Category Archives: Text Analysis

The Beginnings of Remix Data: Poemita, an Experimental Online Writing Project

Figure 1: The five most repeated words from 2010-2013.

Poemita began in 2010. It means little poem in Spanish. The basic premise was to experiment with tweets as new forms of writing. I eventually decided to use it as a resource (think of it as data mulch) for various projects. Some of the tweets  are being repurposed as short narratives, which I have not released. Poemita was actually preceded by writing I developed for my video [Re]Cuts, a project influenced by William Burroughs’s cut-up method. I am in the process of producing a second video that uses actual tweets from Poemita.

I worked on Poemita on and off, sometimes not posting for months at a time. In fact, I don’t have a single post for the year 2011.  But during the month of August 2014, I realized that Poemita has been a project that is closely related to my ongoing remix of Theodor Adorno’s work in Minima Moralia Redux. It could be thought of as a negative version of that project (I am using the term “negative” here in dialectical terms). To allude to this relation, I inverted the color scheme for the word cloud visualizations of Poemita to be the opposite of Minima Moralia Redux’s. Poemita takes the concept of the aphorism as Adorno practiced it and tries to make the most of each tweet. Most of the postings are well under 140 characters, and they all try to reflect critically on different aspects of life and culture.  I try to do this creatively, and write content that may appear difficult to understand, but ultimately may not even make sense; the aim is to create the possibility for the reader  to see things that would not be possible otherwise. In short it is an experiment in creative writing, and this is why the project was titled Poemita.

I may not be able to post consistently, but I will certainly be posting tweets more regularly then before.  And I will eventually be repurposing the tweets in different ways to explore how context and presentation along with selectivity are ultimately  major elements  in the creative act. This will become clear as I release the tweets in different formats in the future. This, in essence, is a way of remixing data.

To reflect on where this project is going, I decided to analyze it as I would other texts to understand how it is constructed, and to evaluate the type of patterns that may be at play in my online writing. What follows, then, is a set of studies of  the tweets for the years 2010, 2012 and 2013. I will be releasing analysis of 2014 later, after the year is over.

First, it is worth looking at word clouds for the three years:

Poemita_2010Figure 2: Word cloud of tweets for 2010

 

Poemita_2012

Figure 3: Word cloud for tweets of 2012

 

Poemita_2013

Figure 4: Word cloud for tweets of 2013

Poemita_2010-13

Figure 5: Word cloud of tweets from 2010-2013.

We can note the top four or five words for each cloud for the respective years of 2010, 2012, and 2013 and consider how they eventually become part of the larger cloud for all of the years of 2010-2013. The number of occurrences could be accounted for yearly, but for the current purpose of this analysis, it should be sufficient to evaluate the number of words in the largest cloud for all three years (figure 5).

In the cloud above (figure 5), then,  there are a total of a 1,712 words and 863 unique words. The most used words besides articles and prepositions appear much larger. These words appear the following number of times in the actual body of the text:

Time: 12
Thought: 11
Sound: 7
Space: 5
Thoughts: 5

The word trend chart at the top of this page (figure 1) shows how these words relate to each other in terms of writing sequence. If you were to choose a particular node, you would be taken to the actual text and shown how the word appears in its context. The tool I used to this word analysis is Voyant. Seeing the words in a diagram provides a visual idea of how they relate to each other within the actual writing.

This gives a sense of repetition, and may even allude to certain interests in terms of content and ideas within the corpus of the text, but it does not provide a clear sense of how the words actually function, or under what context they recur. For this, the way the words are used in actual sentences can be mapped. In the following word trees, the top five words (in order of times repeated), Time, Thought, Sound, Space, and Thoughts are linked to all the phrases that follow them:

PoemitaTime10_13

Figure 6: The word “time” linked to the phrases that come after it. Click on the image to view a larger file.

PoemitaThought_10_13

Figure 7: The word “thought” linked to the phrases that come after it. Click on the image to view a larger file.

PoemitaSound10_13

Figure 8: The word “sound” linked to the phrases that come after it. Click on the image to view a larger file.

PoemitaSpace_10_13

Figure 8: The word “space” linked to the phrases that come after it. Click on the image to view a larger file.

 

PoemitaThoughts10_13

Figure 9: The word “thoughts” linked to the phrases that come after it. Click on the image to view a larger file.

The word trees above show how each of the words are implemented to create particular statements. At this point, it is possible to make certain assessments.  Let’s take the word “thoughts” (figure 9).  We can see that three out of five times it comes at the end of the sentences. We can also note that the exception to this is a reflective statement: “thoughts of grandeur.” Let’s take a look at the word “thought” (figure 7) and we can notice that it is part of a much more complex set of phrases. Two times, the word is part of the branching recurrences “Improvisation fills one with…” and “the very thought of…” But notice that in the last one thought is also followed by a period.

