Tagging My Tears and Fears: Text-Mining the Autoethnography

This article, which was presented for the first time at CSDH/SCHN 2018, outlines a mixed methodology that combines digital text-mining techniques, like XML tagging and topic modelling, with autoethnography. In this article, I describe my intimate and rocky relationship with my data, referred to as my “life-as-text.” As someone who has Complex Post Traumatic Stress Disorder, I have an extremely difficult time putting many of the events of my life into situational and linear context. The process of formally tagging my data requires me to overcome psychological blocks that have been created as defense mechanisms; thus, allowing me to honestly read and create associations in my dissociated “life-as-text.” Conducting a macroanalysis of data that was gathered through both journaling, and the recording and transcription of “let’s play” videos, has given me a new lens through which to view myself and my experiences. I see myself in my data. I see myself in ways that I cannot otherwise put into words. Text-mining my deepest memories and emotions provides me with an odd sense of stability in a seemingly chaotic assortment of words. In this article, I explain my autoethnographic journey and the powerful impact that digital text-mining has had on my relation to myself and my research. Abstrakt Cet article, qui a ete premierement presente a CSDH/SCHN 2018, decrit une methodologie mixte qui marie des techniques numeriques de la fouille de texte, tels que le marquage XML et la modelisation thematique, et l’auto-ethnographie. Dans cet article, je decris ma relation intime et difficile avec mes donnees, ce que je nomme ma « vie en tant que texte ». En tant qu’une personne ayant le Syndrome de stress post-traumatique complexe, j’ai d’enormes difficultes a contextualiser d’une facon situationnelle et lineaire beaucoup des evenements de ma vie. Le processus de marquage formel de mes donnees m’exige de surmonter des blocages psychologiques qui ont ete crees comme mecanismes de defense, me permettant de lire honnetement et de developper des associations dans ma « vie en tant que texte » dissociee. Effectuer une macroanalyse de donnees, qui ont ete recoltees a travers la tenue d’un journal, ainsi qu’a travers l’enregistrement et la transcription de videos « jouons » (“let’s play”), m’a offert une nouvelle perspective par laquelle je me considere et par laquelle je considere mes experiences. Je me vois dans mes donnees. Je me vois d’une facon dont je ne trouve pas les mots pour le decrire. La fouille de textes de mes souvenirs et de mes emotions les plus profondes me fournit une sensation etrange de stabilite au sein d’une gamme de mots qui est chaotique en apparence. Dans cet article, j’explique mon parcours auto-ethnographique et l’impact important qu’a eu la fouille de textes sur ma relation avec moi-meme et avec ma recherche. Mots-cles: Autoethnographie; Modelisation de sujets; XML; Methodologies mixtes; Exploration de texte; Alienation; SSPT-complexe


My tears and fears: Trauma, C-PTSD, and alienation
Before I begin tackling the digital methods of XML tagging, text-mining, and topic modelling, I need to start with an introduction to the life that will be examined in this paper, mine. I have experienced a great deal of trauma in my life! Immediately, this statement is wrought with difficulties. How does one describe the experience of trauma? How is trauma defined? How is it quantified? Trauma can be physical, as occurs through an injury. I have personally experienced seven concussions, two emergency surgeries, and countless cuts and bruises. Trauma can be social, as described by Jeffrey Alexander (2012) who developed a sociological model of trauma that emphasizes its collective nature and cultural transmission. Trauma can be a discourse that explores issues around collective memory, bearing witness, politics and power, and shattered narratives, as explored by Dori Laub and Andreas Hamburger (2017).
Having formally researched trauma since 2014, I continue to struggle to define trauma in a way that can be neatly summarized in an introductory paragraph, such as this. In reality, such a feat would be impossible. Trauma can be physical, emotional, sexual, social, political, cultural, shared or individual. It can appear in one or all of these forms depending on who is telling the story. Whomever I choose to cite, I acknowledge that I am doing an injustice to the field of trauma studies as a whole.

