The popular press is currently rife with speculation that emoji are becoming a global, digitally-mediated language. Sequences of emoji that function like verbal utterances potentially lend strong support to this claim. We employ computer-mediated discourse analysis to analyze the pragmatic meanings conveyed through emoji sequences and their rhetorical relations with accompanying text, focusing on posts by social media influencers and their followers on a popular Chinese social media platform. The findings show that the emoji sequences can function pragmatically like verbal utterances and form relations with textual propositions, although their usage differs from textual utterances in several respects. We also observed user innovations that make the sequences more language like, although there is not as yet a fixed grammar of emoji sequences. We characterize this emoji use as an emergent graphical language, with the caveats that it is not yet a fully-formed language and that the Chinese emoji language that is emerging is different from the English variety, and therefore emoji are not a universal language. In order to promote the further development of emoji language(s), we advance recommendations for emoji design grounded in linguistic principles.
When the Oxford Dictionary chose the Crying Tears of Joy emoji as the 2015 word of the year, it lent official status to the popular perception that emoji (from the Japanese e- ‘image’ and moji ‘character’) have become a new form of graphical language in computer-mediated communication (CMC) — a language that is playful, semantically rich, and emotionally expressive (Danesi, 2016). In addition to using emoji to substitute for words, social media users string them together creatively to form sequences that function like utterances in online conversations (Herring and Dainas, 2017). Here we define an emoji sequence as two or more emoji with different meanings that stand together to form a conceptual unit. The nature of these conceptual units  is the focus of this article.
The creative and complex ways that users employ emoji sequences raise challenges for theoretical and practical understandings of emoji use and for the design of graphical elements in CMC. Unlike user-created ASCII emoticons, emoji are created primarily by professionals, vetted and codified by the Unicode Consortium, and their specific renderings are determined by different social media platforms. In order to meet users’ expanding needs for emoji that facilitate digital communication, designers of emoji and of social media platforms need to keep abreast of user innovations in emoji use. Moreover, if emoji are to develop as a functional language system, emoji are needed that express a range of concepts and relations and that support the expression of varied and complex ideas. Previous studies mainly focus on the role of single emoji, for example, in modifying, illustrating, or substituting for text (Kelly and Watts, 2015; Rodrigues, et al., 2017). The communicative functions of emoji sequences remain largely unexplored. Yet, we suggest, it is through such sequences that emoji contribute most directly to the creation and use of a purely graphical language.
In this study, we identify two ways emoji sequences are starting to function as sentence-like utterances in social media messages. Specifically, we demonstrate that such sequences convey pragmatic meanings and relate rhetorically to the text they accompany in messages posted to Sina Weibo, China’s most prominent microblogging platform. Asian countries, especially China, are ripe for research on emoji sequences. Emoji and their larger relatives, stickers , originated in Japan, and Asian countries have led the West in their adoption (Ljubešić and Fišer, 2016; Russell, 2013). One reason that has been suggested for the greater popularity of emoji and stickers in China than in the West is the Chinese writing system. Modern Chinese characters, while mostly phono-semantic compounds, include a number of pictograms (e.g., ‘雨’ for rain) and ideograms (e.g., ‘一’ for one, ‘二’ for two) (Norman, 1988), and some Chinese emoji include or are based on Chinese characters . Chinese users also report finding it faster and less cumbersome to send emoji and stickers than text, because of the difficulty of typing in Chinese characters on computer and smart phone keyboards that mainly support the Roman alphabet (Ma, 2016; Russell, 2013).
Sina Weibo lends itself especially well to emoji sequence analysis, due to the emoji affordances of the platform and its culture of emoji use. Proprietary, platform-specific sets of emoji tend to be more popular in China than Unicode emoji (de Seta, 2018). Sina Weibo provides a large array of platform-specific emoji to facilitate and encourage user interactions (Zhou and Pan, 2016), including both static emoji (e.g., people, object, animals, symbols) and animated ones (e.g., gestures, actions). In addition, users posting from the Sina Weibo mobile app can use the emoji keyboard provided by their iOS and Android mobile phones to access the full set of emoji categories listed in the Emoji Unicode version 12.0 . In this study we consider both Sina Weibo platform-specific emoji and Unicode emoji, because both types are available to Sina Weibo mobile app users, and both are sometimes used together in a single sequence.
In part due to the abundance of available emoji, their use — including emoji sequences — on Sina Weibo is very common and creative, especially as used by celebrities (Ge and Gretzel, 2018). Users often employ emoji in innovative ways to express their unique personality, create shared and secret meanings, poke fun at one another, and invoke solidarity and create affiliation (Andersen, 2018; Li, 2017). Thus a focus on Chinese emoji use, specifically on Sina Weibo, can advance the CMC and emoji literatures, provide a conceptual map for non-Chinese users to effectively employ emoji to communicate with Chinese social media users (Yan, 2016), and generate new insights to inform emoji design in social media systems.
Our starting premise is that emoji sequences as defined in the first paragraph of this article are functionally analogous to sequences of words in verbal language . Accordingly, we adopt a language-focused approach to analyze Chinese emoji sequences: computer-mediated discourse analysis (Herring, 2004), an approach that adapts methods from linguistics and related fields to analyze digital discourse. Specifically, we employ two complementary discourse-based methods, speech act analysis and rhetorical structure analysis. In what follows, we first situate our study in relation to previous emoji research, after which we provide background information on speech act analysis and rhetorical structure analysis. We then describe our dataset, which consists of Sina Weibo messages posted by celebrities and comments on those messages by general users, along with the analytical procedures we followed.
Overall, we found that more than 90 percent of the emoji sequences were analyzable as conveying speech acts and as expressing rhetorical relations, supporting our premise that such sequences function like verbal utterances. ‘Claim’ (i.e., expressing personal feelings and opinions) was the most common speech act, and ‘restatement’ (i.e., repeating the information presented in the text) was the most prevalent relation with the accompanying text. We also observed creative usages in the composition of emoji sequences that compensate for the lack of a prescribed emoji sequence grammar and that make the sequences more language-like, including the emergence of a pragmatically-motivated ‘stance-last’ word order pattern. Based on these findings, we propose that emoji use on Sina Weibo constitutes an emergent graphical language. To facilitate users’ tendency to create linguistic structure from emoji, we offer recommendations for emoji design grounded in linguistic principles.
To date, the emoji literature has focused primarily on the meanings, functions, and social uses of single emoji. This line of research suggests that emoji function as nonverbal cues and as words in CMC. Emoji indicating facial expressions and body gestures can enhance sentiment, modify the meaning of written text, and add contextual clues (Pavalanathan and Eisenstein, 2016; Zhou, et al., 2017). Some scholars have described emoji as a type of paralanguage (Novak, et al., 2015) which, like gesture, permits users to emphasize, particularize, embellish, and clarify their verbal propositions (Bavelas and Chovil, 2000). Emoji hand gestures (e.g., ) can function like ‘batons,’ that is, non-verbal actions that offer directions or instructions (Bavelas and Chovil, 2000). Relatedly, McCulloch and Gawne (2018) argue that repeated emoji are similar to ‘beat gestures’ in speech. Emoji can also substitute for words, especially those in the non-facial and non-gesture categories (e.g., objects, animals, activities) (Pohl, et al., 2017; Radford, et al., 2016; Schnoebelen, 2012).
Studies of emoji semantics have examined sentiment (Ai, et al., 2017) and similarity of meaning across different emoji (Wijeratne, et al., 2017). A common argument is that emoji allow ideas and emotions to be conveyed vividly (Lu, et al., 2016). Emoji also possess meaning inherently and add a wide array of semantic nuances to written text (Danesi, 2016). As for emoji functions, studies have revealed five main uses that potentially overlap: 1) expression of emotion (Alismail and Zhang, 2018; Novak, et al., 2015; Stark and Crawford, 2015; Vidal, et al., 2016); 2) tone modification (Ge and Gretzel, 2018; Herring and Dainas, 2017); 3) substituting for nonverbal behavior (Pavalanathan and Eisenstein, 2016); 4) opening and closing conversations (Danesi, 2016); and, 5) when the user has nothing to say (Danesi, 2016; Riordan, 2017a). Additional functions were identified by Herring and Dainas (2017) in a study of graphical icons (graphicons) on Facebook, e.g., reaction, mention, and action. Finally, research on the social uses of emoji underscores their role in enhancing interpersonal relationships, sharing secret meanings and fostering social connectedness between users (Kelly and Watts, 2015), expressing one’s identity, and increasing perceived intimacy (Riordan, 2017b; Zhou, et al., 2017).
