Professional Documents
Culture Documents
with the collaboration of Ana Almeida, Hozumi Araki, Cristina del Barrio, Angela Costabile,
Bojan Dekleva, Anastasia Houndoumadi, Kenneth Kim, Ragnar P. Olafsson,
Rosario Ortega, Jacques Pain, Lena Pateraki, Mechthild Schafer, Monika Singer,
Andrea Smorti, Yuichi Toda, Helgi Tomasson, and Zhang Wenxin
The study of school bullying has recently assumed an international dimension, but is faced with difficulties in
finding terms in different languages to correspond to the English word bullying. To investigate the meanings
given to various terms, a set of 25 stick-figure cartoons was devised, covering a range of social situations be-
tween peers. These cartoons were shown to samples of 8- and 14-year-old pupils (N 1,245; n 604 at 8 years,
n 641 at 14 years) in schools in 14 different countries, who judged whether various native terms cognate to
bullying, applied to them. Terms from 10 Indo-European languages and three Asian languages were sampled.
Multidimensional scaling showed that 8-year-olds primarily discriminated nonaggressive and aggressive car-
toon situations; however, 14-year-olds discriminated fighting from physical bullying, and also discriminated
verbal bullying and social exclusion. Gender differences were less appreciable than age differences. Based on
the 14-year-old data, profiles of 67 words were then constructed across the five major cartoon clusters. The
main types of terms used fell into six groups: bullying (of all kinds), verbal plus physical bullying, solely ver-
bal bullying, social exclusion, solely physical aggression, and mainly physical aggression. The findings are dis-
cussed in relation to developmental trends in how children understand bullying, the inferences that can be
made from cross-national studies, and the design of such studies.
aggression as a category was shown by the research inside a room, sent nasty notes, when no-one ever
of Björkqvist and colleagues (Björkqvist, Lagerspetz, talks to them and things like that. These things can
& Kaukiainen, 1992) in Finland. They distinguished happen frequently and it is difficult for the young
direct physical aggression (such as punching) and direct person being bullied to defend himself or herself. It
verbal aggression (such as name calling), as well as is also bullying when a young person is teased re-
indirect aggression. Indirect aggression, characterized peatedly in a nasty way. But it is not bullying when
by its somewhat covert nature and use of third par- two young people of about the same strength have
ties, had principal forms of gossiping and spreading the odd fight or quarrel.
rumors, and social exclusion (deliberately not allow-
This definition specifies that bullying is an aggres-
ing a person into a group). Somewhat similar (al-
sive act. In addition, it suggests an imbalance of
though not identical) concepts have been described
power (the victim finds it difficult to defend himself
by Crick and colleagues (Crick, Casas, & Mosher, 1997;
or herself) and some element of repetition (these
Crick & Grotpeter, 1995, 1996) as relational aggres-
things can happen frequently). These two additional
sion, and by Galen and Underwood (1997) as social
criteria serve to distinguish bullying as a subset of the
aggression; these relate more to the consequences
broader concept of aggression. Notably, two sorts of
of the negative act and the intent to damage relation-
actions are labeled as not being bullying: the odd fight
ships. There are corresponding forms of physical,
or quarrel between two young people of about the
verbal, and indirect or relational bullying (Rivers &
same strength, which is aggression, is explicitly not
Smith, 1994).
bullying because it fails the above two additional cri-
Currently, Olweus (1999, pp. 10–11) states that
teria. The definition also implicitly labels friendly
“bullying is thus characterized by the following three
forms of teasing as not being bullying; teasing is a
criteria: (1) it is aggressive behavior or intentional
rather ambiguous behavior, but in friendly teasing
‘harmdoing’ (2) which is carried out repeatedly and
there would not be an intent to harm, whereas in
over time (3) in an interpersonal relationship charac-
nasty teasing there would be.
terized by an imbalance of power. One might add that
There is now widespread use of the Olweus ques-
the bullying behavior often occurs without apparent
tionnaire, and similar survey instruments, on an in-
provocation,” and “negative actions can be carried
ternational basis (Morita, Smith, Junger-Tas, Olweus,
out by physical contact, by words, or in other ways,
& Catalano, 1999; Smith et al., 1999). A modified ver-
such as making faces or mean gestures, and inten-
sion has been used in a cross-national comparative
tional exclusion from a group.” This kind of defini-
survey of five countries (Japan, England, The Nether-
tion of the term bullying is now accepted by many re-
lands, Norway, and the United States) funded by the
searchers (Farrington, 1993; Smith & Sharp, 1994).
United Nations Educational, Scientific and Cultural
Such a definition labels bullying as a subset of ag-
Organization (UNESCO) and the Japanese Ministry
gressive behavior. Aggressive behavior, in turn, is
of Education, Science, Sports, and Culture (Mon-
often defined as negative acts carried out intention-
busho; Junger-Tas & van Kesteren, 1999; Morita, 1999;
ally to harm another; thus, behavior that accidentally
Smith & Shu, 2000). One aim of such international co-
harms another would not be aggressive. The issue of
operation is to compare the frequency and structural
whether the aggressive behavior is provoked, or in
characteristics of bullying in different societies, to bet-
some sense justified, is less clearly resolved. Olweus
ter understand and then plan interventions to reduce
stated that “the bullying behavior often occurs with-
the harmful consequences that bullying can have.
out apparent provocation,” but some researchers (no-
tably Pikas, 1989) have suggested that there is a cate-
gory of “provocative victims” who play an active part
The Implications of Cross-National Research
in “inviting” or “perpetuating” the bullying behavior.
