Emotion Word Comprehension from 4 to 16 Years Old: A Developmental Survey

Background: Whilst previous studies have examined comprehension of the emotional lexicon at different ages in typically developing children, no survey has been conducted looking at this across different ages from childhood to adolescence. Purpose: To report how the emotion lexicon grows with age. Method: Comprehension of 336 emotion words was tested in n = 377 children and adolescents, aged 4–16 years old, divided into 6 age-bands. Parents or teachers of children under 12, or adolescents themselves, were asked to indicate which words they knew the meaning of. Results: Between 4 and 11 years old, the size of the emotional lexicon doubled every 2 years, but between 12 and 16 years old, developmental rate of growth of the emotional lexicon leveled off. This survey also allows emotion words to be ordered in terms of difficulty. Conclusions: Studies using emotion terms in English need to be developmentally sensitive, since during childhood there is considerable change. The absence of change after adolescence may be an artifact of the words included in this study. This normative developmental data-set for emotion vocabulary comprehension may be useful when testing for delays in this ability, as might arise for environmental or neurodevelopmental reasons.

Whereas the early development of emotion and mental state vocabulary is well documented, studies describing complex emotions corpora (Ortony et al., 1987;Storm and Storm, 1987;Johnson-Laird and Oatley, 1989) have not tested the development of emotion vocabulary in older children or adults. One study tested the development of complex (and basic) emotion labels in toddlers and pre-schoolers aged 18-71 months, using a list of 125 emotions (Ridgeway et al., 1985). However, this was not extended into primary school aged children and adolescents. The aim of the study reported here was to collect normative data from children and adolescents aged 4-16 years old, in order to survey which emotion and mental state terms are acquired earlier than others, and how the emotional lexicon grows with age. We used a large corpus of 366 emotion or mental state words.

SampleS
Data was collected from 6 age groups (4-6, 7-8, 9-10, 11-12, 13-14, and 15-16), corresponding with UK school year bands (Years 1-2, 3-4, 5-6, 7-8, and 9-10, respectively). Confirmation of age bands was checked using the Annual Schools Census for the academic year 2004/5, provided by the Schools Statistical Unit at the Department for Education and Skills, UK. Participants were all attending mainstream education in the Cambridgeshire area (UK), and included a mix of both state and private schools, in order to attempt to achieve a representative sample of the community. The total sample was therefore n = 377, and they comprised an equal number of males and females. There were n = 30 age 4-6 years old, n = 34 age 7-8 years old, n = 26 age 9-10 years old, n = 87 age 11-12 years old, n = 134 age 13-14 years old, n = 66 age 15-16 years old.
The standard view is that there are only a small number of basic emotions that are recognized cross-culturally: happy, sad, angry, afraid, surprised, and disgusted. All other emotions -often labeled "complex emotions" -are either treated as being blends of the basic 6 or as being culture specific (Ekman and Friesen, 1969;Prinz, 2004). When it comes to developmental understanding of emotional and mental state terms, whereas the development of basic emotion understanding has been extensively studied, only a small number of complex emotions (e.g., jealous, proud, disappointed, embarrassed) have been studied developmentally (Harris, 1989;Herba and Phillips, 2004). The study reported here aims to examine a much broader range of emotions and mental states in terms of when children and adolescents understand the words for each of these.
Children's abilities to label emotional expressions start developing from as early as 2 years of age (Izard and Harris, 1995;Denham, 1998). Bretherton and Beeghly interviewed mothers of 28 month old children about their vocabulary, and found that over 60% of the children were familiar with the emotional terms happy, scared, and mad (i.e., angry), and were able to use them in their language. Sad was known to more than 50% of the children at this age, but more than 80% of them knew and used emotion words such as like and love (Bretherton and Beeghly, 1982). Another study reported that more than 75% of 3 year olds use emotion words for feeling good, happy, sad, angry, loving, mean, and surprised (Ridgeway et al., 1985). Toddlers are more likely to first use emotion labels to refer to their own emotional state (at about 20-24 months). Later, such labels are used with reference to others' emotions, and by 3-3.5 years of age, children will use emotion terms for imaginary characters or in reference to those in a story (Reilly et al., 1990).
Emotion word comprehension from 4 to 16 years old: a developmental survey has been found to be reliably correlated with comprehension tested in the lab. Only those items endorsed as Clearly Understood were judged to be within the comprehension of an individual. Finally, words where data was not available from all six age groups were not included in the results, since this was a developmental study 1 .

