Public Response to Suicide News Reports As Reflected in Computerized Text Analysis of Online Reader Comments

Archives of Suicide Research

ISSN: 1381-1118 (Print) 1543-6136 (Online) Journal homepage: https://www.tandfonline.com/loi/usui20

Public Response to Suicide News Reports as Reflected in Computerized Text Analysis of Online Reader Comments

G. Rosen, H. Kreiner & Y. Levi-Belz

To cite this article: G. Rosen, H. Kreiner & Y. Levi-Belz (2019): Public Response to Suicide News Reports as Reflected in Computerized Text Analysis of Online Reader Comments, Archives of Suicide Research, DOI: 10.1080/13811118.2018.1563578 To link to this article: https://doi.org/10.1080/13811118.2018.1563578

Accepted author version posted online: 13 Jan 2019. Published online: 26 Feb 2019.

Submit your article to this journal

Article views: 42

View Crossmark data

Full Terms & Conditions of access and use can be found at https://www.tandfonline.com/action/journalInformation?journalCode=usui20 Archives of Suicide Research, 0:1–17, 2019 # 2019 International Academy for Suicide Research ISSN: 1381-1118 print/1543-6136 online DOI: 10.1080/13811118.2018.1563578 Public Response to Suicide News Reports as Reflected in Computerized Text Analysis of Online Reader Comments

G. Rosen , H. Kreiner, and Y. Levi-Belz

Previous research has documented the rise in rates of suicidal behaviors following media reports of celebrity suicide. Whereas most research has focused on documenting and analyzing suicide rates, little is known about more subtle psychological effects of celebrity suicide on the public, such as despair and feelings of abandonment. The Internet has revolutionized the responses to news reports, enabling immediate and anonymous responses potentially reflecting these psychological processes. Thus, the current study explored the unique psychological impact of a celebrity suicide on the public by analyzing the big data of readers' comments to suicide news reports, using computational linguistics methods. Readers’ comments (N ¼ 14,506) to suicide news reports were retrieved from 4 leading online news sites. The comments were posted in response to 1 of 1 types of reports: a celebrity suicide (Robin Williams), a non-celebrity suicide, and general reports of suicide as a social phenomenon. LIWC software for computerized linguistic analysis was used to calculate the frequency of the various types of words used. Comparison of the responses to the 3 types of suicide reports revealed higher frequency of first-person pronouns and for emotionally charged wordsoncommentstoacelebritysuicide,comparedwithcommentsto the other types of suicide reports. The findings suggest that celebrity suicide news reports evoke the expression of positive emotions, possibly related to the venerated celebrity, alongside negative, internalized emotions, and feelings of social isolation. Theoretical, practical, and methodological implications are discussed.

Keywords big data linguistic analysis, celebrity suicide, LIWC, online reader comments, Robin Williams

1 Public Response to Suicide News Reports

INTRODUCTION According to the vertical identification theory (Stack, 1990), individuals tend to Suicide is one of the leading causes of death identify with those of higher status, such worldwide, with more than 800,000 people as celebrities. Thus, celebrities' actions dying by suicide annually (WHO [World may modulate cognitive processes and Health Organization], 2014). One of the arouse emotional responses among mem- major proximal risk factors for suicide is bers of society (see also Bandura & imitation (Mann et al., 2005;Cheng, Huston, 1961). However, the public’s Hawton, Lee, & Chen, 2007). The rise in expression of subtle social, emotional, and rates of suicidal behaviors following media cognitive psychological responses to reportsofsuicidehasbeentermedthe reports of celebrity suicide in the media is Werther effect (Phillips, 1974), alluding to unclear. The current study addressed this the high rates of suicidal attempts following issue by analyzing users' responses, as the publication of Goethe’srenowned reflected in online readers’ comments to book, The Sorrows of Young Werther. suicide news reports. Our goal was to Imitation of suicidal behavior has been cal- explore the subtle emotional effects that culated to be 14.3 times more likely to celebrity suicide may have on the public occur following reports of celebrity suicide, and its relationship to the Werther and compared with a non-celebrity suicide Papageno effects. (Stack, 2000), and Niederkrotenthaler et al. The Internet has revolutionized media (2012) found this surge to be especially consumption, opened new channels for characteristic of the month immediately fol- news reporting, and created opportunities lowing the suicide. While most of the for the public to express opinions and celebrity suicide studies have focused on the emotions instantly, through readers’ com- increasing suicide rates subsequent to the ments (Bergstr€om & Wadbring, 2015). suicide (Koburger et al., 2015; The vast volume of online comments can Niederkrotenthaler et al., 2012;Pirkis, be considered the big data of the public's Burgess, Francis, Blood, & Jolley, 2006), response to any transpiring event. These more subtle psychological processes, such as raw data waiting to be harvested offers mental pain, despair, and feelings of aban- novel opportunities for a wide spectrum of donment that may be linked to these research fields. Recent studies suggest that reports have been only rarely investigated. the immediacy and anonymity of online Recently, Niederkrotenthaler et al. reader comments facilitate attaining (2010) asserted that media stories about authentic insight into people's opinions suicide may also have a protective impact, and emotions (Back et al., 2010; Henrich conceptualized as the Papageno effect, a ref- &Holmes,2013). Hence, several studies erence to a character from Mozart’s opera, have investigated the public’s reaction to The Magic Flute (Sisask & V€arnik, 2012). suicide by analyzing readers’ comments to This effect was found in media stories online suicide news reports. Most studies with a focus on suicidal ideation unaccom- used human raters to evaluate readers’ panied by a suicide attempt or completed responses (Sisask, Mark, & V€arnik, 2012; suicide. Thus, it would be useful to also Sisask, Varnick, & Wasserman, 2005). examine the possible protective impact of While this approach may help detect spon- media reports on celebrity suicide on the taneous public cognitions and emotions general population. toward suicide, the use of subjective

