Perspective taking in children’s narratives about
Naomi J. Aldrich1, Harriet R. Tenenbaum2*, Patricia J. Brooks1,
Karine Harrison2and Jennie Sines2
1The College of Staten Island and The Graduate Center, City University of
New York, USA
2Kingston University, Kingston upon Thames, UK
This study explored relationships between perspective-taking, emotion understanding,
children generated fictional narratives, using a wordless picture book, about a frog
experiencing jealousy. Children’s emotion understanding was assessed through a
standardized test of emotion comprehension and their ability to convey the jealousy
theme ofthestory.Perspective-takingabilitywas assessed withrespecttochildren’suse
of narrative evaluation (i.e., narrative coherence, mental state language, supplementary
evaluative speech, use of subjective language, and placement of emotion expression).
Olderchildren scored higher than youngerchildren on emotion comprehension and on
understanding the story’s complex emotional theme, including the ability to identify a
rival. They were more advanced in perspective-taking abilities, and selectively used
emotion expressions to highlight story episodes. Subjective perspective taking and
of supplementary evaluative speech, in turn, was predictive of both subjective
perspective taking and narrative coherence.
Jealousy refers to a negative emotion that symbolizes some form of interpersonal rivalry.
Specifically, jealousy has been defined as a complex emotion with its foundation being
the threat of loss of an important relationship to another individual (Smith, Kim, &
Parrott, 1988). Given the social nature of this emotion and its pervasiveness in everyday
life, many researchers have investigated individual differences in the affective
experiences and effects of this emotion from as young as 6 months old (Hart, 2002),
and into older ages: preschool (Bauminger, Chomsky-Smolkin, Orbach-Caspi, Zachor, &
Levy-Shiff, 1998), early adolescence (Parker, Low, Walker, & Gamm, 2005), and
adulthood (Shackelford et al., 2004). Apart from this literature focusing on the external
expression and personal experiences of jealousy, what remains to be understood is how
*Correspondence should be addressed to Harriet R. Tenenbaum, Kingston University, Penrhyn Road, Kingston upon Thames,
Surrey KT1 2EE, UK (e-mail: firstname.lastname@example.org).
British Journal of Developmental Psychology (2010), 00, 1–25
q 2010 The British Psychological Society
children come to understand the concept itself. More specifically, what capabilities
do children have in discerning specific components of jealousy, such as the presence of
a rival or the threat of a relationship, at an age when research shows they are
experiencing a burgeoning awareness of other people’s mentality? The goal of the
present research was to evaluate children’s understanding of jealousy during middle
childhood within the context of narrative ability and to assess the role of perspective
taking in this development.
How children come to understand others in relation to themselves
Over the past two decades, a large amount of research has focused on children’s
understanding of people’s beliefs, feelings, and desires. Whereas many studies have
focused on the emergence of socio-cognitive understanding in children of ages of 3–4
years (Wellman, Cross, & Watson, 2001), research has demonstrated that this capacity is
further elaborated throughout middle childhood (de Rosnay & Hughes, 2006; Flavell,
Green, & Flavell, 1993).
Integral to the ability to understand another person’s thoughts and feelings is the
ability to take their perspective. Selman (1980, 1981) has proposed a hierarchical
organization to children’s development of perspective-taking skill. According to his
model, younger children (around the ages of 3–6) often confuse the thoughts and
feelings of others with the thoughts and feelings of themselves, but during the middle
childhoodyears children are able to consider the perspectives ofothers, and they realize
that others have the capacity to do this as well. Children first grasp that perspectives
differ because people have access to different information; gradually, children more fully
utilize this knowledge in thinking about others’ distinct viewpoints.
Another key feature of children’s understanding of the mentality of persons is their
developing understanding of others’ emotions (Denham, 1998). Beginning at 3–4 years
of age, children’s understanding of emotion hinges on their ability to recognize and
evaluate external features of emotion, i.e., their ability to associate facial features,
causes, and situational cues with particular emotions (Pons, Harris, & de Rosnay, 2004).
Between the ages of 5 and 8 years, children become aware of the psychological aspects
of emotion, such as the role desires and beliefs play in emotional reactions, and they
begin to understand complex or secondaryemotions. Whereas the first stage ofemotion
understanding during the preschool years is associated with the basic, primary
emotions, with a focus on behavioural manifestations of emotion, such as tears with
sadness (Bosacki & Moore, 2004), the second stage from ages 5 to 8 is earmarked by
children’s emerging understanding of complex or secondary emotions that hinge on
self-awareness or self-consciousness (Saarni, 1999), such as pride, embarrassment,
jealousy, and ambivalence.