on Bullying
An earlier form of the Olweus definition (1993), ex-
tended slightly by Whitney and Smith (1993, p. 7) in A difficulty facing cross-national comparisons is
their English language version, prefaces the adminis- the comparability of terminology. Arora (1996) drew
tration of recent versions of his questionnaire: attention to the various terms cognate with bullying
in English and in several other languages. The study
We say a young person is being bullied, or picked of bullying in a number of different countries indi-
on, when another child or young person, or a cates that the word bully is not easy to translate
group of young people, say nasty and unpleasant (Genta, Menesini, Fonzi, Costabile, & Smith, 1996;
things to him or her. It is also bullying when a Morita, 1996; O’Connell, Pepler, & Craig, 1999; Rigby
young person is hit, kicked or threatened, locked & Slee, 1991). Different terms are used both in any one
Smith et al. 1121
language and in different languages. The terms bully difference is primarily located. Even if a fuller explicit
and mobbing are familiar in the Scandinavian and definition is given (as in the Olweus questionnaire),
Germanic languages, including English (the etymol- the favored term (bullying, ijime, prepotenza, and so
ogy of the word bully probably coming from the Mid- forth) is used throughout the questionnaire, and in
dle Dutch word boele meaning first sweetheart, then summaries and wider discussion; and respondents to
fine fellow, then blusterer; Encarta World English Dic- the questionnaire may well refer to their personal def-
tionary, 1999). The word bully is familiar in other inition of this term rather than that given early on by
English-speaking cultures such as the United States, the researcher (Arora, 1996; Madsen, 1997).
although less widely used (Hazler, 1996); in the In the present study’s comparison of the meaning
United States the terms victimization and peer rejec- of different terms, the intention was not to “privilege”
tion are often used to denote negative actions of peers any particular term or language, but rather to show
toward another young person (Asher & Coie, 1990). the kinds of situational meanings attributed to each
Words for bullying are less familiar in the Latin lan- term. However, given a current strong position of En-
guages; for example, the French have no direct trans- glish language in scientific discourse, and the wide-
lation of the word. spread use of the term bullying (including its co-
The importance of the terms used can be illustrated option as, e.g., bullismo in Italian; Fonzi, 1997), there
within one language, English. The terms bully and may be particular interest in comparing other lan-
bullying have been in usage for a long time, to de- guage terms for their similarity/dissimilarity to the
scribe this kind of behavior in children at school (as in English term bullying.
Tom Brown’s Schooldays; Hughes, 1857/1989). The Such information is also important for the study of
term teasing is similar but can have a milder connota- developmental changes in perceptions of bullying.
tion of verbal and possibly playful aggression. An- There is evidence for developmental differences in
other term, harassment, appears similar to bullying the ways in which pupils construe bullying (Madsen,
but tends to be used for adult or adolescent rather 1997; Smith & Levan, 1995; Smith, Madsen, & Moody,
than child behaviors, as in sexual harassment, or ra- 1999). Younger pupils may not distinguish between
cial harassment. Yet another term, abuse, also appears bullying and fighting, broadening the use of the term
similar but tends to be restricted to the family context, bullying to cover nasty kinds of behavior even when
as in parent–child abuse or spousal abuse, or to adult– no imbalance of power is involved. Younger, Schwartz-
child contexts, as in physical abuse or sexual abuse. man, and Ledingham (1985) compared children’s per-
These terms have somewhat different dictionary def- ceptions of aggression, withdrawal, and likability
initions and, more importantly, they may be under- items in Canadian children approximately 6, 9, and 12
stood differently by persons answering questionnaires. years of age. Using multidimensional scaling (MDS)
This issue is also highlighted when questionnaires techniques, they found a shift with age from a simple
such as the Olweus questionnaire are translated into evaluative dimension and low cohesiveness of clus-
different languages. Frequently there is not an exact ters (high structure ratios in the MDS) to a more com-
translation of the term bullying. The Japanese term plex separation of clusters with an additional active–
most equivalent to bullying appears to be ijime, which passive dimension, and higher within-cluster item
has been the subject of a rather independent research cohesiveness (lower structure ratios).
tradition in Japan, through the 1980s and 1990s It is also possible that there are gender differences
(Morita, Soeda, Soeda, & Taki, 1999). However, Morita, in perceptions of bullying and related terms. This is
Soeda, et al. (1999) consider that ijime does differ an issue, given the well-established gender differ-
somewhat from bullying, in having a less physically ences found in the use of direct and indirect or phys-
violent connotation, and a relatively greater emphasis ical and psychological forms of aggression, including
on social manipulation and more female types of ag- bullying (Björkqvist et al., 1992; Crick & Grotpeter,
gressive behavior. In contrast, in Italy, the Italian 1995, 1996; Rivers & Smith, 1994). Although little evi-
words prepotenza and violenza tend to imply more dence has been found thus far for gender differences
physical, violent actions (Fonzi et al., 1999). in the understanding of aggression and bullying
The issue of comparability of terms is central for the (Madsen, 1997), as opposed to their use, the possibil-
accurate interpretation of national and cross-national ity of them emerging should be investigated.
findings. Exact matching of terms across languages is
an unrealizable ideal, but it is necessary to know how
Aims of the Present Study
comparable terms are, and, if they differ, on which
dimensions or criteria (e.g., physical/psychological, This study examined the understanding of a variety
direct/indirect, group/individual, and so forth) the of terms cognate to bullying, in 14 countries, includ-
1122 Child Development
ing 13 major languages. Understanding was opera- Table 1 Captions for the 25 Cartoons (Male Version)
tionalized in terms of applicability of a selection of 25
1. Mike and John don’t like each other and start to fight
stick-figure cartoons that displayed situations that 2. Bill starts a fight with Joey
might or might not be bullying, based on various 3. Martin starts to fight with Akhtar, who is smaller
well-used criteria. Age differences were examined by 4. Sean starts a fight with Ron because he said Sean was stupid
giving the task to primary school children (8 years of 5. Chris starts a fight with Damien every break time
6. David tells Scott that if he doesn’t give him money, he will hit
age) and secondary school children (14 years of age);
him
samples were balanced for gender to also allow for 7. Nick and his friends start to fight Terry
male–female comparisons. 8. Nigel borrows Duncan’s ruler and accidentally breaks it
9. Harry takes Ian’s ruler and breaks it
10. Jim forgot his pen so Kirk lends him one of his
11. Kurt says nasty things to Ben
METHOD
12. Charles says nasty things to Marcus every week
The Cartoons Task 13. Stuart says nasty things to Jeff about the color of his skin
(alternate caption if color of skin is not an important factor
We first developed a series of 25 stick-figure car- in the culture: Stuart says nasty things to Jeff about his
toon pictures (see Table 1) that illustrated different sit- talking in a different way)
14. Joshua has a bad leg and must use a stick, Carl says nasty
uations that might or might not be bullying, based on
things to him about it
elements used in existing definitions of bullying 15. George says nasty things to Derek about his sexual orientation
(Smith, 1999). Most of the cartoons portrayed nega- 16. Ken makes fun of Graham’s hair, they both laugh
tive acts; however, two prosocial cartoons (10, 18) 17. Anthony makes fun about Stan’s hair, Stan is upset
were included. Two other nonaggressive cartoons 18. Mick asks Richard if he would like to play
19. Matt won’t let Lenny play today
were paired with corresponding aggressive ones: one
20. Sebastian never lets Rob play
(8) showed a negative but accidental act and was con- 21. Henry and his friends won’t let Ray play with them
trasted with one (9) in which a similar act was inten- 22. The girls won’t let Mark skip with them because he’s a boy
tional; the other (16) showed friendly verbal teasing 23. The boys won’t let Karen play football because she’s a girl
and was contrasted with one (17) that depicted simi- 24. Gerry tells everyone not to talk to Guy
25. Bill spreads nasty stories about Alan
lar teasing that upset the recipient. One cartoon (4) re-
ferred to provoked aggression. The remaining car-
toons covered physical forms of aggression (1–7),
direct verbal aggression (11–15), social exclusion ag- and avoided use of any general terms such as bully-
gression (19–23), and indirect relational aggression ing. The English captions constituted a reference set.