Sapach: Tagging My Tears and Fears
Art. 4, page 3 of 28 I have therefore chosen to conduct an autoethnographic project that focuses on my own subjective experiences of trauma. My research explores my "life-as-text", also referred to as "life narrative" by Smith and Watson (2010), in order to develop grounded theories connecting a relational theory of alienation, developed by Rahel Jaeggi (2014) and discussed below, to my personal struggles with complex posttraumatic stress disorder (C-PTSD). My research is primarily exploratory though it is guided by an informal hypothesis that techniques for the mediation and resolution of alienation may be found in the ways that I have mediated and partially resolved my own C-PTSD.
In order to ethically protect the people in my life, I am choosing to refrain from specific descriptions of traumatic incidents in this paper. Instead, I will draw your attention to the CDC -Kaiser Permanent Adverse Childhood Experiences (ACE) Study conducted between 1995 and 1997 by Felitti, Vincent, Robert Anda, Dale Nordenberg, David Williamson, Alison Spitz, Valerie Edwards, Mary Koss, and James Marks. The study involved conducting a survey regarding childhood abuse and neglect, the results of which were then connected to health, risk behaviour, and disease in adulthood (Felitti et al. 1998). Important for our purposes is the continued research that has been conducted by the Centers for Disease Control and Prevention, as well as academic researchers (Waite 2019), utilizing adapted versions of the original ACE questionnaire. These studies reinforce the validity of the ACE questions in measuring the impact of adverse childhood experiences.
In a 2010 report, for example, Bynum et al. (2010) administered an 11question version of the ACE questionnaire to 26, 299 adults across five states. The 11 questions covered incidences of verbal, physical, and sexual abuse as well as the relationship status of parents, substance abuse by parents, exposure to mentally ill household members, and witnessing domestic violence, all before the respondent turned 18 years old. The study found that 40.6% of respondents received an ACE score of 0, 22.4% a score of 1, 13.1% a score of 2, 8.8% a score of 3, 6.5% a score of 4, and 8.7% a score of 5 or more. The higher the ACE score, the more exposure the individual had to a diverse range of adverse childhood experiences, which correlates to higher incidences of health, risk behaviour, and disease in adulthood. After taking a selfreport, 10 question version of the ACE questionnaire which can be found on the news site ACESTooHigh (Stevens n.d.), my ACE score is 8. When considering the severity of my own score, please note that my primary reason for mentioning it here is to concretize the intensity and diversity of trauma that I faced growing up. It is the duration and intensity of this trauma that led to my diagnosis of C-PTSD.
As the name suggests, C-PTSD, has left me suffering with long-term symptoms of Post Traumatic Stress Disorder (PTSD), such as flashbacks, nightmares, a heightened startle reflex, and memory difficulties related to the traumatic events. The addition of the word "complex" describes the often difficult to articulate core psychological changes that occurred as a result of long-term exposure to trauma. The following description of people who suffer from C-PTSD is the one that hits closest to home for me: Severe and long-standing trauma introduces a profound destabilization in the day-to-day existence of many victims…They have symptoms that alter their perceptions of their environment, disrupt their cognitive functioning, and interfere with a sense of continuity in their lives [emphasis added]….They experience self-hate and self loathing and feel little kinship with other human beings. They long for a sense of human connection but are profoundly alone, regarding other people with great mistrust and suspicion. They want to feel understood but cannot even begin to find the words to communicate with others about their most formative experiences (Chu 2011, 18).
Destabilization is the most difficult thing to deal with in my life. I have spent my entire life attempting to find a stable footing, a foundation upon which I could build a "normal" life. I have lived, and continue to live, with constant feelings of unpredictability, chaos, and lack of control. Even now, as a PhD candidate at the age of 39, I continue to search for stability. Ultimately, I understand that I may never find the foundation that I am lacking, and I am learning to embrace my destabilization as a part of who I am. I have also learned to identify and accept my fairly consistent