ReSultS
n = 336 emotion words met all inclusion and exclusion criteria, and these are listed in Table 1. We analyzed the percentage of words understood by 75% or more of each group. Seventy-five percent was selected because the Binomial test indicated that the probability of achieving this level by chance was p < 0.001. Of the list of 336 emotion words, 41 words were understood by at least 75% of 4-6 year olds, 88 by 7-8 year olds, 180 by 9-10 year olds, 299 by 11-12 year olds, 320 by 13-14 year olds, and 330 by 15-16 year olds. Thus, between 4 and 8 years old, the number of emotion terms understood doubles. Between 9 and 12 years old, this number doubles again. However, between 13 and 16 years old, emotion vocabulary size hardly changes at all, suggesting the earlier dramatic changes in emotional vocabulary comprehension reach a plateau by early adolescence.
These changes between age-bands are not only apparent by visual inspection (see Figure 1) but are confirmed by statistical analysis. A goodness of fit test conducted for the whole sample revealed a significant difference in the proportion of words recognized by at least 75% of participants in each age band (χ 2 [5] = 999.14, p < 0.0001). Next, five goodness of fit tests were conducted between each adjacent age-band. Using the Bonferroni correction for multiple testing, we chose a p value of <0.002 as a conservative level of significance, to control for the number of comparisons being conducted. The analysis revealed significant differences between 4-6 year olds and 7-8 year olds (χ 2 [1] = 21.19, p < 0.0001), between 7-8 year olds and 9-10 year olds (χ 2 [1] = 52.53, p < 0.0001), between 9-10 year olds and 11-12 year olds (χ 2 [1] = 102.94, p < 0.0001). However, between 11-12 year olds and 13-14 year olds, the difference in number of mental state terms understood by at least 75% of each age-band did not meet our level of significance (χ 2 [1] = 9.03, p < 0.005), and nor did the difference between 13-14 year olds and 15-16 year olds (χ 2 [1] = 4.7, p < 0.05). These results confirm that the size of emotional vocabulary increases significantly with age during childhood, but does not develop significantly after adolescence. This pattern is illustrated in Figure 1.
In order to examine whether development of the emotion vocabulary varies for different emotion categories, the 336 words included in this survey were allocated to the 24 emotion categories, included in the Mind Reading DVD (Baron-Cohen et al., 2004 1 ). Table 2 shows the percentage of words in each emotion category that at least 75% of children understood in each age-band[OG3]. This shows the same developmental trend, of major change across the first 3 age-bands and a leveling off by early teens, especially for the more common (or basic) emotion groups (e.g., happy, sad, afraid, kind). On the other hand, the understanding of words from some emotion groups is more delayed (e.g., disbelieving, touched, romantic), and understanding of words from others (e.g., bored) is incomplete even in the highest age band. This may reflect an ongoing development of the emotional lexicon beyond adolescence years.

method
A search for all "emotion" terms in the English language was conducted, using the electronic thesaurus in Microsoft Word.

WoRd Selection
Inclusion criteria 1. A word was considered to be an emotion term if it described a mental state with an emotional dimension, that could be preceded by the phrase I feel x, or he/she looks x, or he/ she sounds x. This definition was used because this study was linked to the development of an encyclopedia of emotion video or audio clips of emotions, portrayed by actors. The collection of these recordings is available on a separate piece of software called Mind Reading (Baron-Cohen, Golan, Wheelwright, and Hill, 2004 1 ; Baron-Cohen et al., 2003). 2. Mental states that were epistemic but which nevertheless had an unambiguous emotional dimension were included (e.g., doubting).

Exclusion criteria
1. Mental states that could be a purely bodily state (e.g., thirsty) with no necessary mentalistic dimension were excluded, since this was a study of emotion terms. 2. Slang words (e.g., chuffed) were excluded, in order to only collect normative data that could be useful to a broad range of English-speaking cultures. 3. Swear words (e.g., pissed off) were excluded, again because these vary considerably from one sub-culture or sub-dialect to another. 4. Words which the authors judged to be too difficult for all six age-groups were also excluded (e.g., schadenfreude, a word borrowed from German but which appears in the English language and which depicts the emotion of pleasure that one feels at someone else's misfortune). 5. Epistemic states with no emotional dimension (e.g., reasoning) were excluded, since this was a study of emotion terms rather than all mental state terms.

