This article introduces the use of semantic analysis as a technique for uncovering how the terms bullying and harassment are used. We infer the latent meanings from the titles and abstracts of all articles in the Social Science Citation Index as of September 2016 using Symphony Content Analysis Software. That software applies rigorous techniques to qualitative input to obtain relationships among words and phrases. Content analysis is an empirically based, exploratory methodology. Results suggest that when improper behavior is directed toward children, it is usually identified as bullying. When it involves sexual content, it is sexual harassment. When it is more physical, it may more likely be termed bullying. When it involves name calling or related actions, it is labeled harassment. More importantly, the results show that all these forms of improper behavior are so similar that the distinctions may really involve intentions or impacts rather than the actions, actors, or targets. Suggestions are made about further research applications to better understand terms used in describing dysfunctional behavior.
Unacceptable behavior has become an increasingly significant problem in our workplaces. The widespread acceptance and use of social media and networking provide previously inconceivable opportunities for unacceptable behaviors with their own unique social, humanitarian, and productivity issues [
Bullying has been frequently reported in schools [
Bullying and harassment are like baseballs and cricket balls―much alike but also significantly different: both spherical in shape … about 9 in circumference and 5 oz. in weight … fabricated with a cork center and wrapped in yarn or string … covered with leather strips sewn together. Although the terms bullying and harassment are widely used, the lack of precise definitions makes it difficult to distinguish between them [
The purpose of this article is to use qualitative analyses to more clearly understand the similarities and differences between these two words. In an earlier effort to develop a clear understanding of what constitutes bullying, over 450 respondents could completely agree on only one specific behavior as bullying [
Clearly, a lot of information is available dealing with the many forms of improper workplace behavior, especially bullying and harassment. However, while useful to understanding specific research questions, this results to some extent in information overload [
In such overload situations, word or tag clouds have emerged as a tool that enables a visual representation of the most prominent words in a body of literature [
In Symphony© word clouds used in this study, the most frequent word is positioned in the center, using the largest font. Then starting above the center, words are added in a clockwise fashion around the center, creating a spiral. The words are added in the descending order of frequency. As the word frequency goes down, the font size is reduced and the color is changed to accentuate the change in font size. The clouds are more qualitative than quantitative. They do not tell how many more times a word appears than another; they only indicate that a word appears more often. If the font sizes of two words are different from each other, it indicates that, given the range of word frequencies in that set of data, one word appears more often than another.
It was anticipated that clouds focused on the concepts of Bullying, Harassment, and Abusive Management would be different but with some overlap.
Area 1: Words unique to Bullying,
Area 2: Words unique to Harassment,
Area 3: Words unique to Abusive Management,
Area 4: Words common to both Bullying and Abusive Management,
Area 5: Words common to both Harassment and Abusive Management,
Area 6: Words common to both Bullying and Harassment,
Area 7: Words common to all three concepts.
Cloud ResultsIn forming the word clouds, only the 50 most frequent words were used so as to make visualization of the concept clearer (visual clarity increases as the number of displayed words decrease). Figures 2-4, then, are the word clouds for Bullying, Harassment, and Abusive Management.
social, sex, respond, gender, relate, and race; but male, female, and women are also nearby.
To better understand the relationships among these sets of words, the entire set of words in each area in
When the words that occur four or more times in the Bullying and Harassment sets are examined, the situation changes markedly. Again, referring to the areas in
In
Most Frequently Occurring Words That Mention: | |||||
---|---|---|---|---|---|
Bullying Only ( | Harassment Only ( | Overlap ( | |||
Word | Count | Word | Count | Word | Count |
Stutter | 105 | Abort | 150 | Bully | 21,753 |
Preschool | 85 | Plaintiff | 112 | Victim | 11,418 |
Kid screen | 49 | Juror | 70 | Sex | 9699 |
Twins | 42 | Dowry | 53 | Harass | 8727 |
Orphan | 40 | Syringe | 46 | Behave | 7817 |
Breakfast | 25 | Apology | 32 | Social | 5487 |
Unemotional | 23 | Bribery | 31 | Experience | 5362 |
Inhale | 20 | Post deployment | 31 | Associate | 5281 |
Eczema | 19 | Diastole | 29 | Relate | 4891 |
Gelotophobia | 19 | Eeoc | 29 | Health | 4565 |
Aboriginal | 18 | Dod | 26 | Work | 4152 |
Syndemic | 17 | Quid pro quo | 26 | Peer | 3981 |
Drunken | 16 | Woman's | 26 | Effect | 3947 |
Defiant | 14 | Token | 24 | Gender | 3834 |
Noninvolved | 14 | Hill | 23 | Women | 3722 |
Stature | 14 | Despliegue | 21 | Relationship | 3709 |
Auditory | 13 | Sanitize | 21 | Self | 3672 |
Bulimia | 13 | Cadet | 20 | Workplace | 3529 |
Chef | 13 | Dancer | 20 | Perceive | 3500 |
Deaf | 13 | Needle | 20 | Problem | 3444 |
Lunchtime | 13 | Opioid | 20 | Violence | 3365 |
Cool | 12 | Clergy | 19 | Risk | 3322 |
Dyslexia | 12 | Ada | 18 | Participate | 3158 |
Monozygotic | 12 | Cyclist | 18 | Respond | 3157 |
Amygdala | 11 | Repeal | 17 | Age | 3135 |
Dyslexic | 11 | Antigay | 15 | Level | 2997 |
Eyeglass | 11 | City's | 15 | Aggressive | 2946 |
Gelotophobes | 11 | Copulate | 15 | Support | 2838 |
Iud | 11 | Landlord | 15 | Intervene | 2648 |
Preservice | 11 | Episteme | 15 | Physical | 2526 |
The pattern for words only in the Harassment set (
It is no surprise, then, that a mixture of results is observed when examining the words appearing only in overlapping Bullying and Harassment sets (
In examining only action-oriented words, the Bullying Only set (
Noting the sizable overlap between Bullying and Harassment, a word cloud of the entire set of words was obtained. This visualization based on the frequency of words in the database clearly shows bully(ing) as the central word with harass close by. Behave, experience, victim, self, associate, and relate are also nearby as are social, sex, peer, and work. Abuse and aggressive appear but are more distant. The frequency of occurrence is not the complete story, though. Both words occur frequently in this database, but a more interesting question is how often are they used together? Understanding that involves one final step in the analysis.