(24, 25); however, embedded in these were compari- In each other country, a researcher translated these
sons of the defining criteria of bullying, namely repe- captions into the main native language as used by
tition and imbalance of power. Thus, repetition was participants in the task. These captions were then
made explicit in some cartoons (5, 12, 20), and imbal- back-translated into English by another person, and
ance of power in others (3, 7, 21). Themes such as rac- returned to the first author in London, who then
ism (13), sexism (22, 23), and discrimination on the checked them against the original versions and dis-
basis of disability (14) or sexual orientation (15) were cussed any discrepancies until these were resolved.
also included. All 25 cartoons were used identically in all 14
The cartoons were piloted extensively before the countries, with the exception of cartoons 13 and 15. If
set was finalized. Stick figures were used so as to color of skin was not an important factor in a culture,
avoid issues of clothing, which might vary by culture, an alternate caption about saying nasty things about a
and to avoid suggesting any particular ethnic group child talking in a different way was used for cartoon
or skin color. Thus, identical pictures were used 13. In addition, in some countries permission was not
across all cultures. Each cartoon had a caption in the readily forthcoming from schools to use cartoon 15,
native language, as listed in Table 1 for the English which was on sexual orientation (usually with 8-year-
language version (boys). In each language, one set of olds, but occasionally with 14-year-olds as well); thus,
cartoons had captions with typical boys’ names in the the cartoon was omitted from the main cross-national
country concerned, and a corresponding set had cap- analyses.
tions with typical girls’ names; cartoons 22 and 23
were in common to both gender sets. Examples of
General Procedure
four cartoons (3 and 10 from the boys’ set, 14 and 21
from the girls’ set) are shown in Figure 1. Researchers in all 14 countries followed an agreed-
The cartoon captions were descriptive (see Table 1) on three-step procedure.
Smith et al. 1123
Figure 1 Examples of four of the cartoons: Numbers 3 and 10 from the boys’ set, and 14 and 21 from the girls’ set.
Step 1: List and select terms for bullying and social ex- It was recognized that this work could not often be
clusion in the relevant language. Likely terms to use in done on a varied sample across different sites in a
the investigation were taken from dictionaries and country (although some research teams did use two
thesauruses, questionnaires on bullying translated sites). However, dialect terms that sometimes ap-
into the language, and research and popular writings peared, but were local to a region within a country,
on the topic. Some 5 to 10 terms were then chosen, on were not selected for Step 3.
the basis of meaning and applicability across the The intention was to end up with four to six terms
country/culture, to proceed to Step 2. for Step 3. Often, more terms were used and under-
Step 2: Use focus groups with children to check on usage stood by 14-year-olds than 8-year-olds. In such cases,
and broad understanding of terms. At least two focus all terms were used with the 14-year-old sample, but
groups were held, each consisting of at least four (usu- only the subset of those terms that were understood
ally more) 8-year-olds and four (usually more) 14-year- by 8-year-olds were used with the younger sample. As
olds. The ostensible aim for the participants was to an example from the English data, pilot work and focus
generate words that were currently used to describe groups with children established that relevant terms in
antisocial/aggressive behaviors and situations at English, besides bullying, included teasing, picking on,
school. To engage participants in debate, a subset of tormenting, harassment, and intimidation. All these
the cartoons was employed as stimulus materials. For terms were used and understood by 14-year-olds.
the purposes of the study, the aim was to ensure that However, only bullying, teasing, and picking on were
terms selected for Step 3 were spontaneously used by readily understood by 8-year-olds, and thus only
some participants, and were broadly familiar to most these three terms were used with this age group.
participants. Step 3: Sorting task using cartoons to delineate ways in
1124 Child Development
which these terms are used. The work was carried out Method section) as part of their “definition” of each
in school settings, with children withdrawn from term was computed. The similarity or difference be-
class on an individual basis (for all 8-year-olds; in tween any two cartoons could be assessed by com-
some countries the 14-year-old data were obtained on paring their percentage profiles. This permitted an
a class basis). Of the terms selected for investigation, analysis of the structure of the cartoon set, over all re-
each was taken in turn. Children were shown or given spondents. To this end, MDS was conducted, using
the cartoons, with the researcher also reading the cap- datasets from all 14 countries.1
tion in the case of 8-year-olds. Gender-appropriate Genders were combined, but analyses were sepa-
cartoon sets were used. For each term, children were rated for the 8- and 14-year-olds, using the 47 terms
required to either include or exclude each cartoon in that were common to both age groups so that the two
their definition of the term with which they were pre- age groups could be compared on the same set of
sented. This was done by sorting them in a pile under terms. Scores were averaged for each country before
the heading: “this is X” or “this is not X” (where X was aggregating (correcting for minor differences in sample
the term being currently considered); or, in class pre- size). The MDS was run on SPSS, minimizing Young’s
sentation with older children, by checking a standard stress. No transformations were carried out. The
score sheet appropriately. The procedure was repeated Manhattan proximity measure was employed to cre-
until all terms were investigated. The captions were ate a single distance matrix between cartoons, the dis-
given in the order shown in Table 1. This was done to tance between two items being the sum of the abso-
maximize consistency across cultures, and to give a lute differences between the values (percentages) for
“narrative” line to the task, as children moved through the items (cartoons). The matrix is square symmetric.