Sapach: Tagging My Tears and Fears
Art. 4, page 5 of 28 sense of dissociation which can be best described as a feeling of being on the outside looking in.
While the destabilization leaves me feeling out of control, dissociation leaves me feeling distant, separated, and alienated. James A. Chu (2011) describes pathological dissociation as a discontinuous sense of identity resulting from disconnected state changes. Everyone experiences a wide range of state changes daily as they transition between various social roles (that of worker, parent, friend, etc.). While people usually experience these state changes as fluid, unconscious transitions between various parts of a stable identity, dissociation causes me to experience these state changes as a disruptive and uncomfortable process requiring a lot of conscious effort. I am left feeling alienated, somehow separated from myself as I constantly work to adapt to my social and physical environments.
Due to the fact that I often describe my own symptoms as alienating, I am exploring the connection between my own experiences with trauma and C-PTSD, and social experiences of trauma and alienation. While there are a multitude of definitions and theories about alienation throughout the literature, I am focused on the definition provided by Jaeggi (2014) who, after careful consideration of the history of alienation theory, concludes that alienation can ultimately be defined as a "relation of relationlessness" (893). Jaeggi clarifies this definition as follows: If, as I have claimed, alienation is a relation of relationlessness rather than a mere absence of a relation, then giving an account of this relation will be especially complicated: as clear as it is, on the one hand, that we are somehow antecedently connected to ourselves, it is, for precisely this reason, just as unclear how it is possible for this connection to break down….The thesis, then, is that a relation to oneself is a relation that can be disturbed in various ways. And in this relation, too, we find…we are alienated from something that is simultaneously our own and alien, we are involved in relations in which we alienate ourselves, we are in a certain sense at once perpetrator and victim (Jaeggi 2014, 1352).
Jaeggi's conception of alienation is that it is about deficient relations, as opposed to non-existent ones. Alienation from the self, from others, and from the environment, is not a lack of a relation to these things, but an explicit relation that is characterized by an inability to relate. The resolution of alienation, for Jaeggi, involves appropriation: regaining and reworking the relations between the self and the self, the self and the social world, and the self and the material world. To me, C-PTSD also represents a relation of relationlessness, to my inner child, to my memories, and to my daily experiences, through dissociation and destabilization.