Finally, we can consider the words that come before these words. Let’s take the word “thought” for a brief example. For this we can use voyant:


At this point we can get a full sense of how the word recurs and how it functions each time it appears. This approach puts me in the position to evaluate what similarities and differences their implementation may share in order to evaluate particular tendencies I may have in my writing.

We could go on and examine the other top words in the same way, but this is enough to make my point.  It becomes evident that how the word “thought” and its plural “thoughts” are used has much variation in the creative approach in terms of twitting. At least, I, as the actual writer, become aware of the way that I tend to relate to the singular and plural instantiation. This in the end is a reflective exercise that enables me to be critically engaged in understanding my own tendencies as a writer. I plan to use this analytic approach to further the possibilities of writing tweets that can offer a lot more content just under 140 characters.

One of the issues that I assess in all this is the role of repetition.  One may think that repetitive occurrences are bad for creativity, but in practice, it is through repetition that we come to improve our craft and technique in any medium. In terms of how words are used or repeated, with analytical exercises like this one, a writer can come to understand how certain words recur and under what context, to then decide if to implement them differently or omit them altogether in future writing.

I certainly was not thinking that I would use these words the most when I began writing in 2010. They appear to recur and I’m not sure why, but the point is that now I can use this awareness to improve my own creative process.

This analysis can get very detailed, obviously, but this should be enough at this point. This is just a brief sample of how I am data-mining my own writing to also develop other projects  by remixing the content. I will also be mining twitter postings to evaluate how what I learn in this focused project may or may not appear to be at play in the way online communities communicate.

The Beginnings of Remix Data as Research: Preliminary Notes on Analysis of Theodor Adorno’s Minima Moralia

The following post was originally released on Remix Theory on August 14, 2013. It is reposted to give an idea of the research process that led to Remix Data as an online resource.

Detail of Minima Moralia 21 and 22 and their respective remixes

Image 1: Word cloud visualizations of Theodor Adorno’s Minima Morlia, aphorisms 21 and 22 on the left and their corresponding remixes on the right. (Click image for detail)

My first post for Minima Moralia Redux is dated October 16 2011, but I had done much research prior to this date. I had been reading extensively on Theodor Adorno and his work, while also creating visualizations of YouTube viral memes for my post-doc at The Department of Information Science and Media Studies at the University of Bergen in affiliation with The Software Studies Lab in San Diego, now also based in NYC.  As I analyzed meme patterns, it became evident that much of the material that is discussed in terms of remixing in music and video, which is also quite popular across media culture, usually relies on acts of selectivity–meaning that with the ubiquity of cut/copy & paste, people tend to re-contextualize pre-existing material, much how DJs and producers used sampling to remix in dance music culture during the eighties. [1]

Image 2: Word cloud visualization of the first thirty aphorisms in Theodor Adorno’s Minima Moralia. (Click image to view large file)

Minima Moralia Redux is a type of mashup, itself, of art, writing as a literary act, and media research that explores how data visualization is providing new possibilities for understanding creative processes. The project explores the selective remix, which arguably is quite popular across culture since cut/copy and paste became a common act due to daily use of computers. Certainly this is the type of remixing that most people debate over in remix culture. The selective remix consists of evaluating the source material and deciding what to leave and what to omit, as well as what to add, all while making sure that the source material remains recognizable.[2]  This means that large parts are kept as originally produced while others may be radically different. A tension in authorship develops, as the remixer clearly shows creativity quite similar to an “author’s.” At the same time, the remixed work relies heavily on the cultural recognition of the author and his/her work.  Much has been written about such tensions, but it is my hope that the research I am now introducing here in preliminary fashion will be a contribution to understanding how we come to create works that appear to be autonomous and credited to a single person, and how we can move past such conventions to more productive approaches that do justice to the way culture is communicating at an ever increasing pace.

 

Image 3: Word cloud visualization of the remixes of the first thirty aphorisms in Theodor Adorno’s Minima Moralia (Click image for large file)

