See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/237603717

Analysing weblogs of terminally ill patients using the Linguistic Inquiry and
Word Count (LIWC) program

Article

CITATIONS READS

0 156

2 authors:

Stella Mills Paul Kingston

Staffordshire University University of Chester
38 PUBLICATIONS   250 CITATIONS    106 PUBLICATIONS   804 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

A critical analysis of adult safeguarding in NHS mental health services View project

Brightlife Evaluation: addressing social isolation and loneliness in older adults View project

All content following this page was uploaded by Paul Kingston on 21 March 2015.

The user has requested enhancement of the downloaded file.

Analysing weblogs of terminally ill patients using the Linguistic Inquiry and Word
Count (LIWC) program.

Authors: Nothando Ngwenya (Centre for Ageing and Mental Health)

Professor Stella Mills (Faculty of Computing, Engineering and Technology)

Professor Paul Kingston (Centre for Ageing and Mental Health)

Affiliation: Staffordshire University

Correspondence e-mail: [email protected]

ABSTRACT
The phenomenon of death is common to us all, yet the experience of dying is rarely openly discussed.
Patients with terminal illness receiving palliative care are more aware of their imminent death. The
impact of technological advancement has had an effect on the way some terminally ill patients in
American and Australian hospices express themselves through the use of weblogs as a form of
therapeutic writing to journal and help interpret death. According to the WHO’S definition of
palliative care, patients’ psychosocial and spiritual needs have to be addressed alongside their physical
and mental health needs. Using a weblog can help address mental and spiritual needs of terminally ill
patients through therapeutic writing. This paper describes the use of the Linguistic Inquiry and Word
Count (LIWC) program to analyse the emotional, cognitive and structural components present in the
blogs of five terminally ill adult patients. The results of this study show that terminally ill patients use
a high percentage of affective words, and have a high percentage use of the word death as a personal
concern.

INTRODUCTION
Death and Dying
Even though death is a universal human experience, responses to this phenomenon vary across cultures
and societies. Conceptualisation of death in Western society has moved from the eighteenth century
where it was public and visible to the nineteenth century where it became an invisible and private
affair.1In the twentieth century, death became bureaucratised and the conduct of society encouraged
people to place all their faith in medical technologies thus effectively denying death. 2-5 For people to
accept death as part of nature and to deal with dying in a dignified manner, they need to understand the
phenomenon of dying and death .6 The twenty first century has seen a slight change of attitude towards

1 
 
dying and death within society, where death is becoming perceptible once more with a move towards
open awareness of death. Open awareness is when the patient, medical staff and relatives are all aware
of the patient’s impending death and can openly talk about it.7Taylor argued that the twenty first
century has not really embraced open awareness, but instead has moved towards conditional open
awareness.8 Field et al supported this argument by reporting on how medical professionals were not
comfortable about disclosing undesirable information especially when it involved a terminal
prognosis.9 Miyaji provides evidence which supports this argument in research conducted with
physicians in USA, who reported that they modified information given to terminally ill patients
according to the patient’s questions and what they felt the patients needed to know.10 All this research
substantiates the view that open awareness and honesty is still not fully practised when dealing with
dying and death. The increased mortality rate and technological innovations used to prolong life may
be well and good for today’s society but they raise significant questions about what life, dying and
death mean to people. Deconstructing or defining the meaning of death can help people understand and
deal with the practical aspects of the dying process and more so for terminally ill patients facing an
imminent death. Talking about dying and death to terminally ill patients helps them make meaning of
their death as well as informing us about the dying process which can assist in caring for dying
patients.11
Weblogs and Therapeutic Writing
With the reluctance of some professionals to provide open awareness, some terminally ill patients have
found other ways to communicate openly, through the use of technology. 12The internet has become an
important way for patients to find information and support .13 Patients using this form of
communication have reported that use of the Internet encouraged active coping and made them face
their fears and has been a personally empowering experience .14 Some critics have argued against the
use of technology in palliative care suggesting that technology is impersonal and objective.15-17 Seely
et al, go on further to suggest that technology eliminates the personal care involved in palliative care
and does not help to heal the patient.18 In contrast to this stance, a growing body of literature shows
that internet technology does help in the process of healing for patients.19-21 Internet-based
interventions in hospices and palliative care are proving effective for patients, caregivers and palliative
care providers.22, 23Most of the patients use the internet for e-mails, support groups or seeking
information, although there is a growing number of patients using weblogs for communication. The
weblogs referred to in this study are the journal type web pages that have posts in reverse
chronological order and function as a two way communication by affording visitors to post
comments.24 Many blog authors claim their blogs offered them some form of catharsis although there