The question of how children use their understanding of the basic emotions to
progress towards comprehension of complex emotions has led some researchers to
investigate children’s ability to describe situations that would bring about different
emotional reactions. Harris, Olthof, Terwogt, and Hardman (1987) asked 5-, 7-, 10-, and
14-year-olds to describe situations that would bring about 20 emotions, ranging from
basic emotions (e.g., happiness, anger) to more complex emotions (e.g., jealousy,
pride). Different emotion concepts appeared at different ages, with the younger
children largely unable to provide situations evoking the more complex emotions, such
as jealousy. Russell and Paris (1994) asked children (4- to 7-year-olds) to describe a
situation evoking a certain emotion and to decide whether or not the emotion was a
2 Naomi J. Aldrich et al.
good or bad feeling. They proposed that children’s comprehension of basic emotions
and the feelings associated with them become the building-blocks towards
understanding more complex emotions. According to this hypothesis, a child’s
understanding of jealousy would begin with a negative feeling (such as that associated
with anger) and would be elaborated as the child acquires additional characteristics of
the emotion, such as the presence of a rival.
Narratives as a means for assessing socio-cognitive understanding
Storytelling is an important cultural activity in which emotions, morals, and societal
values are conveyed. Narratives provide a means for individuals to explain the world
they live in, make connections with others, and find meaning in their lives. Stories thus
serve as an important cultural tool for expressing socio-cognitive understanding of
feelings and beliefs (Fivush, 1989). Specifically, the knowledge a narrator possesses of
real-world influences on psychological internal states (such as emotion) is reflected in
their narrative constructions and this knowledge is further utilized in children’s
understanding of themselves and others (Trabasso & Stein, 1997). Children learn a
variety of storytelling techniques, including how to construct a personal narrative, in
which they recount a specific experience in their life, and storybook narration, in which
they recount the episodes of a fictional story. Both techniques rely on the assumption
that a child’s internal representation of the episode or event is related to how the child
will construct a story about that event (Fivush & Haden, 1997). Researchers utilizing
personal narratives have suggested that talking about one’s past experiences leads to an
increased awareness of one’s self in relation to others through assessment of different
perspectives (Fivush, Haden, & Reese, 2006; Fivush & Nelson, 2004). Children’s ability
to construct personal narratives has been found to be related to their emotion
understanding (Cutting & Dunn, 1999) and theoryof mind (Kleinknecht& Beike, 2004).
Some researchers suggest that fictional narratives, in comparison to personal
narratives, present advantages as a more controlled form of discourse. Bamberg and
Damrad-Frye (1991) offer that fictional narratives allow for more precise analyses that
the variability in personal narratives, often times, will not allow. Furthermore, telling a
fictional narrative, as opposed to retelling a personal experience becomes more salient
to children as they encounter fictional narratives as a means of school-based instruction
(Ukrainetz et al., 2005). Researchers have found that children are able to attribute
emotion and mental states to story characters within a fictional context (Bamberg &
Damrad-Frye, 1991; Bamberg & Reilly, 1996; Peterson & Slaughter, 2006; Reilly, 1992),
and use different narrative events (i.e., goals, outcomes, reactions) to determine the
causes of emotion for story characters (Bourg & Stephenson, 1997). In addition,
children’s use of mental state language during fictional storytelling is related to their
theory of mind (Symons, Peterson, Slaughter, Doyle, & Roche, 2005).
Despite the importance of perspective taking in children’s socio-cognitive
understanding and the success with which researchers have been able to explore the
emotion component, the assessment of perspective-taking ability apart from narrative
capabilities has largely been neglected in the narrative literature on the development of
socio-cognitive understanding. O’Neill and Shultis (2007) assessed 3-, 4-, and 5-year-olds’
attributions of mental perspectives to story characters through the use of toy models
and narrative vignettes, and found that children at 5 years of age were capable of
tracking the mental perspectives of different characters in their narratives. Further
investigation of how perspective-taking ability develops during middle childhood and
Perspective taking and jealousy3
how it manifests itself within a narrative context is needed. We address this gap in the
literature by focusing on how perspective taking and emotion understanding come
together during middle childhood, and how the ability to take another person’s
perspective contributes to one’s understanding of jealousy.
Narrative ability and its relation to socio-cognitive understanding
Researchers utilizing narratives have focused on how children use linguistic devices to
provide meaning and organization to their stories from a narrator’s perspective. By
acting as gatekeeper, the role of the narrator is to provide the audience with a point of
view through which they reveal the important aspects of a story. Specifically, narrators
do this by telling a coherent and complex narrative through evaluation of information
they consider to be essential to the story. To create a coherent narrative, the narrator
must construct a series of statements that connect the main events of the story
(Peterson & McCabe, 1983; Snow, Tabors, Nicholson, & Kurland, 1995). Narrators
include clauses that relay the sequential ordering of story elements through formalized
storybook schemas. Several types of clauses may be used to emphasize particular
aspects of the story structure; these provide the audience with information about
the setting of the story or story resolution (i.e., events that occur after the climax of the
story), and may utilize appendages such as formalized beginnings or endings of the story
(i.e., Once upon a time, They lived happily ever after; Peterson & McCabe, 1983; Snow
et al., 1995; Willenberg & Kang, 2001). Peterson and McCabe (1983) identified specific
elements of narrative skill, such as the provision of temporal and referential
information, and storybook language. Willenberg and Kang (2001) additionally
emphasized the importance of relating story elements and events to each other to build
a coherent narrative.