Autoethnography
Before moving into the discussion of tagging and text-mining my above outlined tears and fears, as promised in the title of this paper, I want to briefly outline why I felt compelled to add digital text analysis methods to a methodology that is, at its core, subjective and personal. My own experiences with trauma, which have resulted in consistent feelings of destabilization and dissociation, have severely disturbed the relations that I have with myself in many ways. Conducting my autoethnographic research has, as a result, been simultaneously highly therapeutic, in that it has allowed me to take control over the fear of my memories, and extremely difficult, in that I often struggle to solidify specific memories into articulable narratives. As a research methodology, autoethnography, perhaps more than any other, requires an explicit recognition of the deep subjectivity involved (Ellis, Carolyn. 2004). Audrey Giles and DJ Williams (2007) note that the method is a blending of fact and fiction that draws from multiple layers of consciousness and is always partial and situated.
Norman Denzin (2014) discusses the multiple layers of narratives that must be acknowledged in autoethnography: Stories within stories told to groups remind us that every life story is a multiplicity of stories that could be told. There is no single life story or self-autobiography that grasps or covers all that a life is for a person. There are only multiple stories that can be told. Each storyteller can only tell the stories his or her biography allows to be told (Denzin 2014, 55). Fortunately, research has been done, and theories developed, that led me to understand that my trauma may not be "standing in my way" at all. Ron Eyerman (2013) argues that a close reading of the trauma behind social and political theory can reveal a hidden world of meaning through an acknowledgment of "sudden breaks, lapses and absences" (50). This has led me to realize that in order to honestly and accurately develop social theory through my own subjective lens, I need to acknowledge my own breaks, lapses, and absences which have become a significant part of my "life-as-text" due to dissociation and destabilization. Geoffrey H. Hartman (1995) also seeks unpredictability and ruptures in literature as noted in the following particularly poetic and relevant quote: But I share with trauma studies a concern for the absences of intermittences in speech (or of conscious knowledge in speech); for the obliquity or residual muteness of "flowers of speech" and other euphemic modes; for the uncanny role of accidents; for the "ghosting" of the subject; for the connection of voice and identity; for interpretation as a feast not a fast; and for literature as a testimonial act that transmits knowledge in a form that is not scientific and does not coincide with either a totally realistic (as if that was possible) or analytic form of representation (Hartman 1995, 552).
Trauma creates ruptures in lived experience, in meaning development, and in identity formation. My biography contains more intermittences, accidents, breaks, lapses, absences, and self-ghosting than it does coherent narrative. Acknowledging this led me to realize that I would be doing an injustice to my autoethnographic research if I did not somehow incorporate the parts of my biography that seemed inaccessible to me. This is where the XML tagging, text mining, and topic modelling came into play.
In order to find that which seemed inaccessible in my biography, I needed to come up with a way to both expose, and then analyze, that which remained lost in my dissociation and destabilization. I came up with a two step process.
Step two involved the macroanalysis techniques just mentioned. As Matthew Jockers (2013) states: "[t]he macroanalytical approach reveals details about texts that are, practically speaking, unavailable to close readers of the texts" (Jockers 2013, 26). I understood that due to my C-PTSD and traumatic relationship with my life narrative, I would likely dissociate from acknowledging certain themes and patterns during qualitative coding (microanalysis). I wanted to see if text mining and topic modelling would reveal themes and patterns, breaks, lapses, and absences, about my biography that I would consciously or unconsciously avoid when approaching my "life-as-text". Before I could conduct an analysis, I needed something to analyze.
Step one therefore involved expanding my data gathering beyond the limits of writing directly from memory. I will explain step one in more detail below before describing the results of step two.

Step one: Constructing my life narrative/life-as-text
This is a good place to note that I completed an application for ethics approval from the University of Alberta research ethics office and was granted approval on June 5 th , 2017. Although it may seem harmful to refer to myself as a participant or object of study, I felt compelled to write out an ethics application as a form of contract between myself as a "researcher" and myself as a "participant" in order to reinforce the fact that I took all of the necessary steps to ensure my psychological safety. These steps include(d): continuing to attend therapy, continuing to take my medication, and agreeing to remain in control of what I disclosed. The last step was the most important as I had/have to maintain a balance between what is useful to include/analyze, and what is doing more harm than good. Additionally, I am highly The second technique that I used was intended as more of a "traditional" ethnographic method. I carried a notebook with me, basically everywhere that I went, which allowed me to take field notes. I was able to record observations of events, thoughts, symbols, places, and social interactions as they occurred. I actively noted important details such as time, date, location, and my emotional state of mind in an attempt to discover my own breaks, lapses, and absences in daily life.
While the journal writing allowed me to interpret and elaborate on various thoughts and experiences, the field notes provided me with context during episodes of dissociation or ongoing traumatic triggers. One common symptom of C-PTSD is memory problems due to dissociation. The field notes reminded me of where I was, who I was around, and what I was doing in order to compliment the blogs, and to guide my autoethnographic writing.
Context is highly important for an autoethnographic project as noted by Giles and Williams (2007)  In total, I recorded 39 videos that are included in my dataset. I actually recorded many more videos; however, they either focused entirely on my hardware setup and acted as test videos, or I chose to exclude them due to their content as per my ethics agreement with myself. I only ended up excluding one video due to traumatic content, but I did make note of the fact that I was silent and nearly catatonic for over 75% of the video. I later blogged about the experience in order to preserve the trigger that led to the extreme dissociation. There are a total of 24 hours and 34 minutes of video that I eventually transcribed manually, resulting in a 100,038-word corpus of analyzable video data.
There is a speculative question that lies at the heart of this project; namely, how can collective play and participation in video game culture help us cope with, This has resulted in an extremely personal corpus of data that not only explores my childhood trauma, but also provides important insights into how video games, and my academic study of them, currently works to mediate my C-PTSD symptoms.
Step two: Data analysis techniques One aspect of my data analysis involved a qualitative open and axial coding of all blogs, field notes, and let's play video transcriptions. I conducted this coding before running any digital text analysis in order to determine what themes and patterns I could derive from the data. As an exploratory study based in grounded theory, this coding was highly personal and relatively informal. A detailed analysis of my findings is not appropriate for this paper; however, I will say that the process was an emotional and often triggering one. For example, each time I read the name of a particular person who caused me a lot of harm, I noticed myself skimming through the section as quickly as possible. My reaction to reading my own recollection of my trauma was often dissociation in the form of avoidance. This difficulty, which I did word I spoke, including words like "um" which ended up being my most frequently used word. During this second viewing, I also made notes related to mumbling and stuttering by using three asterisks followed by the description followed by three more asterisks; for example: "um um I am talking ***mumble*** I th think ***stutter***." I wanted to have a clear way of making note of my breaks and verbal intermittences while being able to quickly identify them during qualitative coding. The third viewing involved me taking note of emotional reactions, long pauses in speech, and obvious signs of dissociation. For the pauses, I made side notes about how long I sat in silence for. For the emotions and dissociation, I coded them as I perceived them which is another reason that I did not follow any formal guidelines. My C-PTSD alters how I express emotions in various situations. A single tear can represent anger while a smile can represent shame. As an intimately subjective process, I felt that using my own transcription techniques would provide me with the best information to accurately transcribe my life narrative into my "life-as-text" for digital analysis.