2 
 
is no formal evidence of this; thus, this study explores the use of weblogs for terminally ill patients. By
offering a form of catharsis, weblogs are in effect promoting therapeutic writing.
A growing body of literature has demonstrated the positive effects of writing on physical, emotional
and mental well-being.25-27 Writing can help individuals to process and express their emotions. Cancer
patients who have used this mechanism of coping have experienced less cancer-related morbidities,
decreased stress and enhanced health compared to those with low emotional expression.28
Functionalist theories of emotion suggest that emotions can draw individuals’ attentions to their goals
and concerns prompting them to take adaptive action.29 Most research on emotional processing and
expression has utilised writing as a channel through which people can express themselves.30,31 This is
because writing is the most appropriate method of translating emotional experience into words,
therefore being a gateway to understanding people’s underlying emotional and cognitive processes.32
Language use and LIWC
Research evidence suggests that individuals construct the meaning of situations and experiences
through the language they use.33, 34 The study of language use is not a new concept with research as
early as 1901 where Freud purported that errors in speech reflected an individual’s deeper fears.35 The
literature on language use has continued to grow indicating the importance of language use in
understanding individuals. Ricoeur suggested that the way we describe an event defines the meaning
of the event and we use those meanings as a bridge to reality.36 Analysis of text can help to deconstruct
these individual meanings and explore people’s emotional and cognitive functioning .37 There is
empirical evidence that suggests a link between biological activity and function words. Research also
shows that depressed people and individuals experiencing a stressful situation use more first person
pronouns .38 Words can also help in the prediction of physiological and health changes with the use of
emotion and cognitive words being correlated with health changes.39 By analysing the text of
terminally ill patients, the important aspects of their psychological world can be revealed helping us to
deconstruct and understand the meaning of the dying process and their imminent death. Methods of
studying language use vary from those that make an assumption that language is contextual to those
that statistically count the words within text.40 Within the quantitative approach there are different
variations of word analysis, namely judge based thematic content analyses, word pattern analysis and
word count strategies. Thematic content analyses have been used mainly for studying psychological
phenomena whilst the word pattern analysis is useful for determining the co-variance across samples
of text. The word count strategies make the assumption that words convey psychological information
that is independent of their semantic context and is therefore suitable for use within this study.
Computerised text analysis programs have become widely used in the study of language use increasing
the efficiency and accuracy of analysis .41 The LIWC (Linguistic Inquiry Word Count) text analysis

3 
 
software was initially developed in the context of Pennebaker’s work to identify words that could
predict health improvements.42 This program is now widely used in clinical psychology to explore
psychological and linguistic dimensions of language. This paper evaluates and describes the use of the
LIWC program to analyse content of terminally ill patients’ weblogs. In contrast to previous work in
palliative care, where the investigator interviews participants, records and decodes what is said, this
paper focuses on the emotional, cognitive and structural components of written language.

METHOD
The Use of weblogs in palliative care
For clarification purposes within this paper, Schiano et al’s definition of weblogs will be applied - a
weblog is a web page that is frequently updated with posts in reverse-chronological order with the
latest post at the top.43The present study is based on weblogs of five terminally ill patients who have
maintained their weblog for over two years. Writing samples from the patients’ blogs were analysed
using LIWC. From the content of the blogs, the patients indicated how the Internet has enabled them
to obtain social support and express their emotions whilst discussing issues of importance to them
related to their illness and imminent death.
Participants
Participants were five patients with varying terminal illnesses based in North America. Two of the
patients have cancer, one has AIDS, one has leukaemia and the fifth an undisclosed terminal illness.
Three of the participants are male and two are female. The participants are between the ages of 31 and
53 years old with a mean age of 44 (SD = 8 years). One participant is a visionary artist, one is a former
newspaper reporter, one is a student, one is a radio news reporter and the fifth is a journalist.
Participants have been blogging at their own initiative and convenience for at least two years. Within
their blogs, participants reported on their illness, progress of their respective treatments and other
general life issues.
Materials – Software
The Linguistic Inquiry and Word Count software 2007 version was used to identify the emotional,
cognitive and structural categories of the words used within the text of the blogs.44 The default LIWC
2007 dictionary has 4, 500 words that define the word categories. One word can be part of more than
one category. The program analyses text, word by word and calculates the percentage of words that
match each of the language dimensions or categories defined within the dictionary. If a word belongs
to more than one category, each of the categories is incremented if the word is identified in the text.
The output variables include general descriptors (e.g. word count, word per sentence), standard

