#exercise2

Exercise 2

Working with Voyant Tools to Create Data Visualizations

For this exercise, I experimented with Voyant Tools and used Dostoevsky's novella Notes from Underground, sourced from Project Gutenberg.

I've actually read this novella in my first year Philosophy class, I read it again from time to time.

I chose this text because Fyodor Dostoevsky's novella is unique case study. The protagonist, the infamous "Underground Man" is extremely self-aware, but trapped and constantly contradicts himself. So, if there is any literary work that could benefit from having the linguistic pattern quantified and visualized, I thought this would be a good one.

With a document that contains a total of 44,481 total words and 7,479 unique word forms, the corpus was substantial enough to see patterns, but still remain meaningful for my first distant reading method.

First Impressions: The Interface

A feature I found to be distinct in Voyant Tools, is its multi-panel dashboard. Each corner of the site displays a different tool I could reorganize and customize. I actually thought it was just a visual quirk. But now I find it as a philosophical approach to text analysis. The multiple perspectives need to be visible simultaneously, which can help with comparative thinking.

Each panel can be swapped and interactive, which allows me to customize my analytical workspace. Akin to a digital historian's workstation.

The design of the interface is a methodological lesson, and my analysis worked best when I could see the different types of evidence side by side.

Source Criticism

Before any analysis could start, I had to clean the data. The Summary Tool displayed an immediate issue with my corpus and that is the word "gutenburg", which appeared 75 times. This was not in Dostoevsky's lexicon, but a metadata embedded by Project Gutenberg.

My distant reading began with criticising my source by filtering out administrative words. The Stopwords Tool is in charge of this, which affects the word cloud and the frequency statistics in real time. This digital cleaning tool is equivalent to removing stamps or other residues from a physical source or document.

All sources, digital or not, often come with baggage.

Cirrus and Document Terms

After cleaning the data, I analyzed Cirrus (word cloud) and the Document Terms tool to identify which words appeared frequently.

"Man" used 107 times.

"know" used 88 times.

"time used 75 times.

"like" used 75 times.

"love" used 67 times.

I think this highlights the narrators preoccupations: his identity, epistemology, temporality and affect.

The world cloud allowed me to take this data and turn it into a visual map. With the narrator using "man" 107 times, it meant more than just a word. This repetition shows me that questions of identity, self, and what it means to be human are central in the novella. I could narrow my focus for close reading, and philosophical or historical questions.

The Context Tool was like a highlighter, finding every single time that the word "man" was used and displaying sentences it appeared in. For instance, "I am a sick man", found in Part I.

This data no longer showed me a number but a pattern where the narrator used it to describe himself negatively. My analysis concluded that the Underground Man's negative labels for himself might reveal notions of masculinity and identity in 19th-century Russia. The Context Tool didn't directly answer this for me but now I have textual evidence. I could focus on interpretation.

Fancy Visual Voyant Tools: Knots and Dreamscape

After trying the basic tools, I jumped into the fancier tools Voyant offered, such as Knots and DreamScape. These were the most appealing to me.

Knot is a tool that turns the words into spiral or twisty lines. The frequency of the word's usage, the more twists and turns the line will display. The word "know" used 88 times, had more tangles, compared to "man", used 107 times. It's a weird but also cool way of seeing the repetition without the use of numbers.

Discovering the DreamScape Tool was the most exciting for me. It's aim is to find a location's name within the corpus text and dot it in a map. I originally thought it would show me Russia, since the context of the novella takes place there. Instead, it showed multiple locations in North America, places that were not in the book at all.

Before I experimented with the tool it did give me a warning prompt about the tool undergoing experimentation and "do not trust this data" on the Voyant Tools Help page. This taught me to always double-check what the digital tools show me.

Just because a digital tool looks cool, it doesn't speak for its reliability. If Voyant did not show a warning, I would have plainly believed the tool's capabilities of factual data.

For a historian, one shouldn't plainly believe any information at face value. We are surrounded with forged or fabricated documents. It exists in the digital world too.