A manner in which narratives become more complex is through additional
evaluative devices that enable a narrator to use their own perspective to relate
important story aspects for their audience. Building upon previous work (Labov &
Waletzky, 1967; Peterson & McCabe, 1983), Bamberg and Damrad-Frye (1991)
suggested that a narrator attempts to provide the audience with a meaningful
interpretation of the story by connecting temporal aspects of the story and evaluating
important events in the story. Bamberg and Damrad-Frye explored five categories of
evaluative device used in narratives (i.e., frames of mind, story character speech,
hedges, negativestates and actions, and causal connectors) and how and where children
(5- and 9-year-olds) and adults used these devices when narrating a wordless picture
book, Frog Where Are You? (Mayer, 1969). Significant developmental trends were
observed in how evaluative devices were used: children tended to use evaluation as a
way to introduce and arrange events according to the timeline of the story (i.e., children
told the story from a local perspective). Adults, on the other hand, tended to use
evaluation to introduce and arrange events based on their perceived importance to the
story as awhole (i.e., they told the story from a global perspective). Between the ages of
5 and 8 years, children differed in their distribution of evaluative devices (including
references to emotion and mental states), with younger children evaluating only local
details and older children providing more global evaluations.
Bamberg and Damrad-Frye (1991) also reported a developmental progression in how
mental state terms and emotions were used in storytelling. Five-year-olds’ narratives
made references to mental states and emotions to emphasize local perspectives, often
linking a character’s facial expression with the preceding plot point. In contrast, adults,
4 Naomi J. Aldrich et al.
and to a lesser extent 9-year-olds, were better able to link their evaluations of mental
states and emotions to the story as a whole, without relying solely on the immediately
preceding events, thus, relating a global perspective to their audience. Children’s
increasing ability to provide global evaluations has been linked to their ability to
construct an overarching emotional theme for a story (Berman & Slobin, 1994;
Stro ¨mqvist & Verhoeven, 2004).
Perspective taking and children’s narratives about jealousy
The present study is the first to explore the development of children’s understanding of
the concept of jealousy through narrative development in middle childhood. In doing
so, we investigate the interrelationships between two main features of socio-cognitive
understanding – perspective taking and emotion understanding – within the context of
narrative ability. Children (5- to 6-year-olds and 7- to 8-year-olds) were asked to narrate a
wordless picture book about a frog that experiences jealousy when his owner adopts
a new frog (One Frog Too Many, Mayer & Mayer, 1975). The fictional narrative, as
opposed to asking children about their personal experiences with jealousy, was chosen
given the stigma associated with jealousy as being a negative emotion associated with
feelings of rejection, anger, and resentment (Smith et al., 1988). It was our belief that
children would be more willing to talk about someone else’s experience of jealousy than
their own. We specifically explored how children’s perspective-taking ability predicts
their ability to relay this complex emotional theme to their audience.
To assess children’s emotion understanding, we examined the extent to which
children identified the jealousy theme and conveyed different components of the
theme, such as the presence of a rival, anger directed toward rival, and interpersonal
conflict. Children were also given a standardized test of emotion comprehension (TEC;
Pons et al., 2004).
To assess children’s perspective-taking ability, we examined how often children
conveyed the viewpoint of the narrator through their use of subjective speech. We also
evaluated the coherence of their narratives, i.e., the degree to which children conveyed
the full details of the story in an appropriate fashion in order for the story to ‘make
sense’. Furthermore, we looked at children’s use of additional evaluative speech as well
as their placement of emotion labels and emotion explanations within the episodic
structure of the story, exploring whether children conveyed a local versus global
perspective to their audience as described by Bamberg and Damrad-Frye (1991). We
hypothesized that children’s emotion understanding and perspective-taking abilities
would increase with age and would be predictive of their success in conveying the
emotional theme of jealousy in their narratives.
A total of 47 children from the London, UK metropolitan area participated. Children’s
ages ranged from 5;1 (years; months) to 8;6 with a mean of 6.96 years. Children were
dividedintoayoungeragegroup(5–6 years ofage)andan older group(7–8 yearsofage).
