Food Quality and Preference 49 (2016) 119–128

Contents lists available at ScienceDirect

Food Quality and Preference

journal homepage: www.elsevier.com/locate/foodqual

Use of emoticon and emoji in tweets for food-related emotional

expression
Leticia Vidal a,⇑, Gastón Ares a, Sara R. Jaeger b
a
Instituto Polo Tecnológico de Pando, Facultad de Química, Universidad de la República, Gral. Flores 2124, C.P. 11800 Montevideo, Uruguay
b
The New Zealand Institute for Plant & Food Research Limited, 120 Mt Albert Road, Private Bag 92169, Victoria Street West, Auckland, New Zealand

a r t i c l e i n f o a b s t r a c t

Article history: Emotional responses to food and beverages has been established as a significant research topic within
Received 17 August 2015 sensory and consumer science. The current research contributes to this activity by building new insights
Received in revised form 30 November 2015 regarding consumers’ spontaneous expressions of food-related emotional experiences. This was done by
Accepted 2 December 2015
analysing 12,260 tweets about breakfast, lunch, snack and dinner eating situations, previously retrieved
Available online 8 December 2015
by Vidal et al. (2015). A descriptive approach was adopted, wherein focus was direct to capturing fre-
quency and diversity of emoticon and emoji use. It was found that consumers express a wide range of
Keywords:
positive and negative emotions and that emoticon and emoji use is tailored to the content of the tweets.
Twitter
Emoji
Emoji were used more frequently than emoticons to express emotions. While it was rare for tweets to
Emoticons include more than one emoticon or emoji, their use was almost exclusively in addition to other content
Emotion of the tweet. Our results suggest that emoji and emoticon seem to be an easy and intuitive way to express
Consumer research emotions in a food context. This could represent an opportunity for development of non-verbal subjective
methods to measure food-related emotions.
Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction
Emotions, defined as short-term affective responses to the
appraisal of stimuli with reinforcing potential, contribute to the
control of basic human behavioural systems (Frijda, 1986;
Gibson, 2006). Food and emotions have been reported to share a
bi-directional relationship: on the one hand, emotions can shape
food choice, food intake and liking; while, on the other hand, food
consumption can influence consumers’ mood and emotions
(Canetti, Bachar, & Berry, 2002; Macht, 2008).
Studying emotional responses to food and beverages has
become a major research interest within sensory and consumer
science in recent years and has prompted the development of
methodological approaches that seek product discrimination, inde-
pendently of traditional hedonic responses (Meiselman, 2015).
The most common approach for studying food-elicited emo-
tions has focused on explicit emotions, which are consciously per-
ceived and, therefore, can be directly reported by consumers
(Köster & Mojet, 2015). Several food-related emotion question-
naires have been developed by researchers in the last ten years
by reviewing emotion lists or by directly asking consumers to state
⇑ Corresponding author.
E-mail address: lvidal@fq.edu.uy (L. Vidal).
how they feel when consuming specific products (Chrea et al.,
2009; Ferrarini et al., 2010; Gmuer, Nuessli Guth, Runte, &
Siegrist, 2015; King & Meiselman, 2010; Ng, Chaya, & Hort, 2013;
Porcherot et al., 2010; Spinelli, Masi, Dinella, Zoboli, &
Monteleone, 2014). These questionnaires typically include 25–39
emotion words and have been used to characterise a wide range
of product categories (Meiselman, 2015).
Although consumers have been reported to find these emotion
questionnaires easy and intuitive, some of them think it is an odd/
weird task (Jaeger, Cardello, & Schutz, 2013). This suggests that the
use of emotion questionnaires can lead to demand characteristic
bias, encouraging consumers to select emotional terms that are
cognitively associated with the products, even if they are not actu-
ally experiencing them before, during or after consumption
(Thomson & Crocker, 2015).
Therefore, a need exists for research on how consumers sponta-
neously express food-related emotions in their daily life. Deter-
mining if people actually report explicit emotions before/during/
after food consumption can contribute to increasing the ecological
validity of self-reported emotion measurements and demonstrate
the real contribution of these methodologies for explaining con-
sumers’ food choices.
Internet, and particularly social media, represents an opportu-
nity to obtain spontaneous consumer information elicited in real-
life situations (Citrin, Stem, Spangenberg, & Clark, 2003). Twitter

http://dx.doi.org/10.1016/j.foodqual.2015.12.002
0950-3293/Ó 2015 Elsevier Ltd. All rights reserved.
120 L. Vidal et al. / Food Quality and Preference 49 (2016) 119–128
is one of the most popular social media, enabling users to send and
receive text-based messages of up to 140 characters, called tweets,
which can include photos and videos (Barash & Golder, 2010).
Twitter has recently attracted interest from marketing and con-
sumer science researchers (Carr et al., 2015; Fried, Surdeanu,
Kodbourov, Hingle, & Bell, 2014; Ghiassi, Skinner, & Zimbra,
2013; Worch, 2014). The use of the internet for studying how con-
sumers spontaneously express their food-related emotions could
have the potential to change the way in which emotion research
is carried out (Mostafa, 2014).
In a recent study, Vidal, Ares, Machín, and Jaeger (2015)
reported that tweets about eating (breakfast, lunch, snack and din-
ner) included information about what was consumed, when,
where, with whom, and why. These authors reported that con-
sumers included references to mood and emotions in approxi-
mately 25% of the tweets. However, words were less frequently
used than emoticons and emoji for this purpose, which suggests
that the use of these graphical characters to express food-related
emotions deserve further exploration.
Pictographs, such as emoticons and emoji, have been consid-
ered a partial substitute of standard language (Truss, 2004), as well
as an effortless and automatic way of expressing emotions (Cowie
et al., 2001). These symbols are basically abstractions of facial
expressions or bodily gestures, which have been developed to help
communicating emotions or mood in computer-mediated commu-
nications (Walther & D’Addario, 2001). Emoticons are basically
typographic displays created by alphanumerical characters, for
example. :) for happy and :(for sad (Wikipedia, 2015). Graphical
characters are also used to convey emotional expressions, and
these are called emoji, a Japanese word meaning ‘‘picture word”
(refer to tables/Supplementary material for exemplars). Research
has shown that emoticon and emoji are increasingly used in social
networks, blogs and other applications by males and females of dif-
ferent ages (Huang, Yen, & Zhang, 2014; Huffaker & Calvert, 2005;
Wolf, 2000).
Against this background, the present work aimed to better
understand how emoticons and emoji are used by consumers to
spontaneously express food-related emotional experiences. Focus
was placed on eating situations as context and the situational
appropriateness are expected to have a larger impact on emo-
tional reactions than products themselves (Köster & Mojet,
2015). This was done by further analysing tweets about breakfast,
lunch, snack and dinner, previously retrieved by Vidal et al.
(2015). Considering the paucity of information regarding food-
related emoticon use, a descriptive approach was adopted,
wherein focus was on capturing frequency and diversity of emoti-
con and emoji use. The latter considered the different emoticons
and emoji used in tweets, as well as associations between emoti-
con/emoji use and content of the tweets as it related to different
characteristics of the eating situations. It was beyond the scope of
this work to undertake a detailed comparison of emoticon and
emoji use across different eating situations. Results from the pre-
sent work are expected to provide insights on how consumers
spontaneously express emotional reactions to eating and drinking
(and the products they consume) and to support methodological
development in relation to emotion research within the field of
sensory and consumer research.
2. Methodology
2.1. Retrieval of the tweets and content analysis
The present work consists of a re-analysis of a study by Vidal
et al. (2015), who presented a thematic content analysis of tweets
about eating situations. This section summarises the process
followed by these authors for retrieval of tweets and content
analysis (see Vidal et al. (2015) for full details).
Tweets containing each of four English keywords – breakfast,
lunch, dinner, and snack – were retrieved using the twitteR package
(Gentry, 2014) of R software (R Core Team, 2013). Between 16,285
and 20,490 tweets for each eating situation were retrieved using
multiple searches during five working days in September 2013.
Repeated tweets and re-tweets were discarded, which led to
11,016–13,045 tweets being retained for content analysis in each
of the four eating situations.
Due to the time-consuming nature of manual content analysis,
this was performed on 4000 randomly selected tweets within each
eating situation. Using a process of inductive coding (Krippendorff,
2004), two coders fluent in English and with more than 2 years of
experience in consumer research classified the content of the
tweets (and accompanying pictures/videos). The final themes and
sub-themes (see Section 3 for details) were established by consen-
sus and their frequency of mention determined. According to its
content, each tweet was assigned to one or more themes. Tweets
in which the content was not related to the eating situation, that
were posted in languages other than English or tweets by compa-
nies/organizations were not considered for further analysis.
The total number of tweets included in this research was
12,260, roughly evenly distributed across the four eating
situations.
2.2. Identification and classification of emoticons and emoji in tweets
In this research further analysis of the tweets described above
was performed. Focus was directed towards tweets containing
emoticons, created with alphanumerical characters, or picto-
graphic emoji characters.
For alphanumerical emoticons, searches for 242 different
emoticons were performed, comprising the Western style emoti-
con list from Wikipedia (2015) and possible variations (e.g., the
emoticon (: was considered as a variation of :)). The Western list
of emoticons was used because only tweets in English were consid-
ered. For purposes of data simplification, the different alphanumer-
ical emoticons were grouped into 39 categories using the
classification listed in Wikipedia.
Searches for the 631 emoji characters available in Twitter were
also performed. For purposes of data simplification the emoji char-
acters were grouped into seven categories listed in the Emojipedia
(2015). Considering the descriptive focus of this research and com-
parison of emoticon and emoji use being of low priority, it was
considered acceptable to use existing classification schemes rather
than developing a unified classification scheme for both emoticons
and emoji. The twitteR package retrieves emoji characters as speci-
fic character codes, as explained by Vidal et al. (2015). For example,
using examples from Table 3, the emoji named smiling face with
smiling eyes is represented by the code í ½íŠ, while the code í ½í
corresponds to loudly crying face. In order to access the code for
all the 631 emoji characters, they were posted in the Twitter
account of one of the authors, and those tweets were retrieved
using the twitteR package.
The valence of emoticons and emoji was classified by the
authors into positive, negative and neither positive nor negative
according to the facial expression or gesture they intended to con-
vey and/or their written description (Emojipedia, 2015; Wikipedia,
2015). Published emotion classifications were considered (Ekman,
1994; Jiang, King, & Prinyawiwatkul, 2014; Laros & Steenkamp,
2005). Although interpretation of emoticons and emoji can differ
among people, it was beyond the scope of the present paper to
study these individual differences and therefore a single classifica-
tion was considered.
 L. Vidal et al. / Food Quality and Preference 49 (2016) 119–128 121