4 
 
linguistic dimensions (e.g. pronouns, verbs, articles), and psychological constructs (e.g. affective
processes, cognitive process, social references). The results of the output variables are listed as
percentages of word use.
Procedure
A total of 15 blog posts by terminally ill patients were analysed. Three posts from different stages of
blogging were selected from each participant’s blog (early, middle, and late). The posts were loaded
into Microsoft Word and edited for typographical errors. All postings were saved as Word text files
and analysed using the LIWC program. The LIWC data were aggregated by phase and patient, and
therefore for each patient, 3 data sets corresponding to the three blog phases were available. The
construct validity of the program was assessed by comparing the LIWC results obtained from the
participants’ blogs with the norm values from the program authors .45 The LIWC has 90 language
dimensions, but for this study, only 12 dimensions that have been found to correlate with social and
psychological variables were examined in the analysis of function and emotion words. This analysis
was used to explore the participants’ social and psychological profiles.

RESULTS
Table 1 summarises the LIWC categories relevant in this study that were analysed. The numbers
shown are the mean percentages of the total word use from the three different phases of the blog posts
that were analysed. A comparison of this study’s results was carried out against the average values
published by the program’s authors. The LIWC 2007 authors carried out a study to evaluate the
program’s word usage across various setting which included emotional writing, control writing,
science articles, novels, talking and blogs. The three classes of text from Pennebaker’s study are shown
in bold (Emotional writing condition; Control writing condition; Blogs).

5 
 
Table 1: Means for LIWC Categories
Emotional Control
Case 1 Case 2 Case 3 Case 4 Case 5 W W Blogs
Category
Linguistic
Processes

Word Count (mean) 632.33 669.00 306.00 959.67 407.00 443.00 405.00 7,304.00

Words > 6 Letters 15.74 15.83 11.14 19.20 16.27 13.27 13.87 14.12
Total function
words 60.04 56.00 62.13 57.21 58.12 63.93 57.53 55.29

Pronouns 13.29 10.96 18.25 14.18 16.87 20.23 14.29 16.07

Psychological
Processes

Affect 5.81 5.28 5.06 6.27 5.76 6.02 2.57 5.84

Positive emotion 3.36 3.92 3.84 3.10 3.43 3.28 1.83 3.72

Negative emotion 2.33 1.39 1.22 3.10 2.22 2.67 0.71 2.07

Anxiety 0.89 0.13 0.17 0.75 0.06 0.68 0.21 0.3

Anger 0.69 0.30 0.26 0.53 0.25 0.66 0.14 0.76

Sadness 0.79 0.76 0.17 1.29 0.76 0.63 0.14 0.42

Cognitive
Processes 16.37 15.54 17.39 16.23 15.29 19.66 14.42 15.97
Current Concerns

Death 0.14 0.16 0.30 0.90 0.72 0.18 0.03 0.15

Table 1, shows how all the participants used a high percentage of function words with the lowest being
56%. This corroborates with the literature which identifies how the use of function words is diagnostic
of emotional and biological states and can also reveal personality and cognitive styles. It is also of
interest that all the participants have higher percentage use of positive emotion words compared to
negative emotion words. These effects could be due to the low rate use of anger and hostility words
across all the cases analysed.

Figure 1 shows a comparison of the results from this study and the means of three settings listed by the
program’s authors- emotional writing, control writing, and blogs.