There were 11 younger boys (M ¼ 5:86 years, SD ¼ 0:54 years), 12 younger girls
(M ¼ 5:96years,SD ¼ 0:33years),11olderboys(M ¼ 8:05years,SD ¼ 0:34years),and
13 older girls (M ¼ 7:90 years, SD ¼ 0:44 years). There was no statistically significant
Perspective taking and jealousy5
mean age difference between the boys and girls in the younger age group,
Fð1;22Þ ¼ 0:34, p ¼ :57, or the older age group, Fð1;23Þ ¼ 0:91, p ¼ :35. The children
took part in a larger study of children’s emotion understanding reported elsewhere
(citation removed for review), and were selected for this study on the basis of having
narrated awordless picture book, One Frog Too Many (Mayer & Mayer, 1975) during the
initial session ofthatstudy.Thesedatawerenotreportedincitationremovedforreview.
Procedure and materials
The children were recorded telling a story to accompany the wordless picture book,
One Frog Too Many. This 28-page book follows the adventures of a boy who receives a
new frog. His older pet frog engages negatively with the new frog. The older frog kicks
the younger frog into the water resulting in the disappearance of the younger frog. The
pictures end with an illustration of a reunion scene and its aftermath. Children told the
story to an adult experimenter who did not read the story with the child beforehand.
The children were told to ‘read the book by making up a story to go along with the
pictures’. The instructions were left deliberately vague to allow the child to tell the story
as he or she wished. Each child was asked if they had seen the story before; none had.
A few children were concerned that they could not read. They were reassured and told
that the book had no words. During the task, the child held the book and was free to
turn the pages at their own pace. Before telling the story the experimenter thanked the
child for helping them with their work. No other information was given to the child.
Test of emotion comprehension
After the narrative task, the children were tested on a standardized TEC (Pons et al.,
2004). The TEC assesses emotional understanding of 3- to 10-year-olds by presenting
vignettes in which a gender-matched protagonist encounters situations eliciting
different emotional responses. After each vignette, the child is asked to identify one of
four ways the protagonist feels in nine different situations, namely: (1) represented by
faces, (2) caused by external circumstances, (3) involving situations in which feelings
result from desires, (4) stemming from a character’s false belief, (5) elicited by
reminders, (6) when a protagonist tries to control an emotion, (7) hidden,
(8) conflicting, and (9) resulting from self-restraint.
Coding and scoring
Numbers of word tokens were calculated to obtain narrative length. The narrativeswere
coded along two dimensions of storybook narration – emotion understanding and
perspective-taking ability – utilizing the six-step process described below.
Emotion understanding: Thematic understanding
The first step of the coding process examined children’s thematic understanding of the
story, i.e., whether or not, and to what degree, children were able to attribute jealousy in
their narratives. Children were given credit for understanding a variety of negative
6 Naomi J. Aldrich et al.
emotions, cognitions, and behaviours related to jealousy, even if they never labelled
jealously explicitly. Based on jealousy as referring to the feelings, thoughts, and
behaviours that occur when a person believes an important relationship is being
threatened by a rival, each child received points along a continuum of their
understanding of jealousy that were combined to generate an overall Thematic
Understanding score. Children received one point for a basic negative emotion
identified (e.g., angry, sad, dislike, grumpy, unhappy, upset, stroppy) with points
ranging from 0 to 5. If the child explicitly labelled the complex emotion of jealousy, they
received the full five points. Next, children received an additional point for the
identification of the rival in the story (e.g., the new frog). A point was received for
indicating a behavioural reaction indicative of jealousy (e.g., the big frog went off by
himself in the forest). Another point was received for referencing a mental state or
cognition relative to the big frog’s jealousy (e.g., the big frog thought that the boy would
no longer love him). Children also received an additional point for acknowledgement
that the new frog interferes with the big frog’s relationship with the boy. Thus, overall
thematic understanding scores could range from 0 to 9.
Perspective-taking ability: Narrative coherence
The second step of the coding process focused on computing scores for the presence or
absence of a particular story feature. Each story was coded along five aspects
(i.e., events, elements, story structure, temporality and reference, and storybook
language) of story structure (adapted from Uchikoshi, 2005; Willenberg & Kang, 2001)
utilizing a checklist method where the child received a point for the inclusion of a
specific item, regardless of how many times she or he referenced it. The sum of points
received in a story structure category produced the score for that story structure aspect.
The scores from each of the five aspects were then added together to yield a total
combined narrative coherence score. Scores for the combined narrative coherence
measure could range from 0 to 28.
were considered to be significant for the story’s plot and conveying the characters’
actions. The range of possible scores for the event coding was from 0 to 8.
Eight narrative events were identified within One Frog Too Many that
reference to the new frog when it first appeared in the story, statement of the problem,
explanation of the problem, the climax, and the resolution of the story’s problem) were
coded. Possible scores for the element coding ranged from 0 to 5.
Next, children’s inclusion of the key elements of the story (i.e., the
abstract (e.g., giving the story a title), introduction to the story, character orientation,
setting orientation, event orientation, character delineation, and coda or conventional
story closing. Correspondingly, the range of possible scores for this category was 0 to 7.