2 VOLUME 0 NUMBER 0 2019 G. Rosen et al.

ratings to analyze and classify readers’ Psychological analysis of the text is based comments to specific content and emo- on interpretation of the relative counts of tional themes may constrain the objectivity various word categories. For example, a of the results. Novel algorithms based on high percentage of singular compared with computational linguistics offer a more plural first-person pronouns may be indica- objective tool for text analysis. These com- tive of social isolation and lack of social puterized tools also allow big-data analysis support (Cohn, Mehl, & Pennebaker, of online users’ behavior that may be more 2004). Based on analyses of different texts, informative, economical, and reliable. including personal diaries, college assign- ments, chat rooms, and online commun- Linguistic Analysis ities (Pennebaker & King, 1999; Nguyen et al., 2017;Stone&Pennebaker,2002), The notion of linguistic analysis as a LIWC was evaluated and further developed window on psychological processes can be to expand the number of categories and traced back to Freud's notion of the slip- attain a better understanding of the rela- of-the-tongue phenomenon (Freud, 1917). tionship between word categories and spe- This notion was later developed into for- cific psychological dimensions (for a mal projective tests (e.g., Rorschach Test), review, see Tausczik & Pennebaker, 2010). derived from the assumption that con- Most relevant to the current research scious and subconscious psychological are studies utilizing LIWC to analyze texts processes, including thoughts, intentions, written by individuals who died by suicide emotions, and motivations are projected (Barnes, Lawal-Solarin, & Lester, 2007; onto neutral, ambiguous stimuli (Mihura Egnoto & Griffin, 2016; Gunn & Lester, & Meyer, 2015; Murray, 1943) and can 2012; Handelman & Lester, 2007;Li, be interpreted by analyzing the verbal con- Chau, Yip, & Wong, 2014; Lightman, tent. Aiming to develop a more objective McCarthy, Dufty, & McNamara, 2007; way to analyze verbal content, Weintraub Stirman & Pennebaker, 2001). These (1989) used word counts and demon- studies revealed typical linguistic patterns strated the association between counts of of suicidal individuals. For example, different pronouns and depression. Stirman and Pennebaker (2001) showed Building on developments in compu- that texts of poets who died by suicide are tational linguistics, Tausczik and characterized by higher rates of first-person Pennebaker (2010) developed the singular words, such as “I” and “me,” Linguistic Inquiry and Word Count compared with texts of non-suicidal poets (LIWC), a computerized software for psy- over their careers. Interestingly, online dis- chological-linguistic analysis of large cussions of celebrity suicide cases show amounts of text. The LIWC software is some similarity to these patterns. In a based on a dictionary, in which every word study examining such discussions, Kumar, was classified by human judges into one or Dredze, Coppersmith, and De Choudhury more categories, with each category relating (2015) analyzed posts published by users to a different psychological dimension in an online suicide support group, com- (Schwartz et al., 2013). LIWC output dis- paring the responses culled during the two plays the percentage of words from each weeks prior to and the two weeks follow- category out of the total word count of the ing the suicidal act. The authors analyzed text (Tausczik & Pennebaker, 2010). approximately 66,000 posts relating to 10

ARCHIVES OF SUICIDE RESEARCH 3 Public Response to Suicide News Reports

different celebrity suicide cases and meas- person pronouns,”“emotions,”“death ured the change in posting activity and words,” and “word count.” These catego- their content. Their findings showed that ries reflect psychological dimensions that posts during the two weeks following the have been shown to be associated with sui- celebrity suicide reflected a significant cidal behavior and with reactions of indi- increase in the use of first-person singular viduals who were exposed to such pronouns, and in the total word count behavior. Specifically, the use of pronouns per post. may reflect how someone relates to others and interacts with them (Tausczik & The Current Study Pennebaker, 2010). A study investigating married couples showed that the use of The current study aimed to expand on various types of pronouns may be indica- the work of Kumar et al. (2015)intwo tive of the quality of marital interaction important aspects. First, it examined com- (Simmons, Gordon, & Chambless, 2005). ments of the general public posted in For example, higher frequency of the use popular news websites, rather than those of second-person pronouns is associated posted in a closed suicide support group. with more negativity in couple interac- Thus, this study comprises an ecological tions, and the use of first-person pronouns big-data sample of reader comments on is related to better problem-solving abilities suicide reports in online news websites. as a couple (Simmons et al., 2005). Second, to further understand the special Positive and negative emotion words characteristics of celebrity suicide, we com- categories were found to be reliable in pared readers’ comments to a celebrity sui- detecting human emotions (Holmes et al., cide report (CSR) with comments on two 2007; Kahn, Tobin, Massey, & Anderson, types of non-celebrity suicide reports 2007). For example, a study using LIWC (NSR), and general reports of suicide as a software revealed that depressed college social phenomenon (social suicide students used more negative words than reports; SSR). did non-depressed or formerly depressed For CSR, we focused on the case of students (Rude, Gortner, & Robin Williams, whose suicide was one of Pennebaker, 2004). the celebrity suicides examined in the The death words category comprises Kumar et al. (2015) study. Williams, a words relating to death, such as “coffin.” In renowned actor and comedian, who per- their published works, suicidal poets used formed in 103 movies and was awarded more death-related words, compared with numerous prizes, died by suicide at the age non-suicidal poets (Stirman & Pennebaker, of 63 on August 11, 2014. We selected 2001). Narratives containing death words Robin Williams’s case because of his indis- of female assault victims predicted worse putable status as a worldwide celebrity and post-treatment functioning and lower over- the fact that his death was a clear case of all well-being (Alvarez-Conrad, Zoellner, & suicide that was publicly known Foa, 2001). In addition, using more death and reported. words may signify distress (Pennebaker, Although LIWC software has multiple Mayne, & Francis, 1997). categories, for the current study, we ana- Concerning the word count category, lyzed readers' comments as manifested in it seems that when team members use a four major linguistic categories: “first- greater number of words, the chances of