6 
 
Figure 1: Comparison with standard original values
7.00 

6.00 

5.00 

4.00  Terminally ill Patients
Emotional W
3.00 
Control W
2.00  Blogs

1.00 

‐
Affect Positive  Negative  Anxiety Anger Sadness Death
emotion emotion

Construct validity was carried out on the affect processes as they are most relevant to the emotional
components of words. Figure 1 visually demonstrates how the word usage for the cases in this study
are very much similar to those carried out on emotional writing studies and general blogs. In all three
of these texts, there is a higher percentage of words counted in affect, positive emotion and negative
emotion than that counted for the control. It is interesting to note that with the cases in this study,
lower proportion of negative emotion words were expressed compared to Pennebaker’s emotional
writing group. The control group recorded the lowest word count in all categories with more
differences to the terminally ill patients’ group than any of the other texts. The blog group’s pattern
from Pennebaker’s findings is similar to the terminally ill patients’ pattern. It is interesting to note that
the terminally ill patients group had a higher word usage of death compared to all the other groups of
text. These results support the efficacy of the LIWC program in this setting.

DISCUSSION
The results of this study confirm the possibility of using the LIWC in analysing the emotional,
cognitive and structural components present in the writings of terminally ill patients. It is a useful way
of measuring people’s linguistic characteristics. This study is preliminary work on the use of weblogs
in an end of life context, and the results encourage further development within the subject area.
Through the use of a blog as a form of disclosure channel, the terminally ill patients had a high
percentage use of affective words, both negative emotion and positive emotion. The word usage

7 
 
pattern for the terminally ill patients was very similar to the data reported by the program’s authors for
the emotional writing group and for blogs. There were differences observed however from the control
writing group and the terminally ill patients’ word usage results. These results can be interpreted in
many ways, including the suggestion that blogs can be therapeutic and, just like emotional writing,
may have positive health benefits.46-49This is evidenced by the patients’ high word use of affective
words mostly positive emotion words. The LIWC has many more word categories (up to 90) that were
not used in this study that can be encompassed in future work to reveal individual personalities and the
personalities that can benefit positively from maintaining a weblog. Previous research suggested that
linguistic characteristics can predict health outcomes; therefore the application of the LIWC program
can help to identify predictors in patients who may need specific personal attention and possible
interventions.50, 51 Advice or suggestions can be made to people to change their style of writing (e.g.
encourage the expression of positive emotions) if it helps them in a positive way. Different kinds of
information and knowledge can be gained from examining the written words of individuals including
their psychological state and this is confirmed from the results in this study.52

Limitations of this research should be noted. The small and specific sample used in this study means
care should be taken in the generalisation of the results to other populations. Individual differences
were not analysed as there were only five participants. The length of the messages varied between
participants and from time to time whilst the participants wrote about anything within their blogs
which produced a combination of emotional and control writing.53The results however seemed to
corroborate the program’ authors’ “norm” values which concludes that the tool was accurate in
identifying the linguistic dimensions present within the blog content.

CONCLUSION
The LIWC is a simple and very flexible tool which can be modified according to specific research
needs. The dictionary can be modified or a new one installed to suit the purpose of research. The
LIWC program has the potential to be used within other settings to help and give insight to the
psychological and emotional state of individuals. This information could be used within a wider
context for other clinical purposes .54The study has demonstrated that LIWC can be used to analyse
terminally ill patients’ blogs and further evaluation can be carried out according to specific research
needs. Research has suggested that cognitive words can reveal how a person thinks and, from the
cognitive processes word usage in this study, there is an indication that terminally ill patients have a
more reflective manner and approach to life. The high percentage use of cognitive process words may
be interpreted in line with previous research which suggested that during stressful or unpleasant

8 
 
situations, individuals use words that reflect cognitive processing.55This supports evidence that states
that there is higher prevalence of stress among terminally ill patients.

Further work will examine changes in word use over a period of time and their relationship to
psychosocial indicators used to predict health outcomes. More work can be conducted with a larger
sample size of terminally ill patients to establish the generalisability of these findings. Other
characteristics of participants could be profiled to explore if individual differences may have
influenced the results in this study. It may also be useful for future research to explore other categories
and aspects of word usage other than emotion words such as religion, to see if it has an influence on
how individuals understand and construct the meaning of dying and death. Exploring non-emotion
words people used whilst going through intensely emotional experiences can reveal the principal
emotion processing mechanisms. Future work could also use objective biological markers of health to
corroborate the LIWC analysis results.