This coding aspect focused on story structure features, such as:
Temporality and reference.
story character introduction comprised temporality and reference. Children received a
The sophistication of children’s use of connectives and
Perspective taking and jealousy7
score of zero for not using connectives in their story, a score of one for the use of basic
connectives (e.g., and, so, then), and a score of two for the use of more sophisticated
connectives (e.g., before, when, finally). Similarly, children received a score based on
the first introduction of a story character. If the child did not mention a story character
in the beginning of the story, they received a score of zero, a score of one indicated that
the child used an unspecified pronoun (e.g., he, they, one) to refer to the initial story
character, a score of two indicated that they used a presupposed reference utilizing a
definite article and a noun to refer to the first story character (e.g., the pets, the boy),
and a score of three was received if the child referred to the first story character through
use ofa non-presupposing introduction using a indefinite article and a noun or a number
(e.g., a dog, three pets). Temporality and reference scores could range from 0 to 5.
different types of language associated with storytelling. Specifically, children’s use of
quotes (direct or indirect), -ly adverbs, and conjoined phrases (noun, verb, or adverbial)
was examined with scores ranging from 0 to 3.
The final story structure aspect examined children’s use of three
Perspective-taking ability: Narrative evaluation
Building off of work by Bamberg and Damrad-Frye (1991), the next steps of the coding
process centred on coding for instances of evaluative speech that narrators use to
emphasize important events within the story for their audience. Given that one of the
main goals of the study was to address how different components of socio-cognitive
understanding interrelate, we chose to utilize various coding schemes identified in the
literature. In particular, given our primary interest in children’s emotion understanding,
our coding methodology sought to unpack the ‘frames of mind’ category of evaluation
used by Bamberg and Damrad-Frye. In this way, intellectual, perceptual, and emotions
could be examined separately to understand whether children used them in differing
Frames of mind.
(i.e., intellectual, perceptual, and emotive) were included based on frequency of
occurrence with scores calculated for each state term type and pooled to create an
overall Frames of Mind score. Based on Peterson and Slaughter (2006), children’s
narratives were coded for intellectual state terms or words denoting thinking,
knowledge, or mental ability (e.g., he thought, they knew); and for the inclusion of
perceptual terms, referring to characters obtaining information through their senses
(e.g., he heard, he looked). Finally, based on Cervantes and Callanan (1998), the
narrativeswere also codedfor emotion state words in whichthe child received credit for
each occurrence. Emotion state words included pleasure (e.g., chuffed), affection
(e.g., love), surprise (e.g., surprised), fear (e.g., scared), distress (e.g., gutted), concern
(e.g., worry), indifference (e.g., don’t care), anger (e.g., cross), dislike (e.g., hate),
provocation (e.g., stroppy), and jealousy (e.g., jealous). The emotion state words were
then coded as occurring within labels or explanations. Explanations were used with
causal information, but not necessarily with causal language (e.g., because), whereas
causal information was not provided with labels. Adaptation from Cervantes and
Callanan was chosen for this aspect of the frames of mind category because of the
differentiation between children’s labelling and/or explaining of emotions.
For the third step, children’s references to psychological state terms
8 Naomi J. Aldrich et al.
Supplementary evaluative speech.
instances of evaluative devices (adapted from Uchikoshi, 2005; Willenberg & Kang,
2001) based on frequency of occurrence. Included were devices of the type examined
by Bamberg and Damrad-Frye (i.e., character speech, negative, cause), such as: direct
speech (e.g., ‘Stop doing that’, they said.), direct speech markers (e.g., the turtle said
‘yikes’), negatives or defeats of expectations (e.g., they couldn’t find the little frog), and
causal markers (e.g., the little frog fell in the water because the big frog kicked him).
Also included were: intensifiers (i.e., adverbials such as very, really, almost, right, as in
The frog almost got left behind), adjectives (e.g., the little frog, the raft was flat),
references to emotional behaviour (e.g., the little frog was crying), references to
physical states (e.g., the frog was hurt), and words with high evaluative content
(i.e., evaluative words other than adjectives and verbs that have emphatic intrinsic
meaning, such as Wow! or The little frog went crashing into the water).
Fourth, the narratives were coded for additional
Narrator perspective taking.
taking ability not previously used in the literature, in the fifth step children’s narrative
utterances were coded for attribution of subjective or objective points of view based on
their frequencyofoccurrence (based on Wiebe, 1994). Specifically, each sentence of the
narrative was coded for its type of perspective or point of view. The subjective point of
view occurred when the child portrayed the private states of a character or characters,
whereas the objective point of view occurred when the child narrated the fictional
events without reference to a character’s viewpoint. The subjective point of view
utterances were further coded along three dimensions of attribution of a character’s
psychological point of view, corresponding to the frames of mind category above:
intellectual, perceptual, or emotive. For an utterance to be coded as subjective, the child
needed to refer to the character as being the experiencer of the state (e.g., the new frog
is jealous or he was angry), rather than the child as narrator inserting himself or herself
as the evaluator (e.g., the boy looks happy or the baby frog seems frightened). The
subjective intellectual occurred when the child referred to a character as having an
intellectual private state such as one believing, knowing, or wondering something.