BULLYING ONLY (Top counts) | |||||
---|---|---|---|---|---|
Positive | Count | Physical | Count | Verbal | Count |
Console | 4 | Cheat | 7 | Affront | 3 |
Makeup | 3 | Defiant | 14 | Curse | 4 |
Reassess | 3 | Detain | 3 | Deride | 3 |
Reorient | 3 | Glee, grin | 3 | Quarrel | 3 |
Haunt | 3 | Spiel | 4 | ||
Massacre | 6 | ||||
Rampage | 6 | ||||
Remand | 3 | ||||
Shun | 9 | ||||
Snatch | 5 | ||||
Spank | 5 | ||||
Trail | 6 |
HARASSMENT ONLY (Top counts) | |||||||
---|---|---|---|---|---|---|---|
Sex | Count | Labelling | Count | Physical | Count | Verbal | Count |
Abort | 150 | Advertise | 8 | Bash | 4 | Groan | 3 |
Copulate | 15 | Announce | 7 | Batter | 13 | Grunt | 3 |
Dates | 6 | Blackmail | 5 | Clubbing | 2 | Menace | 8 |
Desexualize | 10 | Disgrace | 3 | Fracture | 4 | Scold | 3 |
Flirt | 6 | Insinuate | 4 | Glance | 3 | Scorn | 3 |
Kiss | 4 | Promulgate | 5 | Siege | 2 | ||
Procreate | 4 | Relabeled | 3 | Confiscate | 7 | ||
Seduction | 13 | Quid pro quo | 26 | Startle | 4 | ||
Slander | 3 | Rob(bery) | 2 | ||||
Smear | 5 | Wield | 2 | ||||
Ultimatum | 9 | Shot | 3 |
In addition to developing word clouds, Symphony Content Analysis Software© can perform multidimensional scaling (MDS). MDS has been shown to be useful in exploratory data analysis [
understand through visualization the similarity or dissimilarity of individual words in a database [
MDS analysis identifies each word in a distance matrix of N-dimensional space with similar words near each other and dissimilar ones farther from each other. In our analysis, the distance between two words was determined by the percent of occurrences of one word appearing in an abstract with no match for another word. For example, if Words A and B appear in an abstract once, the distance is zero; if Word A appears but not Word B, the distance is one; or if Word A appears twice and Word B once, the distance is one. These values are added and divided by the sum of the occurrences of the two words with the resulting percent indicating the distance between the two words. The MDS analysis also took an additional step in determining the distances. In addition to how often two words appear together, it also factors in how often all other words appear with either of those words. In other words, the distance is a composite of the differences in the distances word A and word B have from all the other words as well.
The MDS analysis translates the distances into coordinates on a graph. This is done by finding coordinates that minimize how far the points assigned to the words are from their calculated distances. The more dimensions used, the closer they are to being identical and the closer the system is to a zero goodness-of-fit value known as stress. Three dimensions were used, as that is the limit of human comprehension. All the words start at a given coordinate, then one-by-one they are placed in a variety of trial coordinates to find the one that produces the lowest stress for the overall system. This is repeated for all the words over and over until a configuration of coordinates has been discovered such that if a word is moved, the overall stress will not decline.
The resulting 3-dimensional plot is shown in
portraying three dimensions in two-dimensional space). Examining the MDS plot for words in close proximity indicates that bully, harass, violence, women, and suicide are clustered, with aggressive, victim, and cyberbullying clustering nearby. Work, workplace, discriminate and race are also clustered with another nearby cluster of public, management, job, worker, employee, organization, and organizable. Nurse stands alone near bully. Bully(ing) and harass(ment) appear relatively close to one another, suggesting that they frequently occur together and share many words in common. They are separated only by the words women, violence, and suicide which suggests that those are key terms in differentiating bullying and harassment. Nearby are aggressive, victim, cyberbullying, race, discriminate, and work and workplace, suggesting that those terms are related to both concepts.