In contrast, research specifically focusing on emoji sequences is sparse, although several studies mention them. Herring and Dainas (2017) identified ’narrative sequence’ (i.e., the use of a series of consecutive emoji to tell a story) as one of the pragmatic functions of emoji. ‘Narrative sequences’ align with what other researchers call emoji phrases (Danesi, 2016), emoji sentences (Radford, et al., 2016), a cluster of emoji (Park, et al., 2018) and compound multi-emoji expression (Monti, et al., 2016). Moreover, although some other authors do not directly discuss emoji sequences, they include examples that illustrate such sequences, according to our definition (Dürscheid and Siever, 2017; Ge and Gretzel, 2018; Illendula and Yedulla, 2018; Radford, et al., 2016). In one study of the pragmatic intentions underlying emoji use (Cramer, et al., 2016), several users reported using instances of emoji sequences to repeat the text of messages, provide situational context, or propose a fun puzzle for the receiver to decipher. In examining the use of emoji for persuasion, Ge and Gretzel (2018) found that Chinese users deploy consecutive emoji to express emotion, convey facts or specific information, and establish credibility. A study of emoji usage through a gender lens (Chen, et al., 2018) found that while female users are likely to use only one emoji or multiple emoji dispersed throughout a message, male users tend to use consecutive multiple emoji. However, no study to date has analyzed the pragmatic functions of emoji sequences systematically.
A few studies have discussed the syntactic relations within sequences, and between sequences and text, to some extent. For example, Danesi  postulated that emoji-only sequences possess a non-linear “iconic-conceptual” structure, similar to pictographic texts that permit an iconic connection between the forms and their referents to be inferred. Some sequences, however, follow the predominant subject-verb-object syntax of English sentences (Danesi, 2016). (This is also the normal word order in Chinese.) Relatedly, Schnoebelen (2012) observed that emoji users in English CMC have developed a ‘stance-first’ rule, according to which face-emoji that express the writer’s stance toward the content of a message are placed at the beginning of a sequence, before object emoji (see also the examples in Steinmetz, 2014). However, Tatman (2016) claimed that although some emoji sequences express an agent-patient relationship (i.e., subject-object), most do not have a fixed syntax. Dürscheid and Siever (2017) further claimed that ‘narrative sequences’ consisting exclusively of emoji (Herring and Dainas, 2017) are unlikely to possess syntactic structure, because important grammatical elements such as tense and mode cannot be represented through emoji.
Other scholars have commented on the relations between emoji sequences and adjacent text. The relations observed include text replacement (Danesi, 2016), repetition of text (McCulloch and Gawne, 2018), and text complementarity (Cramer, et al., 2016; Pohl, et al., 2017). Danesi (2016) explained that an insertion of emoji into text is often based on the notion of calquing — that is, the emoji are literal, word-for-word ‘translations’ of words or punctuation marks. An example is: I received [emoji representing gifts and flowers] from my friends. However, very few messages involve emoji sequences at the beginning or in-between textual utterances; it is more common to place a string of emoji at the end to repeat or complement the text (Cramer, et al., 2016). These findings align with findings on single emoji (Danesi, 2016; Novak, et al., 2015; Zhou, et al., 2017): Emoji tend to occur at the end of a message. These studies, although few, indicate that emoji sequences have become integrated into textual CMC and may work coherently with text.
As the review of past literature shows, there is considerable evidence that emoji sequences can substitute for words, exhibit word order patterns, fulfill pragmatic functions, and complement accompanying text. These are all properties of verbal utterances. Thus, a working premise of this study is that emoji sequences potentially express verbal meaning (although it may not always be straightforward to translate them into words) and function in some respects like verbal utterances.
To explore this premise systematically, we employ two language-focused approaches — speech act analysis (Searle, 1969) and rhetorical structure analysis (Mann and Thompson, 1988) — to address the following research questions:
RQ1: What types of speech acts do emoji sequences convey?
RQ2: How do emoji sequences relate rhetorically to the text they accompany?
These two approaches are complementary: Speech act analysis is concerned with the pragmatic meaning of each sequence considered as a self-standing unit, and rhetorical structure analysis is concerned with the relationship of each unit to its neighboring text. Thus each emoji sequence in a message that also contains text potentially expresses both a speech act and a rhetorical relation.
The literature relating to emoji sequences suggests that they can function as verbal utterances and convey ideational propositions (Danesi, 2016; Ge and Gretzel, 2018; Steinmetz, 2014). These, in turn, can fulfill various pragmatic functions, such as thanking, congratulation, and informing. The different types of actions that people perform by producing utterances are known as speech acts (Searle, 1969). Speech act analysis is a well-established technique that has been applied to both spoken and written language. In the off-line domain, it has been used to analyze face-to-face and telephone conversations, public speech, and letters (Aijmer, 2014). CMC studies have adopted speech acts to analyze chat room messages (Kapidzic and Herring, 2011), instant messages (Nastri, et al., 2006), Facebook status updates (Ilyas and Khushi, 2012), online travel reviews (Vásquez, 2011), and firm-initiated social media conversations (Ge, et al., 2018).
Recent research has gone further to establish a link between emoji and speech acts. Dos Reis, et al. (2018) examined the pragmatic intentions underlying emoji use, or what they call ‘intenticons,’ and Ge and Gretzel (2018) examined verbal actions — for instance, informing and requesting — conveyed through text and one or more emoji. Building on this body of research, we adopt the CMC act taxonomy developed by Herring, et al. (2005) to analyze the pragmatic meanings of emoji sequences. This taxnomy is a modified and simplified adaptation of the speech act categories developed by Francis and Hunston (1992) for spoken conversation and by Bach and Harnish (1979) for formal and deliberative discourse. Although designed for textual CMC, the CMC act taxonomy offers exhaustive categories of types of acts performed in online communication (Nemer, 2015). The categories of the CMC Act Taxonomy are summarized, defined, and illustrated in Appendix 1.
Rhetorical structure relations
Rhetorical structure theory (RST) (Mann and Thompson, 1988) is concerned with how textual units relate to each other logically and rhetorically. Its primary focus has been on written discourse, especially monologue genres such as expository and hortatory discourse (e.g., Mann and Thompson, 1992), but it has also been applied to spoken dialogue and multimedia discourse (Taboada and Habel, 2013; Taboada and Mann, 2006). In contrast, its application in CMC has focused on discourse parsing. For instance, Conroy, et al. (2015) and Rubin, et al. (2015) deployed RST to detect online fake news through identifying and comparing the coherence of deceptive and truthful messages.
Here we adopt the list of basic rhetorical relations from RST (Mann and Thompson, 1988) and postulate that emoji sequences stand in a rhetorical relation with the text that they accompany. That is, we treat an emoji sequence and its accompanying text as separate units and identify the rhetorical relation between them. The advantages of using RST relations are that they are well-defined and have been tested extensively for textual materials and multimodal documents such as text and figures (Das, et al., 2017; Taboada and Habel, 2013), although this is the first study to apply them to analyze emoji. The RST categories and their definitions are listed in Appendix 2.
Our data are Sina Weibo posts that contain emoji sequences, operationalized as instances of two or more semantically different emoji positioned together with no intervening text or punctuation. Because the pragmatic and rhetorical functions of emoji sequences is a new line of inquiry, we sought information-rich samples that captured the phenomena we were interested in (Bauer and Gaskell, 2000). This led us to sample emoji sequences as used by Chinese celebrities. Abidin (2015) reports that celebrities in Singapore often reply to their followers on Instagram with smiley faces and heart-shaped emoji as a way to express their acknowledgment and appreciation. Similarly, Ge and Gretzel (2018) found that Chinese celebrities are proficient emoji sequence users who incorporate multiple emoji in a single post to engage other users. Chinese celebrities are also active social media users in general, adept at producing relevant and interesting content (involving personal life, humor, social issues).
These celebrities can be conceptualized as lead users (Hippel, 1989) or social media influencers and thus as particularly important in informing understanding of emoji sequence use. Online celebrities influence other social media users and their fans as regards fashion, lifestyle, and ways of expression in online spaces (Koetse, 2017). It is reasonable to assume that this influence extends to emoji use. Indeed, we observed informally that general Weibo users tend to use emoji when responding to emoji posts initiated by popular celebrities. To examine this tendency further, we collected both 1) posts published by social media influencers (recognized celebrities or entertainment professionals, including actors/actresses, singers, and TV show hosts); and, 2) general user-posted comments responding to those posts.