4 VOLUME 0 NUMBER 0 2019 G. Rosen et al.

success on a group task increase (Sexton & commits suicide”), and 18 articles regard- Helmreich, 2000). The use of more words ing SSR (e.g., “Number of suicides in UK indicates increased involvement that can increases, with male rate highest lead to more success on a task. Moreover, since 2001”). total word count has been shown to be a good predictor of self-disclosure levels of Reader Comments the text’s author (Levi-Belz & Kreiner, 2016). An abundance (14,506) of spontan- In sum, the words people use may dis- eous comments were collected from the close their current mental state, feelings, three kinds of suicide news reports: (a) and cognitions. By analyzing online com- 7,501 reader comments to a celebrity sui- ments in the wake of suicide news reports, cide report (CSR); (b) 1,074 reader com- we aim to shed more light on their ments to several non-celebrity suicide authors’ emotions and cognitions concern- reports (NSR); and (c) 5,931 reader com- ing the suicide cases. Thus, in this study, ments to a social suicide report (SSR). we hypothesized that readers’ comments to the CSR will be characterized by higher levels of the linguistic categories of first- Instruments person pronouns, emotions, and death words, as well as by higher levels of total LIWC computer program, 2015 ver- word count, compared with readers’ com- sion (Pennebaker, Boyd, Jordan, & ments regarding non-celebrity suicide Blackburn, 2015) was used to calculate the reports and general reports of suicide as a percentage of each specific word type in social phenomenon. specific categories included in the full text of the comment. Eight linguistic categories were analyzed: first-person singular pro- METHOD nouns, first-person plural pronouns, negative emotion words (e.g., nasty, ugly, Suicide News Reports hurt), positive emotion words (e.g., love, nice, sweet), as well as more specific cate- Forty-nine online suicide news reports gories of anger (e.g., hate, kill), sadness were retrieved from the New York – (e.g., grief, crying), anxiety (e.g., worried, Times U.S. Edition, Yahoo! News, Google fearful), and death words (e.g., coffin, kill). News (news search engine), and The Following several studies employing Guardian, from April 2016 to July 2016. LIWC analysis (e.g., del Pilar Salas-Zarate These news sites were selected because et al., 2014; Holtgraves, 2011; Sexton & they support reader comments and are 1 Helmreich, 2000; Tausczik & Pennebaker, ranked among the top 10 news sites on 2010;), we also included each comment’s the Web. The 49 reports included 14 total word count as a ninth category. articles regarding Robin Williams’ suicide, published within two weeks of his suicide (August 11–24); 17 articles regarding NSR Procedure (e.g., “Harassed by seniors, banker The comments were extracted from the various news sites, either manually or 1Alexa.com—top sites by category: News, 2016 imported from the webpage script. From

ARCHIVES OF SUICIDE RESEARCH 5 Public Response to Suicide News Reports

the raw comments data, we retrieved the relating to the role of celebrity status when comments’ text and date (when possible) analyzing comments. The Statistical and sorted them onto a spreadsheet con- Package for the Social Sciences (SPSS, ver- taining the comment serial number, type sion 20.0 for Windows) was used for all of news report, article number, comment analyses. The level of statistical significance placement within the specific article, date, was set at p =.05. and content. The sorted comments were analyzed using LIWC software. The program scanned each comment separately RESULTS and classified each word into the relevant linguistic categories, yielding an output of Factor Analysis the percentage of words used in each selected linguistic category per each com- At first, we examine the relationships ment. Subsequently, the full data were between the study variables, using the statistically analyzed. Pearson correlation test. The results are pre- sented in Table 1. The matrix shows that most variables were significantly and weakly Data Analysis related to each other. A relatively high posi- The analysis was carried out on a data tive association was found between the set of 14,506 reader comments, such that death and anger categories. A relatively each comment represented a single case, high negative association was found one independent variable, namely the between the word count and positive emo- news report type (CSR, NSR, SSR), and tions categories, as well as between the first- the nine linguistic categories noted in the person singular and plural word categories. Instruments section. In the first step, a ser- Following that, we conducted an ies of Pearson correlation tests were con- exploratory factor analysis with Varimax ducted, followed by an exploratory factor rotation of all the reader comments, ena- analysis on the linguistic characteristics to bling us to reveal categories loading onto reveal the main linguistic categories emerg- the same factor. Our analysis revealed a ing from the data. Based on these results, four-factor structure explaining 56.7% of we calculated weighted factors to be used the variance of all variables. Table 1 shows in subsequent analyses. the variance explained by each factor and In the second step of the analysis, a the variable loadings to the factors. The generalized linear model (GLM) was per- minimal loading value to each of the fac- formed to determine group differences in tors was 0.25. the original LIWC parameters (the propor- As shown in Table 2, the LIWC cate- tion of each variable), with a Poisson gories were loaded on one factor each, log-linear function appropriate for propor- with one exception: the words charged tions. Moreover, GLM was also conducted with positive emotion category, which for the word-count category and the loaded onto Factors 1 and 2, with a low weighted factors that emerged from the negative loading (.30) on the former. factor analysis. We also compared two spe- Four specific factors were extracted: Factor cific groups’ differences: CSR vs NSR 1 comprised three LIWC categories: words comments and NSR vs SSR comments, expressing anger, words concerning death, enabling us to identify the differences and words charged with positive emotion

6 VOLUME 0 NUMBER 0 2019 G. Rosen et al.

(negatively). This factor represents the expression of negative feelings and a higher score on Factor 1 indicated a higher frequency of negative emotions. .011 1 1 Factor 2 comprised two LIWC categories: words charged with positive emotion and word count per comment, which were both negatively correlated, indicating that .04 .18

positive comments were shorter. Thus, Factor 2 represents a low engagement of readers with positive emotions. Higher scores on Factor 2 indicate a higher fre- .00 1 .02 1 quency of positive emotions as well as shorter comments. Factor 3 comprised two positively correlated LIWC categories: words expressing .03 .07 .01 .07 .08 sadness and words expressing anxiety. As these words reflect internalized feelings, we interpreted this factor as representing nega-

tive internalized emotions, as opposed to 02 01 . .02 1 .03 .00 . the negative feeling of anger and death 1 (Factor 1). Higher scores on Factor 3 indicated a higher frequency of negative internalized emotions. 10 Factor 4 includes two negatively corre- .02 . .04 .02 .05 lated LIWC categories: first-person singu- 03. lar and first-person plural pronouns. This

14,506) factor represents isolation from society, 5

N and thus, higher scores on Factor 4 indi- 10 01 cated a higher frequency of first-person .00 1 .07 .07 .02