Final Thoughts

I think Voyant is a helpful pair of glasses for near-sightedness, but not a magnifying glass. If I take the glasses off, I can read the small prints myself, analyse the tone and feel what the underground man is trying to tell me.

I think Voyant is a tool that can be used at the beginning of any project that could help me decide where my focus should be. But the "thinking" job is for me to do on my own with the data it can provide for me.

I recorded the first audio in an extremely quiet environment, and the second audio played the first audio in different rooms and collected the noise from other rooms.

outa, pull down outa's pants

position your body over the wine barrel

outa, pull down outa’s pants

position your body over the wine barrel

position your body over the wine barrel

the Boer wants to discipline you

outa, please we want to go home

During the class we worked on three exercises/inputs I proposed to the students. Several remarkable results of these exercises are published here. Here are the three exercises as they have been proposed at the beginning of the class:

Exercise 1

“Vito Acconci’s exercise: write in exploded axonometry”.

When I invited the American artist Vito Acconci to my poetry class at the Ecole Normale Supérieure in Paris in 2009, he came up with several fantastic exercises, one of which was to try to write a text having in mind the model of the exploded axonometry, often used in design and architecture. The response to this input didn’t necessarily have to be visual, but also conceptual, narrative, towards a multidimensional perception of writing. Given its implications, I decided to re-propose this exercise to the Multilingualmediadimensional class. 

Exercise 2

“Create a score/map as a piece and as a notational system for a piece to come, such as a performance, a sound work, an installation. How do the score or the map differ from the piece?”.

We discussed widely the poetic possibilities of nonconventional maps and notational systems, in terms of multilingualism (maps and scores as languages, as ways of translating actions and pieces), multimedia (maps and scores for multimedia projects to come), and multidimensionality (maps and scores as multiple narratives and ways of poetically and artistically reinventing space and performance).

Exercise 3

“Give words to the dancers in the following sequence from Sascha Waltz’s choreography NoBody: http://www.ubu.com/dance/waltz_nobody.html, minutes 29′ to 33′. Make a poem out of their movements. What would the dancers say? Would they all speak simultaneously, or one after the other? Would they join their voices in just one message, or would they say different things? How could this collective poem add something to the space of choreography and to the dancers’ movements, but also to the film of the performance rather than just the performance itself?”.

This exercise was also a way of exploring forms of translation among poetry, performance, space, multiplicity, collectivity, multidimensionality, voice, movement, body, perception, and so on and so forth, in the spirit of our class.

Soojung Lee, n9667598

This week I am started to modelling my second character, which is a nine-tailed fox. 

^ Figure 1: Import source images

The first thing I have to do was to import the source images to each perspective as I did in the previous modelling.

^ Figure 2: Create the body

After I import the source images, I apply the basic mesh. I decided to start with cube mesh as my character has angled body shape.

^ Figure 3: Extrude the faces

^ Figure 4: Cut the mesh into half

I extrude the cube as body length, but before I go any further process, as I have symmetrical character design, I have to cut it into half to make my work easier. Therefore, I added the edge loop to the centre of the cube mesh and delete the faces, as you can check in Figure 4.

^ Figure 5: Change the shape

I changed to the front view and started to change the shape of the mesh as source images. I moved the vertex of the mesh to fit as source images. The reason I choose to work with vertex because this is much easy to transform the mesh compare to edges or faces.

^ Figure 6: Extrude the faces to create the legs

After I transform the mesh in front view, I moved back to the side view of the mesh to model the legs. I have to extrude the faces to model the legs and to do this, I have to add edge loops to where the legs need to be placed. As I added the edge loops, I grab the face for each place and extrude the faces. However, as you can see in Figure 6, the legs are not square. It was much close to the triangle. Therefore, I select the vertex and scale down to narrow the area.

^ Figure 7: Extrude the faces to create the neck

^ Figure 8: Extrude the faces to create the head

The next step I moved on was to model the neck and head. This was done in similar steps with the legs. I added extra edge loops, grab faces and extrude the face.

