First Monday

Assessing the cognitive and communicative properties of Facebook Reactions and Likes as lightweight feedback cues by Erin M. Sumner, Rebecca A. Hayes, Caleb T. Carr, and Donghee Yvette Wohn

The emergence of Facebook Reactions provides new opportunities to explore the nature of paralinguistic digital affordances (PDAs; lightweight one-click social media response cues). Guided by adaptive structuration theory and the concept of cognitive automaticity, a survey of 255 individuals aged 18–24 assessed the cognitive processes and communicative meanings associated with the provision of Facebook Reactions and Likes. Although Like and Reaction cues (excluding Angry) were all identified as more literal in meaning than not, specific results indicated: (a) Likes were perceived more faithfully than Reactions; (b) the Like and Love cues were labeled as the most faithful; and (c) Reactions were perceived as more deliberate and less automatic communicative behaviors than Likes. Collective results suggest social media platforms that offer multiple one-click response cues (e.g., Facebook) can afford different communicative opportunities than platforms with a single PDA response option, presenting challenges for future cross-platform research addressing lightweight response cues.


Paralinguistic digital affordances




Social media use is a routine part of most people’s online behavior (Greenwood, et al., 2016), and facilitates masspersonal interactions in which users communicate via one-to-many postings of a personal nature (O’Sullivan and Carr, 2018). One common form of interaction involves the provision of a one-click response cue — a paralinguistic digital affordance (PDA) — common to most social media and exemplified by the Facebook Like button.

Paralinguistic digital affordances are lightweight forms of nonverbal communication provided in response to another person’s social media content (Hayes, et al., 2016a). Minimalist in design and represented by a single symbol, PDAs provide more to users than their simplicity would imply. PDAs are used to convey an array of content-based messages (e.g., I appreciate or saw this post) and relational-based sentiments (e.g., I support you; Sumner, et al., 2018). This built in ambiguity is a huge affordance of PDAs. Users can click a single button and let others interpret their own meanings.

Individual PDAs have been examined as a form of social currency that afford various interpersonal benefits (Wohn, et al., 2016); yet most interpersonal communication scholarship predates or omits Facebook’s introduction of Reactions in 2016. The limited published research regarding Facebook Reactions has focused on social media users’ engagement with news and other mass media (Badache and Boughanem, 2017; Larsson, 2018), as well as how people employ these seemingly benign cues to incite hatred and display social outrage (Matamoros-Fernández, 2018). It is therefore prudent to examine the more interpersonal implications of Facebook’s expansion from a single Like button to a set of Reaction cues.

Compared to the iconic Like, Facebook Reactions require an extra step to use (i.e., a long press or hovering action, followed by selecting one of five icons), and afford greater breadth and nuance of communicative and emotional meaning (i.e., Love, Happy, Wow, Angry, and Sad). These distinctions might undermine the ambiguous nature of a single PDA interface, and thereby provoke increased user cognition and greater communicative clarity in ways that challenge previous conceptualizations of PDAs. We therefore examine both cognitive (i.e., perceived automaticity) and communicative (i.e., perceived faithfulness) factors pertinent to the provision of Facebook Reaction cues in comparison to the Like button.



Paralinguistic digital affordances

The social value of PDAs

PDAs — like most reduced-cue online communication — are somewhat ambiguous and ambivalent in nature. Ambivalence, as discussed by Phillips and Milner (2017), refers to the paradoxical nature of many online behaviors. They are interpreted within larger cultural systems, yet also give rise to new systems of meaning. For example, a particular meme might be seen as an amusing source of connection for one group of users, while simultaneously interpreted as an antisocial action against a different audience. It’s all a bit ambivalent, and interpretation requires in-depth contextual knowledge of large and small scale cultural and communicative systems.

PDAs seemingly fit Phillips and Milner’s (2017) notion of ambivalence. PDAs possess explicit verbiage (e.g., Like or Favorite), yet their intended and interpreted meanings are diverse and vary across users and contexts. For example, receiving PDAs can provide users with a sense of social support and inclusion, yet social comparisons and unmet expectations based on PDAs can make users feel ostracized (Wohn, et al., 2016; Hayes, et al., 2018). After all, it can feel hurtful to receive fewer Likes than your friends do, or have your post overlooked by a loved one. So while social media users appreciate PDAs as a form of communication, many also feel pressure regarding how to interpret and provide these cues in an appropriate way.

Rather than discussing broad notions of internet ambivalence, Hayes, et al. (2016a) used the more specific concept of phatic communication in their conceptualization of PDAs. Phatic communication encompasses an array of relatively vague communicative behaviors (e.g., waving or saying hello) that allow individuals to engage in prosocial behavior without articulating a nuanced or informative message (see Malinowski, 1972; Miller, 2008). Online communicators might wish to send forms of phatic communication that are difficult to put into words, and the use of nonverbally inspired social media cues (stickers, emojis, PDAs, etc) can enable more fluid and seamless emotional communication (Lim, 2015).