We selected the celebrity accounts from the ‘Sina Weibo Influencer Popularity List.’ The list comprises 200 influencer accounts and was published on 6 April 2017 (http://data.weibo.com/top/hot/famous). We selected all of the accounts that met the criterion that the most recent 10 posts had at least one post containing an emoji sequence. This produced 87 accounts owned by 87 celebrities, 52 female and 35 male. We were not able to identify the general users’ gender, because they all used pseudonyms and avatars for their profile photos.
The use of manual methods of analysis and the exploratory nature of this study limited the amount of data that could be analyzed, because both call for close, iterative analysis (Bauer and Gaskell, 2000; White and Marsh, 2006). With this in mind, we conducted a three-step data collection procedure. First we collected the three most recent posts published by each of the 87 celebrities. We selected messages containing both text and emoji sequences, rather than messages containing only emoji sequences, in order to be able to evaluate how emoji sequences relate to their accompanying text. As a result of the first step, 289 posts were selected. In the second step, we removed sequences that did not satisfy our emoji sequence definition, including sequences consisting of only emoji with similar meanings (e.g., ) and sequences consisting of only two emoji in which the second emoji simply indicates skin tone (e.g., ); this left 211 posts. In the third step, we selected the three most recent user comments involving emoji sequences that were posted in response to the 211 posts posted by the celebrities, resulting in 58 posts. In total, our data set consists of 269 posts containing emoji sequences that were posted between 11 April and 28 April 2017.
We used computer-mediated discourse analysis, or “language-focused content analysis” (Herring, 2004), to analyze the communicative functions of emoji sequences at the levels of pragmatic and rhetorical meaning. Given that emoji sequences can convey propositions (Herring and Dainas, 2017), it is possible a priori that any and all of the speech act and RST categories could be found in our data. At the same time, emoji sequences do not (yet) possess a complete grammatical system, and their functioning like verbal utterances is an emergent phenomenon (Danesi, 2016). Thus it would not be surprising to find that there is not a perfect fit between the analytical categories and the emoji sequences. It is also possible that the sequences function in ways that are not reflected in the sets of acts and relations drawn from the literature. For this reason, we employed a mixed approach, coding each sequence for the categories in Appendices 1 and 2, while also adopting an inductive approach to let categories emerge from the data. Both types of analytical categories were successively refined through an iterative coding process (Bauer and Gaskell, 2000; White and Marsh, 2006). All 269 posts involving emoji sequences in the dataset were manually coded.
In order to code the sequences for speech acts and rhetorical relations, it was first necessary to identify the ideational meaning of the emoji sequences, i.e., translate them approximately into Chinese words. Because emoji translation is often subjective, we conducted emoji sequence translation in three steps. First, the authors translated and coded emoji sequences collaboratively. The first author, a native Chinese speaker and an emoji researcher, coded the entire data set; the second author, an emoji researcher and native English-speaking linguist with knowledge of the linguistic structure of Chinese, coded a large amount of the data together with the first author until consensus on the most likely interpretation of each emoji sequence in context was reached . The emoji definitions provided by Weibo helped during this process.
At the end of the collaborative coding, 35 ambiguous sequences remained. We then conducted a second step by recruiting seven crowdsourced coders on Witmart (http://www.witmart.com/cn/, a Chinese crowdsourcing platform similar to Amazon’s Mechanical Turk) to provide interpretations of the 35 ambiguous emoji and emoji sequences. We recruited coders by posting a job description and the required qualifications on Witmart, and we selected competent coders by asking them to code 10 test items. This step resolved 26 ambiguous emoji sequences. For the remaining nine unclear cases, in a third step, we consulted three proficient emoji users recommended by a Chinese graphicon design firm. These emoji users worked as emoji consultants for the firm and thus can be considered expert sources. As a result of this step, all nine remaining ambiguous cases were resolved.
In order for an emoji sequence to be coded as a speech act using the CMC taxonomy, it must function like a verbal utterance, meaning that it has to include a verbal or predicating element. In our data, this was usually an emoji indicating an action ( ‘hug’), activity ( ‘dance’), or gesture ( ‘come here’). Sometimes, though, there was no explicit verbal element, and we interpreted emoji representing people or things as implied predicates. For example, in the sequence ( ‘do not have beer and ice cream’), the first emoji ‘woman gesturing no’ was treated as a predicating element. In another sequence, a celebrity used an emoji microphone to signify the action of singing ( ‘sing a song’). Emoji sequences for which no verbal or predicating element was explicit or inferable were coded as ‘n/a’ (e.g., ‘different kinds of fruit’).
To code rhetorical relations, we treated an emoji sequence and its accompanying text as separate units and attempted to determine the relation of the former to the latter based on the RST categories from the literature. Some examples could not be coded by applying RST; we grouped these into two categories. Emoji sequences that substituted for words in a text utterance were coded as ‘partial substitution,’ and sequences that repeated certain words in the text were coded as ‘partial repetition.’ The results of the coding are presented in the next section.
The emoji represented in our data include all of the categories in Unicode Emoji 12.0 (e.g., smileys and people, animals and nature, objects, activities, food and drink, travel and places, symbols, flags). The mean number of emoji used across all sequences is 4 , with a mode of 3 and a standard deviation of 2.52. In what follows, we present the frequency distribution results of the speech act and the rhetorical structure analyses first, followed by examples illustrating each category.
Types of speech acts
In the Weibo emoji sequences, we identified the speech acts ‘claim,’ ‘desire,’ ‘explain,’ ‘thank,’ and ‘repair’ from the CMC act taxonomy. In addition, the act ‘congratulate’ (example 1) emerged as important in our data; we defined it as “praise someone for an achievement; convey good wishes when something special or pleasant has happened to someone.” Based on our data, we also narrowed the ‘react’ act from the CMC taxonomy to ‘endorse,’ defined as “show approval or support of someone or something.” Further, the act ‘manipulate’ was adopted from Kapidzic and Herring (2011) as a collective term for the acts ‘direct,’ ‘request,’ and ‘invite,’ which occurred relatively infrequently in our data. As Table 1 shows, the frequencies of these acts in the celebrities’ and in the general users’ posts are similar overall.
Table 1: Emoji sequence acts: Frequency distribution of celebrities and general users. Categories Number Percent Number Percent Number Percent Celebrities’ posts General users’ posts All posts All posts Claim 91 43.1% 26 44.9% 117 43.5% Desire 25 11.8% 12 20.7% 37 13.8% Explain 25 11.8% 3 5.2% 28 10.4% Congratulate 20 9.5% 6 10.3% 26 9.7% Manipulate 17 8.1% 5 8.6% 22 8.2% Endorse 17 8.1% 2 3.4% 19 7.1% n/a 8 3.8% 3 5.2% 11 4.1% Thank 6 2.8% 1 1.7% 7 2.6% Repair 2 1.0% 0 0% 2 0.6% Total 211 100% 58 100% 269 100%
The users in our dataset often employed emoji sequences to express feelings and opinions (‘claim’); to express hope, wishes, or promises (‘desire’); and to comment on or explain something (‘explain’). Conversely, they rarely used emoji sequences to express gratitude (‘thank’) or to clarify what was said in the text (‘repair’). The ‘endorse’ and ‘manipulate’ acts were also relatively infrequent. Emoji sequences in the ‘n/a’ category lack a predicting element and therefore do not convey speech acts. Examples of each act are presented further below.
Compared to the celebrities, the general users expressed more ‘desire’ acts, typically to communicate that they hoped or promised to watch their favorite celebrities’ performances. General users also produced fewer ‘explain’ acts, in part because their comments were relatively short compared to the celebrities’ posts.
Types of rhetorical relations
Emoji sequences appeared most often at the end of posts, followed by in-between two segments of text. Sequences were inserted in-between text either as an integral component of a textual utterance (substituting for words) or as an independent utterance between two independent textual utterances. The celebrities placed emoji sequences between text segments more than the general users did. Conversely, the general users more often placed emoji sequences at the beginning of their posts, although that position was least favored overall (Table 2).
Table 2: Locations of emoji sequences. Position Number Percent Number Percent Number Percent Celebrities’ posts General users’ posts All posts All posts Beginning of post 13 6.2% 8 13.8% 21 7.8% In-between text 46 21.8% 4 6.9% 50 18.6% End of post 152 72.0% 46 79.3% 198 73.6% Total 211 100% 58 100% 269 100%
The rhetorical structure analysis revealed six ways the emoji sequences relate to their accompanying text: ‘restatement,’ ‘evaluation,’ ‘elaboration,’ ‘motivation,’ ‘result,’ and ‘enablement.’ In addition, we identified two categories to which the RST categories cannot apply, by definition: ‘partial substitution’ and ‘partial repetition.’ The users in our dataset sometimes inserted an emoji sequence to substitute for words in the text, that is, partial substitution. Moreover, some of the emoji sequences merely repeat certain words presented in the text, that is, partial repetition. The emoji sequences in these two categories do not convey independent, complete propositions and therefore cannot stand in a rhetorical relation with other propositions. As Table 3 shows, the findings for the celebrities and the general users are similar overall, except for the ‘result’ and ‘partial substitution’ relations.