physical, verbal, and more indirect/relational scenar- An ordinal MDS model was specified, using Kruskal’s
ios. The possible drawback of order effects was recog- (1964a, 1964b) least squares monotonic transforma-
nized, but pilot work suggested that a random order tion. The Euclidean distance model was used. The
was more confusing for the children to follow. Kruskal stress values (Formula 1) for one-, two-, three-,
and four-dimensional solutions, respectively, were
.08, .05, .03, and .01 for 8-year-olds; and .14, .08, .03,
Participants
and .02 for 14-year-olds. These low values suggested
The sorting task in Step 3 was given to a minimum two-dimensional solutions were adequate for both
of 20 boys and 20 girls each at 8 years and 14 years of age groups. The MDS solutions for each age group are
age (total N 1,245; n 604 at 8 years, n 641 at 14 shown in Figures 2 and 3.
years). They were selected from schools that were The MDS solutions for the 8- and 14-year-olds, re-
deemed reasonably representative of the education spectively, were similar in overall structure; both
system in that country; that is, they were not drawn showed on Dimension 1 (horizontal axis in figures)
from extreme groups in terms of academic ability, or those cartoons that were not aggressive (prosocial,
socioeconomic status. In all cases there were nearly friendly teasing, accidental damage) at one end (left-
equal numbers of boys and girls at each age group hand side of figures) and cartoons of increasing ag-
(exactly equal unless stated). Sample sizes are shown gression toward the other end (right-hand side of fig-
in Table 2, together with the district of the country ures), whereas the second dimension (vertical axis in
from which they were recruited via local schools. figures) opposed physical cartoons (bottom of fig-
Table 2 also shows the number of terms used for both ures) and verbal and social exclusion cartoons (top of
8-year-olds and 14-year-olds (only one term was figures). The somewhat lower stress value for the 8-
available from the French team; due to a misunder- year-olds (.05) indicates a better fit, and a simpler
standing of instructions they chose the one term most structure, than for the 14-year-olds (.08). This is seen
similar to bullying, from those terms used in the focus
groups with French pupils).
1 Multidimensional scaling attempts to find the structure in a
set of distance measures between objects or cases. This is accom-
RESULTS plished by assigning items to specific locations in a conceptual
space (usually two or three dimensional) such that the distances
SPSS 10.0 for Windows was used for statistical purposes. between the points in space match the given dissimilarities as
closely as possible. Points are arranged so that similar items are
represented by points that are close together (proximities) and
Structure of Responses to the Cartoons dissimilar items are represented by points that are far apart. The
stress value of an MDS solution shows the degree of fit within
The percentage of participants who included each the number of dimensions used, with smaller stress values
of the 24 cartoons (cartoon 15 was excluded, see meaning a better fit.
Smith et al. 1125
Table 2 Details of the Samples, and Number of Terms Used, in each of 14 Countries
Figure 4 Multidimensional scaling solution of cartoon struc- Figure 5 Multidimensional scaling solution of cartoon struc-
ture for boys. ture for girls.
by the 14-year-olds, however, were nonaggressive (8, For both genders, there were clusters of nonaggres-
10, 16, 18), social exclusion (19, 20, 21, 22, 23), verbal: sive (8, 10, 16, 18), social exclusion (19, 20, 21, 22, 23),
directindirect (11, 12, 13, 14, 17, 24, 25; and nine and physical aggression (1, 4); both genders also had
placed more distantly), physical aggression (1, 4), and clusters of verbal:directindirect (11, 12, 13, 14, 17, 24,
physical bullying (2, 3, 5, 6, 7). 25), and physical bullying (2, 3, 5, 6, 7), with the only
difference being that cartoon 9 (nonaccidentally
breaking a ruler) was included in the physical bully-
Gender Differences
ing cluster for boys, but in the verbal:directindirect
An analysis of the structure of the cartoons, in both cluster for girls.
MDS (corresponding to Figures 2 and 3) and hierar-
chical cluster analysis solutions, was carried out sep-
Comparison of Terms on the Cartoon
arately for boys and girls. For this purpose, data
Cluster Structure
from 8- and 14-year-olds were combined, and the anal-
yses were run on 45 terms used by both age groups, To compare terms on the cartoon cluster structure,
from 13 countries (Norway was not included because a similar cluster analysis to that described above was
labeling by gender had been omitted in data collec- used, but was based on data for all 67 terms defined
tion). The Kruskal stress values for one-, two-, by 14-year-olds, and not just the 47 also used by the
three- and four-dimensional solutions, respectively, 8-year-olds. This cluster analysis produced the same
were .12, .07, .03, and .01 for males; and .09, .05, .03, five main clusters, and a plot very similar to that
and .02 for females. These low values indicated that shown in Figure 3: a nonaggressive cluster (8, 10, 16,
two-dimensional solutions were adequate for both 18) that included two prosocial items, plus friendly
genders. teasing and accidental damage; a social exclusion
The MDS solutions for both genders are shown in cluster (19, 20, 21, 22, 23) that included the five so-
Figures 4 and 5. The overall structures were very sim- cial exclusion cartoons; a verbal:directindirect
ilar; and, as described for Figures 2 and 3, the hori- cluster (9, 11, 12, 13, 14, 17, 24, 25) that included the
zontal axis discriminated prosocial from aggressive verbal cartoons, plus nasty teasing, and the two in-
cartoons, and the vertical axis opposed physical direct verbal cartoons, together with the accidental
(bottom of figures) and verbal (middle of figures) damage cartoon (9), which was intermediate between
aggression from social exclusion (top of figures). this cluster (where it technically fell) and the phys-
To establish the extent of similarity more objec- ical bullying cluster (see Figure 3); a physical ag-
tively, two hierarchical cluster analyses were con- gression cluster (1, 4) that included the two physi-
ducted, one for each gender, on the same data used in cal cartoons that most clearly did not fit a definition
the MDS solutions above. The specifications followed of bullying—an even-handed dispute and a pro-
were identical to those used for the earlier age com- voked retaliation; and a physical bullying cluster
parisons. At the level of getting a five-cluster solution (2, 3, 5, 6, 7) that generally implied some repetition
as discussed above, configurations were very similar. or power imbalance.