The complete lists of emoticons considered in this research and confidence that the results would generalise rather than being
their classification can be found in Tables 1 and 2 of the Supple- specific to one meal situation.
mentary material (emoji and emoticons, respectively). Frequencies of emoticon and emoji use were also calculated in
relation to tweet content, as categorised by the themes previously
2.3. Data analysis developed in Vidal et al. (2015). A chi-square test was used to iden-
tify significant differences in the percentage of tweets including
In accordance with the descriptive focus of this research, fre- emoticons/emoji among the different categories. A chi-square per
quency of use of each emoticon and emoji was determined by cell test was used to identify the source of variation of the global
counting the number of tweets that contained it. Frequencies were chi-square (Symoneaux, Galmarini, & Mehinagic, 2012).
calculated at the aggregate level and separately for each of the four
eating situations. Although differences by eating situation were not
3. Results
central to this research, these analyses were performed in increase
3.1. Overview of emoticon and emoji use in tweets
Table 1
Percentages and characteristics of the emoticons and emoji for the 12,260 tweets Table 1 presents a summary of emoticon and emoji use in the
analysed in this research. Average frequency and range (i.e., minimum and maximum
analysed tweets. As was reported for this data by Vidal et al.
percentages) across the four eating situations (breakfast, lunch, dinner and snack) are
shown. (2015), a significant proportion of the tweets studied here included
emoticons or emoji (24.0%). People tended to use emoji characters
Average Range
more frequently than emoticons (68.1% vs. 30.9%). Moreover, these
Type of pictograph* two types of pictograph emoticons were rarely used concurrently
Emoticon 30.9 20.9–44.9
in a single tweet (1%).
Emoji characters 68.1 54.8–78.4
Both 1.0 0.3–1.6 Regarding the number of different emoticons and emoji
included in the tweets, only about 1-in-4 contained more than a
Number of different emoticons or emoji*
1 76.4 74.1–79.4 single pictograph (Table 1). However, the use of multiple emoti-
2 15.2 12.6–16.6 cons or emoji was markedly more frequent for emoji than for
3 5.0 3.7–5.8 emoticons: 31.1% vs 4.8%, respectively. Besides, the maximum
4 or more 3.4 1.0–2.6 number of different emoticons and emoji characters included in
Maximum number of different pictographs included in an individual tweet the tweets was 4 and 16, respectively.
Emoticons 4 2–4 The tweets contained many different emoticons and emoji. As
Emoji characters 16 10–16
shown in Table 1, a total of 50 different emoticons were identified.
Total number of different pictographs identified in the tweets The range of expressions introduced by emoji characters was even
Emoticons 50 23–40
Emoji characters 254 116–161
higher: a total of 254 different characters were identified. Results
pertaining to the use of different emoticons and emoji are pre-
*
Within a column percentages sum up to 100. sented in Section 3.2.

Table 2
Exemplar tweets in which different types of emoticons and emoji were used to convey information not expressed using words or to emphasise written information.

Emoticons/emoji used to
Convey information not
expressed in words (92.4%)
Type of emoticon/emoji expression
( ⁄)
Positive emotional expression
(66.0%)
Negative emotional expression
(15.1%)
Neither positive nor negative
emotional expression (1.3%)
Not related to expression of mood or
emotions (17.6%)
Exemplar emoticon or emoji use in Tweets
Blue namiiiii for dinner
Breakfast in bed! :) #chill #fruitplatter #relaxed
@xxx let’s have a lunch date ;)
Ever since i work here i always skip breakfast. I have no time for breakfast :(
@xxx @xxx I’ll have to go dear! I’m going to lunch very late :/ ! Lol 5 pm bye! Take care!!
Just ate way more than a late night snack
Just now eating dinner
My snack is a salad...I’m soo damn healthy
I honestly don’t know how everyone’s going to have lunch at the same time like the pla-
zas will be so damn packed:|
Midnightsnack alone in bed
#lunch #kauai

SNACK http://t.co/9fAseivNYu
Emphasise information Positive emotional expression Good food for good mood. Happy breakfast!:D
expressed in words (7.6%) (75.2%) I love when my dad surprises me by making a huge dinner for me after work

My mom got taco bell for dinner ! YAYY !