XML tagging
As this is a paper about an exploratory project aimed at developing grounded theory, it is important to address the things that both worked, and did not work, in allowing me to find the patterns and themes that qualitative coding missed. XML tagging is one of the things that did not work for me…as it was intended to work. Admittedly

Sapach: Tagging My Tears and Fears
Art. 4, page 14 of 28 a novice in this area, my original plan did not include XML tagging as an option.
During a discussion with one of my dissertation supervisors, we began contemplating the best ways to organize my life narrative into a cohesive, easy to navigate, format.
He suggested that I incorporate structured, hierarchical metadata into my corpus through the use of XML. Having worked with HTML previously, I did not anticipate a problem with the learning process.
I installed Oxygen XML Editor 19.1, and began researching the best way to tag my transcripts and blogs. I read through the book XML Pocket Reference 3 rd ed (St Laurent and Fitzgerald 2005) and discovered that RELAX NG seemed to be the appropriate schema to utilize due to its simplicity and flexibility. I began by following a simple hierarchy, including the use of a <head> element which contained the date of the entry as well as the time of recording and video game information for video transcriptions (see Figure 1). I then grouped paragraphs thematically. For video recordings, I used the category "setup" when I was talking about recording issues, Figure 1: Example of XML tagging for a video game recording transcript using a basic RELAX NG schema with personalized tags.

Sapach: Tagging My Tears and Fears
Art. 4, page 15 of 28 "description" when I was talking about the video game being played, "memory" when I was talking about memories, "methodology" when I was breaking the fourth wall and talking about the process of recording as a research method, and "interpretation" when I brought up academic theory or insights about my experience replaying the game. Figure  XML tagging brought me face-to-face with things that I might otherwise ignore or shut-out due to my C-PTSD. Ultimately, the process became more harmful than beneficial in that I was "forcing" myself to attempt to categorize my symptoms as emotional responses. In my lived experience, the primary emotion I can genuinely identify is fear. I do feel things like love, happiness, sadness, and anger, but they are more often than not accompanied by some sense of fear, or shame. My trauma created this confusion of emotions, and I have learned to live with and navigate them. When conducting XML tagging, I was asking myself to pick an emotion, and concretize it as either positive or negative. I eventually acknowledged that I was incapable of doing it, so I stopped.
Before moving on, I would be doing a disservice to this article if I did not mention the following research. My focus on attempting to categorize my emotions using XML as a precursor to potential sentiment analysis stemmed from work that has been done on utilizing text-analysis and sentiment analysis to identify PTSD related information for the purposes of identifying potential PTSD sufferers.
Vadim Kagan, Edward Rossini, and Demetrios Sapounas (2013)  I was inspired by the research and had hoped to find indicators of C-PTSD through an analysis of my own "life-as-text." Not in a formally scientific way as the above study did: my project is not a formal scientific study. What I had hoped was that there would be patterns to my emotions that might provide more insight into my own symptoms. As I noted however, as I began confounding my symptoms with emotions, I realized that the "complex" part of C-PTSD is thus named for a reason.