Minima Moralia Redux has various layers of significance. First, I wanted to explore, as I already explained, how the selective remix functions. I decided to do this by embedding myself in the process, as opposed to studying another person’s remix. In this project, I examine each entry carefully, do research on it, and eventually re-write it to make it relevant to issues that are taking place in contemporary times. While doing this, I keep in mind that it is the voice of Adorno that is at play here. This means that I need to make sure that Adorno’s theories remain his.  In other words, it is not necessarily my opinion that is expressed in the remixes, although I do take creative license and adjust– even critique Adorno’s views within his own writing. This is no different than a music remixer who often times will create a different piece of music, one which nevertheless, is not credited to him/her as author/artist, but only as a person who remixed the author’s work. In the case of music this is done in the commercial sector for increasing sales, but in remix culture, it is done because people may simply love doing it, and/or are fans of the artist/author.  Taking this approach with Adorno’s work, I argue, is only fair given that Adorno himself believed in revising one’s view on life and the world. In the 1960s, he admitted that some of his critical analysis in Dialectic of Enlightenment, which he co-wrote with Horkheimer, no longer stood their ground in 1969. He considers the book “a piece of documentation.” By this Adorno and Horkheimer let the book be part of history. [3]  Based on this critical position on his part, it is very unlikely, for instance, that in 2013, he would use the word “savage” as he did when he wrote aphorism 32.[4]  The result of this approach in Minima Moralia Redux is a new text that is clearly still in large part Adorno’s, but which I hope resonates with the language and issues of the twenty first century.

I rewrite each aphorism  one sentence at a time, evaluating it word for word. I study the history of particular words, and evaluate the sentence’s relevance during the times when the book was written. I then consider how it may be understood and at play in contemporary times. When I rewrite the aphorisms I am conscious of the way remixing functions in music and video, and apply it to writing to see what the results may be. At the same time, I become immersed in the creative process based on intuition as I am also interested in exploring aesthetics.  I use two translations for the rewriting of each entry. The first is by Dennis Redmond, available on Marxists.org, and the other is the official English publication of Minima Moralia translated by E. F. N. Jephcott for Verso Press. I combine parts from both sources, while adjusting sentence structure, and I add and delete material to come up with a statement that is relevant to contemporary times.

For the word cloud visualizations I use Many Eyes, an online resource developed by Martin Wattenberg for IBM. The clouds are useful to evaluate how often words are repeated in the original entries. The visualization of the original text appears at the top of each blog entry. The main section of each post consists of the remixed text with a link to the original source available on Marxists.org. At the bottom is a thumb image of the same visualization along with a second visualization of the actual remix. These thumb images are presented with each post to provide a quick understanding of how key terms are reused and others omitted, while others are added in accordance to the principles of selective remixing. The reader can click on each thumb image to view a detailed version and compare them. I provide two visualizations of aphorisms  at the top of this entry (image 1).

The visualizations expose the constant usage of particular words, and when comparing the original entries to the remixed versions, it becomes evident how selectivity is at play. For instance, one can notice in aphorisms 21 and 22 that some of the words that are more pronounced in the original entry are still repeated often in the remixed versions, while others disappear and others are added (larger words means more repetition, smaller, less frequent). This is similar to how remixing functions in music as well.  I am also evaluating sentence structure and actual number of word repetition for each visualization. I will be releasing a concrete analysis of all this in the future in connection to viral memes, as well as a set of YouTube video mashups. The latter research I have not made available online at all, but two of the videos part of this research can be found on page 106 in my book Remix Theory.  My research of the selective remix as found in the thirty entries that I share on this post is part of my examination of selectivity in other forms of online media production. The idea to look at how remixing functions in text developed out my research in analyzing video. My findings so far have been that there are patterns that certainly crossover among image, music and text, which enables the viewer or reader to sense how remixing is at play in particular pieces.

So far I have remixed thirty-five aphorisms, and provide visualizations of thirty of them as part of this post. Image 2 offers an overall sense of the originals, and image 3 a comparative sensibility of how they were changed after they were remixed.  The process behind the production of each remixed entry takes quite some time to perform, so it will be a while before I can release my final version of this project. This brief entry should at least provide some details on the process that makes Minima Moralia Redux possible.

Below I provide a two column comparative visualization of the first thirty aphorisms(image 4). On the left are the original entries, and on the right appear the remixes. Examining one next to the other provides an idea of how different patterns are at play within and across the originals and the remixes, while looking at them as a large group gives a sense of the aesthetics of writing as a creative act–something that certainly cannot be fully measured, but one could hope can be appreciated.

Image 4: A two column comparison of the first thirty aphorisms of Theodor Adorno’s Minima Moralia and their remixed versions. Comparing each aphorism with its corresponding remix shows the process of selectivity that takes place in remixing text, which is deliberately performed, in this case, along the line of music remixing.

 

[1] I go over much of this in my book: Remix Theory: The Aesthetics of Sampling.

[2] If  too much material is omitted, then the remix may start to lean towards other types of remixes which will not be discussed in this instance. See chapter three inRemix Theory.

[3]Mark Horkheimer & Theodor Adorno, Dialectic of Enlightenment, trans. Edmund Jephcott (Stanford: Stanford University Press, 1987), xi -xii.

[4] See my remix, which is an extensive critique of Adorno’s conflicted bourgeois position, by using his own words:http://minimamoraliaredux.blogspot.com/2013/06/minima-moralia-32.html