Negative emotional expression
(11.2%)
Neither positive nor negative
emotional expression (1.3%)
Not related to expression of mood or
emotions (12.3%)
Life without milk is sad:(#breakfast
hate it when theres nothing to snack on in the house
@xxx yes. I miss having lunch with you
Still indifferent on curry chicken for dinner :S
Indifferent about this going away dinner
Chocolate cake for breakfast courtesy of @xxx
Lunch with luke
Very #decent day! Another win for @AthleticOadbyFC #TopOfTheLeague Sunday lunch
with white watch and 12 holes of golf. #SuperSunday

Usernames have been replaced by xxx due to privacy issues. (⁄) Percentages were calculated based on the number of individual emoticons and emoji, regardless of whether
they were included in the same tweet or not. A total of 12,260 tweets were analysed.
122 L. Vidal et al. / Food Quality and Preference 49 (2016) 119–128

Consideration of differences in emoticon and emoji use across Table 3

the four eating situations (breakfast, lunch, dinner and snack) was Frequency of use (>0.2% shown) of emoji for the ‘‘People” category. Emoji are
identified by names in accordance with Emojipedia (2015) and drawn from the list of
beyond the scope of this research. For this reason only the minimum 631 emoji available on Twitter. Average frequencies (based on 12,260 tweets) and
and maximum frequencies for eating situations were presented in range (i.e., minimum and maximum percentages) across the four eating situations
Table 1. These values make a small contribution towards general- (breakfast, lunch, dinner and snack) are shown.
ising the findings by showing that the patterns of emoticon and Category Emoji Average Range
emoji use were fairly consistent across the different types of eating
People 21.0 16.4–
situations. Readers interested in further details about differences in 24.2
the frequency of use of individual emoticons and emoji across 1.9 0.8–3.2
Face savouring delicious food
eating situations are referred to the Supplementary material.
Smiling face with heart-shaped eyes 1.6 1.2–2.1
Predominately, Twitter users included emoticons or emoji to
convey information which was not expressed using words. This Face with tears of joy 1.6 0.7–2.3
was observed in 92.4% of the analysed tweets (Table 2). Words 1.3 0.8–1.6
Ok hand sign
were used to describe the eating situation, and emoticons or emoji
Smiling face with smiling eyes 1.1 0.7–1.5
were used to express feelings about that specific situation. For
example, a brief tweet stating that the person posting the tweet Red heart 0.8 0.5–1.3
was eating dinner, followed by a smiley face emoji. Less frequently 0.7 0.2–1.2
Face throwing kiss
(7.6%), emoticons or emoji were used to stress information written
Grinning face with smiling eyes 0.7 0.6–0.8
in the tweets, such as the type of food being eaten or specific emo-
tions experienced in the eating situation. As shown in Table 2 an Thumbs up sign 0.7 0.6–0.7
example of this use was ‘‘life without milk is sad :(#breakfast”. 0.7 0.5–0.7
Unamused face
Table 2 also shows that emoticon and emoji were mainly used
Smiling face 0.6 0.4–0.7
to express positive emotional reactions/associations. Regardless
of whether emoticons and emoji were used to convey information Smirking face 0.6 0.5–0.7
not expressed in words or add emphasis to information expressed 0.6 0.3–1.0
Weary face
in words, more than 2-in-3 instances of emoticon and emoji use
Loudly crying face 0.6 0.2–1.1
was in connection with positive emotional expression. Emoticon
and emoji valence is considered further below in connection with Pensive face 0.5 0.3–0.7
description of the different emoticons and emoji used in the tweets. 0.4 0.2–0.6
Two hearts

Person raising hands in celebration 0.4 0.2–0.5

3.2. Diversity of emoticon and emoji use in tweets
3.2.1. Emoticon use for emotional expression in tweets
The strong tendency seen in Table 2 for emoticons and emoji to
be used in tweets to express positive rather than negative reac-
tions/associations (66.7% vs. 14.8%), was also seen in Tables 3
and 4. These tables present the frequency of use of different emoji
and emoticons, respectively. Considering only emoji from the cat-
egory ‘‘People”, which is the primary category for emotional
expression, Table 3 revealed that 9 of the 10 most frequently used
emoji expressed positive emotions (unamused face being the
exception). The most frequently used emoji was face savouring deli-
cious food (1.9%), which clearly conveys a positive hedonic reaction.
The other emoji in the top-5 most frequently used list (>1%: smiling
face with heart-shaped eyes, face with tears of joy, ok hand sign, and
smiling face with smiling eyes) also expressed positive reactions/
associations.
Regarding emoticons (Table 4), smiley and happy faces were the
most frequently used; in particular the simple smiley sign :) (3.5%).
This category of positive emoticons (i.e., ‘‘Smiley, happy face or
laughing”) was used more than five times more frequently than
the category ‘‘Frown, sad” (5.2% vs. 0.9%). In the latter category :
(was the most frequently used (0.7%).
Aside from the tendency to use emoticons and emoji to express
positive rather than negative reactions/associations, it was also
evident that there was greater diversity in positive than negative
emotional expression. Table 3 illustrates this well. In the list of
the most frequently used emoji (>0.2%), only 4 pertained to nega-
tive emotional reactions: unamused face, weary face, pensive face
and loudly crying face.
Emoticons and emoji related to neither positive nor negative
emotional reactions associations were used least frequently and
corresponded to 1.3% of the total number of emoticons and emoji
included in the tweets. Within this category, the most frequently
used emoji were expressionless face and face without mouth,
whereas the most frequently used emoticons were :| and :S.
Flushed face 0.3 0.1–0.4
0.3 0.1–0.3
Clapping hands sign
0.2 0.1–0.3
Winking face
Relieved face 0.2 0.0–0.3
0.2 0.2–0.2
Face with stuck-out tongue and wink
0.2 0.1–0.3
Victory hand
Frequencies of emoticon and emoji use for emotional expres-
sions were similar across the four eating situations. The emoji face
savouring delicious food is one example of the differences that did
exist. It was less frequently used in tweets related to lunch than
snack (0.8% vs. 3.2%: Table 3). Conversely, the emoji loudly crying
face was more frequently used in tweets relating to lunch than
tweets about dinner (1.1% vs. 0.2%: Table 3). With a frequency of
1.2%, the emoji face throwing kiss was more often used in
connection with dinner than snack and breakfast eating situations
(0.2–0.4%). It is beyond the scope of this research to cover these
similarities/differences in detail and interested readers are referred
to the Supplementary material.
For completeness, we note at the closing of this section, that
emoticons and emoji use was also related to consumption of speci-
fic products in the four eating situations. As exemplified in Table 5,
emoticons and emoji were used to convey positive hedonic reac-
tions to foods (e.g., :), face savouring delicious food) but also differ-
ent emotional reactions related to both positive emotions (e.g.,
smiling face with smiling eyes, smirking face) and negative emotions
(e.g., unamused face).
3.2.2. Use of emoji to communicate what was consumed, when, where,
with whom, and why
While this research is focused on exploring emoticon and emoji
use for emotional expression, the analysis revealed that 17.2% of
 L. Vidal et al. / Food Quality and Preference 49 (2016) 119–128 123