9 
 
Notes
1
P Aries, The Hour of Our Death, Oxford University Press, Oxford, 1991.
2
R Blauner, ‘Death and Social Structure’, Psychiatry, vol 29, 1966, pp.378 – 394.
3
B G Glaser, and A L Strauss, Awareness of Dying, Aldine, New York, 1965.
4
J W Riley, ‘Dying and the meanings of Death: Sociological Inquiries’, Annual Review Sociology,
1983, pp.191 – 216.
5
ibid., pp. 191 – 216
6
B G Glaser, and A L Strauss, 1965, op. cit.
7
B G Glaser, and A L Strauss, 1965, op. cit.
8
K M Taylor, ‘Telling bad news: physicians and the disclosure of undesirable information’, Sociology
of Health & Illness, vol 10, 1988, pp. 109 – 133.
9
D Field and G Copp, ‘Communication and awareness about dying in the 1990s’, Palliative Medicine,
vol 13, 1999, pp. 459 – 468.
10
N J Miyaji, ‘The power of compassion: truth telling among American doctors in the care of dying
patients’, Social Science & Medicine, vol 36, 1993, pp. 249 – 264.
11
E L Emanuel, D L Fairclough, P Wolfe and L L Emanuel, ‘Talking with terminally ill patients and
their caregivers about death, dying and bereavement’, Archives of Internal Medicine, vol 164, 2004,
pp. 1999 – 20004.
12
D Field and G Copp, 1999, loc. cit.
13
A Meier, E J Lyons, G Frydman, M Forlenza and B K Rimer, ‘How cancer survivors provide support
on cancer-related internet mailing lists’, Journal of Medical Internet Research, vol 9, 2007
14
A Meier, E J Lyons, G Frydman, M Forlenza and B K Rimer, 2007, ibid.
15
D Thorne, ‘Adding technology to care – is this progress?’ Progress in Palliative Care, vol 7, 1999,
pp53 – 54.
16
J C Hays, ‘High-technology and hospice home care’, Nursing Clinics of North America, vol 23,
1988, pp. 329 – 340.
17
D M Dush, ‘High tech, aggressive palliative care: in the service of duality of life’, Journal of
Palliative Care, vol 9, 1993, pp. 37 – 41.
18
J F Seely and B M Mount, ‘Palliative medicine and modern technology’, JAMC Canadian Medical
Association Journal, vol 161, 1999, pp. 1120 – 1121.
19
L Willis, G Demiris and D Parker-Oliver, ‘Internet use by hospice families and providers: a review’,
Journal of Medical Systems, vol 31, 2007, pp. 97 – 101.

10 
 
20
K T Washington, G Demiris, D Parker- Oliver and M Day, ‘Home Internet Use among Service
Recipients: Recommendations for Web-Based Interventions’, Journal Of Medical Systems, vol 31,
2007, pp. 385-389.
21
G W Alpers, A J Winzelberg, C Classen, H Roberts, P Dev, C Koopman and C Barr-Taylor,
‘Evaluation of computerised text analysis in an Internet breast cancer support group’, Computers in
Human Behaviour, vol 21, 2005, pp. 361 – 376.
22
L Willis, G Demiris and D Parker-Oliver, 2007, loc. cit.
23
K T Washington, G Demiris, D Parker- Oliver and M Day, 2007, loc. cit.
24
M A Stefanone and C Y Jang, ‘Writing for friends and family: the interpersonal nature of blogs’,
Journal of Computer mediated Communication, vol 13, 2007, viewed on 27 July 2008,
http://jcmc.indiana.edu/vol13/issue1/stefanone.html.
25
C Hunt, Therapeutic dimensions of autobiography in creative writing, Jessica Kingsley, London,
2000.
26
G Bolton, ‘Writing or pills: therapeutic writing in primary care’, in The self on the page: Theory and
practice of creative writing in personal development, Jessica Kingsley, London, 1998.
27
J W Pennebaker, ‘Writing about emotional experiences as a therapeutic process’, Psychological
Science, vol 8, 1997, pp. 162 – 166.
28
A L Stanton, S Danoff-Burg, C L Cameron, M Bishop, C A Collins, S B Kirk and L A Sworowski,
‘Emotionally expressive coping predicts psychological and physical adjustment to breast cancer’,
Journal of Consulting and Clinical Psychology, vol 68, 2000, pp. 875 – 882.
29
N H Frijda, ‘Emotions are functional, most of the time’, in The nature of emotion: Fundamental
questions, P Ekman & R J Davidson (eds), Oxford University Press, New York, 1994, pp. 112 – 122.
30
J W Pennebaker, ‘Writing about emotional experiences as a therapeutic process’, 1997, loc. cit.
31
J M Smyth, ‘Written emotional expression: Effect size, outcome types, and moderating variables’,
Journal of Consulting and Clinical Psychology, vol 66, 1998, pp. 174 – 184.
32
J W Pennebaker and C K Chung, ‘Expressive writing, emotional upheavals, and Health’, in
Handbook of health psychology, H Friedman, & R Silver, (eds), Oxford University Press, New York,
2007, pp. 263 – 284.
33
C J Groom and J W Pennebaker, ‘Words’, Journal of Research in Personality, vol 36, 2002, pp. 615
– 621.
34
J W Pennebaker, R J Booth and M E Francis, ‘Linguistic Inquiry and Word Count: LIWC, 2007’,
viewed on 12 February 2008, www.liwc.net.
35
S Freud, Psychopathology of everyday life, Penguin, New York, 1975.

