People, Places and Emotions:
Visually Representing Historical Context in Oral Testimonies
Annie T. Chen, Ayoung Yoon, Ryan Shaw
University of North Carolina at Chapel Hill
School of Information and Library Science
E-mail: atchen, email@example.com, firstname.lastname@example.org
This paper presents visualizations to facilitate users’ ability to understand personal narratives in the historical and sociolinguistic
context that they occurred. The visualizations focus on several elements of narrative – time, space, and emotion – to explore oral
testimonies of Korean “comfort women,” women who were forced into sexual slavery by Japanese military during World War II.
The visualizations were designed to enable viewers to easily spot similarities and differences in life paths among individuals and also
form an integrated view of spatial, temporal and emotional aspects of narrative. By exploring the narratives through the interactive
interfaces, these visualizations facilitate users’ understandings of the unique identities and experiences of the comfort women, in
addition to their collective and shared story. Visualizations of this kind could be integrated into a toolkit for humanities scholars to
facilitate exploration and analysis of other historical narratives, and thus serve as windows to intimate aspects of the past.
Keywords: historical narratives, personal narratives, visualizations, emotion identification, comfort women
As individuals travel through life, the experiences they
partake in, their inner states, and their interactions with
the world become unique life journeys. However, their
life paths also cross, and they themselves are set in
particular social ecological contexts. Thus, personal
narratives can be invaluable sources for understanding
individuals in the past and present, as well as acquiring a
broader sense of their common experiences in different
social and cultural settings through the passage of time.
Theories of narrative construction and coherence have
often examined the importance and yet, difficulty, of
making sense of time in narrative. Narration is
distinguished by ordering and sequence; narrators create
plots from disordered chronological experience (Cronon,
1992, p.1349). Personal narratives link temporal
properties to spatial ones; “they look back on and
recount lives that are located in particular times and
places… the narratives themselves are produced in
particular times and places” (Laslett, 1999, p.
392). Space and time are also key dimensions along
which we construct our understanding of narratives
While space and time structure our personal narratives,
emotions give them meaning. As Jones (2005) writes,
“Life is inherently spatial, and inherently emotional…
Each spatialized, felt, moment or sequence of the
now-being-laid-down is … mapped into our bodies and
minds to become a vast store of past geographies which
shape who we are and the ongoing process of life” (p.
205-206). Without these emotions there is perhaps a
question as to whether the traces of our experience
would continue to linger with us: “There is also
inevitably in each memory the expression of emotion; it
is almost as if these memories could not exist if there
had not been strong emotion felt and then expressed in
the face, body, or gut” (Singer & Salovey, 1993. p. ix).
These emotional attachments are what distinguish
personal narratives from simple chronologies of life
events. The narrator plays a critical role in the shaping
of narrative: “With narratives, people strive to configure
space and time, deploy cohesive devices, reveal identity
of actors and relatedness of actions across scenes. They
create themes, plots, and drama. In doing so, narrators
make sense of themselves, social situation, and history”
(Bamberg & McCabe, 1998: ). Personal narratives
and the emotions that are reflected within them “serve as
a window to identity” (Horrocks & Callan, 2006). Thus,
through narrative, readers can come to see how
individuals make sense of themselves and their world.
However, in perusing narrative, it is not always easy for
readers to make meaning of what they read. This
difficulty may arise from a variety of factors such as
space and time discontinuities and the inherent diversity
of stories and life experiences. It is with regard to these
difficulties that text mining and visualization methods
may be of assistance.
This paper proposes a number of visualizations to
facilitate users’ ability to understand personal narratives
in the historical and sociolinguistic context that events
unfolded. The visualizations focus on several elements
of narrative – time, space, and emotion – to explore a
particular corpus: oral testimonies of Korean “comfort
women,” women who were forced into sexual slavery by
Japanese military during World War II. The methods
also leverage shared resources that are often used in text
mining, emotion detection and sentiment analysis.
1.1 Personal Narratives in Historical Context:
Oral Testimonies of Korean “Comfort Women”
The Japanese military sexual slavery system, or the
“comfort women” system, was in operation from 1932
to 1945, during the period of the Manchurian and Pacific
wars (for more detail, see Yoshiaki, 2000; Stetz & Oh,
2001; Chung, 1997). The exact number of women who
were drafted into the sexual slavery system is still
controversial, but it is generally estimated at 200,000 or
more. The wide mobilization of military sex slaves was
in the context of mobilizing human resources from
occupied territories as part of the war effort. The
majority of them were Korean, aged 14-19, from the
rural lower classes (Chung, 1997), but women from
China, Taiwan, and the Philippines were also forced to
serve as “comfort women.”