Table 4 ‘‘Activity”, ‘‘Travel & Places” and ‘‘Nature”. In these instances differ-
Frequency of use (%) of emoticons. Category names are in accordance with Wikipedia. ences in emoji use across eating situations were more noticeable
Average frequencies (based on 12,260 tweets) and range (i.e., minimum and
maximum percentages) across the four eating situations (breakfast, lunch, dinner
than those pertaining to emotion expression. For example, in the
and snack) are shown. ‘‘Food & Drinks” category the emoji cooking (an egg on a frying
pan), was only found in tweets about breakfast and hot beverage
Category Exemplar Average Range
emoticons
(a cup of tea or coffee) was used most frequently in this eating
situation (0.4%). The emoji soft ice cream, red apple and cookie only
Smiley, happy face or laughing 5.2 3.2–7.4
:) 3.5 2.2–4.9
featured in tweets about snacks, while spaghetti was most common
:D 0.6 0.3–0.8 in tweets about dinner eating situations (0.3% vs. 60.1%).
(: 0.4 0.3–0.6 Emoji from other categories provided contextual information
:-) 0.2 0.1–0.4 about the focal situation to the tweets, and it fitted expectations
:3 0.2 0.1–0.6
that some differences in frequency of use existed across eating sit-
Frown, sad 0.9 0.5–1.6 uations (Table 6). For example, the emoji representing a set of
:( 0.7 0.3–1.2
books was used more frequently than other emoji in the ‘‘Non-
:-( 0.1 0.0–0.2
): 0.0 0.0–0.1 food objects” category, particularly in connection with tweets
:c 0.0 0.0–0.1 about lunch. Emoji from the ‘‘Celebration” category were never
Wink, smirk 0.8 0.6–1.1 used in connection with snack eating situations, and were found
;) 0.6 0.5–0.9 three-times more frequently in tweets about dinner than lunch
(; 0.1 0.0–0.2 (0.3% vs. 0.1%). Emoji from the ‘‘Travel & Places” category were only
;-) 0.1 0.0–0.1 used in connection with lunch (0.1%), while nature themed emoji
Tongue sticking out, cheeky/playful 0.3 0.2–0.3 were found in 1.3% of tweets in connection with snack eating occa-
:p 0.1 0.0–0.3 sions. Tentatively, this reflects the on-the-go/outdoors characteris-
:P 0.1 0.0–0.1
:b 0.0 0.0–0.1
tics of snacking. Please refer to Supplementary material for
additional detail.
Sceptical, undecided, uneasy, hesitant 0.2 0.1–0.3
:/ 0.2 0.1–0.2
/: 0.0 0.0–0.1 3.3. Associations between emoticon and emoji use and the content of
Kiss :⁄ 0.2 0.0–0.6 the tweets
Others 0.4 0.1–0.7
Vidal et al. (2015) performed a content analysis of the textual
information in the tweets analysed here. This enabled us to explore
emoticons and emoji were not related to emotions or moods. the relationships between emoticon and emoji use and the content
Primarily their use conveyed something about what was of the tweets. A contingency table between themes and emoticon
consumed, when, where, with whom, and why. use was calculated (see Table 7), considering the 20 most fre-
Table 6 presents results in this regard, capturing emoji use in quently used pictographs (of which 16 were emoji, and all 20 con-
the categories ‘‘Food & Drink”, ‘‘Non-food Objects”, ‘‘Celebration”, veyed a positive or negative emotional expression).

Table 5
Exemplar tweets in which different types of emoticon and emoji were used to convey emotional-reactions to specific food products in the four eating situations.

Eating situation Exemplar tweet

Breakfast Breakfast for dinner one of the best ideas ever!!French toast and bacon
Eggs and ketchup is so good” that’s what I had for breakfast
How’s my life going? Oh just eating pizza rolls for breakfast and wondering if ill actually make it to December
Spaghetti for breakfast. Thanks cousin tercinta x) LOL
Pancake and Milk for Breakfast
Life without milk is sad :(#breakfast
Lunch Lunch time!! Finally get to eat my subway!

Mmmm spaghetti would be great for lunch ” that’s what I’m having

Had the best Ramen in Bugis for lunch today!:)

Beef teppanyaki and cold ocha for lunch.. Let’s eat..
I had Chinese for lunch and it was bomb
Salmon spread on bread for lunch :)
Dinner ugh I wish I could actually cook I always resort to cereal for dinner when I’m home alone, not good
Shrimp & rice for dinner. Super spiceh

Carne Asada Fries for dinner !(: I feel like I’m back at home #WhoKnewKansasWouldHaveCarneAsadaFries
Making burgers for dinner haha (:
Steak fried pickles green beans & stuffed peppers for dinner. Now some peach cobbler & ice cream.
Snack Midnightsnack: pretzels and creme soda
my new favourite snack is peanut butter and pretzels

My yuuummy late night snack I looooove me some ice cream! #cookies

Eating apples and peanut butter is my favourite snack

Yummy late nite snack, watermelon with lime and chile (:
pizza rolls and tiramisu are tonight’s late night snack
124 L. Vidal et al. / Food Quality and Preference 49 (2016) 119–128

Table 6
Frequency of use (>0.1% shown) of emoji in 6 different categories. Emoji are identified by names in accordance with Emojipedia and drawn from the list of 631 emoji available on
Twitter. Average frequencies (based on 12,260 tweets) and range (i.e., minimum and maximum percentages) across the four eating situations (breakfast, lunch, dinner and snack)
are shown.

Category Exemplar emoji Average Range

Food & Drinks 3.0 1.7–4.5
Fork and knife 0.5 0.3–0.7

Slice of pizza 0.1 0.0–0.3

French fries 0.1 0.1–0.2

Spaghetti 0.1 0.0–0.3

Cooking 0.1 0.0–0.6

Hot beverage 0.1 0.0–0.4

Hamburger 0.1 0.1–0.1

Soft ice cream 0.1 0.0–0.4

Bread 0.1 0.0–0.3

Cookie 0.1 0.0–0.4

Poultry leg 0.1 0.0–0.1

Red apple 0.1 0.0–0.2

Non-food objects 0.5 0.1–1.0

Revolver , Bomb , Books , Hocho
Celebration 0.2 0.0–0.3
Party popper , Sparkles , Confetti ball
Activity 0.1 0.0–0.2
Dancer , Swimmer , Surfer , Runner

Travel & Places 0.1 0.0–0.1

Flag of italy , Wedding , Airplane ,
Minibus
Nature 0.8 0.5–1.3
Black sun with rays , Rainbow

New moon with face , Pig face , White medium star , Fire , Crescent moon

Frequency of use of emoticons and emoji significantly differed
according to the content of the tweet (v2 = 229.2, p < 0.0001). As
shown in Column 5 of Table 7 (percentage of tweets including
emoticons or emoji), emoticon and emoji use was significantly
more frequent in tweets describing: the context of food consump
tion/purchase/preparation, particularly tweets related to the
company of other people, special occasions and late timing,
and in tweets stressing the unhealthfulness of the consumed
products (29.3–43.2%). Conversely, tweets describing specific
foods/beverages, negative hedonic associations, food craving, hun-
ger, satiation, eating at restaurants or work/school or not being
able to eat tended to include emoticons or emoji less frequently
than average (14.8–23.9%). This points to positive emotion expres-
sion being more likely prompted by memorable (i.e., with others or
special situations) and/or naughty (i.e., late and unhealthy) eating
occasions.
Frequency of use of individual emoticons and emoji tended to
differ with the content of the tweet. As seen in Table 7, in tweets
with content relating to food consumption, positive emotional
expression was dominant, and the most frequently used pic-
tographs were the emoji face savouring delicious food eating and
smiling face with heart-shaped eyes, together with the emoticon :).
When people were tweeting about looking forward to eating the
emoji face with tears of joy was more frequently used. On the con-
trary, when the tweets related to not being able to eat (i.e., theme
‘‘Not eating”), users frequently included emoticons or emoji
expressing negative emotional reactions, such as unamused face,
pensive face and :(. When consumers referred to craving foods,
the most frequently used emoji were unamused face, face savouring
delicious food and face with tears of joy.
When considering tweets describing the context of food con
sumption/purchase/preparation, frequency of use of the emoticons
and emoji depended on whether the users were alone or with
other people. As shown in Table 7, the most frequently used emoti-
cons and emoji when users referred to the company of other peo-
ple were related to positive emotional reactions, such as smiling
face with heart-shaped eyes, face with tears of joy, ok hand sign
and:). However, when the tweets stressed eating, preparing or
purchasing food alone emoticons or emoji related to negative
emotional associations, such as loudly crying face, pensive face and
:(, were frequently used. Something similar was observed
when comparing tweets which described eating at restaurants
and eating at work/school. The first type of tweets mainly included
emoticons or emoji associated with positive emotional and
hedonic reactions, whereas the latter also included some emoti-
cons or emoji related to negative emotional reactions, such as :
(and loudly crying face.
Finally, differences were observed in the frequency of use of the
emoticons and emoji in tweets with content relating to associa-
tions with foods or consequences of food consumption. For exam-
ple, Twitter users tended to include emoticons or emoji expressing
negative emotional reactions when describing negative hedonic
associations or feeling hungry (e.g. :(, unamused face, weary face
and loudly crying face), but used emoticons or emoji related to pos-
itive emotional associations in tweets describing positive hedonic
experiences or feeling satiated/full.
Table 7
Percentage of tweets (n = 12,260) including emoticons and emoji for each of the themes identified in the content analysis and percentage of tweets within each theme including each emoticon. Percentages in each row do not sum up to
100% because only the most frequently used emoticons and emoji are shown.