50.8665.15 2.25 4.38singular .83 2.21 and a 4.71 lower 8.28 frequency .50 2.44 of 1.48 4.87 first-per- 1.62 4.65 1.13 3.09 son plural pronouns. According to the loading of the variables to each factor, we calculated weighted scores for each comment on the four extracted factors. We .001. used these four factors in our subsequent <

p analysis concerning the differences between

the suicide reports groups. .01, Measure 1 2 3 4 5 6 7 8 <

p Differences between the Three Groups of Reader Comments Inter-correlations among study variables ( .05, < M SD p A generalized linear model (GLM) was . carried out to examine differences between 3 First-person plural words .07 12 Word Count First-person singular4 words 5 Words charged6 with positive emotion Words expressing anxiety Words expressing anger . 1 . TABLE 1. 7 Words expressing sadness 8 Words concerning death Note the groups on the LIWC categories and on

ARCHIVES OF SUICIDE RESEARCH 7 Public Response to Suicide News Reports

TABLE 2. Results of the factor loadings based on an exploratory factor analysis with varimax rotation for all Reader Comments in Eight Categories of the Linguistic Inquiry and Word Count (N 5 14,506)

Factor 1 Factor 2 Factor 3 Factor 4 Words expressing Anger .74 2.066 2.048 2.024 Words concerning Death .704 .080 2.048 .046 Words charged with positive emotion 2.299 2.707 2.208 .040 Word count per comment 2.227 .746 2.197 .080 Words expressing Sadness 2.069 2.117 .707 2.002 Words expressing Anxiety 2.020 .102 .707 .007 First-person singular words 2.189 .090 2.075 2.746 First-person plural words 2.154 .125 2.066 .715 Eigen Value 1.25 1.16 1.08 1.03 % of Variance 15.67 14.55 13.56 12.92

the four weighted factors revealed in the Upon examining the differences in the factor analysis. for each category or factor, weighted factors, significant group effects we calculated the model effect size and sig- were found for all factors, excepting Factor nificance as specific odds ratios to the dif- 1. The specific comparisons revealed that ferences between the group. Two CSR comments were higher on Factor 3 comparisons were carried out: CSR vs and lower on Factors 2 and 4, compared NSR and NSR vs SSR. with the other groups. As shown in Table 3, the GLM analysis yielded a significant group effect of all of the linguistic variables. The specific comparisons revealed that, relative to the other groups, CSR comments were signifi- DISCUSSION cantly higher in first-person singular pronouns, words charged with positive and In the present study, we examined the negative emotion, words expressing anger effects of a celebrity suicide report on the and sadness. CSR comments were lower in public's cognition and emotional state, as words concerning death, and general word reflected in online reader comments. We count per comment. Interestingly, the used a naturalistic approach, aiming to odds ratio for sadness and anger words in obtain the public's authentic and spontan- CSR comments were around fourfold that eous reactions to news reports concerning of the other groups, emphasizing the dif- Robin Williams’ suicide, as compared with ferences in the character of these com- non-celebrity suicide reports and with gen- ments. Moreover, the odds ratio for low eral articles about suicide as a phenom- number of words in CSR comments were enon. Consistent with our hypotheses, our eightfold that of the other groups, high- findings revealed differences among com- lighting that CSR comments were signifi- menter groups in two main aspects: first- cantly shorter than the other person pronouns and emotionally groups’ comments. charged words.

8 VOLUME 0 NUMBER 0 2019 TABLE 3. Descriptive statistics, results of generalized linear models and group differences estimation of LIWC Word Categories and weighted Factors (N 5 14,506)

estimate estimate CSR Group NSR Group SSR Group 1 vs. 2 2 vs. 3 (1) (2) (3) Model Wald B B n 5 7501 n 5 1074 n 5 5931 Chi-Square Exp(B) Exp(B) First-person singular words M .73 .32 .30 693.17 .88 .064 SD .01 .02 .01 2.41 1.07 CI .70–.75 .28–.37 .28–.32 First-person plural words M .07 .12 .07 20.07 .035 .51 SD .00 .01 .00 1.03 1.67 RHVSO UCD RESEARCH SUICIDE OF ARCHIVES CI .07–.08 .10–.15 .06–.08 Words charged with positive emotion M 3.20 1.61 2.05 895.27 .45 2.24 SD .03 .06 .03 1.57 .79 CI 3.15–3.27 1.5–1.73 1.99–2.1 Words charged with negative emotion M 3.6 2.03 1.52 2132.08 .86 .29 SD .03 .07 .03 2.37 .1.33 CI 3.54–3.67 1.9–2.17 1.47–1.57 Words expressing anger M .83 .36 .21 1441.75 1.35 .51 SD .01 .02 .01 3.88 1.67 CI .81–.86 .32–.40 .20–.23 Words expressing sadness

M .76 .30 .18 1403.40 1.45 .55 al. et Rosen G. SD .01 .02 .01 4.27 1.73 CI .74–.78 .27–.35 .16–.19 Words expressing anxiety M .11 .04 .08 60.06 .35 2.75 SD .00 .01 .00 1.42 .473 CI .10–.12 .03–.05 .07–.09 9 (Continued) 0VLM 0 VOLUME 10 TABLE 3. (Continued). estimate estimate CSR Group NSR Group SSR Group 1 vs. 2 2 vs. 3 (1) (2) (3) Model Wald B B n 5 7501 n 5 1074 n 5 5931 Chi-Square Exp(B) Exp(B) Words concerning death M .26 .24 .19 49.03 .32 .237 SD .01 .02 .01 1.37 1.27 CI .24–.27 .20–.27 .73–.20 Word count per comment (WC) M 34.68a 61.67a 69.37a 5624.5 2.34.69 27.71 SD .07 1.92 .82 0.00 CI 33.26–36.11 57.9–65.43 67.77–70.98 8.57 Negative expression (Factor 1) 2 M .01 .02 .01 2.36 .02 .03 Reports News Suicide to Response Public

SD .01 .03 .01 1.02 1.03 UBR0 NUMBER CI 2.01–.03 2.04 –.08 2.04–.01 Positive expression (Factor 2) M 2.26 .22 .29 1119.76 2.54 2.07

SD .01 0.03 .01 .58 .93 2019 CI 2.28 –.24 .16–.28 .26–.31 Negative internalization (Factor 3) M .10 2.05 2.12 177.13 .23 .08 SD .01 .03 .01 1.25 1.08 CI .08–.13 2.11–.01 2.15 –.1 Isolation (Factor 4) M 2.09 0.14 0.08 119.95 2.17 .06 SD .01 .03 .01 .84 1.06 CI 2.11 –.06 .08–.20 .06–.11 Note. CI-Confidence Interval. Groups: CSR: Celebrity suicide response; NSR: Non-celebrity suicide response; SSR: Suicide as social phenomena response. p<.001; p<.01; p<.05. aSignificance values for all the values that are significant. G. Rosen et al.