Smith et al. 1127
The above clusters were used to simplify the mean- the distance matrix computed between terms. Six
ing profiles of the 67 terms. A mean percentage score main groups or clusters were identified (see Figure 6).
was computed for each term on each cluster. The The clusters can be interpreted by referring back to
meaning of each term could thus be expressed in 5 Table 3, where the “meaning” of each term (with re-
percentage scores, instead of 25. The outcome is pre- gard to the cartoons) can be inferred.
sented in Table 3. The range of scores for each cluster The first cluster consisted of 19 terms, which were
was 0 to 100. For example (taking the fourth line in generally higher on the physical bullying, verbal,
Table 3), the term angreifen was only applied in 3% of and social exclusion clusters than on the physical ag-
responses to the nonaggressive cartoons (i.e., aver- gression cluster; thus, these terms tended to be clos-
aged over these four cartoons individually), and only est to the definition of bullying, and did indeed in-
7% of the social exclusion and 11% of the verbal bul- clude the English term bullying, as well as the term
lying cartoons, but it was applied to 74% of the phys- picking on.
ical aggression cartoons, and 83% of the physical bul- The second cluster consisted of six terms, which
lying cartoons; this indicates that the term angreifen scored most highly on the social exclusion cluster.
is used for physical aggression and bullying but The third cluster had seven terms; these all scored
not for nonphysical forms. As expected, none of the much more highly on the verbal:directindirect
terms scored highly on the nonaggressive cluster; cluster than on the physical bullying or social exclu-
however, there was considerable variation in weight- sion clusters. The English term teasing appeared in
ing on the other four clusters, which showed differ- this cluster.
ences in meaning—sometimes obvious, sometimes The fourth cluster had 19 terms; these were weighted
more subtle. most highly on the verbal:directindirect cluster, but
moderately on the physical bullying cluster, and less
on the social exclusion cluster. The English terms ha-
Conceptual Structure of Terms Used
rassment, intimidation, and tormenting appeared in
in Different Languages
this cluster.
For each term, the percentage of participants who The fifth cluster contained three items, which were
included each of the 24 cartoons as part of their “def- only weighted highly on physical aggression and
inition” of that term was computed. The meaning of physical bullying. The sixth cluster had 13 items,
each term was operationalized with regard to the which loaded highest on either the physical aggres-
mean percentage for the cartoons in each of the five sion or the physical bullying cluster, but moderately
clusters. The similarity or difference in meaning be- on one or two other aggressive clusters.
tween any two terms could be assessed by comparing Corresponding terms from different languages
their percentage profiles across the five clusters. This were not necessarily close together; thus, although
permitted a comparison of the meaning of the terms different countries did contribute different numbers
across languages and cultures. of terms to the analyses (range, 1–7; see Tables 2 and
To examine similarities and differences in their 3) there was no reason to suppose that this had af-
meanings, MDS was first conducted on all 67 terms fected the cluster structure obtained. For example, the
for the 14-year-olds. The specifications for the analy- four terms used in Austria all fell in different clus-
sis were identical to those applied in the earlier MDS ters (Table 3, Figure 6): gemein sein in the bullying
analyses, except that in this case, a distance matrix cluster, sekkieren in the verbal cluster, ärgern in the ver-
was computed between terms instead of between car- bal physical cluster, and angreifen in the physical
toons as was done in the earlier analyses. Stress only cluster.
values for one- to four-dimensional solutions were, Figure 6 shows that the initial bifurcation of clus-
respectively, .34, .18, .09, and .06. This suggests that a ters of terms was between the “physical” terms (Clus-
three-dimensional solution was needed. Examining ters 5 and 6) and the rest. Clusters 5 and 6 effectively
this, the first dimension opposed the aggressive and included terms corresponding to violence (such as
the nonaggressive terms, the second dimension op- Portuguese violência, Italian violenza). The next split
posed the physical and social exclusion terms, and was to separate out the bullying terms (Cluster 1)
the third dimension differentiated the verbal and from those that referred more specifically to verbal
social exclusion terms. Because a three-dimensional and social exclusion (Clusters 2, 3 and 4). The third
solution is difficult to visualize or portray, a hierachi- split separated social exclusionary terms in Cluster 2
cal cluster analysis was conducted to aid the identifi- (such as Portuguese rejeição) from the verbal terms
cation of subgroups of terms. The specifications for (Clusters 3 and 4). The fourth split led to the six clus-
the analysis were identical to those used before, with ters described above.