Subjective perceptual utterances were coded when the child referred to a character as
perceiving someone or something (e.g., seeing, hearing). Subjective emotive
utterances occurred when the child referred to a character as experiencing or having
had experienced an emotion (e.g., he hates the new frog). The subjective and objective
point-of-view codes were considered mutually exclusive of each other, with the
subjective point of view taking priority over the objective point of view. The subjective
intellectual, subjective perceptual, and subjective emotive codes were not mutually
exclusive and a child was given credit for each occurrence.
To include an additional assessment of perspective-
work by Bamberg and Damrad-Frye (1991), the final step of the coding process noted
where in the episodic structure of the story references to emotion labels and emotion
To further assess children’s use of narrative evaluation, following
Intercoder reliability was attained separately for the six steps of coding and was
evaluated with kappa coefficients. Reliability for each coding step was attained by the
Perspective taking and jealousy9
first author and a trained research assistant independently coding 20–30% of the data
set, apart from the emotion expression coding, which was attained by the second and
fourth authors. For each coding scheme, we interpreted the kappa coefficient according
to Landis and Koch (1977): poor agreement (k , 0), slight agreement (k ¼ :0–:20), fair
agreement (k ¼ :21–:40), moderate agreement (k ¼ :41–:60), substantial agreement
(k ¼ :61–:80), and almost perfect agreement (k ¼ :81–1:00). For each coding step, all
disagreements between the coders were resolved through discussion.
The two coders reached substantial agreement (k ¼ :76) when coding for thematic
understanding. As far as the fiveaspects of narrative coherence, almost perfect reliability
was achieved for story events (k ¼ :88) and story elements (k ¼ :82), perfect agreement
for story structure features (k ¼ 1:0), and almost perfect agreement for temporality and
reference (k ¼ :84) and storybook language (k ¼ :79). Almost perfect agreement was
attained for the frames of mind coding scheme (intellectual, k ¼ :87, perceptual,
k ¼ :86, specific emotion words, k ¼ :89; emotion labels, k ¼ :83; emotion
explanations, k ¼ :83). When coding for the use of supplementary evaluative speech,
the coders attained substantial agreement (k ¼ :72) and almost perfect agreement was
obtained for the narrator perspective-taking coding scheme (k ¼ :92). When coding for
the placement of emotion expression (emotion labels, k ¼ :91; emotion explanations,
k ¼ :85) almost perfect agreement was reached.
Test of emotion comprehension
Following the standard TEC scoring procedure, children received a point when they
answered correctly for each of the nine different components of emotion assessed. Each
child received an overall TEC score that was computed by adding the points together
from the nine components of emotion with possible scores ranging from 0 to 9.
We conducted five sets of analyses. First, we examined age differences1across measures
using a series of one-way ANOVA models. Second, we examined relations between
measures through correlational analyses. Third, we investigated predictors of children’s
perspective taking, narrative coherence, and thematic understanding through a series of
multiple regression analyses. Fourth, we examined children’s placement of emotion talk
within the episodic structure of the story using chi-square analyses (Bamberg &
Damrad-Frye, 1991). Finally, to more fully evaluate children’s understanding of jealousy,
we conducted qualitative analyses of children’s thematic understanding of the story.
Table 1 includes the means and standard deviations for each dependent variable, along
with the results of one-way ANOVAs comparing older and younger children. We provide
eta-square estimates to indicate the proportion of variance accounted for by age group:
values between .01 and .09 indicate a small effect size, values between .09 and .24
1Initially, we also explored gender differences within each set of analyses. Given that there were no effects for sex, we dropped
sex from all subsequent analyses.
10 Naomi J. Aldrich et al.
indicate a medium effect, and values greater than .25 indicate a large effect (Cohen,
1988). Older children produced significantly longer narratives and scored higher on the
measures of emotion understanding as measured by the TEC and thematic
understanding than younger children. Similarly, older children scored higher on the
measures of perspective-taking ability than their younger counterparts. For instance,
children’s scores on the combined narrative coherence measure showed a significant
advantage for older children.2As far as including references to frames of mind, older
children used more intellectual, perceptual, and overall mental state terms than did
An overall age effect was also found for supplementary evaluative speech, with older
children using more of these evaluative devices than younger children. Specifically,
older children used significantly more intensifiers, adjectives, and negatives than did
Table 1. Mean scores by age group (standard deviations in parentheses)
(N ¼ 23)
(N ¼ 24)
Test of emotion comprehension
Frames of mind
Overall frames of mind
Supplementary evaluative speech
Highly evaluative content
Direct speech markers
Overall Supp. evaluative speech
Narrator perspective taking
*p , :05;**p , :01;***p , :001.