population SuRvey
Every emotion term was listed in a vocabulary checklist. Groups of volunteers from the general population (see "Participants" above) were asked if they knew the meaning of each word. The volunteers ranged from age 4 to 16 years old and were stratified by age. For individuals aged 12-16, self-report was accepted. For individuals age 4-11 years old, parent-or teacher-report was collected. This questionnaire survey method invited the participant to choose one of three response options for each word: Clearly Understood, Not Understood, and Possibly Understood. This method (parent-or teacher-report) was used in the development of the Communication Development Inventory (Fenson et al., 1994) and

diScuSSion
Comprehension of 336 emotion words was assessed using either self-report (from teenagers) or parent-or teacher-report (for younger children), ranging from 4 to 16 years old. The aim of the study was to collect normative data to document when different emotion terms are understood in development, and to measure differences in the proportion of each age group who understand each term. This was undertaken because of gaps in knowledge in how such competence grows with age. Results confirm that complex emotion vocabulary develops dramatically during childhood, doubling in size every 2 years between 4 and 11 years old, but that between 12 and 16 years old there was no significant increase in the number of emotion words understood. Apart from measuring differences between age-bands, this study also provides detailed data on each emotion-term so that these can be ordered according to level of difficulty, for future studies to exploit. The finding that there is no significant development in this ability after adolescence may be an artifact of three factors. First, that we set a conservative probability level (p < 0.002) to control for multiple-testing, and the differences between the adolescent age groups, whilst meeting more relaxed criteria (p < 0.05), did not reach these levels. Secondly, we limited the word list to 336 words, and it may be that if a longer word list had been used, with more difficult words, changes between 12 and 16 years old would also have become evident. Lastly, since comprehension of the emotions was judged by different respondents in the younger vs. the older age bands, the lack of significant increase in emotion vocabulary comprehension amongst the adolescents could be related to their reluctance to admit they were not familiar with some emotion words. For these reasons, we do not wish to draw any conclusions about underlying mechanisms (brain development, hormonal changes, etc.) at the onset of adolescence having any special role in the rate of development of comprehension of the emotional lexicon.
One obvious limitation of this study is that the data was collected by self-report (in the case of 12-16 year olds) and parent/ teacher report (in the case of children aged 4-11 year olds). There is thus no independent confirmation of comprehension, such as a performance test. This would not have been feasible in a large-scale study of this kind, but the reliability of self or parent/teacher report could be tested in future studies. We chose parent-or teacherreport for the pre-adolescent years because other instruments (such as the McArthur Checklist for Vocabulary Development) have used this method and shown that parent-report correlates well with actual performance measures of language comprehension (Fenson et al., 1994). For this reason, we suggest that the data reported here is likely to be reliable, but this should be tested formally using laboratory measures. Future studies could also look at effects of sex (with larger samples), word frequency, and emotion valence.
Secondly, this study only assessed the comprehension of emotion words. Ongoing studies are using this corpora to test the ability to recognize complex emotions and mental states from video-clips of faces and audio-clips of voices, to document Disbelieving (  how recognition of emotions from facial or vocal expressions develops with age (Golan et al., 2006;Golan and Baron-Cohen, submitted). We conclude first that this study documents significant changes in rate of emotional lexicon comprehension between 4 and 11 years of age, which should be taken into consideration by researchers and educators when designing developmentally sensitive tests and teaching materials. Whereas at age 4-6 years old only 41 emotion words are understood by more than 75% of children, by 11 years old children understand 299 such words. Secondly, this study allows emotion terms to be ordered developmentally with respect to their level of difficulty.
Finally, the normative data reported here may be useful not only for studies of typical emotion comprehension but also against which to assess emotion-comprehension delays or deficits that may arise either for environmental reasons (e.g., following neglect or abuse; Pollak et al., 2000), or for neurodevelopmental reasons (e.g., as occurs in conditions such as autism and Asperger Syndrome). Although individuals on the autistic spectrum may not show deficits in basic emotion recognition (Loveland et al., 1997;Grossman et al., 2000), they do have deficits in recognizing complex emotions (Yirmiya et al., 1992;Baron-Cohen et al., 2001). The use of a graded developmental scale of emotion comprehension may help reveal such deficits in clinical populations, should they exist, and enable researchers to quantify the degree of developmental delay in emotion lexicon comprehension relative to typical agematched controls.