available to the public. Looking towards the above list of dominant factors per year, it can be gathered that from 1917 onwards, there was a larger distribution of factors that we correlated to positive emotional responses as compared to earlier in the war. Therefore, we can conclude that as the war marched on the factors that contributed to a negative emotional response became less prevalent. Nearly half of the letters from 1917 trend positive, and more than half from 1918 do the same. While we cannot make significant judgements about 1919 or 1920, it does stand to reason that they would also follow this trend. Therefore, we are confident in saying that across our set of data, there is an observable link between the wartime factors we have identified and the mental wellbeing of soldiers serving in the first world war, and that link leads to a positive trend in their mental wellbeing as the war came closer to its end.