2For brevity, we report only the combined narrative coherence score, as older children scored significantly higher on all five
aspects of story structure (events, elements, story structure features, temporality and reference, and storybook language), and
the five measures correlated significantly with each other.
Perspective taking and jealousy11
younger children. In addition, older children discussed emotional behaviour
significantly more often than the younger children and incorporated more highly
evaluative content (e.g., Wow! or Bang!) in their stories than younger children. There
were no significant differences between younger and older children in their references
to physical states, causal markers, direct speech, or direct speech markers.
Comparisons of children’s use of narrator perspective markers indicated that older
children marked a subjective point of view within their narratives significantly more
often than younger children. Whereas older children provided more instances of the
subjective intellectual point of view than their younger counterparts, there were no
significant differences between the age groups in their use of the subjective emotive
point of view or subjective perceptual point of view. Comparison of children’s usage
of the objective point of view also revealed no significant difference between the
Correlations between emotion understanding and perspective-taking ability
We examined correlations between TEC scores, thematic understanding, perspective-
taking ability (i.e., narrative coherence, overall frames of mind, overall supplementary
evaluative speech, overall use of subjective point of view) while controlling for narrative
length (word tokens; see Table 2). After Bonferroni correction (alpha level set at .003),
children’s emotion understanding (TEC) scores correlated significantly with their
narrative coherence. Narrative coherence correlated with supplementary evaluative
speech. As expected, given the overlapping nature of the content, frames of mind was
found to be significantly related to children’s overall use of the subjective point of view.
Predictors of children’s subjective perspective taking, narrative coherence, and thematic
A simultaneous regression analysis, which determines the contribution of each
predictor over and above all the other predictors, was conducted to predict children’s
subjective perspective-taking ability. We included three variables as predictors:3age,
TEC score, and overall use of supplementaryevaluativespeech. The results are shown in
Table 3. Supplementary evaluative speech was highly predictive of children’s use of
Table 2. Correlations between children’s emotion understanding and perspective-taking ability (after
partialling out effects of narrative length and after Bonferroni correction; N ¼ 47)
2. Thematic understanding
3. Narrative coherence
4. Frames of mind
5. Supplementary evaluative speech
6. Subjective perspective taking
*p , :003.
3Because instances of subjective perspective taking corresponded to children’s use of mental state terms also coded within the
frames of mind category, individual or overall frames of mind utterances were not included as predictors.
12 Naomi J. Aldrich et al.
subjective perspective taking; the children who used more evaluative devices in their
narratives also produced more instances of subjective perspective taking.
Next, we conducted a simultaneous regression analysis with narrative coherence as
the outcome variable, and with age, TEC score, supplementary evaluative speech, and
subjective perspective-taking ability as predictors. As shown in Table 3, age and
supplementary evaluative speech jointly accounted for 57% of the variance in narrative
Finally, we conducted a simultaneous regression analysis with thematic under-
standing as the outcome variable, with TEC score, subjective perspective taking, and
narrative coherence as predictors. Given that age and supplementary evaluative speech
strongly predicted narrative coherence, we did not enter age or supplementary
evaluative speech into the model. Both subjective perspective-taking ability and
narrative coherence were positively related to children’s thematic understanding,
together accounting for 48% of the variance in thematic understanding.
Children’s episodic placement of emotion expressions
Next, we explored children’s placement of emotion expression within the episodic
structure of the story (see Appendix). We investigated how well the children’s
placement of emotion labels and emotion explanations corresponded to the
multinomial distribution of expected frequencies of occurrence for each type of
emotion talk. Based on previous work by Bamberg and Damrad-Frye (1991), we
calculated the expected frequencies for each type of talk per episode based on the
number of pictures in the episode sequence, divided by the total number of pages in the
book, and multiplied by the total sum of occurrence for type of talk by age group.
Initial analyses revealed that the distribution of emotion labels in the story
significantlydiffered from theexpected distribution for younger,x2ð5;N ¼ 23Þ ¼ 11:93,
p , :05, and older children, x2ð5;N ¼ 24Þ ¼ 34:32, p , :001. We subsequently
conducted individual episode chi-squares to reveal which episodes contributed to the
significant differences found between the observed distributions and the expected
Table 3. Standardized regression coefficients obtained from multiple regressions with subjective
perspective taking, narrative coherence, and emotion understanding (thematic understanding) as
outcome variables and age, emotion understanding (TEC), supplementary evaluative speech, subjective
perspective taking, and narrative coherence as predictor variables
Supplementary evaluative speech
Subjective perspective taking
*p , :05;**p , :01;***p , :001.