Table 3: Rhetorical relations: Frequency distribution of celebrities and general users. Categories Number Percent Number Percent Number Percent Celebrities’ posts General users’ posts All posts All posts Restatement 79 37.4% 21 36.2% 100 37.2% Evaluation 67 31.8% 20 34.5% 87 32.3% Result 12 5.7% 10 17.3% 22 8.2% Elaboration 15 7.1% 3 5.2% 18 6.7% Partial substitution 16 7.6% 1 1.7% 17 6.3% Motivation 12 5.7% 1 1.7% 13 4.8% Enablement 6 2.8% 1 1.7% 7 2.6% Partial repetition 4 1.9% 1 1.7% 5 1.9% Total 211 100% 58 100% 269 100%
Overall, ‘restatement’ and ‘evaluation’ are the most two prominent categories identified in the Weibo users’ posts. All of the users inserted emoji sequences (expressing acts such as ‘claim’ and ‘manipulate’) to repeat propositions stated in the text and to offer subjective comments on entities or ideas mentioned in the text. In contrast, ‘elaboration’, ‘motivation’, and ‘enablement’ are little-utilized categories. The users in our data only rarely used emoji sequences (such as explanation acts) to explain what was said in the text or to provide a reason for, or details about, performing the action described in the text.
Compared to the celebrities, the general users had a higher percentage of ‘result’ sequences. For instance, they often inserted emoji sequences expressing congratulations and thanks as a consequence of celebrities’ achievements and to express their gratitude for the inspiration and encouragement received from the celebrities. Examples of each rhetorical relation are presented and discussed in the following section.
Examples of emoji sequences
An emoji sequence may both express a speech act (if the sequence includes a predicating element) and stand in a rhetorical relation with its accompanying text. The following examples illustrate some of the most common co-occurrences of the two functions found in the Weibo data.
Example 1. An emoji sequence posted by a general user illustrating a ‘congratulate’ act and in a ‘motivation’ relation with the text.
A free English translation of the Chinese text in example 1 is: ‘We should go out for a drink.’ The first emoji in the sequence means ‘receive’; the following four emoji depict ‘four awards’; the last emoji means ‘celebration’ or ‘congratulations.’ The overall meaning is something like ‘Congratulations on receiving four awards.’ This emoji sequence presents a reason for going out for a drink and therefore stands in a motivation relationship with the preceding text.
Example 2. An emoji sequence posted by a celebrity illustrating a ‘manipulate’ act and in a ‘restatement’ relation with the text.
A free translation of the text in example 2 is: ‘How can you take a pretty selfie? Please click the following image.’ The first emoji in the sequence functions like a deictic verb, pointing to the image; the second emoji represents the speaker or message author (Ge and Gretzel, 2018). The overall meaning is that the celebrity asks readers to ‘go (click the image) below, I (say).’ Therefore, it repeats what was said in the text.
Example 3. An emoji sequence posted by a celebrity expressing a ‘claim’ and in an ‘evaluation’ relation with the text.
A free translation of the text in example 3 is: ‘The food is ready.’ The first emoji represents ‘food’ (rice); the second emoji means ‘spicy’; and the last emoji means ‘love.’ The overall meaning of the sequence is ‘I love spicy food.’ It indicates the speaker’s subjective assessment of the food mentioned in the text.
Example 4. An emoji sequence posted by a celebrity expressing ‘desire’ and in an ‘enablement’ relation with the text.
A free translation of the Chinese text in example 4 is: ‘We need to lose lots of weight and become very thin this summer.’ The emoji sequence consists of an emoji for ‘running,’ the female gender symbol (indicating the speaker herself), a person saying no, a glass of beer, and a dish of ice cream. The overall meaning is something like ‘I will run and stay away from beer and ice cream.’ This sequence presents specific details that facilitate performing the action of losing weight described in the text.
Example 5. An emoji sequence posted by a celebrity illustrating an ‘explain’ act and in an ‘elaboration’ relation with the text.
The text in example 5 can be loosely translated as: ‘Ok, I did what you said.’ The emoji sequence is made up of four emoji: a happy face (defined by Weibo as ‘haha’), getting a haircut, light skin tone , and the male gender symbol (indicating the speaker himself). The overall meaning is something like, ‘LOL I got a haircut.’ The sequence adds more specific information to the text about what the speaker did.
Example 6. An emoji sequence posted by a general user expressing a ‘thank’ act and in a ‘result’ relation with the text.
A free translation of the Chinese text in example 6 is: ‘Your music encouraged me when I was feeling low.’ The first emoji expresses thanking; the second one (a dog face) indicates the speaker. The overall meaning is ‘I thank you.’ The sequence is a result caused by the past situation described in the text.
Example 7. An emoji sequence posted by a celebrity expressing ‘repair’ and in an ‘elaboration’ relation with the text.
A loose translation of the text in this example is: ‘ Friendship is over.’ The emoji sequence that follows consists of three emoji: a chuckling face, a kiss, and a dog face (indicating the speaker). The overall meaning is something like, ‘I’m just kidding; kiss.’ The sequence comments on the statement presented in the text, communicating that the celebrity is joking.
Example 8. An emoji sequence posted by a general user expressing an ‘endorse’ act and in an ‘evaluation’ relation with the text.
The Chinese text in example 8 means, ‘This is sister Na’s first book.’ The first emoji in the sequence indicates support or approval; the second emoji (defined by Weibo as ‘onlookers’) means ‘we’ in this context. The overall meaning is ‘We approve (of it).’ The emoji sequence thus indicates the subjective evaluation by the poster (representing himself or herself and other fans of the celebrity) of the book mentioned in the text.
Example 9. An emoji sequence posted by a celebrity that does not convey a speech act (‘n/a’) and is in a ‘partial substitution’ relation with the text.
The Chinese text loosely translates as: ‘To successfully lose weight, I am determined to only eat these for the next two weeks .’ The five fruit emoji (watermelon, tangerine, apple, strawberry, banana) are functionally equivalent to nouns and do not convey a speech act that is separate from the speech act expressed by the text. The sequence also partially substitutes for words in the text.
Example 10. An emoji sequence posted by a general user that does not convey a speech act (‘n/a’) and is in a ‘partial repetition’ relation with the text.
The text in this example can be freely translated as: ‘A singer needs to be in tears in front of the audience in order to be popular.’ The two emoji ( man, microphone) are nouns and together express the concept ‘singer’; they do not convey a speech act. They merely repeat the word ‘singer’ in the text that precedes them.
Innovations in emoji use
The above examples illustrate several user innovations that are relevant to the claim that emoji are becoming a language. Celebrities and general users on Weibo adapt emoji that represent objects as verbal elements (e.g., a microphone to signify singing or broadcasting) and incorporate conventional abstract symbols (e.g., ‘awesome’) and gender signs (e.g., ). We also observed that users sometimes repeat action emoji as a means of amplification (e.g., ‘thanks a lot’) and repeat object emoji as a way of indicating plurality (e.g., ‘3 bowls of rice plus a glass of beer’).
Moreover, while the literature suggests that the syntactic structure of emoji sequences in English is linear and follows basic English word order (stance or subject precedes the object, e.g., ‘(I) dislike taking the bus’) (Danesi, 2016; Schnoebelen, 2012), the sequences in our data more often occur in the opposite order. That is, the stance-expressing subject tends to follow the object, e.g., ‘I like coffee,’ literally ‘coffee I-like.’ This can be seen in examples 2, 6, 7, and 8. The male gender symbol in example 5 also seems to illustrate this pattern.
Research question revisited
The emoji sequences in our Weibo data very often function like stand-alone utterances and express illocutionary forces, similar to verbal speech acts. Both the celebrities and the general users who follow them on Weibo employ emoji sequences to express a variety of act types. Our first research question asked, what speech acts do emoji sequences convey?
The most frequent act category (‘claim’ or expressing feelings and opinions) is consistent with the findings of many previous studies that emoji are emotionally expressive (e.g., Danesi, 2016). Emoji sequences also convey other pragmatic meanings, however, such as expressing wishes, congratulating, and explaining textual content. This shows that their role in CMC is more complex than simply expressing feelings, which is also the case for single emoji (e.g., Illendula and Yedulla, 2018).