Due to various complexities including the power
relationship in East Asia, diplomatic relations between
Korea and Japan, and efforts by the Japanese
government to keep the military sexual slavery system
secret, the existence of “comfort women” was not
revealed until 50 years after the war ended. In addition,
as the experience was a “shameful” part of an
individual’s personal past, the victims were reluctant to
identify themselves or to be formally identified as
“comfort women” (Chung, 1997). While feminists,
human rights activists, and historians have worked to
raise public awareness of this chapter of history, the
individual stories of “comfort women” are neither part
of the official national histories of the countries involved,
nor exist as part of their collective memories.
As the testimonies of the comfort women include
experiences, perceptions, and emotions, testimonies can
be seen as one form of personal narrative, trauma
narratives, in a historical context. However, they are also
different from personal narratives, as they are known to
often contain more political tendencies, engagements of
readers’ sympathy, and more possibilities of intentional
narrator intervention (Stephen, 1994; Beverley, 1991;
Kaplan, 1991; Sommer, 1988).
The “comfort women” narratives are similar to other
personal narratives in that they may be fragmented, and
that they may also have factual errors, omissions, and
contradictions. At the same time, their narratives are
surprisingly detailed, including the names of ships that
transported them from place to place, the names of their
companions, and the names of the small towns by which
they passed – their memories particularly vivid,
persistent, and somatic, as has often been observed with
trauma narratives (Misztal, 2003). In addition, extreme
events connected their experiences to certain emotions,
which reflect how they see and understand those
experiences and actions.
1.2 Techniques for Mining People, Places and
Text mining techniques have previously been used for
extracting information about historical events and
displaying them using maps and timelines (e.g. HiTiME,
Yamamoto et al., 2011). The Historical Timeline
Mining and Extraction (HiTime) Project has developed a
text analysis system for the recognition and extraction of
historical events and facts from primary and secondary
historical sources such as biographies, brochures, letters
and old newspaper articles (http://ilk.uvt.nl/hitime/).
ThemeRiver employs a river metaphor to depict changes
in thematic variations over time in a large document
collection (Havre, Hetzler, & Nowell, 2000).
There has also been substantial research on automated
methods for identifying emotional expression in
narrative. Many studies employed the Linguistic
Inquiry and Word Count software, which provides
statistics on the presence of words representing
emotional and cognitive processes, as well as various
linguistic patterns (Pennebaker & Francis, 1996) (e.g.
Bantum & Owen, 2009; Liess et al., 2008).
SentiProfiler incorporates Wordnet-Affect to support the
visual examination of sentiment in Gothic literature
(Kakkonen & Kakkonen, 2011). Plaisant et al. (2006)
demonstrated how text mining and visualization could
be used to explore erotics in a corpus of letters between
Emily Dickinson and her sister-in-law. Pennebaker and
Gonzales (2009) illustrated how linguistic patterns in
blog posts might comprise historical memories and
reflect the social dynamics of traumatic events.
Considering the literature, it becomes apparent that
though there has been previous work in highlighting
temporal, spatial and emotional aspects of narrative,
extant systems do not readily support the visual
integration of these three elements of narrative. However,
it is also evident that these elements are inextricably
intertwined, both in experience and memory. Tools that
facilitate visual synthesis of these aspects of historical
narrative could be of invaluable assistance to
scholars. Thus, the aim of this paper is to propose
methods for textual analysis in the spirit of casting light
upon the historical and sociolinguistic context of the
narratives, as well as the life course of the individuals
whose stories are being told.
The narratives employed in this analysis were compiled
from two anthologies of translated interview content,
compiled by the Korean Council for the Women Drafted
for Military Sexual Slavery by Japan and Korea
Chongshindea’s Institute (Howard, 1995; Schellstede &
Yu, 2000). To render the content to digital format, the
content was scanned and visually inspected to correct
any errors. Person and place names were identified
using the Stanford Named Entity Recognizer (Finkel,
Grenager, & Manning, 2005).
Emotional content was identified using a lexicon that
was constructed based on Wordnet-Affect (Strapparava
& Valitutti, 2004). In order to understand more about
what the women in the testimonies thought and how
they viewed themselves, two categories were added:
Cognition and Self-reflexivity. Sentences were
identified as involving cognitive processes if at least one
of the following words appeared in any tense: feel, think,
believe, and wonder. Sentences were labeled
self-reflexive if there was reference to “myself” within
the sentence. The selection of these words was partially
guided by Raskovsky, Slezak, Wasser, and Cecchi’s
(2010) study of instropection in texts, and by the authors’
own reading of the narratives. Following the extractions,
visualizations were generated using PHP scripts.