1128 Child Development
Table 3 Mean Percentage of 14-Year-Olds Who Included the Cartoons in Each Cluster as Part of Their Definition of that Term
Verbal:Direct
Nonaggressive Physical Aggression Physical Bullying Indirect Social Exclusion
Austria
A1: sekkieren 16 40 51 73 34
A2: ärgern 21 56 60 85 58
A3: gemein sein 23 65 78 94 90
A4: angreifen 3 74 83 11 7
China (romanized terms)
C1: lingru 2 28 57 71 51
C2: qifu 3 13 94 85 68
C3: qiling 2 25 82 72 52
C4: qiru 2 22 79 80 53
C5: qiwu 2 23 67 89 58
C6: qiya 3 17 85 55 60
C7: wuru 2 24 39 84 45
England
E1: bullying 4 34 94 91 62
E2: harassment 10 42 88 84 49
E3: teasing 15 43 35 83 51
E4: intimidation 7 42 78 82 46
E5: tormenting 14 59 76 84 64
E6: picking on 8 39 95 96 67
France
F1: violence 3 56 83 60 26
Germany
GE1: ärgern 7 85 61 65 46
GE2: angreifen 2 89 91 72 30
GE3: gemein sein 4 28 85 93 85
GE4: schikanieren 6 20 58 80 55
Greece
G1: kano to magha 8 40 84 70 65
G2: miono 10 70 74 85 78
G3: taleporo 13 76 92 85 75
Iceland
IS1: radast a 3 71 95 42 25
IS2: hrekkja 9 40 75 88 60
IS3: skilja ut undan 1 4 12 27 97
IS4: strida 13 41 52 90 64
IS5: taka fyrir 4 31 81 73 65
IS6: einelti 4 11 86 81 75
Italy
IT1: aggressività 10 91 96 68 63
IT2: fare il duro 12 68 80 74 72
IT3: prepotenza 10 71 92 86 90
IT4: violenza 10 93 96 63 59
IT5: approffitarse 12 48 82 84 75
IT6: cattiveria 11 75 96 91 86
IT7: scorretto 15 85 94 93 89
Japan
J1: ijime 4 9 50 87 39
J2: ijiwaru 8 27 56 94 56
J3: iyagarase 8 30 56 97 43
J4: fuzake 27 26 38 56 19
J5: nakamahazushi 4 8 29 55 65
Norway
N1: erting 18 29 28 67 36
N2: mobbing 6 17 46 70 39
N3: plaging 11 35 81 77 58
N4: krangling 9 88 45 22 18
(Continued)
Smith et al. 1129
Table 3 Mean
Continued
Percentage of 14-Year-Olds Who Included the Cartoons in Each Cluster as Part of Their Definition of that Term
Verbal:Direct
Nonaggressive Physical Aggression Physical Bullying Indirect Social Exclusion
Portugal
P1: abuso 3 10 71 55 30
P2: armar-se 1 18 72 56 75
P3: insulto 4 35 17 58 20
P4: provoção 3 60 69 78 44
P5: rejeição 1 8 38 54 81
P6: violência 1 18 74 24 16
Slovenia
SL1: nadlegovanje 13 51 89 67 42
SL2: nasilništvo 6 88 96 48 33
SL3: trpinčenje 7 58 93 75 65
SL4: ustrahovanje 4 45 79 49 31
SL5: zavračanje 7 43 54 66 92
SL6: zlorabljanje 4 40 84 59 49
Spain
S1: maltrato 8 49 96 85 65
S2: meterse con 15 66 63 86 40
S3: rechazo 6 28 42 81 96
S4: abuso 5 21 86 75 58
S5: egoismo 6 21 54 66 94
Thailand
T1: nisai mai dee 15 77 79 81 69
T2: klang 19 32 64 86 67
T3: tum raai 10 38 76 61 48
Note: The five clusters consist of the following cartoons: nonaggressive (8, 10, 16, 18), physical aggression (1, 4), physical bullying (2, 3, 5,
6, 7), verbal:directindirect (9, 11, 12, 13, 14, 17, 24, 25), and social exclusion (19, 20, 21, 22, 23).
directly experience more physical bullying, and girls late 1990s, still did not fully include social exclusion.
may experience more indirect bullying, there is con- Thus the Portuguese term abuso might be taken as
siderable overlap. In addition, both boys and girls are reasonably similar in common usage to the English
likely to observe a considerable amount of bullying term bullying, being high on physical and verbal bul-
involving both same-sex and opposite-sex children lying, lower on social exclusion, and very low on
(O’Connell et al., 1999). fighting. However, the Portuguese term armar-se could
The cartoon methodology allowed for the ability to be seen as a better correspondent to the scientific def-
compare 67 terms from 14 countries, with regard to inition, because it is higher on social exclusion. This
their weighting on the five major clusters obtained imperfect correspondence between popular and sci-
from Figure 3 (in a slightly modified version embod- entific definition of terms such as bullying is an issue
ying 67 not 47 terms). The results (Table 3) and the that researchers must remain aware of.
cluster analysis (Figure 6) show something of the con- Table 3 may also help in the interpretation of cross-
ceptual structure captured in different languages, and national comparisons. For example, Genta et al.
show how closely other terms relate to the term bul- (1996) compared rates of bullying (using the Olweus
lying. Naturally, the results are limited by the choice questionnaire) in Italy, with those in the United King-
of terms used in each country, and the number of dom reported by Whitney and Smith (1993). Gener-
terms available. The relatively few items in the social ally high rates of bullying were found in both central
exclusion cluster (six items) may reflect an uninten- and southern Italy that were notably higher than the
tional lesser emphasis on these terms in the instruc- rates reported in the United Kingdom. In the Italian
tions given to participating teams for generating questionnaire, the term prepotenza was used for bully-
terms to use. Indeed, the English data did not include ing. Table 3 shows that prepotenza does indeed load
such terms as social exclusion or rejection. Thus, little highly on physical, verbal, and social exclusionary
attention can be given to the absence of a term in the bullying, but also on fighting (Cluster 2). Thus, the
clusters in Table 3, for a particular country or lan- higher rates may reflect a more inclusive response in-
guage. However, it is possible to compare profiles of corporating even-handed fighting, as well as bullying
those terms, which are present. with an imbalance of power. Using the term prepo-
The English term bullying loaded highly on both tenza in questionnaires in Italy may therefore lead to
physical and verbal bullying, moderately highly on an overestimation, if the findings are then referred to
social exclusion, less highly on an even-handed dis- as yielding rates of bullying. Other Italian terms such as
pute or a provoked retaliation, and, of course, mini- violenza, cattiveria, or scoretto suffer a similar problem.