Theme Sub-theme Category Number Percentage of :) :D (: ;) :(

of tweets including
tweets emoticons/
emoji ⁄
Food consumption Consuming Eating 5345 25.7 10.4 8 4.2 6.6 2.5 3.9 2.7 2.6 3.4 5.7 1.9 1.5 1.5 1 0.8 16.2 2.7 1.3 2.5 1.4
food Looking 252 26.6 7.5 3 10.4 1.5 4.5 3 3 0 1.5 3 3 4.5 4.5 1.5 4.5 11.9 3 1.5 1.5 3
forward to

L. Vidal et al. / Food Quality and Preference 49 (2016) 119–128

eating
Not eating 378 21.2 (-) 0 0 5 1.3 0 0 0 3.8 0 1.3 0 17.5 2.5 2.5 12.5 2.5 0 1.3 1.3 12.5
Craving Craving 469 17.9 (-) 8.3 4.2 7.1 1.2 0 1.2 0 3.6 3.6 0 2.4 3.6 13.1 2.4 4.8 6 6 1.2 3.6 3.6
food food
Food preparation 1083 26.9 10.7 9.3 4.1 7.6 2.4 4.8 1.7 3.8 3.1 3.4 3.1 3.1 2.7 1 0.7 13.1 1 3.1 2.7 3.8
Food purchase Food 437 23.6 4.9 2.9 9.7 3.9 2.9 3.9 6.8 4.9 7.8 9.7 1.9 3.9 3.9 2.9 1.9 9.7 1 1.9 2.9 0
purchase
Context of food Company Alone 165 22.4 2.7 0 5.4 10.8 0 0 0 0 2.7 0 0 2.7 0 8.1 5.4 2.7 2.7 0 0 21.6
consumption/ Other 3248 31.1 (+) 4.1 7.9 8.8 4.7 3.6 5.3 6 2.4 2.6 7.4 2 2.2 2.1 2.9 2 13.6 2 2.8 2.3 2.4
preparation/ people
purchase Occasion Occasion 847 29.3 (+) 7.7 10.9 5.6 7.3 3.6 5.2 7.3 1.6 2 7.7 2.8 2 1.6 1.2 0.8 11.7 1.2 3.2 2.4 0.8
Place Restaurant 1249 20.9 (-) 8 7.7 6.9 5.4 3.1 2.7 3.1 3.4 1.1 4.6 2.3 1.9 2.3 0.8 1.1 16.9 1.5 1.5 1.5 1.9
Work/ 1066 20.2 (-) 2.8 2.8 10.7 6 3.3 3.3 0.9 3.7 2.3 3.3 2.8 3.3 2.8 4.7 3.3 10.2 1.4 2.3 2.8 8.8
school
Timing Early 214 26.6 5.3 3.5 1.8 1.8 1.8 1.8 0 0 3.5 1.8 3.5 3.5 3.5 0 1.8 14 1.8 1.8 3.5 0
Late 984 36.4 14 6.4 1.4 5.9 1.7 3.1 1.1 1.4 1.7 3.4 2 1.1 2 1.4 1.1 6.1 1.1 0.6 1.1 1.1
Specific foods/ 2858 23.9 (-) 14.5 10.8 5 6.7 2.2 3.9 0.9 3.5 3.7 4.7 2.3 1.2 2.3 1.5 0.7 12.6 2.9 1.8 1.9 1.9
beverages
Associations with food Health Healthful 262 18.3 (-) 6.3 10.4 2.1 16.7 2.1 4.2 2.5 2.1 4.2 4.2 2.1 0 0 2.1 2.1 10.4 0 2.1 0 2.1
/ consequences of associations Unhealthful 111 43.2 (+) 8.3 6.3 6.3 0 2.1 0 0 0 0 4.2 2.1 0 0 0 2.1 0 2.1 16.7 0 2.1
consumption Hedonic Hedonic 155 14.8 (+) 0 0 8.7 0 0 0 0 0 0 0 4.3 8.7 4.3 8.7 4.3 4.3 0 0 0 17.4
associations negative
Hedonic 1668 25.8 14.2 9.8 1.9 9.3 3.3 5.8 4.2 2.1 4 5.8 2.8 0.7 0.9 0.7 0 14.4 2.3 1.9 3.5 0.7
positive
Hunger/ Full/ 281 16.4 (+) 13 6.5 10.9 0 0 6.5 0 2.2 2.2 6.5 6.5 2.2 2.2 6.5 2.2 10.9 2.2 0 0 4.3
satiety Satiated
Hungry 322 18.0 (+) 5.2 3.4 6.9 0 0 0 2.4 6.9 1.7 1.7 1.7 8.6 10.3 0 5.2 1.7 0 1.7 3.4 5.2
⁄
Effect of the chi-square per cell test for comparing the theme of the tweets: (+) or (–) indicate that the observed value is higher or lower than the expected theoretical value for p 6 0.05. Both Global chi-square test and chi-square
per cell test were performed only on the number of tweets including emoticons or emoji within each of the categories identified in the content analysis.

