A question that seemed to develop from the first sampling of text analysis for this project brought out the need to delve into comparing more published letters to collections of unpublished letters. Although I was able to support my theory that the word 'war' would certainly show up more often in the published letters than the unpublished letters, I could not wholeheartedly stop there. The reason being, was that just one sample from each is not enough to clearly say this was a overarching theme from all letters unpublished or published.
The next published volume I looked at was "Letters of A Soldier, 1914-1915." This too was taken from the Gutenberg.org site (www.gutenberg.org/ebooks/17316). As I commented on an earlier post, just taking the text of the letters in a grouping did not consistently give me the same graph that I received when I uploaded the individual transcribed letters from one of my collections. The challenge here was to take and make an individual file for each of the letters in this volume to make a consistent graphing analysis of words.
There were approximately 130 letters from a French painter and I had to make 130 files. This volume had been published in 1917 with an introduction written by A. Clutton-Brock (British Officer I believe). Doing the individual copying of the letters forced me to take a more closer look at the text. This was something I did not do with the Paul Jones letters. I just dumped into the Voyant-tools.org program to see what type of word cloud would be produced.
Being forced to review the letters gave me two things - 1) a basic understanding of what was included in the printed letters and 2) a closer scrutiny of the introduction and its publication date, along with the dates of the letters. Going back to the basics of research helped me to understand how this program could work in giving hints in the direction of what to study in order to understand how the volume may have been a piece of propaganda.
For example this is the word cloud that was produced from the anonymous painter's published letters:
The next published volume I looked at was "Letters of A Soldier, 1914-1915." This too was taken from the Gutenberg.org site (www.gutenberg.org/ebooks/17316). As I commented on an earlier post, just taking the text of the letters in a grouping did not consistently give me the same graph that I received when I uploaded the individual transcribed letters from one of my collections. The challenge here was to take and make an individual file for each of the letters in this volume to make a consistent graphing analysis of words.
There were approximately 130 letters from a French painter and I had to make 130 files. This volume had been published in 1917 with an introduction written by A. Clutton-Brock (British Officer I believe). Doing the individual copying of the letters forced me to take a more closer look at the text. This was something I did not do with the Paul Jones letters. I just dumped into the Voyant-tools.org program to see what type of word cloud would be produced.
Being forced to review the letters gave me two things - 1) a basic understanding of what was included in the printed letters and 2) a closer scrutiny of the introduction and its publication date, along with the dates of the letters. Going back to the basics of research helped me to understand how this program could work in giving hints in the direction of what to study in order to understand how the volume may have been a piece of propaganda.
For example this is the word cloud that was produced from the anonymous painter's published letters:
My first thought was this doesn't fit with my theory. The word 'war' shows up as a frequently used word, but 'dear' and 'mother' seem to stand out more often. Had I not been forced to review these letters as I created the 130 individuals files, I would not have know that this is how he addressed many of the letters and was also an endearment he used within the text of the letters for his mother.
In the Voyant-tools program, certain words can be added to the Stop List of common words. I just added 'dear' and 'mother.' This is the second word cloud created.
In the Voyant-tools program, certain words can be added to the Stop List of common words. I just added 'dear' and 'mother.' This is the second word cloud created.
'War' is still a frequent word but not the most frequent. Looking at the individual letters gave me insight into how they were presented in the published volume. What I mean here, is that reading them, it is clear to me that portions of the letters have been omitted. References to things like God and LIfe appear to be the most important part of the letters. I say this because when data mining, you need to look at all aspects surrounding the text being imported into the word cloud program. For instance, the letters were written at the beginning of the war, (1914-1915). This volume was not published until 1917. This was a time when moral was very low in the French regiments due to the massive amounts of casualties. A clue about the propose of this volume could be determined by who wrote the introduction and what he said:
"These letters reveal to us a new type of soldier, a new type of hero, almost a new type of man...who can endure without any romantic illusions..."
With that in mind, it can be determined that this volume was published as a tool of propaganda. Doing a basic graph analysis of the two words I previously honed upon in the past blog posts will support that, even though the word 'war' was not the most frequently used word as in the letters of this Frenchman, it was still used more than the word 'home.'
"These letters reveal to us a new type of soldier, a new type of hero, almost a new type of man...who can endure without any romantic illusions..."
With that in mind, it can be determined that this volume was published as a tool of propaganda. Doing a basic graph analysis of the two words I previously honed upon in the past blog posts will support that, even though the word 'war' was not the most frequently used word as in the letters of this Frenchman, it was still used more than the word 'home.'
An interesting side note: the same misspelled word in Chester Cole's letters of the word 'today,' appeared across the ocean in these Frenchman's letters. The same spelling 'to-day' show up in the graph analysis. While not really having anything to do with the main research, it is an perplexing item that could lead to other analysis regarding the evolution of spellings over time.