Some of these results suggest that bullying and harassment are different, yet other results point to their sameness. Satan and the Devil may be the same but people interpret and discuss them in different ways. Similarly, bullying and harassment may be interpreted and discussed in different ways. Multidimensional Scaling would show them far apart if not for the inclusion of the similarities that they have with other frequently used words. When improper behavior is directed toward children, it is usually identified as bullying. When improper behavior involves sexual content, it is sexual harassment. When improper behavior is more physical, it may more likely be termed bullying. When improper behavior involves name calling or related actions, it is labeled harassment. When improper managerial behavior involves verbal criticism and is done publicly, it is bullying [
All these forms of improper behavior, while different, are similar. Does that matter? If so, why? First, the label that is applied by one researcher, employee, company policy, etc. may not apply at all to the behavior that was given the same label by another person. Like baseball or cricket ball, the label may connote a different act from what was intended. Indeed, the distinctions may really require identifying intentions of the actor and impacts on the target rather than the actions, the actors, or the target. For example, what motivated the actor or what was the actor consciously or unconsciously trying to achieve? How does it affect the individuals involved, others, or the organization? It is important to note, for instance, that both bullying and harassment involve actions that suggest the exercising of control over another individual. So, the important question becomes not so much their common or different techniques but rather why the actor feels a need to exercise control.
Are repeated incidents a necessary aspect of either bullying or harassment? MacKinnon [
As with all studies, this one has limitations. First, our inclusive sampling method included articles in languages other than English and articles dealing almost exclusively with schools and/or children. The latter may reflect more than just improper behavior in the workplace. Second, words can have different meanings so the interpretations and classifications which were used here may not correspond to those used by others. Thus, future research is needed.
Because both bullying and harassment involve the actor exercising control over the target, future research should focus on the “why.” Why do some individuals feel the need to control others in this way? What is the intent of the bullying? What impact or result does the actor want? Assuming that sexual favors are the intent of sexual harassment, what is the intent of other forms of harassment? Understanding the intent or purpose of bullying and harassment should enable better approaches for dealing with such behaviors.
Future research also should identify the factors which lead teasing or joking to morph into bullying or harassment and when this can be sublimating or cathartic. The most salient aspects of elements of the workplace, actors, targets, and external influences that lead to improper behavior should be specified [
Questions that might be addressed in future research also could include:
1) What leads to understanding, acceptance, and change so that effective training and education can occur?
2) How can individuals or organizations intervene in specific situations to curtail negative outcomes?
3) When is such intervention necessary? The line would seem to be drawn by the actor’s impact on the target and all the ambient effects.
The fundamental conclusion of this study is that a clearer understanding of the similarities and differences between the concepts of bullying and harassment can be gained through qualitative analysis. In particular, the use of word clouds and Multidimensional Scaling to examine the large set of words used by researchers has identified many of those similarities and differences.
This analysis indicates that just like baseballs and cricket balls, bullying and harassment are similar but not identical. Where does one draw the line to separate them? Like the Sorites Paradox of trying to decide the precise moment when a tadpole becomes a frog [
Nevertheless, our word cloud formations and analyses do help us visualize more clearly the concepts of bullying and harassment as mentioned by researchers.
Bullying. Bullying includes name calling, pushing and shoving physically or verbally, and threatening. Bullying is perceived as being done by someone more powerful emotionally, physically, or organizationally. It appears that when the focus is on a specific target, the improper behavior is likely identified as bullying, especially when children are the targets. While most often associated with children, bullying also occurs among adults acting in childish behavior. In addition, bullying behavior seems to be displayed more willingly than harassment in an open or public setting.
Harassment. Harassment includes “labeling,” name calling, or threatening to label the target. Harassment also is associated with organizations and workplaces. Harassment may also be associated with violent actions including suicide. Non-sexual harassment is likely to be perceived as behavior that is repeated over some period of unstated time including stalking. The term is used in legal matters; and when sexual references are involved, it is differentiated by the label sexual harassment.
Using this or similar qualitative analyses to further reflect the meaning of these terms, current perceptions and imprecise definitions can be refined to obtain more acceptable, inclusive definitions. On the other hand, because these two concepts have so much in common, separate policy or legislative statements need not be developed. Modeling the EEOC definition of sexual harassment in the workplace (https://www.eeoc.gov/eeoc/publications/fs-sex.cfm), a more inclusive definition might be:
Unwelcome verbal or physical conduct constitutes bullying or harassment when this conduct explicitly or implicitly affects an individual's employment, unreasonably interferes with an individual's work performance, or creates an intimidating, hostile, or offensive work environment.
Using this more inclusive definition in policy statements, organizations can train personnel to recognize and deal with improper behavior, whatever the label.
Van Fleet, D., White, L. and Van Fleet, E.W. (2018) Baseballs or Cricket Balls: On the Meanings of Bullying and Harassment. Journal of Human Resource and Sustainability Studies, 6, 131-148. https://doi.org/10.4236/jhrss.2018.61032