125
126 L. Vidal et al. / Food Quality and Preference 49 (2016) 119–128
In summary, there was evidence of emoticon and emoji use
being determined by tweet content (Table 7). Emoticons or emoji
denoting negative emotional reactions, such as unamused face,
pensive face and:(were associated with tweets describing negative
hedonic reactions, eating alone, not being able to eat or being
hungry. Emoticons and emoji use for positive emotional expres-
sion was more diverse, in terms of the number of different emoti-
cons and emoji used, and the aspects of eating situations they
were associated with. Briefly, face throwing kiss, smiling face, smil-
ing face with smiling eyes, face with tears of joy and :) were asso-
ciated with tweets describing food purchase, special occasions,
the company of other people or eating unhealthful food. Further-
more, tweets describing positive hedonic associations and specific
food/beverages were associated with the emoji face savouring
delicious food, face with heart-shaped eyes, ok hand sign, thumbs
up sign and :D.
4. Discussion
4.1. Emoticon and emoji use to express emotions
Our analysis of emoticon and emoji use in 12260 tweets previ-
ously retrieved by Vidal et al. (2015) provided new insights about
consumers’ emotional expression. Although emoticons and emoji
from categories not related to emotion expression were found in
the data, their frequency was quite low (17.2%) and we focus this
discussion on the results relating to emoticon use for emotional
reactions/associations.
Results confirmed that consumers spontaneously express food-
related emotions in their daily life. Emoticons and emoji were
more frequently used than words for this purpose (cf. Vidal et al.,
2015), in agreement with several authors that have stated that
these pictographs have become an effortless way for expression
(Cowie et al., 2001; Truss, 2004). In a similar vein, Köster and
Mojet (2015) stated that consumers rarely name the emotions
evoked by foods/drinks when they experience them. These previ-
ous findings may also help explain why emotion expression in
tweets rarely accompanied expression of emotion in words. In only
7.6% of tweets were emoticons or emoji used to emphasise infor-
mation expressed in words. To exemplify: Tweets like ‘‘I have no
time for breakfast” accompanied by a sad face emoji were much
more common that tweets such as ‘‘Hate it when there is nothing
to snack on in the house” being accompanied by a sad face emoji.
These findings raise concerns about the ecological validity of emo-
tion questionnaires.
It was most frequently observed that tweets included only a
single emoticon or emoji (76.4%). Expression of multiple emotions
in reference to a focal eating situations was thus rare. Tentatively,
this could be linked to the limited length of tweets (141 charac-
ters). However, we note that this finding aligns with Jaeger et al.
(2013), who found that the mean number of terms freely elicited
by consumers to express food-related emotions (typically 1–3) is
much lower than the number of terms included in emotion ques-
tionnaires (typically 16–59).
Lexicons of human emotions often contain 200+ different
emotions (Chrea et al., 2009; Clore, Ortony, & Foss, 1987; Laros &
Steenkamp, 2005; Ortony, Clore, & Foss, 1987; Rousset, Deiss,
Juillard, Schlich, & Droit-Volet, 2005) and evidence of a similar
magnitude of different emoticons and emoji in the analysed tweets
(254 emoji and 50 emoticons) had face validity and supported the
view that many different emotional associations/reactions are
experienced by consumers in their food-related activities
(Meiselman, 2015). Considering that emoji were used about 5 times
more frequently than emoticons, it was not surprising that a greater
number of different emoji were identified. In part, this finding is
also likely linked to the number of available emoji in Twitter (631).
Emoticon and emoji use was primarily associated with positive
emotional expression (66.7%). This corresponds with the hedonic
asymmetry proposed by Desmet and Schifferstein (2008).
According to these authors healthy people usually have a positive
affective disposition towards eating. The dominance of positive
emotions in eating/drinking is also found in the majority of
emotion questionnaires used in new product development
(Meiselman, 2015).
In the analysed tweets, several of the most frequently used
emoji are identical to those identified by Swiftkey (2015) as being
among the most popular emoji worldwide. According to this
company, which provides keyboard apps for phones and tablets,
the most frequently used emoji happy faces worldwide are: face
with tears of joy, face throwing kiss, smiling face with heart-shaped
eyes and smiling face with smiling eyes. For sad emoji faces, the most
frequently used worldwide are: loudly crying face, face screaming in
fear, unamused face and pensive face. With the exception of face
screaming in fear, these 8 emoji feature in the list of the 20 most
frequently used emoji in this research (Table 3). Perhaps it is not
surprising that people use popular emoji, which they are likely to
be literate about/have familiarity with. However, it could perhaps
also indicate that searching for the best emoji to express a partic-
ular emotion is not the highest priority for Twitter users. As long as
the focal emoji is ‘‘about right” it is ‘‘good enough.” Our data does
not allow an examination of this hypothesis, but it is clear that
emoticons and emoji are matched to tweet topic and content. For
example, the emoji face screaming in fear was only included in
0.06% of the analysed tweets, but its low frequency of use had face
validity considering that fear does not feature as a specific emotion
in emotion questionnaires targeted to foods/beverages or fragrance
products (Chrea et al., 2009; Ferdenzi et al., 2013; King &
Meiselman, 2010; Nestrud, Meiselman, King, Lesher, & Cardello,
2013; Richins, 1997; Spinelli et al., 2014). Furthermore, the most
frequently used emoji character in the present study (face savour-
ing delicious food) does not feature on the Swiftkey list, which can
be explained by the specificity of this emoji for food-related
situations.
The meaningful relationships between tweet content and
emoticon and emoji use (Table 7) further demonstrated that
emoticons and emoji were deliberately used to express certain
experienced emotions. Twitter users tended to include emoticons
and emoji more frequently when they described specific details
of the context of the eating situation, such as the presence of other
people, special occasions and timing. Highlighting the importance
of context in shaping food-related emotions, emoticon and emoji
use was significantly higher in tweets related to these content
themes than in tweets describing specific foods or positive hedonic
experiences. Köster and Mojet (2015) have (re)-stressed this point
and empirically it was recently demonstrated in a series of studies
by Piqueras-Fiszman and Jaeger (2014a, 2014b, 2014c), who
showed that consumers’ response to emotional questionnaires
when evaluating food products varies significantly across (evoked)
consumption contexts. Piqueras-Fiszman and Jaeger (2015) have
also shown that ‘‘memorable meals”, which often involve special
occasions and eating in the company of other people whose com-
pany one enjoys is associated to positive emotions. Emotion
expression was also frequent when tweets related to the unhealth-
fulness of what was consumed, and in these instances the most fre-
quently used emoticons and emoji conveyed a positive reaction.
This has merit considering the relationship between foods rich in
sugar and fat, hedonic experiences and reduction of aversive feel-
ings towards stress (Pool, Delplanque, Coppin, & Sander, 2015).
Although the present work was focused on eating situations, it
is interesting to highlight that in many tweets emoji and emoticons
were used to express product-related emotions. As exemplified in
Table 5, emoji and emoticons were used to convey a wide range of
 L. Vidal et al. / Food Quality and Preference 49 (2016) 119–128
emotional reactions, including both positive and negative emo-
tions. In particular, emoji conveyed a wider range of emotional
reactions/associations than emoticons. These results suggests that
emoji may be a promising tool for evaluating emotional responses
to food products.
4.2. Limitations and suggestions for further research
4.2.1. Emoticon and emoji use in Twitter data
Results from the present work encourage several avenues for
further research on emoticon and emoji use in Twitter. The Twitter
data used in this research pertained to breakfast, lunch, dinner and
snack situations, but an in-depth analysis of these eating situations
was not performed. This would be relevant to better understand
similarities and differences in emoticon and emoji use. In future
research, Twitter data could also be used to study emotional