11 
 
36
P Ricoeur, Interpretation Theory: Discourse and the surplus of Meaning, Texas Christian University
Press, 1976.
37
J W Pennebaker and A Graybeal, ‘Patterns of natural language use: Disclosure, personality and
social integration’, Current Directions in Psychological Science, vol 10, 2001, pp. 91 – 93.
38
C Chung and J W Pennebaker, ‘The psychological function of function words’, in Social
Communication, Friedler (eds), Psychology Press, New York, 2007, pp. 343 – 359.
39
J W Pennebaker and A Graybeal, ‘Patterns of natural language use: Disclosure, personality and
social integration’, 2001, pp. 91 – 93.
40
J W Pennebaker, M R Mehl, and K G Niederhoffer, ‘Psychological aspects of natural language use:
Our words, our selves’, Annual Review of Psychology, vol 54, 2002, pp. 547 – 577.
41
G W Alpers, A J Winzelberg, C Classen, H Roberts, P Dev, C Koopman and C Barr-Taylor, 2005,
loc. cit.
42
J W Pennebaker, ‘Writing about emotional experiences as a therapeutic process’, 1997, loc. sit.
43
D J Schiano, B A Nardi, M Gumbrecht and L Swartz, ‘Blogging by the Rest of us’, CHI, 2004,
ACM.
44
J W Pennebaker, R J Booth and M E Francis, ‘Linguistic Inquiry and Word Count’, 2007, loc. cit.
45
J W Pennebaker, R J Booth and M E Francis, ‘Linguistic Inquiry and Word Count’, 2007, loc cit.
46
J W Pennebaker, ‘Writing about emotional experiences as a therapeutic process’, 1997, loc. cit.
47
G Bolton, Writing or pills: therapeutic writing in primary care, 1998, op. cit.
48
J M Smyth, 1998, loc .cit.
49
C Hunt, 2000, op. cit.
50
C J Groom and J W Pennebaker, ‘Words’, 2002, loc. cit.
51
J W Pennebaker, R J Booth and M E Francis, ‘Linguistic Inquiry and Word Count’, 2007, loc. cit.
52
J W Pennebaker, M R Mehl, and K G Niederhoffer, ‘Psychological aspects of natural language use:
Our words, our selves’, 2002, loc. cit.
53
J H Kahn, R M Tobin, A E Massey and J A Anderson, ‘Measuring emotional expression with the
Lingusitic Inquiry and Word Count’, American Journal of Psychology, vol 120, 2007, pp. 263 – 286.
54
G W Alpers, A J Winzelberg, C Classen, H Roberts, P Dev, C Koopman and C Barr-Taylor, 2005,
loc. cit.
55
S W Stirman and J W Pennebaker, ‘Word use in the poetry of suicidal and non suicidal poets’,
Psychosomatic Medicine, vol 63, 2001, pp. 517 – 522.

12 
 

View publication stats