What is perhaps more surprising is that only 4.1 percent of the sequences did not lend themselves to interpretation as speech acts (i.e., were coded as n/a), meaning that the vast majority (95.9 percent) of the emoji sequences in our data function pragmatically like verbal utterances. Yet we found only eight act types compared to the 16 acts in the CMC taxonomy, which could be taken to suggest that the pragmatic expressive potential of emoji sequences is limited compared to that of verbal utterances (or that our sample is not large enough to manifest the less-common act types). In these data, at least, it appears that emoji sequences specialize for certain pragmatic meanings.
Our second research question asked how emoji sequences relate to the text they accompany. The Weibo findings support our assumption that emoji sequences, like verbal utterances, form rhetorical relations with adjacent utterances. Placing an emoji sequence at the end of a post is a common user practice, consistent with the placement of single emoji (Cramer, et al., 2016; Danesi, 2016; Novak, et al., 2015; Zhou, et al., 2017). The most-used relation type, restatement, also partially aligns with previous studies that found that users often insert emoji to mention or emphasize what was said in the text (Herring and Dainas, 2017; Pohl, et al., 2017). Meanwhile, the second-most frequent category, evaluation, shows that emoji sequences do not merely repeat what was said in the preceding text, in contrast to the English-language findings of McCulloch and Gawne (2018).
Interestingly, while rhetorical structure theory (RST) suggests that the restatement component and the accompanying text are of roughly equal size, we found that emoji sequences are typically shorter than the text they accompany. This supports the idea that emoji can condense information (Benenson, 2010; Danesi, 2016). Moreover, restatement and evaluation — the most prominent categories in this study — are the least-used categories in rhetorical relations in textual materials and multimodal documents (e.g., text and figures/tables) in previous RST studies (Mann and Thompson, 1988; Das, et al., 2017). Finally, we found only six out of the 25 rhetorical relations in RST, suggesting that the range of emoji sequence relations is limited compared to purely textual relations. These results show that although emoji sequences can function like verbal utterances and form relations with textual propositions, their usage differs from textual utterances in several respects.
The identification of the ‘partial substitution’ and ‘partial repetition’ relations, although together accounting for less than 10 percent of the sequences, points to another difference between emoji sequences and textual utterances. Unlike in traditional RST, where the propositions (typically, clauses) and the relations between them are governed by the syntactic structure of the language (Mann and Thompson, 1988), the patterning of emoji sequences is looser and more fluid (Danesi, 2016). Thus sequences might not always have coherent relations with their neighboring text. Taken together, these observations indicate that emoji sequences are not as conventionalized as verbal clauses in terms of their structure and functions.
At the same time, Weibo users are clearly developing distinctive patterns of emoji use. They repeat emoji to indicate amplification and plurality and use object emoji as verbal elements. Furthermore, they tend to put stance-indicating emoji after emoji that represent action/activity (similar to Verb-Subject word order) (examples 5, 6, and 8). Pragmatically, this stance-final structure parallels the common practice of using a single emoji that expresses attitude or emotion at the end of a textual utterance to indicate the tone of the utterance as a whole (Kelly and Watts, 2015; Novak, et al., 2015). Syntactically, the subject-final emoji structure resembles the phenomenon of ‘backgrounding’ a clausal constituent by postposing it to the right of the rest of the clause, as a way of assigning to it a de-emphasized status (Herring, 1994). If this interpretation is correct, the emoji sequence users in our dataset use emoji order to foreground actions, activities, and objects and to background their own presence, while expressing an attitude or stance that colors the interpretation of the sequence as a whole. This tendency to background the self is consistent with a Chinese cultural tendency to be self-effacing (Oliver, 1969).
The implications we derive from the aforementioned user practices are twofold: First, Weibo emoji sequence users are not simply calquing the Chinese language, as Danesi (2016) observed for English emoji, because Chinese word order is similar to English (Subject-Verb-Object). Rather, Sina Weibo users are innovating their own emoji order pattern. Second, it has been claimed that “emojis have become a universal language that is used across apps, across platforms, and across cultures” . The language-particularity of the patterns in our data suggest that while emoji may be evolving into a language, emoji language is not the same across cultures. In other words, it may be more accurate to speak of “emoji languages” than a single, universal “emoji language.”
The findings of this study have implications for emoji design. First, given the organic way in which emoji usage is evolving, designers of emoji and social media platforms should be guided by emergent user practices, as well as by established usage. The emerging trend on Sina Weibo is to create emoji sequences that function like verbal utterances. Second, in order to make emoji a useful and effective form of digitally-mediated language, in keeping with this trend, emoji design should consider not only the aesthetic and representational aspects of emoji, but also their linguistic aspects.
Current emoji sets, including Emoji Unicode, lack many of the components necessary to constitute a fully-functioning language. Therefore, we recommend that emoji sets add emoji that can be used to express grammatical categories and grammatical relations. The use of emoji objects to express verbal ideas shows that there is a need for more emoji that indicate actions and activities, as well as for verbal inflections such as tense and aspect. An example of the latter is the emoji ‘soon’ , but emoji that indicate ‘past,’ ‘now,’ ‘future,’ ‘habitually,’ and so forth are also needed to expand the narrative potential of emoji. Moreover, emoji are needed that can function as connectives (e.g., but, and, therefore), because connectives provide an explicit means of indicating rhetorical relations. Further, Weibo users like to use animal emoji to refer to themselves and their friends (Ge and Gretzel, 2018). The adoption of animal emoji and gender symbols to represent the self points to a need for emoji that function like personal pronouns. Finally, the use of hearts, roses, and smiling face emoji as utterance endings suggests a trend toward using emoji as punctuation (a trend that is already well established for emoticons; see Dresner and Herring, 2010). While some explicit punctuation emoji exist (e.g., ), the graphical nature of emoji is ideally suited for representing less standard punctuation, such as the interrobang (), the rhetorical question mark (), the love point (), the doubt point (), and the sarcastic mark ().
We also recommend that social media platforms provide preformulated emoji sequence options. The use of emoji sequences to express ‘thank,’ ‘congratulate,’ and ‘endorse’ acts suggests that a convenient shorthand for typing sequences expressing common expressions such as “Happy New Year,” “Congratulations on your marriage,” and “Thank you for your support” would be appreciated by users, rather than having to search through emoji menus to construct the desired sequences. Such preformulated sequences would build on two of the main advantages of emoji use, that is, efficiency and convenience (Danesi, 2016).
Finally, given that some emoji sequences in both English and Chinese exhibit a preferred ordering of elements, particularly in relation to the position of the subject or writer’s stance, we recommend that social media platforms add an option for users to sort emoji in menus according (broadly) to syntactic function. Verb emoji, noun emoji, and emoji that function as grammatical markers could be grouped separately in order to make constructing emoji utterances more efficient . This suggestion does not contradict our conclusion above that emoji grammar differs across languages, since all languages have nouns, verbs, words (or parts of words) that function as connectors, time modifiers, and so forth, and not all syntactic emoji categories would need to be selected for every language.
The popular press (at least in the English- and Chinese-speaking worlds) is rife with speculation that emoji are becoming a new, global language (Cheung, 2017; Cohn, 2015; Thompson, 2016). The most compelling evidence in support of this view is emoji sequences. It is increasingly common for social media users to create strings of utterance-like emoji to convey ideas and information, express opinions and emotions, and even tell stories of sorts (Herring and Dainas, 2017; Steinmetz, 2014). Our study largely supports the idea that emoji are developing into an independent language: They can substitute for words, and emoji sequences can resemble complete utterances with subject, verb, and object. In addition, we have shown that emoji sequences can fulfill certain communicative functions that were previously associated only with verbal utterances.
At the same time, emoji have not yet reached the point of being a fully functioning language, even on Sina Weibo. The most common types of rhetorical and pragmatic functions emoji sequences express differ from those for text sequences, and their expressive range is more limited. Emoji sets also lack words to express grammatical functions such as tense and number, as well as articles and conjunctions, and the syntactic structure of sequences is emergent and flexible, rather than strictly rule governed. Therefore, at this time we believe it is most accurate to describe the linguistic status of emoji as an emerging graphical language.
The emergence of an emoji language appears to be further advanced in China than in the West. McCulloch and Gawne (2018) found that emoji sequences in their English data primarily repeated the text they follow, whereas our results show that emoji sequences on Sina Weibo have additional functions, being used also to evaluate and elaborate on the text. And whereas Danesi reported that emoji sequences in English-speaking contexts tend to follow the basic word order of English (and often simply substitute emoji for English word-for-word), we found evidence of an innovative, emerging emoji order pattern in the Chinese sequences on Sina Weibo. These findings argue against the view that emoji are a “universal” language.