Perspective taking and jealousy13
The main difference between the two age groups in their emotion labelling occurred
in the Prelude to the story. Older children tended to label emotions as the story opened
(e.g., the boy is happy that he has a present), x2ð5;N ¼ 24Þ ¼ 13:14, p , :05, whereas
younger children did not (see Figure 1). Younger children largely neglected Episodes 2
(i.e., the big frog kicks the new frog and is punished), and 3 (i.e., after being punished,
the big frog sneaks back to the group and continues being mean to the new frog),
x2ð5;N ¼ 23Þ ¼ 5:12, ns, and tended to emphasize the ‘happy ending’ at the
Completion of the story, x2ð5;N ¼ 23Þ ¼ 4:15, ns. Similarly, older children chose to
disregard Episodes 2 and 3, x2ð5;N ¼ 24Þ ¼ 5:79, ns, as well as Episode 4 (i.e., the big
frog continues to bully the new frog and the new frog is lost), x2ð5;N ¼ 24Þ ¼ 4:02, ns,
and also chose to focus their emotion labelling on the ‘happy ending’ Completion
episode, x2ð5;N ¼ 24Þ ¼ 10:23, ns.
Older children’s placement of emotion explanations differed significantly from the
expected distribution, x2ð6;N ¼ 24Þ ¼ 30:20, p , :001 (see Figure2), whereas younger
children’s placement did not, x2ð6;N ¼ 23Þ ¼ 6:95, ns. In other words, older children
were more selective than younger children in their placement of emotion explanations.
In examining the pattern of placement that contributed to the overall significant
difference from the expected distribution, older children provided fewer emotion
explanations during Episode 2 (i.e., the big frog kicks the new frog and is punished),
x2ð6;N ¼ 24Þ ¼ 4:93, ns, and instead chose to focus their explanations on Episode 1
(i.e., the big frog first bites the new frog), x2ð6;N ¼ 24Þ ¼ 6:89, ns, Episode 5
(i.e., supporting characters react to the loss of the new frog), x2ð6;N ¼ 24Þ ¼ 6:89, ns,
and the ‘happy ending’ Completion episode, x2ð6;N ¼ 24Þ ¼ 4:87, ns.
Qualitative analysis of thematic understanding
Qualitative comparisons of the narratives revealed that older children (19.15%) tended
to provide an emotional theme more often than the younger children (6.38%), a
difference that approached significance, x2ð1;N ¼ 47Þ ¼ 3:70, p ¼ :06. Further
qualitative analyses based on children who provided an emotional theme revealed
that older children were the only ones to explicitly reference the complex emotion of
jealousy (i.e., 25% of the older children who identified an emotional theme talked about
Prelude E1E2 E3 E4 E5Completion
Placement of emotion labels
Figure 1. Percentage of children’s total references to emotion labels used per story episode per age
group (relative to total number used per age group).
14 Naomi J. Aldrich et al.
jealousy). Older children also mentioned unhappiness/sadness, annoyance (grumpy, Download full-text
stroppy), anger, and dislike of new frog as emotional themes. Younger children, in
contrast, referred solely to anger as the theme of the story. Furthermore, of the older
children who provided an emotional theme, 33.3% provided a secondary emotional
theme within their narratives (e.g., unhappiness and annoyance, anger and sadness,
anger and dislike).
Finally, we examined whether older and younger children differed in mentioning any
of the individual elements of the thematic understanding score. These analyses revealed
an age difference in children’s identification of the rival in the story, with significantly
more of the older children referencing the new frog as an adversary than their younger
counterparts, x2ð1;N ¼ 47Þ ¼ 5:12, p , :05. No other comparisons were significant.
The present study examined relations between emotion understanding, perspective-
taking skill, and children’s ability to narrate a story about jealousy. We utilized measures
of emotion understanding, perspective taking, and narrative ability developed by other
researchers (Bamberg & Damrad-Frye, 1991; Cervantes & Callanan, 1998; Pons et al.,
2004; Uchikoshi, 2005; Willenberg & Kang, 2001) and examined how these measures
related to each other, and to children’s ability to convey the jealousy theme of the story.
Our goal was to obtain a more thorough conception of how narrative abilities reveal
children’s unfolding grasp of complex emotions during middle childhood.
Differences in perspective-taking ability and narrative skill
With age, children’s narratives increased in length, a finding that is consistent with
previous research (Reilly, 1992). Simultaneously, children’s abilities to provide coherent,
elaborated, and emotionally strategic stories also improved. The narratives of 7- to
8-year-olds were more detailed than those of 5- to 6-year-olds, with older children
referring to a greater number of story elements and events, incorporating more story
structure features, storybook language, and references to time than the younger
children. Olderchildren told more complex stories containing more evaluative language
PreludeE1 E2 E3E4E5 Completion
Placement of emotion explanations
Figure 2. Percentage of children’s total references to emotion explanations used per story episode per
age group (relative to total number used per age group).
Perspective taking and jealousy 15