expression in relation to specific food/beverage/personal-care
products and/or specific brands. The need to better understand
emotional associations in relation to branded products has been
highlighted by Thomson (2010), and is attracting attention (e.g.,
Gutjar et al., 2015; Spinelli, Masi, Zoboli, Prescott, & Monteleone,
2015; Thomson & Crocker, 2015; Thomson, Crocker, & Marketo,
2010). Another interesting application of Twitter for studying
consumers’ emotional reactions to products is the analysis of the
individual food-related emotional changes with contextual
variables, such as season, day of the week or time of the day.
Using Twitter, Golder and Macy (2011) have shown daily, weekly
and seasonal changes in the peoples’ emotional reactions across
the globe.
The main advantage of Twitter data is their spontaneous nature,
by virtue of being generated in real life situations. This make them
less susceptible to memory bias or experiment demand effects
(Golder & Macy, 2011). However, as discussed by Vidal et al.
(2015), there are also limitations of Twitter for studying con-
sumers’ food-related behaviour. Key among these is the fact that
Twitter users are not representative of the general population, as
they are mainly composed of people aged between 15 and 34 years
old (comScore, 2011a, 2011b; Mislove, Lehmann, Ahn, Onnela, &
Rosenquist, 2012). Neither is detailed information about users’
background and the context in which the tweets are posted avail-
able. Another important drawback of Twitter data is related to the
time needed for analysing the large quantity of information which
is usually gathered for free in short time frames. According to Vidal
et al. (2015) manual content analysis is required to interpret Twit-
ter data in context and to avoid potential misinterpretation derived
from the use of words with multiple meanings, informal writing
and the use of accidental or deliberate misspellings.
4.2.2. The use of emoticons and emoji for non-verbal subjective
emotion measurement
The findings of the current research point to the possibility that
emoticons and emoji may be used as a more general tool for non-
verbal subjective emotion measurement in food-related research.
In particular emoji appear to hold promise in this regard, consider-
ing that they were used about 5 times more frequently than emoti-
cons (Table 1). Such non-verbal approaches could supplement the
current reliance on surveys featuring lists of named emotions and
would align with similar developments elsewhere. For example,
Ortigosa, Martín, and Carro (2014) have recently proposed the con-
sideration of emoticons when performing sentiment analysis in
social media.
Compared with other non-verbal methods such as Product
Emotion Measuring Instrument (PrEmo), which uses cartoon fig-
ures to represent different emotions (Desmet, 2003), emoji charac-
ters have the advantage of being more familiar to consumers, who
increasingly use them in computer-mediated communications and
127
see them in marketing communications (Huang et al., 2014; Truss,
2004). Results from the present work could therefore be used as a
starting point for selecting emoji to include in questionnaires elic-
iting food-related emotions. An advantage of the current results,
relative to the 14 emotions included in PrEmo, which were selected
using non-food products (Köster & Mojet, 2015), is that they orig-
inate in tweets about eating and drinking.
However, it would be prudent to consider levels of emoji literacy
in the consumer populations that such non-verbal questionnaires
are to be used with. Consumers may not share a common under-
standing of what different emoji characters express, and some
may be uncertain about what an emoji is expressing. The latter
could be associated with incidences of misinterpretation of emoji.
Research on this topic is incipient (Moore, Steiner, & Conlan,
2013) but necessary to ensure that new methods are easy and intu-
itive for participants. At least some respondents do not consider
verbal emotion surveys to have these characteristics (Jaeger et al.,
2013), but a similar negative response to the research task is not
unlikely if respondents struggle to interpret the meaning of the
emoji. Cross-cultural research on similarities and differences in
the understanding and conceptualisation is also warranted. Consid-
ering that emotions in facial expressions, gestures and body pos-
tures are similarly perceived by people in different cultures due
to empathic reactions (Ekman, 1994; Ekman & Friesen, 1971),
cross-cultural differences in how consumers interpret emoji could
be smaller than the influences culture and language exert on verbal
emotion tasks (van Zyl & Meiselman, 2015). Comparison of tweets
retrieved in different languages or from different specific geograph-
ical locations, which can contribute to our understanding of the
relationship between emotions and food, would be premature
without such confirmation/knowledge.
5. Conclusions
Building on Vidal et al. (2015), who illustrated the potential of
Twitter in food-related consumer research, this work showcased
the use of Twitter data for research into food-related emotional
expression. Considering their spontaneous nature, tweets are
advantageous in this regard, as compared with completion of emo-
tion surveys. Yet, this data source is also significantly constrained
by inability to tailor data collection to specific products and obtain
background information about the people who posted the tweets.
Evidence was obtained in this research, which considered 12,260
tweets about breakfast, lunch, dinner and snack eating situations
that consumers express a wide range of positive and negative emo-
tions and that emoticon and emoji use is tailored to the content of
the tweets. Emoji were used more frequently than emoticons to
express emotions. While it was rare for tweets to include more
than one emoticon or emoji, their use was in addition to other con-
tent of the tweet. This supported previous reports that consumers
do not easily name food-related emotions, but perceive emoticons
and emoji as an easy and intuitive way to express emotions. We
suggest this represents an opportunity for development of non-
verbal subjective methods to measure food-related emotions.
Author contributions
GA and SRJ planned the research and wrote the paper. GA and
LV analysed the data.
Acknowledgements
The authors are indebted to The New Zealand Institute for Plant
& Food Research Limited (PFR) for financial support. Sok Leang
Chheang is thanked for support in preparing this paper.
128 L. Vidal et al. / Food Quality and Preference 49 (2016) 119–128
Appendix A. Supplementary material
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.foodqual.2015.12.
002.
References
Barash, V., & Golder, S. (2010). Twitter: Conversation, entertainment, and
information, all in one network! In D. Hansen, B. Shneiderman, & M. Smith
(Eds.), Analyzing social media networks with NodeXL: Insights from a connected
world (pp. 143–164). Burlington, Massachusetts: Morgan Kaufmann Publishers.
Canetti, L., Bachar, E., & Berry, E. M. (2002). Food and emotion. Behavioural Processes,
60, 157–164.
Carr, J., Decreton, L., Qin, W., Rojas, B., Rossochacki, T., & wen Yang, Y. (2015). Social
media in product development. Food Quality and Preference, 40, 354–364.
Chrea, C., Grandjean, D., Delplanque, S., Cayeux, I., Le Calvé, B., Aymard, L., et al.
(2009). Mapping the semantic space for the subjective experience of emotional
responses to odours. Chemical Senses, 34, 49–62.
Citrin, A. V., Stem, D. E., Spangenberg, E. R., & Clark, M. J. (2003). Consumer need for
tactile input: An internet retailing challenge. Journal of Business Research, 56,
915–922.
Clore, G. L., Ortony, A., & Foss, M. A. (1987). The psychological foundations of the
affective lexicon. Journal of Personality and Social Psychology, 53, 751–755.
comScore (2011a). 2010 Europe digital year in review. Reston, Virginia: comScore Inc.
comScore (2011b). 2010 US digital year in review. Reston, Virginia: comScore Inc.
Cowie, R., Douglas-Cowie, E., Tsapatsoulis, N., Votsis, G., Kollias, S., Fellenz, W., &
Taylor, J. G. (2001). Emotion recognition in human-computer interaction. IEEE
Signal Processing Magazine, 18, 32–80.
Desmet, P. M. A. (2003). Measuring emotions. In M. A. Blythe, K. Overbeeke, A. F.
Monk, & P. C. Wright (Eds.), Funology: From usability to enjoyment (pp. 111–125).
Dordrecht, The Netherlands: Kluwer.
Desmet, P., & Schifferstein, H. N. J. (2008). Sources of positive and negative emotions
in food experience. Appetite, 50, 290–301.
Ekman, P. (1994). Strong evidence for universals in facial expressions: A reply to
Russel’s mistaken critique. Psychological Bulletin, 115(2), 268–287.
Ekman, P., & Friesen, W. P. (1971). Constants across cultures in the face and
emotion. Journal of Personality and Social Psychology, 17(2), 124–129.