First-Person Singular vs. First-Person compelling other members of the group to Plural Pronouns re-think their values. Above all, a celebrity suicide is an assault on a major component We found that readers who com- of societal integration. Thus, given the mented on a celebrity suicide report (CSR) bond between the individual and the used more first-person singular pronouns, celebrity by means of identification proc- such as “I” and “me,” and fewer first-per- esses, feelings of social isolation may ensue. son plural pronouns, such as “we,” than Do these patterns characterize did the NSR and the SSR groups. This responses to celebrity deaths of any kind, was also reflected in the corresponding dif- or are they are unique to celebrity suicides? ferences between the groups on the isola- In a study examining the language use of tion factor (Factor 4). These findings are users in AOL chat rooms following the consistent with Kumar et al.’s(2015) find- death of Princess Diana (Stone & ings regarding suicidal support groups and Pennebaker, 2002), an opposite linguistic may be further generalized to the online pattern emerged: Greater use of first-per- behavior of the general public. Together, son plural words, like “we,” and less fre- these results may suggest that isolation quent use of first-person singular words feelings appear to be universal responses, were found in the first days following not limited to suicidal support groups. Diana's fatal accident. A similar pattern Moreover, these results echo previous find- was revealed in an analysis of online users’ ings from writings of suicidal individuals journal texts after 9/11 (Cohn et al., (Barnes et al., 2007; Egnoto & Griffin, 2004). Together, these findings may be 2016; Gunn & Lester, 2012; Handelman interpreted as resulting from a greater & Lester, 2007;Lietal.,2014; Lightman sense of togetherness and social support, in et al., 2007; Stirman & Pennebaker, contrast to experiencing a sense of disinte- 2001), which were characterized by more gration and anomie in response to suicide frequent use of first-person singular words, (Durkheim, 1897/1951). Thus, our study reflecting their experienced isolation. is innovative in that its findings suggest While we anticipated isolation idea- that the general public’s response to celeb- tion among individuals preoccupied with rity suicide diverges considerably from suicide, we were surprised to find this responses to celebrity non-suicide deaths response pattern in the current big-data and, surprisingly, is comparable to the psy- sample of news-reading Internet users. chological process that suicidal individu- Thus, we suggest that there may be a als endure. unique property to a celebrity suicide that affects the public’s sense of affiliation. This Emotionally Charged Words quality may be understood as the result of vertical identification processes (Stack, Comments on news reports regarding 1990) that characterize relationships with a celebrity suicide were also characterized venerated figures. This response may be by more emotionally charged words (both associated with abandonment feelings that positive and negative), compared with the arise following the loss of a symbolic father other groups. Odds ratio levels for anger (Freud, 1917). It can be speculated, then, words and sadness words were fourfold in that when a suicide occurs, it may disrupt comparison with NSR. Another distinc- the social fabric in a brutal way, tion between CSR and the other groups

ARCHIVES OF SUICIDE RESEARCH 11 Public Response to Suicide News Reports

was manifested in general positive place for grief on the loss of their expressed emotions and the negative inter- beloved celebrity. nalized emotions. Interestingly, in general, Interestingly, it would be plausible to comments to the CSR were also signifi- suggest that reader comments, representing cantly shorter (odds ratio = 8.37) and were grieving and emotion processing “in pub- characterized by a lower frequency of lic,” can also serve as a facilitator of death-related words. belongingness (for the authors of the com- These findings are generally consistent ments as well as for their readers) and with those of Kumar et al. (2015), who thus, can be seen at least partly as a buffer showed that users’ responses to celebrity against suicide. In other words, these com- suicide reports in an online suicide support ments may also generate a Papageno effect group were associated with more negative (Niederkrotenthaler et al., 2010), by which feelings, such as anger, sadness, and anx- comments regarding celebrity suicide that iety (see also Sisask et al., 2012), compared share positive coping and emotions may with support group comments prior to the have a protective effect on suicide rates. suicide event. These differences highlight Future studies should examine this the effect of celebrity suicide on the gen- assumption empirically. eral public, revealing stronger emotional responses to CSR than to NSR or to SSR. Limitations and Future Research The higher incidence of negative emotion words for CSR may be interpreted as Several important limitations of the reflecting parasocial interaction. Parasocial current study warrant mentioning. One interaction has been defined as a special limitation concerns our ecological sample. feeling of bonding of a common person Reader comments allow for anonymity, toward a celebrity, characterized by emo- thus limiting the information we can tions of intimacy, closeness, and admir- acquire on participants. Consequently, no ation (Horton & Wohl, 1956). Studies data were available to detect any demo- have shown that parasocial loss evokes feel- graphic or personal bias regarding those ings of grief, indicating that people who choosing to comment. For example, could were more parasocially bonded to Robin individuals choosing to comment on a sui- Williams expressed a greater amount of cide news report have a preoccupation grief following his suicide than did those with the subject of suicide? Future studies feeling less bonded to him (Cohen & may select specifically defined study Hoffner, 2016). Online platforms help groups, such as suicide support groups or users link together with others who share celebrity fan groups, thus allowing their affection for a celebrity (Black, 2009; researchers to characterize the study partic- Williams, 2007). In addition, the online ipants more broadly. Using the compari- environment has emerged as a place for son groups in our study allowed us to mourning (Feigelman, Gorman, Beal, & differentiate between the effects generated Jordan, 2008). Thus, users can utilize by a celebrity suicide compared with a online platforms for parasocial grieving, non-celebrity suicide. However, in order mourning for the loss of the parasocial to differentiate between the relative interaction (Sanderson & Hope Cheong, impacts of a regular vs. suicide death of a 2010). Our findings suggest that the pub- celebrity, we incorporated findings of pre- lic uses the online platform as a legitimate vious studies of text analysis in our

