Reflection blog

For my corpus I have chosen poems. I collected texts from four authors, I include Russian, English and Latin translation of the texts. In total I’m working with nine texts, it should give the representative sample. The range of my texts starts with Ancient Greece and Horace, then the Russian Empire in the XVIII-th century with Lomonosov, after the Russian Empire in the XIX-th century with Pushkin and it ends with the Soviet Union and Vladimir Vysotskiy.

They are very small 1-3 KB.

There a lot of poems with the idea when poet is building monument in his honor to make his work immortal. This tradition begun with Horace and I took Latin, Russian and English translation of his Monument.  Since I’m working with Russian literature I decided to start with Lomonosov who arise this topic in Russian poetry in the XVIII-th century. The next poem is by Pushkin and with Pushkin Russian language changes a lot, he is the father of the Russian poetry and he reformed the language, the style a lot. Even thought it’s one century difference the language and writing style is totally different and I’ wondering if the program can these differences and show it to us. The last but not the least is the song by Vysotskiy where he’s building a monument for his work. Vysotskiy represents another style, we call it bard song. Also he lived in the XX-th century during the Soviet Union time when the government changed the language to its modern state we speak now.

My research questions: to find out which words or phrases are repeated, are these repetitions synonyms or not and according to that how does the translation change the original text, its meaning, its atmosphere. Also it’s interesting to compare poetic metric systems because it’s different for every language and how the translators were solving these problems. The main result of this work should show how the idea of  building the monument has changed through the centuries.

Cleaning.

I deleted all unnecessary signs that were disturbing for the program. Working with Voyant I found out that there is no Russian stop-list words. I start creating my own stop-list, and then I realized that the program was reading the same word but with capitalized and small letter as a two different words. I went back to my texts, changed to capital letters to a small ones and only after that I created a stop-list.

Voyant

The best thing about Voyant is that it reads Russian!

v1

v2

 

 

 

In English translation the main words are: monument, alive, death, muse. In Russian – die, muse, Alexander, lire. Word collocation is very useful, we see how words are surrounded by each other and it shows us the context where we can read the words.  For this sort of work Voyant is amazing to use.

v3

v4

Here we can see that links in English give us more information than Russian one.

From Voyant we can make  a conclusion that translations a different and they show us different sides of the poems.

Antconc

Antconc has also a lot of advantages, for instance, search with the key words, word collocation, cluster analysis, n-grams and etc. That shows us the position of the word in the text, which very useful because looking up at the position of the word we can make already the analysis without wasting time searching for that word in the text, the platform does it for you. Concordance and file view where you can search the word in the text (almost the same as Voyant). Doesn’t read Russian.

a1

a2

a4

Collocate shows us  which words your term is collocated by and it helps to understand what atmosphere author is creating, how he places the words. Then you can click on it and it will show the position of the word in the text.

a3

Jigsaw

For my corpus more convenient was creating a word tree and circular graph to see the word dependence.

j1

As you can see here the word “monument” actually is not that much mentioned in the text, which is surprising.

j2

Poemage

The best tool for analyzing poems is  Poemage. However it doesn’t like Russian too.

I divided poems into three groups: Horace and Lomonosov, Pushkin, Vysotskiy.

p1

p2

Here we have high style text, author talks with muses, his lire, very poetic and abstract.

p3

Pushkin changes the style (4foot yambic) and makes it more about himself: his work raised higher than Alexander’s columned stone, he is more important than the czar,   he is writing about freedom, about peasants…his poems  are his monument.

p4

Vysotskiy changes the idea totally, he is still talking about life and death, but his poem is more sharp, rhythmic and politically oriented.

I enjoyed working with Poemage, for my future work I am planning to modify my corpus so I can more results with next tools we are going to work with.

Corpus Reflection


TextAnalysisMidPresentation

General Reflection of Distant Reading

Digital Humanities is one of the most influential and meaningful fields to emerge in the past few years. Distant Reading allows for an extraction of the text off the page and illustrated through digital tools. This provides a clear overall reading that is not always evident through close reading, the traditional process of reading materials.  Distant Reading is the ability to draw our attention from what traditional reading teaches and uncover the relation of patterns that emerge at a distance and close up.

At First Glance…

The process of Distant Reading begins by asking which general question you would like to research. Then, obtaining the documents needed for that research. Corpus creation could be a full time job. It is neither a project that has to be completed in any set interval of time, nor a project that must answer one distinct question. However, there must be an initial research question to begin an investigation. From that initial question and further analysis one can spawn new inquiries and explorations unbeknownst prior to the research. Indubitably, the best term to describe creating a corpus is the term “iterative”. It is important to realize the magnitude of one’s corpus, which might require narrowing the focus of the research.

My Corpus 

I decided to create a corpus of all the readings from my Comparative Humanities class, HUMN 150. Throughout the duration of our class, HUMN 100, I intend to compare the 2015 syllabus with the first syllabus from 2000.

My research questions are:

  1. How did the HUMN 150 syllabus from 2000 change in terms of genres and authors (gender differences) to the syllabus in 2015?
  2. Also, the course was originally titled “Art, Nature, and Knowledge” and is now “Enlightenments,” what is the most accurate title, or what could be another title for the course?

Slide4I was able to get the books from Project Gutenberg and one from Professor Faull; however, I acquired the supplementary readings in PDF form through my HUMN 150 instructor, Professor Shields.