Finally, we applied speech act analysis and extended rhetorical structure analysis for the first time to analyze graphicons in CMC. The findings clearly show that the functions of emoji sequences extend beyond expressing emotion, nonverbal behavior, or indicating tone, as has been claimed for single emoji (e.g., Pavalanathan and Eisenstein, 2016; Zhou, et al., 2017). Rather, we provided evidence that such sequences are taking on characteristics of verbal language in their pragmatic and rhetorical functions, as well as through user innovations that compensate for the lack of a formal emoji grammar and that constitute an emergent grammar of sorts. We believe that this trend is likely to continue, increasing the complexity and expressive potential of emoji sequences and rendering them more language-like in the future.
About the authors
Jing Ge is a postdoctoral research fellow in the Anthropology Department at the University of California, Berkeley. She has a Ph.D. in marketing communication from the UQ Business School at the University of Queensland, Australia and has close to ten years of online communication industry experience. Her research focuses on computer-mediated communication, the language businesses and consumers use on social media, and visual communication.
Direct comments to jingge [at] berkeley [dot] edu
Susan C. Herring is Professor of Information Science and Linguistics and Director of the Center for Computer-Mediated Communication at Indiana University Bloomington. She received her M.A. and Ph.D. in linguistics from the University of California, Berkeley. She has researched structural, pragmatic, interactional, and social phenomena in digital communication and is the founder of the Computer-Mediated Discourse Analysis (CMDA) paradigm. A past editor of the Journal of Computer-Mediated Communication, she currently edits the online journal Language@Internet. Her current interests include multimodal CMC and telepresence robot-mediated communication..
E-mail: herring [at] indiana [dot] edu
1. This definition excludes sequences comprised only of the same emoji repeated for emphasis, as well as sequences comprised only of emoji with closely-related meanings, such as ‘heart’ and ‘face with heart-eyes.’ On the frequency of such occurrences in English-language messages, see McCulloch and Gawne (2018).
2. The distinction between emoji and stickers is sometimes conflated in Asian scholarship. In this article, we distinguish emoji (small, colorful graphics that represent faces, objects, actions, and symbols) from stickers (larger, more complex graphics that represent specific characters and characteristics); see Herring and Dainas (2017).
3. For example, the emoji (meaning ‘embarrassed’) replicates the Chinese character 囧, which resembles an embarrassed face.
4. Emoji Unicode 12.0 Web site: https://emojipedia.org/emoji-12.0/.
5. The expression ‘verbal language’ is used here to include both speaking and writing.
6. Danesi, 2016, p. 81.
7. In this process, the first author orally translated the jointly-coded sequences into English for the second author, and the second author iteratively queried the first author until mutual understanding was achieved.
8. The higher mean is caused by some sequences that contain emoji repeated multiple times for emphasis.
9. The skin tone image is supposed to be integrated into the preceding emoji, but it displayed separately on our devices.
10. Ai, et al., 2017, p. 2.
11. Emoji that can function as both nouns and verbs would have multiple category designations in such a system.
Crystal Abidin, 2015. “Communicative ❤ intimacies: Influencers and perceived interconnectedness,” Ada: A Journal of Gender, New Media, and Technology. number 8, at https://adanewmedia.org/2015/11/issue8-abidin/, accessed 23 August 2017.
doi: https://doi.org/10.7264/N3MW2FFG, accessed 25 October 2018.
Wei Ai, Xuan Lu, Xuanzhe Liu, Ning Wang, Gang Huang, and Qiaozhu Mei, 2017. “Untangling emoji popularity through semantic embeddings,” Proceedings of the Eleventh International AAAI on Web Social Media, pp. 2–11, and at https://aaai.org/ocs/index.php/ICWSM/ICWSM17/paper/view/15705, accessed 25 October 2018.
Karin Aijmer, 2014. Conversational routines in English: Convention and creativity. New York. Routledge.
Sarah Alismail and Hengwei Zhang, 2018. “The use of emoji in electronic user experience questionnaire: An exploratory case study,” Proceedings of the 51st Hawaii International Conference on System Sciences, pp. 3,366–3,375.
doi: https://doi.org/10.24251/HICSS.2018.427, accessed 25 October 2018.
Margaret Andersen, 2018. “How feminists in China are using emoji to avoid censorship,” Wired (30 March), at https://www.wired.com/story/china-feminism-emoji-censorship/, accessed 18 September 2018.
Kent Bach and Robert M. Harnish, 1979. Linguistic communication and speech acts. Cambridge, Mass.: MIT Press.
Martin W. Bauer and George Gaskell (editors), 2000. Qualitative researching with text, image and sound: A practical handbook. London: Sage.
Janet B. Bavelas and Nicole Chovil, 2000. “Visible acts of meaning: An integrated message model of language in face-to-face dialogue,” Journal of Language and Social Psychology, volume 19, number 2. pp. 163–194.
doi: http://dx.doi.org/10.1177/0261927X00019002001, accessed 25 October 2018.
Fred Benenson (editor and compiler), 2010. Emoji Dick; or the 🐳, at http://www.emojidick.com, accessed 25 October 2018.
Zhenpeng Chen, Xuan Lu, Wei Ai, Huoran Li, Qiaozhu Mei, and Xuanzhe Liu, 2018. “Through a gender lens: Learning usage patterns of emojis from large-scale Android users,” WWW ’18: Proceedings of the 2018 World Wide Web Conference on World Wide Web, pp. 763–772.
doi: http://dx.doi.org/10.1145/3178876.3186157, accessed 25 October 2018.
Rachel Cheung, 2017. “How emojis became the modern world’s status symbols — and how they’ve crossed from messaging apps to real life,” Southern China Morning Post (2 April), at http://www.scmp.com/lifestyle/article/2083504/how-emojis-became-modern-worlds-status-symbols-and-how-theyve-crossed, accessed 19 March 2018.
Neil Cohn, 2015. “Will emoji become a new language?” BBC (13 October), at http://www.bbc.com/future/story/20151012-will-emoji-become-a-new-language, accessed 19 March 2018.
Niall J. Conroy, Victoria L. Rubin, and Yimin Chen, 2015. “Automatic deception detection: Methods for finding fake news,” Proceedings of the Association for Information Science and Technology, volume 52, number 1, pp. 1–4.
doi: https://doi.org/10.1002/pra2.2015.145052010082, accessed 25 October 2018.
Henriette Cramer, Paloma de Juan, and Joel Tetreault, 2016. “Sender-intended functions of emojis in US messaging,” MobileHCI ’16: Proceedings of the 18th International Conference on Human-Computer Interaction with Mobile Devices and Services, pp. 504–509.
doi: https://doi.org/10.1145/2935334.2935370, accessed 25 October 2018.
Marcel Danesi, 2016. The semiotics of emoji: The rise of visual language in the age of the Internet. London: Bloomsbury.
Debopam Das, Maite Taboada, and Manfred Stede, 2017. “The good, the bad, and the disagreement: Complex ground truth in rhetorical structure analysis,” Proceedings of the 6th Workshop on Recent Advances in RST and Related Formalisms, pp. 11–19, and http://aclweb.org/anthology/W17-3602, accessed 25 October 2018.
Gabriele de Seta, 2018. “Biaoqing: The circulation of emoticons, emoji, stickers, and custom images on Chinese digital media platforms,” First Monday, volume 23, number 9, at http://firstmonday.org/article/view/9391/7566, accessed 18 September 2018.
doi: http://dx.doi.org/10.5210/fm.v23i9.9391, accessed 25 October 2018.
Julio Cesar dos Reis, Rodrigo Bonacin, R., Heiko Horst Hornung, and Maria Cecília Calani Baranauskas, 2018. “Intenticons: Participatory selection of emoticons for communication of intentions,” Computers in Human Behavior, volume 85, pp. 146–162.
doi: https://doi.org/10.1016/j.chb.2018.03.046, accessed 25 October 2018.
Eli Dresner and Susan C. Herring, 2010. “Functions of the nonverbal in CMC: Emoticons and illocutionary force,” Communication Theory, volume 20, number 3, pp. 249–268.
doi: https://doi.org/10.1111/j.1468-2885.2010.01362.x, accessed 25 October 2018.
Christa Dürscheid and Christina M. Siever, 2017. “Jenseits des alphabets — Kommunikation mit emojis,” Zeitschrift fü'r Germanistische Linguistik, volume 45, number 2, pp. 256–285.
doi: https://doi.org/10.1515/zgl-2017-0013, accessed 25 October 2018.