3. Visualizing Personal Narratives
3.1 Juxtaposition of Life Paths
The purpose of the first visualization is to assist the
viewer to examine the life courses of individuals as
compared to others (Fig. 1). Each row represents the
life course of one woman, and the constituent elements
are places that she mentions in her testimony. The
places appear in order of appearance in the text. The
paths are aligned based on the places selected by the
viewer. In Figure 1, the focal point of “Shinuiju” is
selected. This visualization enables the user to identify
and peruse testimonies that share commonalities.
Figure 1: Juxtaposition of Life Paths
Some of the challenges in modeling these paths were
differing levels of granularity in place names, as well as
the lack of place names. In some cases, the names of the
places that women were located were never mentioned,
perhaps because they were unclear about where they
Given that many of the places that the comfort women
stayed over the course of their lives may be unfamiliar
to the reader, and that they traveled from place to place
so often that it would be difficult to grasp even for those
who are familiar, a map representation was generated
using Google Maps API V.3. This representation allows
the reader to see the paths taken by the women
throughout the narrative (Fig. 2).
Figure 2: Spatial Depiction of a Life Path
3.2 People, Places and Emotion
As the literature review demonstrated, the experience of
being is inherently spatial, temporal and
emotional. Thus, this visualization was conceived to
facilitate user exploration of this multidimensional
landscape. Scanning the interface below from left to
right, one can quickly acquire an overview of the
affective content of the text, as well as significant people
and places (Fig. 3). As the user hovers over the circles,
the sentences that contain affective content are displayed
in an info-bubble.
Figure 3: People, Places and Emotions
4.1 Contextualizing Experiences Lived
As Cronon (1991) writes, “to recover narratives people
tell themselves about the meanings of their lives is to
learn a great deal about their past actions and about the
way they understand those actions” (p. 1369). The
visualizations presented in this paper can facilitate users’
exploration of narratives; assist them to make
connections between temporal, spatial, emotional and
cognitive elements; and help them to understand the
comfort women as individuals, as well as in terms of
their collective experience.
It goes without saying that there are similarities in their
experiences. As the visualizations show, many of those
who later became comfort women were taken to
Shimonoseki, where the women were dropped off before
being assigned to other locations. It is also possible to
quickly see that fear is by far the most common emotion
felt by the women, not surprising given their
experience. However, the consistent presence of other
emotions such as sadness, regret, anger and hopelessness
also contextualize their experiences.
Aside from discovering similarities, the visualizations
can assist the reader in other ways as well. For instance,
if one flips through the People, Places and Emotions
visualization (Fig. 3) for several women, one might
notice that sadness and regret are present at the end of
many narratives, particularly those generated from
Howard (1995). Through this, we can perhaps see the
work of the editor to end each testimony with the parting
thought from the focus individual regarding their past,
their desires for reparations or apologies, and so on.
Examining the sentences that appear as one hovers over
Sadness and Regret at the ends of the narratives, the
reader may realize that though there are similarities in
the women’s attitudes, their different attitudes also shine
through: “It was bad enough that I had to suffer what I
did. I am bitter when I think of this, but I am not going
to blame others any longer”1 (Yi Yongsuk), “Of course
Japan is to blame, but I resent the Koreans who were
their instruments even more than the Japanese they
worked for” (Kim Tokchin), and “Who would be able to
guess what inner agony I suffer in my heart?” (Choe
Myongsun). In the words of Kim Haksun, “Once I am
dead and gone, I wonder whether the Korean and
Japanese governments will pay any attention to the
miserable life of a woman like me.”
By highlighting the women’s thoughts and references to
themselves, the Cognition and Self-Reflexivity
categories provide yet another view of the individual
characters of the women. For example, the People,
Places and Emotions visualization enables users to
follow Kang Tokkyong through the narrative,
experiencing her abandonment with her as she finds
herself alone in a truck, and then witnessing her defiant
spirit with utterances such as these: “If such a thing
happened now, I would kill myself by biting my tongue
off,” and “I tried to throw myself off of the ship as we
crossed the sea to Korea, but this woman sensed what
was going on and followed me everywhere, making it
impossible for me to take my own life.”
As the above utterances demonstrate, though the women
featured in these testimonies share similarities of
experience, there also aspects of their experiences and
their reactions to them that are different. In principle,
simply by moving one’s mouse over the People, Places
and Emotions visualization for each woman, the user
can acquire a taste of these differences, and then click
into the testimonies for a deep perusal.