Sapach: Tagging My Tears and Fears
Art. 4, page 17 of 28 I still find the above study fascinating and will pursue extension of the results to C-PTSD in the future.

Python/NLTK/R
I originally began my descriptive analysis using the Python NLTK library but I gradually transitioned over to R largely due to the fact that I was required to learn and work with R for an unrelated project. I will first describe the interesting issue around the use of stopwords I encountered while analyzing word frequencies using Python followed by a discussion about dictionary/sentiment analysis conducted using R. With the understanding that standard practice is to exclude stopwords from natural language processing (Bird, Klein, and Loper 2009), I cleaned my corpus accordingly.

My initial descriptive analysis using
Upon running my word frequency analysis, I realized that the word "I" was excluded as it is one of the stopwords in the NLTK library. As I was analyzing a first-person life narrative, I was dissatisfied with the exclusion of such an important word, so I reran the analysis on a version of the corpus with the majority of the stopwords reincorporated.
I made the choice to exclude the words "the" and "and" as I was not particularly interested in exploring their frequency and had found no significant patterns of their usage in my qualitative analysis. Again, as an exploratory study, I was fundamentally playing with my data to see what would happen. Figure 2 shows the frequency distribution of the reintroduced stopwords only as I was interested in understanding how often "I", and other singular first-person pronouns, actually appeared.
The fact that the word "I" appears 7,129 out of 128,252 times is not overly surprising. And yet, for an autoethnography, I actually expected the word "I" to occur more than 5% of the time, as I not only spoke in the first person during my videos