Gill Francis and Susan Hunston, 1992. “Analyzing everyday conversation,” In: Malcolm Coulthard (editor). Advances in spoken discourse analysis. London: Routledge, pp.123–161.
Jing Ge and Ulrike Gretzel, 2018. “Emoji rhetoric: A social media influencer perspective,” Journal of Marketing Management (10 July).
doi: https://doi.org/10.1080/0267257X.2018.1483960, accessed 25 October 2018.
Jing Ge, Ulrike Gretzel, and Yuanxia Zhu, 2018. “Humour in firm-initiated social media conversations: A conceptual model,” International Journal of Digital Culture and Electronic Tourism, volume 2, number 4, pp.273–293.
doi: https://doi.org/10.1504/IJDCET.2018.092180, accessed 25 October 2018.
Susan C. Herring, 2004. “Computer-mediated discourse analysis: An approach to researching online behavior,&edquo; In: Sasha Barab, Rob Kling, and James H. Gray (editors). Designing for virtual communities in the service of learning. New York: Cambridge University Press, pp. 338–376.
doi: https://doi.org/10.1017/CBO9780511805080.016, accessed 25 October 2018.
Susan C. Herring, 1994. “Afterthoughts, antitopics, and emphasis: The syntacticization of postverbal position in Tamil,” In: Miriam Butt, Tracy Holloway King, and Gillian Ramchand (editors). Theoretical approaches to word order in South Asian languages. Stanford, Calif.: Center for the Study of Language and Information, pp. 119–152.
Susan C. Herring and Ashley R. Dainas, 2017. “‘Nice picture comment!’ Graphicons in Facebook comment threads,” Proceedings of the 50th Hawaii International Conference on System Sciences, pp. 2,185–2,194.
doi: https://doi.org/10.24251/HICSS.2017.264, accessed 25 October 2018.
Susan C. Herring, Anupam Das, and Shashi Penumarthy, 2005. “CMC Act Taxonomy,” at http://ella.slis.indiana.edu/~herring/cmc.acts.html, accessed 12 June 2015.
Eric von Hippel, 1989. “New product ideas from ‘lead users’,” Research-Technology Management, volume 32, number 3, pp. 24–27.
doi: https://doi.org/10.1080/08956308.1989.11670596, accessed 25 October 2018.
Anurag Illendula and Manish Reddy Yedulla, 2018. “Learning emoji embeddings using emoji co-occurrence network graph,” arXiv (20 June), at https://arxiv.org/abs/1806.07785, accessed 2 July 2018.
Sanaa Ilyas and Qamar Khushi, 2012. “Facebook status updates: A speech act analysis,” Academic Research International, volume 3, number 2, pp. 500–507, and at http://www.savap.org.pk/journals/ARInt./Vol.3(2)/2012(3.2-63).pdf, accessed 25 October 2018.
Sanja Kapidzic and Susan C. Herring, 2011. “Gender, communication, and self-presentation in teen chatrooms revisited: Have patterns changed?” Journal of Computer-Mediated Communication, volume 17, number 1, pp. 39–59.
doi: https://doi.org/10.1111/j.1083-6101.2011.01561.x, accessed 25 October 2018.
Ryan Kelly and Leon Watts, 2015. “Characterising the inventive appropriation of emoji as relationally meaningful in mediated close personal relationships,” at https://core.ac.uk/download/pdf/38148839.pdf, accessed 8 June 2017.
Manya Koetse, 2017. “Chinese celebrities with most Weibo followers” (11 March), at https://www.whatsonweibo.com/chinese-celebrities-weibo-followers-top-10-2017/, accessed 8 May 2018.
Wenda Li, 2017. “‘Face With Tears of Joy’ is most popular emoji in China,” gbTimes (4 December), at https://gbtimes.com/face-with-tears-of-joy-is-most-popular-emoji-in-china, accessed 18 September 2018.
Nikola Ljubešić and Darja Fišer, 2016. “A global analysis of emoji usage,” Proceedings of the 10th Web Corpus Workshop (WAC-X) and the EmpiriST Shared Task, pp. 82–89, and at at http://www.aclweb.org/anthology/W16-2610, accessed 25 October 2018.
Xuan Lu, Wei Ai, Xuanzhe Liu, Qian Li, Ning Wang, Gang Huang, and Qiaozhu Mei, 2016. “Learning from the ubiquitous language: An empirical analysis of emoji usage of smartphone users,” UbiComp ’16: Proceedings of the 2016 ACM International Joint Conference on Pervasive and Ubiquitous Computing, pp. 770–780.
doi: https://doi.org/10.1145/2971648.2971724, accessed 25 October 2018.
Xiaojuan Ma, 2016. “From Internet memes to emoticon engineering: Insights from the Baozou comic phenomenon in China,” In: Masaaki Kurosu (editor). Human-computer interaction. Novel user experiences. Lecture Notes in Computer Science, volume 9733. Cham: Springer, pp. 15–27.
doi: https://doi.org/10.1007/978-3-319-39513-5_2, accessed 25 October 2018.
William C. Mann and Sandra A. Thompson (editors), 1992. Discourse description: Diverse linguistic analyses of a fund-raising text. Philadelphia: John Benjamins.
William C. Mann and Sandra A. Thompson. 1988. “Rhetorical structure theory: Toward a functional theory of text organization,” Text & Talk, volume 8, number 3, pp. 243–281.
doi: https://doi.org/10.1515/text.1.19126.96.36.199, accessed 25 October 2018.
Gretchen McCulloch and Lauren Gawne, 2018. “Emoji grammar as beat gestures,” Proceedings of the First International Workshop on Emoji Understanding and Applications in Social Media, at http://ceur-ws.org/Vol-2130/short1.pdf, accessed 28 June 2018.
Johanna Monti, Federico Sangati, Francesca Chiusaroli, Martin Benjamin, and Sina Mansour, 2016. “Emojitalianobot and EmojiWorldBot: New online tools and digital environments for translation into emoji,” Proceedings of Third Italian Conference on Computational Linguistics, pp. 211–215, at http://ceur-ws.org/Vol-1749/paper37.pdf, accessed 25 October 2018.
Jacqueline Nastri, Jorge Peña, and Jeffrey T. Hancock, 2006. “The construction of away messages: A speech act analysis,” Journal of Computer–Mediated Communication, volume 11, number 4, pp. 1,025–1,045.
doi: https://doi.org/10.1111/j.1083-6101.2006.00306.x, accessed 25 October 2018.
David Nemer. 2015. “Celebrities acting up: A speech act analysis in tweets of famous people,” Social Networking, volume 5, number 1, pp. 1–10.
doi: https://doi.org/10.4236/sn.2016.51001, accessed 25 October 2018.
Jerry Norman, 1988. Chinese. Cambridge: Cambridge University Press.
Petra Kralj Novak, Jasmina Smailović, Borut Sluban, and Igor Mozetič, 2015. “Sentiment of emojis,” PloS ONE, volume 10, number 12, e0144296.
doi: https://doi.org/10.1371/journal.pone.0144296, accessed 25 October 2018.
Robert T. Oliver, 1969. “The rhetorical tradition in China: Confucius and mencius,” Today’s Speech, volume 17, number 1, pp. 3–8.
doi: https://doi.org/10.1080/01463376909368859, accessed 25 October 2018.
Oxford English Dictionary, n.d. “Is emoji a type of language?” at https://en.oxforddictionaries.com/explore/emoji-language, accessed 12 January 2018.
Ji Ho Park, Peng Xu, and Pascale Fung, 2018. “PlusEmo2Vec at SemEval-2018 Task 1: Exploiting emotion knowledge from emoji and# hashtags,” arXiv (23 April), at https://arxiv.org/abs/1804.08280, accessed 11 July 2018.
Umashanthi Pavalanathan and Jacob Eisenstein, 2016. “More emojis, less:) The competition for paralinguistic function in microblog writing,” First Monday, volume 21, number, 11, at http://firstmonday.org/article/view/6879/5647, accessed 14 December 2017.
doi: http://dx.doi.org/10.5210/fm.v21i11.6879, accessed 25 October 2018.
Henning Pohl, Christian Domin, and Michael Rohs, 2017. “Beyond just text: Semantic emoji similarity modeling to support expressive communication,” ACM Transactions on Computer-Human Interaction (TOCHI), volume 24, issue 1, article number 6.
doi: http://dx.doi.org/10.1145/3039685, accessed 25 October 2018.