Emojipedia (2015). List of emoij meanings. <https://http://emojipedia.org> (Accessed
28.07.15).
Ferdenzi, C., Delplanque, S., Barbosa, P., Court, K., Guinard, J. X., Guo, T., et al. (2013).
Affective semantic space of scents. Towards a universal scale to measure self-
reported odor-related feelings. Food Quality and Preference, 30, 128–138.
Ferrarini, R., Carbognin, C., Casarotti, E. M., Nicolis, E., Nencini, A., & Meneghini, A. M.
(2010). The emotional response to wine consumption. Food Quality and
Preference, 21, 720–725.
Fried, D., Surdeanu, M., Kodbourov, S., Hingle, M., & Bell, D. (2014). Analyzing the
language of food on social media. In 2014 IEEE International Conference on Big
Data, October 27–30th, 2014. Washington, EEUU.
Frijda, N. (1986). The emotions. Cambridge: Cambridge University Press.
Gentry, J. (2014). Package ‘twitteR’. <http://cran.r-project.org/web/packages/
twitteR/twitteR.pdf> (Accessed 17.11.14).
Ghiassi, M., Skinner, J., & Zimbra, D. (2013). Twitter brand sentiment analysis: A
hybrid system using n-gram analysis and dynamic artificial neural network.
Expert Systems with Applications, 40, 6266–6282.
Gibson, E. L. (2006). Emotional influences on food choice: Sensory, physiological and
psychological pathways. Physiology & Behavior, 89, 53–61.
Gmuer, A., Nuessli Guth, J., Runte, J., & Siegrist, M. (2015). From emotion to
language: Application of a systematic, linguistic-based approach to design a
food-associated emotion lexicon. Food Quality and Preference, 40, 77–86.
Golder, S. A., & Macy, M. W. (2011). Diurnal and seasonal mood vary with work,
sleep, and daylength across diverse cultures. Science, 333(6051), 1878–1881.
Gutjar, S., Dalenberg, J. R., de Graaf, C., de Wijk, R. A., Palascha, A., Renken, R. J., &
Jager, G. (2015). What reported food-evoked emotions may add: A model to
predict consumer food choice. Food Quality and Preference, 45, 140–148.
Huang, A. H., Yen, D. C., & Zhang, X. (2014). Exploring the potential effects of
emoticons. Information & Management, 45, 466–473.
Huffaker, D. A., & Calvert, S. L. (2005). Gender identity, and language use in teenage
blogs. Journal of Computer-Mediated Communication, 10(2). http://dx.doi.org/
10.1111/j.1083-6101.2005.tb00238.x.
Jaeger, S. R., Cardello, A. V., & Schutz, H. G. (2013). Emotion questionnaires: A
consumer-centric perspective. Food Quality and Preference, 30, 229–241.
Jiang, Y., King, J. M., & Prinyawiwatkul, W. (2014). A review of measurement and
relationships between food, eating and emotion. Trends in Food Science &
Technology, 36, 15–28.
King, S., & Meiselman, H. L. (2010). Development of a method to measure consumer
emotions associated with foods. Food Quality and Preference, 21, 168–177.
Köster, E. P., & Mojet, J. (2015). From mood to food and from food to mood: A
psychological perspective on the measurement of food-related emotions in
consumer research. Food Research International, 76, 180–191.
Krippendorff, K. (2004). Content analysis: An introduction to its methodology.
Thousand Oaks, CA: Sage Publications.
Laros, F. J. M., & Steenkamp, J. B. E. M. (2005). Emotions in consumer behavior: A
hierarchical approach. Journal of Business Research, 58, 1437–1445.
Macht, M. (2008). How emotions affect eating: A five-way model. Appetite, 50, 1–11.
Meiselman, H. L. (2015). A review of the current state of emotion research in
product development. Food Research International, 76, 192–199.
Mislove, A., Lehmann, S., Ahn, Y.-Y., Onnela, J.-P., & Rosenquist, J. N. (2012).
Understanding the demographics of Twitter users. In Proceedings of the fifth
international AAAI conference on weblogs and social media (pp. 554–557). Menlo
Park, California: AAAI Press.
Moore, A., Steiner, C. M., Conlan, O. (2013). Design and development of an empirical
smiley-based affective instrument. In 1st Workshop on emotions and personality
in personalized services. Rome, Italy: UMAP 2013.
Mostafa, M. M. (2014). More than words: Social networks’ text mining for consumer
brand sentiments. Expert Systems with Applications, 40, 4241–4251.
Nestrud, M. A., Meiselman, H. L., King, S. C., Lesher, L. L., & Cardello, A. V. (2013). A
method for development of EsSense25 – A shorter version of the EsSense(R)
Profile. In 11th Pangborn sensory science symposium, 11–15th August 2013. Rio
de Janeiro, Brazil.
Ng, M., Chaya, C., & Hort, J. (2013). Beyond liking: Comparing the measurement of
emotional response using EsSense profile and consumer defined check-all-that-
apply methodologies. Food Quality and Preference, 28, 193–205.
Ortigosa, A., Martín, J. M., & Carro, R. M. (2014). Sentiment analysis in Facebook and
its applications to e-learning. Computers in Human Behavior, 31, 527–541.
Ortony, A., Clore, G. L., & Foss, M. A. (1987). The referential structure of the affective
lexicon. Cognitive Science, 11, 341–364.
Piqueras-Fiszman, B., & Jaeger, S. R. (2014a). The impact of evoked consumption
contexts and appropriateness on emotion responses. Food Quality and
Preference, 32, 277–288.
Piqueras-Fiszman, B., & Jaeger, S. R. (2014b). Emotion responses under evoked
consumption contexts: A focus on the consumers’ frequency of product
consumption and the stability of responses. Food Quality and Preference, 35, 24–41.
Piqueras-Fiszman, B., & Jaeger, S. R. (2014c). The impact of the means of context
evocation on consumer’s emotion associations towards eating occasions. Food
Quality and Preference, 37, 61–70.
Piqueras-Fiszman, B., & Jaeger, S. R. (2015). Emotions associated to mealtimes:
Memorable meals and typical evening meals. Food Research International, 76,
243–252.
Pool, E., Delplanque, S., Coppin, G., & Sander, D. (2015). Is comfort food really
comforting? Mechanisms underlying stress-induced eating. Food Research
International, 76, 207–215.
Porcherot, C., Delplanque, S., Ravot-Derrien, S., Le Calvé, B., Chréa, C., Gaudreau, N.,
et al. (2010). How do you feel when you smell this? Optimization of a verbal
measurement of odor-elicited emotions. Food Quality and Preference, 21, 938–947.
R Core Team (2013). R: A language and environment for statistical computing. Vienna,
Austria: R Foundation for Statistical Computing.
Richins, M. L. (1997). Measuring emotions in consumption experience. Journal of
Consumer Research, 24, 127–146.
Rousset, S., Deiss, V., Juillard, E., Schlich, P., & Droit-Volet, S. (2005). Emotions
generated by meat and other food products in women. British Journal of
Nutrition, 94(4), 609–619.
Spinelli, S., Masi, C., Dinella, C., Zoboli, G. P., & Monteleone, E. (2014). How does it
make you feel? A new approach to measuring emotions in food product
experience. Food Quality and Preference, 37, 109–122.
Spinelli, S., Masi, C., Zoboli, G. P., Prescott, J., & Monteleone, E. (2015). Emotional
responses to branded and unbranded foods. Food Quality and Preference, 42,
1–11.
Swiftkey (2015). Swiftkey emoji report. London: Swiftkey.
Symoneaux, R., Galmarini, M. V., & Mehinagic, E. (2012). Comment analysis of
consumer’s likes and dislikes as an alternative tool to preference mapping. A
case study on apples. Food Quality and Preference, 24, 59–66.
Thomson, D. M. H., & Crocker, C. (2015). Application of conceptual profiling in
brand, packaging and product development. Food Quality and Preference, 40,
343–353.
Thomson, D. M. H., Crocker, C., & Marketo, C. G. (2010). Linking sensory
characteristics to emotions: An example using dark chocolate. Food Quality
and Preference, 21, 1117–1125.
Thomson, D. (2010). Going beyond liking: Measuring emotional and conceptual
profiles to make better new products. In S. R. Jaeger & H. MacFie (Eds.),
Consumer-driven innovation in food and personal care products (pp. 219–274).
Cambridge: Woodhead Publishing.
Truss, L. (2004). Eats, shoots and leaves: A zero-tolerance approach to punctuation.
New York, NY: Penguin.
van Zyl, H., & Meiselman (2015). The roles of culture and language in designing
emotion lists: Comparing the same language in different English and Spanish
speaking countries. Food Quality and Preference, 41, 201–213.
Vidal, L., Ares, G., Machín, L., & Jaeger, S. R. (2015). Using Twitter data for food-
related consumer research: A case study on ‘‘what people say when tweeting
about different eating situations’’. Food Quality and Preference, 45, 58–69.
Walther, J. B., & D’Addario, K. P. (2001). The impacts of emoticons on message
interpretation in computer-mediated communication. Social Science Computer
Review, 19, 324–347.
Wikipedia (2015). List of Emoticons. In Wikipedia, the free encyclopedia. <https://en.
wikipedia.org/wiki/List_of_emoticons#Western>. (Accessed 28.07.15).
Wolf, A. (2000). Emotional expression online: Gender differences in emoticon use.
Cyber Psychology & Behavior, 3, 827–833.
Worch, T. (2014). What should you know about analysing social media data using
twitteR: The experience of a practitioner. In 6th European conference on sensory
and consumer research, 7–10 September 2014. Copenhagen, Denmark.