mally on nonaggression (see Table 3). This is consis- An alternative to using global terms such as bully-
tent with the general adult understanding of the term; ing in questionnaire surveys is to ask for information
bullying is different from aggression, and is generally on the experience of particular acts. For example, the
thought to be both physical and verbal, but is not al- Life in School checklist (Arora, 1994) asks children if
ways thought to include more psychological forms of they have experienced acts such as being hit, threat-
behavior such as social exclusion. Boulton (1997) ened, teased, or called names. This avoids issues of
found that less than 50% of teachers and only one in children’s understanding of the term bullying. How-
five pupils in English schools defined psychological ever, if researchers wish to make generalizations
or emotional abuse as “bullying.” about the occurrence of bullying, they face the prob-
Table 3 might be used to help choose the most suit- lem of deciding which acts should be included as bul-
able term, when translating a questionnaire such as lying. This amounts to a decision by the researchers,
the Olweus questionnaire into another language. not the children. In addition, such questionnaires nor-
Even if the questionnaire gives an extended definition mally do not specify that there should be an imbal-
near the beginning, it is usual that one term is used in ance of power in the act experienced, so straightfor-
most of the questions (e.g., “How often have you been ward fighting and aggression (rather than bullying)
bullied this month?”), and it is correspondingly likely may be included by researchers. These drawbacks
that children may use their natural understanding of mean that there is likely to continue to be a place for
this term (“bullied”) rather more than the longer def- questionnaires about bullying (or ijime, abuso, or
inition read earlier. whatever term is used in a particular country), and it
However, the English term bullying, as used by is then important to know how children understand
schoolchildren, does not match perfectly with the def- these terms.
initional concept used by many researchers and now Researchers who investigate the phenomenon of
widely accepted in the scientific community; this is bullying are involved, as are adult and child members
because the term bullying, at least in England in the of the community, in the process of constructing its
1132 Child Development
meaning in a social and historical context. Histori- Houndoumadi, Deree College, Athens, Greece; Ken-
cally, meanings of words change. Even the core con- neth Kim, University of East Anglia, Norwich, U.K.;
cept of the term bully has changed dramatically over Ragnar P. Olafsson, Institute for Educational Research,
several centuries. More subtle changes have taken Reykjavik, Iceland; Rosario Ortega, Universidad de
place in the past 5 years, with the incorporation of Sevilla, Seville, Spain; Jacques Pain, Universite Paris
more indirect and relational forms of bullying into X Nanterre, Paris, France; Lena Pateraki, Deree Col-
current definitions. In addition, the term bullying is lege, Athens, Greece; Mechthild Schafer, Universität
now commonly used in the workplace and is not München, Munich, Germany; Monika Singer, Gold-
solely confined to the school context. smiths College, University of London, London, U.K.;
The current study therefore represents an histori- Andrea Smorti, Universitá di Firenze, Florence, Italy;
cal snapshot of the meaning of terms cognate to bul- Yuichi Toda, Osaka University of Education, Osaka,
lying, at the turn of the second millenium. This snap- Japan; Helgi Tomasson, University of Iceland, Reyk-
shot is moreover limited by the particular samples of javik, Iceland; Zhang Wenxin, Shandong Teacher’s
terms used, and of choice of respondents, in the 14 University, Jinan, People’s Republic of China.
countries selected. However, the findings should as-
sist in the design and interpretation of comparative
cross-national studies of bullying at the present time REFERENCES
and for some years to come. In addition, the findings Ananiadou, K., & Smith, P. K. (in press). Legal requirements
concerning the greater discrimination of criteria at 14 and nationally circulated materials against school bully-
years than 8 years, and the lack of gender differences ing in European countries. Criminal Justice.
in understanding and use of terms despite the gender Arora, T. (1994). Measuring bullying with the “Life in
differences in behavior, may be of considerable gener- School” checklist. Pastoral Care in Education, 12, 11–15.
alizability over time and in different countries. Arora, T. (1996). Defining bullying. School Psychology Inter-
national, 17, 317–329.
Asher, S. R., & Coie, J. (Eds.). (1990). Peer rejection in child-
ACKNOWLEDGMENTS hood. New York: Press Syndicate of University of Cam-
bridge.
The authors thank Hidefumi Arimoto, Julius K. Bjorns-
Björkqvist, K., Lagerspetz, K. M. J., & Kaukiainen, A. (1992).
son, Cheryl Blackadder, Pamela Burton, Isabel Fernan- Do girls manipulate and boys fight? Aggressive Behavior,
dez, Rosa Fera, Hector Gutierrez, Stefan Korn, Jessica 18, 117–127.
Mahdavi, Ersilia Menesini, Vicky Panagiotidou, Vicky Boulton, M. J. (1997). Teachers’ views on bullying defini-
Pavlidis, Colin Pritchard, Rosario del Rey, Ebba Staven, tions, attitudes and ability to cope. British Journal of Edu-
Mitsuru Taki, and Thanes Wongyannava for help in cational Psychology, 67, 223–233.
data collection. This study was supported by contract Crick, N. R., Casas, J. F., & Mosher, M. (1997). Relational and
ERBFMRX-CT-970139 from the European Commission overt aggression in preschool. Child Development, 33,
on “The Causes and Nature of Bullying and Social Ex- 579–588.
clusion in Schools, and Ways of Preventing Them” Crick, N. R., & Grotpeter, J. K. (1995). Relational aggression,
gender, and social–psychological adjustment. Child De-
(http://www.gold.ac.uk/tmr).
velopment, 66, 710–722.
Crick, N. R., & Grotpeter, J. K. (1996). Children’s treatment
ADDRESSES AND AFFILIATIONS by peers: Victims of relational and overt aggression. De-
velopment and Psychopathology, 8, 367–380.
Corresponding author: Peter K. Smith, Department of Encarta World English dictionary. (1999). London: Bloomsbury.
Psychology, Goldsmiths College, New Cross, London Farrington, D. P. (1993). Understanding and preventing bul-
SE14 6NW, U.K.; e-mail: p.smith@gold.ac.uk. Helen lying. In M. Tonry & N. Morris (Eds.), Crime and justice:
Cowie is at the University of Surrey Roehampton, An annual review of research (Vol. 17, pp. 381–458). Chi-
London, U.K.; Ragnar F. Olafsson is at the Institute for cago: University of Chicago Press.