12 VOLUME 0 NUMBER 0 2019 G. Rosen et al.

interpretations. Supplementing a non-sui- and its holistic aspects. This would facili- cide celebrity comparison group in future tate an enhancement of our understanding studies would facilitate a better under- of the big data of human behavior that is standing of the unique properties elicited abundant on the net. by an act of suicide, while controlling the celebrity status factor. Implications Moreover, this study investigates read- ’ ers comments, so the population exam- The current study promotes our ined is one who has chosen to express understanding of the effects of celebrity themselves through this medium. Thus, suicide on the public. Theoretically, the the representativeness of our sample is in findings indicate that a celebrity suicide “ question, due to the fact that talk-back- event manifests itself in further subtle con- ” ers to news articles may be a distinctive sequences on the public. The findings group and not reflective of the general show that a celebrity suicide may have a population. Thus, generalizing our find- distinctive psychological impact on the ings to the general population needs to be public not found in non-celebrity suicide approached cautiously. cases. Thus, our study may serve to bolster Another limitation concerns the a new research approach for exploration of method of analysis used by LIWC the various implications of celebrity suicide –– software counting and comparing words as well as for identifying specific factors to those of the internal dictionary. that facilitate the various effects of a celeb- Computerized linguistic analysis using rity suicide event, such as the contribution LIWC has proven reliable in capturing the of the celebrity’s personal characteristics psychological dimensions of words (e.g., personality traits). (Tausczik & Pennebaker, 2010)and,in More practically, our findings suggest the current study, an efficient tool for ana- the possibility of elevated distress and iso- lyzing big data. The program examines lation feelings among individuals from the word use, but lacks the capacity to capture general population following a celebrity the context in which it was written suicide. These results may clarify the need (Crespo & Fernandez-Lansac, 2016; Hirsh for proactive assessment and detection & Peterson, 2009). Thus, comparing the processes for the general population that words to the predefined internal dictionary may help identify groups or individuals at may miss important linguistic features, higher risk for emotional crisis after a such as disambiguation, irony, sarcasm, celebrity's suicide. Increased awareness of and idioms. For example, a popular phrase psychological dynamics following a celeb- like “Oh my God!” would be assigned to rity suicide could enlighten crisis hotline the religion category, disregarding its emo- volunteers, as well as community leaders tional idiomatic connotation. These limita- and in formal and informal educa- tions notwithstanding, our findings are tional settings. consistent with previous studies in the field Moreover, as online platforms have of suicide, claiming that potential errors of emerged as legitimate spaces for expressing classification are only minor hindrances. emotions, cognitions, and attitudes, reader Further efforts should be made to expand comments carry valuable information automated linguistic analysis features, ena- about the writer’s mental state. Thus, bling a fuller understanding of the text using automated computerized linguistic

ARCHIVES OF SUICIDE RESEARCH 13 Public Response to Suicide News Reports

analysis on social media such as Facebook AUTHOR NOTE or online forums may help identify online users who are in distress and are candidates G. Rosen, Clinical Psychology for professional help. However, the fact Department, Ruppin Academic Center, that celebrity suicide stories facilitated Emek Hefer, Israel. more emotionally charged words than did H. Kreiner, Behavioral Sciences non-celebrity suicide stories may also sug- Department, Ruppin Academic Center, gest that online readers tend to disclose Emek Hefer, Israel. more feelings when encountering celebrity Y. Levi-Belz, The Lior Tsfaty Center suicides, an inclination that may facilitate for Suicide and Mental Pain Studies, personal and interpersonal emotional proc- Emek Hefer, Israel. esses (Levi-Belz, 2016; Levi-Belz et al., Correspondence concerning this article 2014); this phenomenon, then, can be should be addressed to Y. Levi-Belz seen as protective behavior in the spirit of Department of Behavioral Sciences, the Papageno effect (Niederkrotenthaler Ruppin Academic Center, Emek Hefer, 40250, Israel. Email: [email protected] et al., 2010). Thus, another implication may be the importance of encouraging ’ readers comments on suicide stories, ORCID which may act as a buffer against actual suicides. G. Rosen http://orcid.org/0000-0002- An additional implication relating to 6712-1304 the methodology of the current study is Y. Levi-Belz http://orcid.org/0000- ’ linking readers textual comments to com- 0002-8865-5639 puterized linguistic analysis, providing a novel and appealing framework for analyzing not only the public response to suicide, REFERENCES but also to other types of events. Readers’ comments are spontaneous and authentic Alvarez-Conrad, J., Zoellner, L. A., & Foa, E. B. reactions to events. They allow users to (2001). Linguistic predictors of trauma path- express their ideas and emotions anonym- ology and physical health. Applied Cognitive ously, leaving their psychological defenses Psychology, 15(7), S159–S170. doi:10.1002/ aside. Consequently, analyzing such texts acp.839 could reveal an authentic reflection of Back, M. D., Stopfer, J. M., Vazire, S., Gaddis, S., users’ thoughts and feelings. Analyzing Schmukle, S. C., Egloff, B., & Gosling, S. D. (2010). Facebook profiles reflect actual personal- them using a computerized, automatic tool ity, not self-idealization. Psychological Science, enables us to harvest the big data of reader 21(3), 372–374. doi:10.1177/0956797609 comments and reflect the public atmos- 360756 phere, particularly the impact of an event Bandura, A., & Huston, A. C. (1961). beyond the individual, at a societal level. Identification as a process of incidental learning. The Journal of Abnormal and Social Psychology, Thus, using this methodological – framework on users’ online behavior could 63(2), 311 318. doi:10.1037/h0040351 Barnes, D. H., Lawal-Solarin, F. W., & Lester, D. provide a meaningful portrayal of the pub- (2007). Letters from a suicide. Death Studies, lic reaction to almost any momen- 31(7), 671–678. doi:10.1080/07481180701 tous event. 405212