This visualization is meant to support Wertsch’s
distributed approach to collective remembering, in
which, though collective memory is inherently social,
there is not “a single system of uniform knowledge and
belief,” but rather, a need for “collaboration between
those focusing on individual remembering and those
concerned with collective phenomena” (Wertsch, 2009,
p. 132). It may also stem the tide that Greene (2004) has
observed of the focus of memory studies shifting away
from individual remembering.
4.2 Implications for Shared Resources and
The visualizations discussed in this paper potentially
contribute to the dialogue on shared resources in various
ways. First, this study employed extant resources for the
mining and visualization of a particular type of narrative,
and therefore serves as an example of the applicability
of these tools to this type of narrative. In the case of
emotions, there were a significant number of false
positives due to words in the lexicon that could take on
different meanings. Future work could integrate a
mechanism for word sense disambiguation or a machine
learning approach to emotion identification.
1 This excerpt and all following excerpts are from Howard
The visualizations in this paper primarily facilitated user
exploration of the nexus of time, space and emotion.
Various other aspects of narrative might be visualized in
a similar fashion.
For example, topic modeling
techniques might be used to extract common themes and
motifs from the narratives, and then the motifs could be
juxtaposed with the other elements of time, space and
emotion. Other aspects of the narrative such as
active/passive voice, frequency of pronouns, etc., might
also be integrated to provide additional methods of
exploring context and mood.
In addition, the testimonies visualized in this paper were
obtained from translated interview content. The
techniques used in the visualizations might be applied to
content in other languages, such as Korean and Chinese.
An interface facilitating comparisons of testimonies in
multiple languages might enable researchers to explore
differences in representation due to translation, editorial
style, linguistic structure and culture.
The techniques used in these visualizations could also be
applied to other narratives. Historical testimonies serve
as a memory of the experiences of particular groups,
such as Holocaust survivors2, Iraqi refugees3, and
survivors from other genocides4. As “the notion of
testimony expresses urgency, a story that must be told
because of the struggles it represents” (Stephen, 1994, p.
224), testimonies have increasingly gained attention
from various fields as windows to unknown or
little-known “truths,” and to promote social
justice. Visually representing individuals’ life traces
could be a way of representing collective experiences
involving marginalization, repression, and oppression,
thus granting access to intimate aspects of the past.
This paper sought to design visualizations that would be
helpful for analyzing historical narrative. These
visualizations enable viewers to easily see the sequence
of places for any one individual, spot similarities and
differences in their life paths, and form an integrated
view of spatial, temporal and emotional aspects of
narrative. These types of visualizations could be
integrated into a toolkit for humanities scholars to assist
them in exploring and analyzing narratives.
This work was partially supported by National Science
Foundation grant IIS 0812363.
2 United Holocaust Memorial Museum:
imony/; USC Shoah Foundation Institute:
3 Refugee Council USA:
4 The Holocaust Memorial Day Trust:
Bamberg, M., & McCabe, A. (1998). Editorial.
Narrative Inquiry, 8(1), pp. iii--v.
Bantum, E. O., & Owen, J. E. (2009). Evaluating the
validity of computerized content analysis programs
for identification of emotional expression in cancer
narratives. Psychological Assessment, 21(1), pp.
Beverley, J. (1991). The margin at the center: On
testimonio (Testimonial Narratives), In Smith, S. and
Watson, J. (Eds.). De-colonizing The Subject.
Minneapolis: University of Minnesota Press.
Cronon, W. (1992). A place for stories: Nature, history,
and narrative. Journal of American History, 78(4), pp.
Chung, C. S. (1997). The Origin and Development of
the Military Sexual Slavery Problem in Imperial Japan.
Positions, 5(1), pp. 219--255.
Finkel, J. R., Grenager, T., & Manning, C. (2005).
Incorporating non-local information into information
extraction systems by Gibbs Sampling. In
Proceedings of the 43nd Annual Meeting of the
Association for Computational Linguistics (ACL 2005),
Greene, A. (2004). Individual remembering and
‘collective memory’: Theoretical presuppositions and
contemporary debates. Oral History, 32(2), pp.
Havre, S., Hetzler, B., & Nowell, L. (2000). ThemeRiver:
Visualizing theme changes over time. Paper presented
at the Information Visualization, 2000. InfoVis 2000.
IEEE Symposium on, pp. 115--123.
Horrocks, A. & Callahan, J. (2006). The role of emotion
and narrative in the reciprocal construction of identity.