^ Figure 9: Changing the shape of the mesh as source images

As I extrude the face to modelling the head, I have to check all the perspectives to check is it matching with the source images, and anything went wrong. During this step, I also change the place of the vertex to model as the source images.

^ Figure 10: Extrude the faces to create ears

^ Figure 11: Add extra edge loops to the side of the face

^ Figure 12: Change head to be more round

I finished with modelling the head shape and I moved to model the ears. First thing I have done was to extrude the face, however, the extruded faces were square rather than triangle as showing in source images. Therefore, I merge the vertex to triangle shapes. After merging, I realised that the shape of the face does not look right. It was too angled compare to ears, so I add extra edge loops to the mesh and pull out the vertex to make the head more round and natural.

^ Figure 13: Add extra edge loops

^ Figure 14: Model the ears

I moved to more detail for the ears. The initial ears were angled, but the actual ears on the source image are more round and soft. Therefore, I add edge loops the ear model, grab the vertex, and pull it out to match with the source images. This result more rounded ears.

^ Figure 15: Extrude the faces for tails

^ Figure 16: Tails

I was having trouble with modelling the tails because I have to make nine of them and they are rounded rather than angled like triangle or cube. I tried various options that I can think of, but this final one is the best idea I can think of. I extruded the faces for four tails, and when I extruded the face of the last tail, I deleted the half of the face after I extrude the face. The reason I delete the half of the face was I have to mirror the mesh at the end, which means I only need to model half of the tail for the ninth tail.

Figure 15 is showing after the extruded the faces, and Figure 16 is showing the steps to modelling the tail. I merge the vertex to create the triangle mesh as I did to ear mesh, and I pull it out to create the end of the tail.

^ Figure 17: Finished all tails

Figure 17 is showing the end of the tail modelling. I added edge loops to the tails to model the round shape with square mesh. I grab the edges and scale up the gave them a volume, which will make more tail looking mesh.

^ Figure 18: Mirror

^ Figure 19: Finish modelling

After I finished all the modelling that I need to be done for this character, I mirror the mesh to create a complete model.

This was complicated modelling, but this was also interesting and challenging. This was a great experience to research the story, set the character profile, design the character and modelling with my own designed character. However, through this process, I realised that I only like until drawing turnaround pose of the character. I also like 3D modelling, but personally, I am more comfortable and suitable for 2D drawing.

Sketchfab

https://skfb.ly/6IPMY

I sent the model to the tutor and received the feedback. He told me that the character is looking angle and broadening out the tail as they get close to the tips, like the mood board in the previous post. I will go to fix the tail as the feedback, but I might keep the boxy shape of the character because as I designed rounding and soft character before, I want to design the opposite character to the blue tongue lizard I have done. Therefore, I designed the fox with angles.

^ Figure 20: Fixed the tail

As the feedback I received, I started to change the shape of the tail. First thing I have done was to spread them out more for space. After I thought it now has enough space between each other, I started to angle and scale up the size of the tails by grabbing the faces and edge loops on the tails.

^ Figure 21: Angled tail and have more volume

This is the front view of the mesh after angling the tail and scale up the size of the tails’ edge loops. It now looks more natural compared to Figure 19. According to the feedback, the tails are now broadening out the tail as they get close to the tips.

^ Figure 22: Create the ear hole

After I fixed the tails, I thought the ear need more detail, such as ear hole so it will look more like the ear. Therefore, I add extra edge loops and grab the edges and pull it in. Now, I can add extra detail to the ear.

^ Figure 23: Mirror the character

As I think I finished with the modelling, I mirror the mesh to create the one mesh.

Sketchfab

https://skfb.ly/6ITG9

It was also difficult as the previous blue tongue lizard modelling even I decided to work as more boxy form. This is because of the tails. I have a lot of problem with modelling the tails with balance, the shape of the tails and place the tails. However, I can end up with final modelling with the tutor’s feedback.