14 VOLUME 0 NUMBER 0 2019 G. Rosen et al.

Bergstr€om, A., & Wadbring, I. (2015). Beneficial Death and Dying, 57(3), 217–243. doi:10.2190/ yet crappy: Journalists and audiences on OM.57.3.a obstacles and opportunities in reader comments. Freud, S. (1917). Mourning and melancholia. In European Journal of Communication, 30(2), the standard edition of the Complete psychological 137–151. doi:10.1177/0267323114559378 works of Sigmund Freud, Vol. XIV (1914-1916): Black, R. W. (2009). English-language learners, fan On the history of the psycho-analytic move- communities, and 21st-century skills. Journal of ment, papers on metapsychology and other Adolescent and Adult Literacy, 52(8), 688–697. works (pp. 237–258). London, UK: Hogarth doi:10.1598/JAAL.52.8.4 Press and Institute of Psycho-Analysis. Cheng, A. T., Hawton, K., Lee, C. T., & Chen, Gunn, J. F., & Lester, D. (2012). Twitter postings T. H. (2007). The influence of media reporting and suicide: An analysis of the postings of a fatal of the suicide of a celebrity on suicide rates: A suicide in the 24 hours prior to death. population-based study. International Journal of Suicidologi, 17(3), 28–30. Epidemiology, 36(6), 1229–1234. doi:10.1093/ Handelman, L. D., & Lester, D. (2007). The con- ije/dym196 tent of suicide notes from attempters and com- Cohen, E. L., & Hoffner, C. (2016). Finding pleters. Crisis, 28(2), 102–104. doi:10.1027/ meaning in a celebrity’s death: The relationship 0227-5910.28.2.102 between parasocial attachment, grief, and shar- Henrich, N., & Holmes, B. (2013). Web news ing educational health information related to readers' comments: Towards developing a meth- Robin Williams on social network sites. odology for using on-line comments in social Computers in Human Behavior, 65, 643–650. inquiry. Journal of Media and Communication doi:10.1016/j.chb.2016.06.042 Studies, 5,1–4. doi:10.5897/JMCS11.103 Cohn, M. A., Mehl, M. R., & Pennebaker, J. W. Hirsh, J. B., & Peterson, J. B. (2009). Personality (2004). Linguistic markers of psychological and language use in self-narratives. Journal of change surrounding September 11, 2001. Research in Personality, 43(3), 524–527. doi: Psychological Science, 15(10), 687–693. doi: 10.1016/j.jrp.2009.01.006 10.1111/j.0956-7976.2004.00741.x Holmes, D., Alpers, G. W., Ismailji, T., Classen, Crespo, M., & Fernandez-Lansac, V. (2016). C., Wales, T., Cheasty, V., … Koopman, C. Memory and narrative of traumatic events: A lit- (2007). Cognitive and emotional processing in erature review. Psychological Trauma: Theory, narratives of women abused by intimate part- Research, Practice, and Policy, 8(2), 149–156. ners. Violence against Women, 13(11), doi:10.1037/tra0000041 1192–1205. doi:10.1177/1077801207307801 del Pilar Salas-Zarate, M., Lopez-Lopez, E., Holtgraves, T. (2011). Text messaging, personality, Valencia-Garcıa, R., Aussenac-Gilles, N., and the social context. Journal of Research in Almela, A., & Alor-Hernandez, G. (2014). A Personality, 45(1), 92–99. doi:10.1016/ study on LIWC categories for opinion mining j.jrp.2010.11.015 in Spanish reviews. Journal of Information Horton, D., & Wohl, R. (1956). Mass communi- Science, 40(6), 749–760. doi:10.1177/ cation and para-social interaction: Observations 0165551514547842 on intimacy at. A Distance. Psychiatry, 19(3), Durkheim, E. (1897/1951). Suicide: A study in soci- 215–229. doi:10.1080/00332747.1956.1102 ology. J. A. Spaulding & G. Simpson, trans. 3049 Glencoe, IL: Free Press. Kahn, J. H., Tobin, R. M., Massey, A. E., & Egnoto, M. J., & Griffin, D. J. (2016). Analyzing Anderson, J. A. (2007). Measuring emotional language in suicide notes and legacy tokens. expression with the Linguistic Inquiry and Word Crisis, 37(2), 140–147. doi:10.1027/0227-5910/ Count. The American Journal of Psychology, 120, a000363 263–286. doi:10.2307/20445398 Feigelman, W., Gorman, B. S., Beal, K. C., & Koburger, N., Mergl, R., Rummel-Kluge, C., Jordan, J. R. (2008). Internet support groups for Ibelsh€auser, A., Meise, U., Postuvan, V., … suicide survivors: A new mode for gaining Hegerl, U. (2015). Celebrity suicide on the rail- bereavement assistance. Omega - Journal of way network: Can one case trigger international