Human Resource Development International, 9(1), pp.
Howard, K. (Ed.) (1995). True Stories of the Korean
Comfort Women. Y. J. Lee (Trans.). London: Cassell.
Jones, O. (2005). An ecology of emotion, memory, self
and landscape. In Davidson, J., Smith, M., & Bondi, L.
(Eds.), Emotional Geographies. Hampshire, England:
Ashgate Publishing Limited, pp. 205--218.
Kakkonen, T., & Kakkonen, G. G. (2011). SentiProfiler:
Creating comparable visual profiles of sentimental
content in texts. In Proceedings of Language
Technologies for Digital Humanities and Cultural
Heritage Workshop, pp. 62--69.
Kaplan, C. (1991). Resisting autobiography: Out-law
genres and the transitional feminist subject. In Smith,
S. and Watson, J. (Eds.). De-colonizing The Subject.
Minneapolis: University of Minnesota Press.
Laslett, B. (1999). Personal narratives as sociology.
Contemporary Sociology, 28, pp. 391--401.
Liess, A., Simon, W., Yutsis, M., Owen, J. E., Piemme,
K. A., Golant, M., & Giese-Davis, J. (2008).
Detecting emotional expression in face-to-face and
online breast cancer support groups. Journal of
Consulting and Clinical Psychology, 76(3), pp.
Misztal, B. A. (2003). Chapter 6: Studying memory. In
Theories of Social Remembering. Maidenhead: Open
University Press, p. 142.
Pennebaker, J. W., & Francis, M. E. (1996). Cognitive,
emotional, and language processes in disclosure.
Cognition & Emotion, 10(6), pp. 601--626.
Pennebaker, J. W. & Gonzales, A. (2009). Making
history: Social and psychological processes
underlying collective memory. In Boyer, P., &
Wertsch, J. V., Memory in Mind and Culture.
Cambridge: Cambridge University Press, pp.
Plaisant, C., Rose, J., Yu, B., Auvil, L., Kirschenbaum,
M. G., Smith, M. N., Clement, T., & Lord, G. (2006).
Exploring erotics in Emily Dickinson’s
correspondence with text mining and visual interfaces.
JCDL ‘06, pp. 141--150.
Raskovsky, I., Slezak, D. F., Wasser, C. D., & Cecchi, G.
(2010). The emergence of the modern concept of
introspection: A quantitative linguistic analysis.
NAACL Young Investigators Workshop 2010.
Schellstede, S. C., & Yu, S. M. (Eds.). (2000). Comfort
Women Speak: Testimony by Sex Slaves of the
Japanese Military: Includes New United Nations
Human Rights Report. New York: Holmes & Meier.
Singer, J. A, & Salovey, P. (1993). The Remembered Self:
Emotion and Memory in Personality. New York: Free
Sommer, D. (1988). ‘Not Just a Personal Story’:
Women's testimonios and the plural self. In Brodzki,
B., & Schenck, C. (Eds.), Life/Lines: Theorizing
Women's Autobiography. Ithaca: Cornell University
Press, pp. 170--130.
Stephen, L. (1994). Hear My Testimony: Maria Teresa
Tula, Human Rights Activist of El Salvador. Boston:
South End Press.
Stetz, M. & Oh, B.C. (Eds.). (2001). Legacies of the
Comfort Women of World War II. London: M. E.
Strapparava, C., & Valitutti, A. (2004).Wordnet-affect:
An affective extension of wordnet. In Proceedings of
the 4th International Conference on Language
Resources and Evaluation (LREC 2004), pp.
Wertsch, J. V. (2009). Collective memory. In Boyer, P.,
& Wertsch, J. V., Memory in Mind and Culture.
Cambridge: Cambridge University Press, pp.
Yoshiaki, Y. (2000). Comfort Wom en . New York:
Columbia University Press.
Yamamoto, M., Takahashi, Y., Iwasaki, H., Oyama, S.,
Ohshima, H., & Tanaka, K. (2011). Extraction and
geographical navigation of important historical events
in the Web. In K. Tanaka, P. Fröhlich, and K. S. Kim
(Eds.). W2GIS 2011, LNCS 6574, pp. 21--35.
Zwaan, R. A. (1999). Five dimensions of narrative
comprehension: The Event-Indexing Model. In
Goldman, S. R., Graesser, A. C., Broek, P. V. D, (Eds.).
Narrative Comprehension, Causality and Coherence:
Essays in Honor of Tom Trabasso. Mahwah, NJ:
Lawrence Ehrlbaum Associates, Inc., pp. 93--110.