Will Radford, Andrew Chisholm, Ben Hachey, and Bo Han, 2016. “Telephone. Person. Sailboat. Whale. Okhand or ‘Call me Ishmael’ — How do you translate emoji?” Proceedings of Australasian Language Technology Association Workshop, pp. 150–154, and at http://www.aclweb.org/anthology/U16-1018, accessed 25 October 2018.
Monica A. Riordan, 2017a. “The communicative role of non-face emojis: Affect and disambiguation,” Computers in Human Behavior, volume 78, pp. 75–86.
doi: https://doi.org/10.1016/j.chb.2017.07.009, accessed 25 October 2018.
Monica A. Riordan, 2017b. “Emojis as tools for emotion work: Communicating affect in text messages,” Journal of Language and Social Psychology, volume 36, number 5, pp. 549–567.
doi: https://doi.org/10.1177/0261927X17704238, accessed 25 October 2018.
David Rodrigues, Diniz Lopes, Marília Prada, Dominic Thompson, and Margarida V. Garrido, 2017. “A frown emoji can be worth a thousand words: Perceptions of emoji use in text messages exchanged between romantic partners,” Telematics and Informatics, volume 34, number 8, pp. 1,532–1,543.
doi: https://doi.org/10.1016/j.tele.2017.07.001, accessed 25 October 2018.
Victoria L. Rubin, Niall J. Conroy, and Yimin Chen, 2015. “Towards news verification: Deception detection methods for news discourse,” Proceedings of Hawaii International Conference on System Sciences., pp. 5–8.
Jon Russell, 2013. “Stickers: From Japanese craze to global mobile messaging phenomenon,” TNW (12 July), at https://thenextweb.com/asia/2013/07/12/stickers/, accessed 18 September 2018.
Tyler J. Schnoebelen, 2012. “Emotions are relational: Positioning and the use of affective linguistic resources,” Ph.D. dissertation, Stanford University, at https://stacks.stanford.edu/file/druid:fm335ct1355/Dissertation_Schnoebelen_final_8-29-12-augmented.pdf, accessed 23 June 2016.
John R. Searle, 1969. Speech acts: An essay in the philosophy of language. London: Cambridge University Press.
Luke Stark and Kate Crawford, 2015. “The conservatism of emoji: Work, affect, and communication,” Social Media + Society (8 October).
doi: https://doi.org/10.1177/2056305115604853, accessed 25 October 2018.
Katy Steinmetz, 2014. “Time exclusive: Here are rules of using emoji you didn’t know you were following,” Time (17 July), at http://time.com/2993508/emoji-rules-tweets/, accessed 17 November 2017.
Maite Taboada and Christopher Habel, 2013. “Rhetorical relations in multimodal documents,” Discourse Studies, volume 15, number 1, pp. 65–89.
doi: https://doi.org/10.1177/1461445612466468, accessed 25 October 2018.
Maite Taboada and William C. Mann, 2006. “Applications of rhetorical structure theory,” Discourse Studies, volume 8, number 4, pp. 567–588.
doi: https://doi.org/10.1177/1461445606064836, accessed 25 October 2018.
Rachael Tatman, 2016. “Do emojis have their own syntax?” (7 December), at https://makingnoiseandhearingthings.com/2016/12/07/do-emojis-have-their-own-syntax/, accessed 8 August 2018.
Clive Thompson, 2016, “The emoji is the birth of a new type of language (no joke),” Wired (19 April), at https://www.wired.com/2016/04/the-science-of-emoji/, accessed 21 August 2018.
Camilla Vásquez, 2011. “Complaints online: The case of TripAdvisor,” Journal of Pragmatics, volume 43, number 6, pp. 1,707–1,717.
doi: https://doi.org/10.1016/j.pragma.2010.11.007, accessed 25 October 2018.
Leticia Vidal, Gastón Ares, and Sara Jaeger, 2016. “Use of emoticon and emoji in tweets for food-related emotional expression,” Food Quality and Preference, volume 49, pp. 119–128.
doi: https://doi.org/10.1016/j.foodqual.2015.12.002, accessed 25 October 2018.
Marilyn D. White and Emily E. Marsh, 2006. “Content analysis: A flexible methodology,” Library Trends, volume 55, number 1, pp. 22–45.
doi: https://doi.org/10.1353/lib.2006.0053, accessed 25 October 2018.
Sanjaya Wijeratne, Lakshika Balasuriya, Amit Sheth, and Derek Doran, 2017. “A semantics-based measure of emoji similarity,” WI ’17: Proceedings of the International Conference on Web Intelligence, pp. 646–653.
doi: https://doi.org/10.1145/3106426.3106490, accessed 25 October 2018.
Dongjie Yan, 2016. “When chatting with Chinese, know your emojis,” China Daily (2 February), at http://www.chinadaily.com.cn/china/2016-02/02/content_23356908.htm, accessed 3 March 2017.
Huiquan Zhou and Quanziao Pan, 2016. “Information, community, and action on Sina-Weibo: How Chinese philanthropic NGOs use social media,” Voluntas, volume 27, number 5, pp. 2,433–2,457.
doi: https://doi.org/10.1007/s11266-016-9685-4, accessed 25 October 2018.
Rui Zhou, Jasmine Hentschel, and Neha Kumar, 2017. “Goodbye text, hello emoji: Mobile communication on WeChat in China,” CHI ’17: Proceedings of the 2017 CHI Conference on Human Factors in Computing Systems, pp. 748–759.
doi: https://doi.org/10.1145/3025453.3025800, accessed 25 October 2018.
Appendix 1: CMC Act Taxonomy (adapted from Herring, et al., 2005). Categories Definitions Examples Inquire Seek information What time is your meeting? Request Seek action politely Could you close the door? Direct Attempt to cause action (e.g., require, prohibit, permit, strongly advise) Calm down.
You shouldn’t do that
Invite Seek participation/acceptance by the addressee (e.g., solicit input, suggest) Let’s get together.
What do you think of him?
Inform Provide “factual” information that is verifiable in principle. July 4 is a public holiday.
It rained here yesterday.
Claim Make a subjective assertion; unverifiable in principle. I love Star Wars!
He does it just to annoy me.
Desire A future, hypothetical, or counterfactual situation (e.g., hope, wish, dream, speculate, promise). I intend to diet this summer.
If I’d known, I would’ve called earlier.
Explain Comment on or elaborate. (We can’t work today.)
Accept Concur; agree; acquiesce. You’re right. / Agreed. / Okay. Reject Disagree; dispute; challenge. No, you can’t. / That’s wrong. React Show a positive, negative, or neutral response; indicate listenership. Cool! / Ugh, you’re kidding. / Very interesting. / Really? Repair Return; clarify; correct misunderstanding. I was just joking. What did you mean by ...? Apologize Humble oneself; self-deprecate. Sorry, this is my fault. Thank Appreciate; express gratitude. I appreciate your help. Greet Greet; inquire formulaically about/wish for someone’s well-being. Hi / Good morning. / How are you? / Happy New Year!
Appendix 2: RST Categories (Mann and Thompson, 1988). Categories Definitions Background Introduces the topic of the text; helps readers understand the content. Antithesis Emphasizes the importance of the content through connectives (e.g., but, rather, instead). Concession Emphasizes the importance of the content through connectives (although, still). Evidence An objective statement that the reader is likely to accept. Reason A subjective statement/thesis/claim. Justify A statement of the fundamental attitude of the writer. Evaluation A subjective evaluation or assessment (e.g., positive/negative, desirable/undesirable). Motivation Presents a reason for performing the action described in the text. Enablement Provides instructions/information for performing the action described in the text. Circumstance An event or state that actually occurs or has occurred. Condition A hypothetical, future, or in other ways unreal situation (e.g. if, in case). Otherwise A hypothetical, future, or in other ways unreal situation (e.g. otherwise). Unless A hypothetical, future, or in other ways unreal situation. Elaboration Provides details or more information on the items described in the text. Interpretation Presents ideas that are not involved in the knowledge presented in the text. Means Provides information or an instrument to facilitate action described in the text. Cause A state or event in the world (e.g. because, since). Result A state or event in the world (e.g. therefore, thus). Purpose A hypothetical, future, or in other ways unrealized situation (e.g. in order to). Solutionhood Provides a solution to the problems presented in the text. Preparation Sets a topic for the text. Restatement Repeats the information presented in the text. Summary Makes a conclusion. Contrast Compares the specific things presented in the text. Sequence Describes states of affairs that occur in a particular temporal order.
Received 5 September 2018; revised 8 October 2018; accepted 27 October 2018.
This paper is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.
Communicative functions of emoji sequences on Sina Weibo
by Jing Ge and Susan C. Herring.
First Monday, Volume 23, Number 11 - 5 November 2018