Educational Research, Reykjavik, Iceland; and Andy Fonzi, A. (Ed.). (1997). Il bullismo in Italia. Firenze, Italy:
P. D. Liefooghe is at Birkbeck College, London, U.K. Giunti.
Fonzi, A., Genta, M. L., Menesini, E., Bacchini, D., Bonino,
Collaborating authors and their affiliations are: Ana
S., & Costabile, A. (1999). Italy. In P. K. Smith, Y. Morita,
Almeida, Universidade do Minho, Braga, Portugal;
J. Junger-Tas, D. Olweus, R. Catalano, & P. Slee (Eds.),
Hozumi Araki, Ritsumeikan University, Kyoto, Ja- The nature of school bullying: A cross-national perspective
pan; Cristina del Barrio, Universidad Autonoma de (pp. 140–156). New York: Routledge.
Madrid, Madrid, Spain; Angela Costabile, Universitá Galen, B. R., & Underwood, M. K. (1997). A developmental
della Calabria, Cosenza, Italy; Bojan Dekleva, Uni- investigation of social aggression among children. Devel-
versity of Ljubljana, Llubljana, Slovenia; Anastasia opmental Psychology, 33, 589–600.
Smith et al. 1133
Genta, M. L., Menesini, E., Fonzi, A., Costabile, A., & Smith, Olweus, D. (1978). Aggression in the schools: Bullies and whip-
P. K. (1996). Bullies and victims in schools in central and ping boys. Washington, DC: Hemisphere Press (Wiley).
southern Italy. European Journal of Psychology of Education, Olweus, D. (1993). Bullying at school: What we know and what
11, 97–110. we can do. Oxford, U.K.: Blackwell.
Hazler, R. (1996). Breaking the cycle of violence: Interventions Olweus, D. (1999). Sweden. In P. K. Smith, Y. Morita, J.
for bullies and victims. Bristol, PA: Accelerated Develop- Junger-Tas, D. Olweus, R. Catalano, & P. Slee (Eds.), The
ment, Inc. nature of school bullying: A cross-national perspective (pp. 7–
Heinemann, P. P. (1973). Mobbning: Gruppvald blant barn og 27). New York: Routledge.
vokane (Bullying: Group violence among children and Pikas, A. (1989). A pure conception of mobbing gives the best
adults). Stockholm: Natur och Kultur. results for treatment. School Psychology International, 10,
Hughes, T. (1989). Tom Brown’s schooldays. Oxford: Oxford 95–104.
University Press. (Original work published 1857) Rigby, K., & Slee, P. T. (1991). Bullying among Australian
Junger-Tas, J., & van Kesteren, J. N. (1999). Bullying and de- school children: Reported behaviour and attitudes to-
linquency in a Dutch school population. Leiden, The Neth- wards victims. Journal of Social Psychology, 131, 615–627.
erlands: Kugler. Rivers, I., & Smith, P. K. (1994). Types of bullying behavior
Kruskal, J. B. (1964a). Multidimensional scaling by optimiz- and their correlates. Aggressive Behavior, 20, 359–368.
ing goodness of fit to a nonmetric hypothesis. Psy- Smith, P. K. (1999). Comparison of bullying definitions
chometrika, 29, 1–27. within and across cultures. In S. Sharp (Ed.), Bullying be-
Kruskal, J. B. (1964b). Nonmetric multidimensional scaling: haviour in schools (pp. 37–38). London: NFER-Nelson.
A numerical method. Psychometrika, 29, 115–129. Smith, P. K., & Levan, S. (1995). Perceptions and experiences
Lagerspetz, K. M. J., Björkqvist, K., & Peltonen, T. (1988). Is of bullying in younger pupils. British Journal of Educa-
indirect aggression more typical of females? Gender dif- tional Psychology, 65, 489–500.
ferences in aggressiveness in 11- to 12-year-old children. Smith, P. K., Madsen, K., & Moody, J. (1999). What causes
Aggressive Behavior, 14, 403–414. the age decline in reports of being bullied at school? To-
Madsen, K. C. (1997). Differing perceptions of bullying. Un- wards a developmental analysis of risks of being bullied.
published doctoral thesis, University of Sheffield, Educational Research, 41, 267–285.
U.K. Smith, P. K., Morita, Y., Junger-Tas, J., Olweus, D., Catalano,
Morita, Y. (1996). Bullying as a contemporary behaviour R., & Slee, P. (Eds.). (1999). The nature of school bullying: A
problem in the context of increasing “societal privatiza- cross-national perspective. New York: Routledge.
tion” in Japan. Prospects, 26, 311–329. Smith, P. K., & Sharp, S. (Eds.). (1994). School bullying: In-
Morita, Y. (1999). Ijime [bullying]: A cross-national investi- sights and perspectives. London: Routledge.
gation [Final report to Monbusho]. Tokyo: Ministry of Smith, P. K., & Shu, S. (2000). What good schools can do about
Education. bullying: Findings from a survey in English schools after a
Morita, Y., Smith, P. K., Junger-Tas, J., Olweus, D., & Catalano, decade of research and action. Childhood, 7, 193–212.
R. (Eds.). (1999). Sekai no ijime [Bullying across the Whitney, I., & Smith, P. K. (1993). A survey of the nature and
world]. Tokyo: Kaneko Shobou. extent of bullying in junior/middle and secondary
Morita, Y., Soeda, H., Soeda, K., & Taki, M. (1999). Japan. In schools. Educational Research, 35, 3–25.
P. K. Smith, Y. Morita, J. Junger-Tas, D. Olweus, R. Cata- Younger, A. J., Schwartzman, A. E., & Ledingham, J. (1985).
lano, & P. Slee (Eds.), The nature of school bullying: A cross- Age-related changes in children’s perceptions of aggres-
national perspective (pp. 309–323). New York: Routledge. sion and withdrawal in their peers. Developmental Psy-
Niedl, K. (1996). Mobbing and well-being. European Journal chology, 21, 70–75.
of Work and Organizational Psychology, 5, 239–249. Younger, A. J., Schwartzman, A. E., & Ledingham, J. (1986).
O’Connell, P., Pepler, D., & Craig, W. (1999). Peer involve- Age-related changes in children’s perceptions of social
ment in bullying: Insights and challenges for interven- deviance: Changes in behavior or in perspective? Devel-
tion. Journal of Adolescence, 22, 437–452. opmental Psychology, 22, 531–542.