Cleaning & Parsing 

I used OCR (Optical Character Recognition/Reader) on Adobe Acrobat Pro to convert the PDFs into text files (.txt), and I saved them in my google drive. My google drive is organized by “Renaissance&Enlightenment” texts, “Text Files from Gutenberg,” and “PDFs.” With these folders, I can keep track of which text files I am using. Also, I keep notes on my corpus construction that indicate what I keep and delete in each text file. Then, I cleaned each file by using Spellcheck.net, text fixer.com, and text cleansr.com. Using these websites, I removed line breaks, paragraph breaks, HTML script, and extra white spaces. Additionally, I manually cleaned each file correcting spelling and removing footnotes, some chapter titles, names of authors, and page numbers.

There were some odd mispellings through OCR, where a “.” would replace an “f”. Words would be randomly split apart, or there would be hyphens from a word split between lines. To manually fix this problem, I utilized the “Control Find” trick on my laptop to search for any letters that were directly after a period. (Crtl + F)

Modifying the CorpusSlide8

Previously above, I mentioned the “iterative” process of corpus creation. This becomes increasingly apparent when one realizes it is time to modify, update, and even begin an analysis of their corpora. Each platform helps to pick out mistakes as well; different views may show a KWIC (Keyword In Context) and the surrounding words might help find some misspellings.

For example:Screenshot (76)

Upon seeing this, I went into my clean text file and noticed that the text was perfectly parsed at this spot. I even ran it through an html script checker. From this, I am able to find faults in the programs as well. Although, it is important to remember that nothing is perfect, and that is why we associate the term “iterative” with this research.

Gender Bias: Voyant, Antconc, & Jigsaw

Voyant was the first platform I used to start my analytical searches. The word cloud automatically defaults the amount of words in the box to 25. I decided to set the amount of words to the maximum, 500. Screenshot (45)

This word cloud depicts the gender bias throughout the HUMN 150 texts. The prominent terms are both “he” and “him”. Also, the word cloud shows a myriad of other terms like “motion,” “people,” “science,” “man,”, and “mind” that demonstrate some of the key concepts of the European Renaissance to the beginnings of modernity in the 19th century.

Antconc showed the statistics that were not always present on Voyant and Jigsaw. For example, I did a collocation of both “he” and “him” compared to a collocation of “she” and “her”.Screenshot (53)Screenshot (52)

Overwhelmingly, it is mostly a negative lexicon surrounding the terms “she” and “her”. There are words like “mistress,” “bitter,” and “folly”. Taking this into consideration, I had to click on “she” or “her” depending on the sentence to see if the KWIC had a negative connotation or if it only seemed negative in that context through the collocation. Frequently, “she” and “her” were used to replace “nature” – purity, “law” and “deliberations” – mutable. The ‘sense’ of these words already is indicating weakness in women. In fact, this supports the claim that in most cases throughout my texts, women are viewed as an object rather than a human being. I question where the “humanist” thought is in this misogynistic view.

For “he” and “him”, the lexicon alluded to power. Some of the lexical items surrounding these terms were “God,” and “Lord.”

Antconc: Collocation of “she|her”

Screenshot (72)There were many other terms, some unique and  a few repetitive, that were used to describe “she” and “her”.

Some of these words include: “submission,” “scrutinizing,” “microscopic,” “prostituting,” and “virginity”.

This lexicon gives the impression that women are represented in my texts only as pure or “dirtied” objects. Throughout history, these words are commonly found when men are describing women. Consequently, all my texts for the Enlightenment and Renaissance era were written by males.

I have not used all of Jigsaw’s viewing platforms to their full potentials yet; however, I used the Jigsaw Tree view to allow me to illustrate a different collocation of the negative lexicon and the difference in the number of usage of “he” and “she”.

Screenshot (74)Screenshot (75)

This clearly indicates the numerous times “he” was used compared to “she”. However, this is only representative of five of my texts. Moreover, the Jigsaw Tree of “he” only shows 17.273% which is 38/220 hits.

True Humanists

Screenshot (60)

Screenshot (61)

These are bubblelines I made on Voyant illustrating the frequency of “people” throughout the Renaissance and Enlightenments texts. It is clear that more than half of the writers seem to focus on the people in general. On the other hand, the other writers are specifically writing about a topic that is important to the world and people as a whole.

¿ “Art, Nature, and Knowledge” ? 

The original title of the course “Art, Nature, and Knowledge” appears to have a consistent presence in all the texts. Using Voyant’s graphs and bubblelines, I was able to take a closer look.

Screenshot (51)Screenshot (66)Screenshot (47)Screenshot (48)

All of the terms are fairly consistent throughout the Renaissance and Enlightenment texts. Once I finish my corpus, I will be able to see if the title has validity or if I could come up with a better title for the course.

The FUTURE of My Corpus

Presenting my corpus to the class allowed me to focus on some of the key points I am trying to divulge in my analyses. Additionally, I was able to foster some new ideas with the help of my fellow classmates’ questions.

Now, I plan to search for both “women” and “men” with respect to the term “knowledge” in my corpus. I want to see if I can uncover any other biases. Also, I would like to add the translations of at least five major texts to see if the translations we are using in class are consistent with the original texts.

Finally, I really want to incorporate sentiment analysis into my research as well. I have theorized using Antconc to discover the negative lexicon in specific passages within my texts, thus being able to see the exact spots I could break the texts to fit Jigsaw to look for sentiment. Also, using Antconc to make an entity class is a brilliant idea.

I am eager to learn Mallet and further my knowledge in the Digital Humanities. Hopefully I will be able to add sentiment along the way and draw a plethora of intriguing conclusions towards the end of the semester.