Sapach: Tagging My Tears and Fears
Art. 4, page 19 of 28 said, I have stated above that a driving force behind my use of computer mediated text analysis was to seek out and acknowledge the breaks, lapses, and absences in my "life-as-text." Put into context, my XML tagging and qualitative coding indicated that the word "um" most often occurred before a significant pause, right around the time that I seem to begin experiencing some sort of emotion or connection to my trauma. These significant pauses are usually accompanied by a switch to an intense focus on whichever game I happen to be playing. XML tagging also seemed to suggest that I often switch to talking about in-game events after my "um's" and long pauses, indicating that I am using the game to re-focus and remove myself from the traumatic feelings associated with recollection. I am fascinated by the discovery of a pattern of behaviour that I was not consciously aware of prior to this research. My use of the word "um" as an indicator that I am becoming emotionally uncomfortable or unsafe is a result worthy of deeper analysis in my autoethnography.
Moving away from word frequencies, I was curious about the appearance of categories of words as found in the Harvard IV-4 and associated dictionaries (2020).
Using R, I once again cleaned my corpus of data and divided it into two sections: blogs and games (representing my journal entries and video recording transcriptions     I wanted an unsupervised, unguided perspective on the themes that might be found in my life narrative. In the previously mentioned analysis techniques I controlled the code, the visualizations, and the selection of dictionaries. I wanted to try to obtain a digital analysis of my "life-as-text" with the least amount of interference from me as possible, and topic modelling seemed like the best fit based on my skillset.
Using Mallet I ran an initial set of topic models, 10, 20, and 50, on the complete corpus of all of my transcribed videos and blogs. I did remove the standard NLTK stopwords, though you may notice some of them in Figure 7, as I struggled to Figure 7: 10-topic model created using Mallet. Note that due to a character set conversion error, some of the NLTK stopwords remain, though most were removed.
correctly remove the punctuation due to novice mistakes around character sets.
For example, the word "don't" is included, but it actually read "don" in the output. I left it in however, as it carries certain contextual meaning related to negativity. The unintentional inclusion of the words "I'll" and "I've" also reinforce the importance of singular first-person pronouns in my corpus.
There are two "outliers" in this example that stand out like sore thumbs to me. The 7 th one, "light, road, back, feel, room, follow, hmmmm, find, tinder, story, sanity, accident, dreams, creepy, door, interesting, car, missing, photos" speaks to my combined playthroughs of the games Trauma and Amnesia: The Dark Descent.
Most of the words describe in-game objects and activities. The 10 th one, "world, truth, people, post, facts, science, sonja, real, gonna, play, facts, comfort, passion, identity, social, important, sort, emotion, fantasy" comes from a single video that I recorded as part of another project. I have included the transcription of the video in my corpus as it does contain important autoethnographic points and was technically recorded while playing a game during my data collection timeline.
The remaining 8 topics are very interesting in that they make sense to me. Each topic reminds me of a specific aspect of my life. I see myself in my data. I see myself in ways that I cannot otherwise put into words. The topics provide an odd sense of stability in a seemingly chaotic assortment of words. To be honest, the actual statistical significance of each topic is relatively unimportant to me. How I use the topics is largely related to story-telling. As I pieced together my autoethnographic narrative and connect my "life-as-text" to various themes and theories, the topic models have helped me group things together that I would otherwise have a difficult time accessing and connecting due to dissociation and the emotional attachment I have to my data. These topics encouraged me to acknowledge my ingrained tears, fears, guilt, shame, and isolation. I have been able to access and examine things that have remained hidden due to my C-PTSD. It is through the computer mediated analysis of my data that I have been able to step back from my emotional attachment and ingrained assumptions, and view the data through a somewhat logical, orderly lens: a lens temporarily cleared of the fog of trauma. The topic modelling in particular has allowed me to step back from my dissociation and destabilization, to view my experiences as connected parts of a larger whole.

Conclusion
My "life-as-text" is more than just the corpus of words that I have gathered. It is a series of interconnected narratives, experiences, and emotions. My relation to my "life-as-text" can be seen as a deficient one, an alienated one. My life narrative comes from me, it is me, and yet I feel separated from it in many ways. My C-PTSD makes me afraid of my own story. My acceptance of my own destabilization and dissociation makes the process of writing an autoethnography a monumental undertaking. I have found, as described above, that the addition of digital text analysis to my autoethnographic process has allowed me to regain and rework my relationship to myself and my narrative. By allowing me to step back and view my "life-as-text" through a somewhat objective lens, methods like topic modelling are aiding in the process of appropriation that Jaeggi (2014) argues is so central to the resolution of alienation. Recall that appropriation needs to be an active and involving process of integration and transformation of meaning and experiences. Analyzing my life narrative using digital methods has allowed me to take an active role in first understanding, and then transforming the meaning of my trauma.
Shining a light on my breaks, lapses, and absences has allowed me to better understand myself despite my C-PTSD and related dissociation. Writing the autoethnography, which in some ways is more of an autobiography, has fostered the mediation of my own alienation. By allowing me to embrace and examine my "tears and fears" through an unsupervised and, in some ways impartial, digital lens, text-mining has provided me with a foundation upon which the reconstruction and acceptance of a haunting and painful past is possible.

Competing Interest
This article draws on research supported by the Social Sciences and Humanities Research Council.

Editorial Contributors
Congress 2018 Special Editor: Constance Crompton, University of Ottawa, Ottawa, Ontario, Canada.
Section/Copy Editor: Shahina Parvin, University of Lethbridge Journal Incubator.