ARCHIVES OF SUICIDE RESEARCH 15 Public Response to Suicide News Reports

effects? Journal of Affective Disorders, 185, Multimedia Tools and Applications, 76(8), 38–46. doi:10.1016/j.jad.2015.06.037 10653–10676. doi:10.1007/s11042-015-3128-x Kumar, M., Dredze, M., Coppersmith, G., & De Niederkrotenthaler, T., Fu, K. W., Yip, P. S., Choudhury, M. (2015). Detecting changes in Fong, D. Y., Stack, S., Cheng, Q., & Pirkis, J. suicide content manifested in social media fol- (2012). Changes in suicide rates following media lowing celebrity suicides. In Proceedings of the reports on celebrity suicide: A meta-analysis. 26th ACM Conference on Hypertext & Social Journal of Epidemiology and Community Health, Media (pp. 85–94). ACM. 66(11), 1037–1042. doi:10.1136/jech-2011- Levi-Belz, Y. (2016). To share or not to share? The 200707 contribution of self-disclosure to stress-related Niederkrotenthaler, T., Voracek, M., Herberth, A., growth among suicide survivors. Death Studies, Till, B., Strauss, M., Etzersdorfer, E., … 40(7), 405–413. doi:10.1080/07481187.2016. Sonneck, G. (2010). Role of media reports in 1160164 completed and prevented suicide: Werther v. Levi-Belz, Y., & Kreiner, H. (2016). What you say Papageno effects. The British Journal of and how you say it: Analysis of speech content Psychiatry, 197(3), 234–243. and speech fluency as predictors of judged self- Pennebaker, J. W., & King, L. A. (1999). disclosure. Social Psychological and Personality Linguistic styles: Language use as an individual Science, 7(3), 232–239. doi:10.1177/ difference. Journal of Personality and Social 1948550616632575 Psychology, 77(6), 1296–1312. doi:10.1037// Levi-Belz, Y., Gvion, Y., Horesh, N., Fischel, T., 0022-3514.77.6.1296 Treves, I., Or, E., … Apter, A. (2014). Mental Pennebaker, J. W., Boyd, R. L., Jordan, K., & pain, communication difficulties, and medically Blackburn, K. (2015). The development and serious suicide attempts: A case-control study. psychometric properties of LIWC2015. Austin, Archives of Suicide Research, 18(1), 74–87. doi: TX: University of Texas at Austin. 10.1080/13811118.2013.809041 Pennebaker, J. W., Mayne, T. J., & Francis, M. E. Li, T. M. H., Chau, M., Yip, P. S. F., & Wong, (1997). Linguistic predictors of adaptive bereave- P. W. C. (2014). Temporal and computerized ment. Journal of Personality and Social Psychology, psycholinguistic analysis of the blog of a Chinese 72(4), 863–871. doi:10.1037//0022-3514.72. adolescent suicide. Crisis, 35(3), 168–175. doi: 4.863 10.1027/0227-5910/a000248 Phillips, D. P. (1974). The influence of suggestion Lightman, E. J., McCarthy, P. M., Dufty, D. F., on suicide: Substantive and theoretical implica- & McNamara, D. S. (2007). Using computa- tions of the Werther effect. American Sociological tional text analysis tools to compare the lyrics of Review, 39(3), 340–354. doi:10.2307/2094294 suicidal and non-suicidal songwriters. In Pirkis, J. E., Burgess, P. M., Francis, C., Blood, Proceedings of the 29th Annual Meeting of the R. W., & Jolley, D. J. (2006). The relationship Cognitive Science Society (pp. 1217–1222): between media reporting of suicide and actual Citeseer. suicide in Australia. Social Science and Medicine, Mann, J. J., Apter, A., Bertolote, J., Beautrais, A., 62, 2874–2886. doi:10.1016/j.socscimed. Currier, D., Haas, A., … Marusic, A. (2005). 2005.11.033 Suicide prevention strategies. A Systematic Rude, S., Gortner, E. M., & Pennebaker, J. Review. JAMA, 294, 2064–2074. doi:10.1001/ (2004). Language use of depressed and depres- jama.294.16.2064 sion-vulnerable college students. Cognition and Mihura, J. L., & Meyer, G. J. (2015). Rorschach Emotion, 18, 1121–1133. doi:10.1080/ inkblot test. Hoboken, NJ: John Wiley & Sons. 02699930441000030 Murray, H. A. (1943). Thematic Apperception Test Sanderson, J., & Hope Cheong, P. (2010). manual. Cambridge, MA: Harvard University Tweeting prayers and communicating grief over Press. Michael Jackson online. Bulletin of Science, Nguyen, T., O’Dea, B., Larsen, M., Phung, D., Technology and Society, 30(5), 328–340. doi: Venkatesh, S., & Christensen, H. (2017). Using 10.1177/0270467610380010 linguistic and topic analysis to classify sub- Schwartz, H. A., Eichstaedt, J. C., Kern, M. L., groups of online depression communities. Dziurzynski, L., Ramones, S. M., Agrawal, M.,

16 VOLUME 0 NUMBER 0 2019 G. Rosen et al.

… Ungar, L. H. (2013). Personality, gender, Stack, S. (1990). A reanalysis of the impact of non and age in the language of social media: The celebrity suicides. Social Psychiatry and open-vocabulary approach. PLoS One, 8(9), Psychiatric Epidemiology, 25(5), 269–273. doi: e73791. doi:10.1371/journal.pone.0073791 10.1007/BF00788648 Sexton, J. B., & Helmreich, R. L. (2000). Stack, S. (2000). Media impacts on suicide: A Analyzing cockpit communications: The links quantitative review of 293 findings. Social – between language, performance, error, and Science Quarterly, 81, 957 971. workload. Human Performance in Extreme Stirman, S. W., & Pennebaker, J. W. (2001). Environments, 5(1), 63–68. doi:10.7771/2327- Word use in the poetry of suicidal and nonsuici- – 2937.1007 dal poets. Psychosomatic Medicine, 63, 517 522. Simmons, R. A., Gordon, P. C., & Chambless, doi:10.1097/00006842-200107000-00001 D. L. (2005). Pronouns in marital interaction: Stone, L. D., & Pennebaker, J. W. (2002). Trauma in real time: Talking and avoiding online conver- What do “you” and “I” say about marital health? sations about the death of Princess Diana. Basic Psychological Science, 16(12), 932–936. doi: and Applied Social Psychology, 24(3), 173–183. 10.1111/j.1467-9280.2005.01639.x doi:10.1207/153248302760179101 Sisask, M., & V€arnik, A. (2012). Media roles in Tausczik, Y. R., & Pennebaker, J. W. (2010). The suicide prevention: A systematic review. psychological meaning of words: LIWC and International Journal of Environmental Research computerized text analysis methods. Journal of – and Public Health, 9(1), 123 138. doi:10.3390/ Language and Social Psychology, 29(1), 24–54. ijerph9010123 doi:10.1177/0261927X09351676 € Sisask, M., Mark, L., & Varnik, A. (2012). Weintraub, W. (1989). Verbal behavior in everyday Internet comments elicited by media portrayal life. New York, NY: Springer Publishing Co. of a familicide-suicide case. Crisis, 33(4), WHO (World Health Organization). (2014). 222–229. doi:10.1027/0227-5910/a000134 Suicide fact sheet N 398. Retrieved from http:// Sisask, M., V€arnik, A., & Wasserman, D. (2005). www.who.int/mediacentre/factsheets/fs398/en/ Internet comments on media reporting of two Williams, M. (2007). The cult of the Mohicans: adolescents' collective suicide attempt. Archives American fans on the electronic frontier. The of Suicide Research, 9(1), 87–98. doi:10.1080/ Journal of Popular Culture, 40(3), 526–554. doi: 13811110590512985 10.1111/j.1540-5931.2007.00407.x

ARCHIVES OF SUICIDE RESEARCH 17