Like other forms of phatic communication, PDAs can convey an abundance of functional and emotional meanings (Hayes, et al., 2016a), ranging from simply showing that a message was received, to signaling concern for another person. PDAs have been found to play a role in important interpersonal processes, such as: impression management, self-presentation, and meeting social obligations (e.g., Lowe-Calverly and Grieve, 2018; Ozanne, et al., 2017); relational facilitation, self-presentation, and metacommunication (e.g., Scissors, et al., 2016; Sumner, et al., 2018); as well as social comparison and self-evaluation (e.g., Carr, et al., 2018; Rosenthal-von der Putten, et al., 2019). The phatic nature of PDAs is apparent across all of the above mentioned studies. For example, participants in Sumner and colleagues (2018) study reported they might Like a friend’s sad post because they wanted show that they cared, yet weren’t really sure what to say about the situation. Users gleaned meaning from the psychological and communicative context in which a PDA was provided, rather than feeling limited by the Like button’s exact verbiage.

As previously noted, the facilitation and maintenance of relationships are prominent functions of PDAs (Ahmadi, et al., 2016; Hayes, et al., 2016a; Sumner, et al., 2018). Clicking Like is an easy way to signal interest in — or concern for — another person. From the perspective of receivers, getting PDAs has been found to correlate with perceptions of bonding social capital (Burke, et al., 2010), social support (Wohn, et al., 2016), and overall satisfaction with the message sharing experience (Bazarova, et al., 2015; Zell and Moeller, 2018). That said, additional nuance is beneficial when it comes to extrapolating interpersonal benefits from PDAs. The social value of PDAs, for example, can vary depending on the relationship between the receiver and provider. Individuals appear to place greater value on PDAs received from stronger social ties (Carr, et al., 2016), as well as providers deemed as more similar and socially relevant to the user (Hayes, et al., 2016b). A friend’s PDAs are therefore more relevant than those of a celebrity with millions of followers.

In sum, the phatic nature of PDAs does not appear to prevent social media users from gleaning a great deal of social significance from lightweight one-click cues. Although much is known about Liking and other single-cue PDA systems, relatively little is understood regarding how users make choices within the more nuanced set of Facebook Reaction cues.

Differentiating Facebook Reactions and Likes

Facebook introduced Reactions in 2016 as supplemental to the iconic one-click Like button. Reaction cues (see Figure 1) include a heart (accompanied with the text description “Love”), and four facial expression emojis labeled Happy, Wow, Angry, and Sad. The Facebook News Feed interface displays the total number of Reactions and Likes a post has received; but users are able to see a breakdown of each cue received by hovering over that particular symbol. In grouping them together, the interface treats Reactions as functionally comparable to Likes.


Reaction cues
Figure 1: Facebook Reaction cues.

Reaction cues, however, present conceptual and methodical problems for continued research involving PDAs broadly and Liking specifically: Should Reactions be treated as akin to Likes, or are they separate communicative mechanisms represented by similar tools? Though Reactions possess many of the characteristics of Likes (e.g., simple use and iconic nonverbal communication), the differences between them (e.g., additional interface steps for provision and more nuanced emotional lexicon) may create distinct patterns of use and reduced likelihood for ambiguous interpretation. As such, it is important to understand how users employ and perceive these emergent social media response tools. In doing so, scholars can provide increased conceptual clarity around the applicability of the concept of PDAs to the expanded set of Reaction cues. Two key elements tied to the use and interpretation of PDAs will therefore guide the present study: the perceived communicative faithfulness (Hayes, et al., 2016a) and perceived cognitive automaticity (Carr, et al., 2016) of a cue’s provision.

Communicative faithfulness. Adaptive structuration theory (AST; DeSanctis and Poole, 1994) considers the bidirectional relationship between technology use and social processes. In short, AST asserts that technologies (and other social structures) are imbedded with certain norms and expectations that will influence — but not fully determine — how people use it. Social groups are labeled as active agents who come up with their own rules and norms for a given technology. In this sense, AST concurs with Phillips and Milners’ (2017) notion of ambivalence in online communication, while offering a more clear theoretical framework.

Within the framework of AST, DeSanctis and Poole (1994) assert that users may appropriate a technology either faithfully (i.e., consistent with the designer’s original intent) or ironically (i.e., differing from the literal designer-intended purpose, and determined by the user). AST has been used to explain ironic uses of other technologies. For example, the ring of a telephone can be interpreted as a faithful signal that someone has an incoming call that should be answered. That said, Donner (2007) found that teenagers in Rwanda had ironically appropriated the ringer as a sort of modern day Morse code; in which a series of rings and hang ups could communicate pre-set messages without incurring the expense of an answered phone call.

AST’s notion of faithful and ironic adaptation is well aligned with the phatic nature and ambivalent uses of PDAs. Social media users no doubt “Like” things that they actually appreciate (e.g., Lowe-Claverly and Grieve, 2018), but this is not always the case. Using qualitative data, Hayes and colleagues (2016a) argued the Like button is employed faithfully when used to communicate appreciation or genuine liking of posted content, and ironically when appropriated to communicate anything else (e.g., to show relational support or signal receipt of the message). Other quantitative research has found that faithful and ironic uses of PDAs are almost equally prevalent. Sumner and colleagues (2018) examined 365 unique Likes by asking participants to copy/paste or describe the content of their most recently Liked Facebook post, and provide commentary about why they chose to provide a Like cue. They found that 52 percent of Likes were intended as faithful content-based responses to a post (i.e., the user literally appreciated or liked the content of a post), whereas the other 48 percent included ironic forms of relational messaging (i.e., the user wanted to show social support or display positive relational regard for someone).

In sum, existing research suggests that Likes are interpreted in ways that are highly-subjective and often ironic (per AST) to their assigned verbiage. This lack of communicative faithfulness might reflect the generally ambivalent nature of Internet communication (Phillips and Milner, 2017), as well as the phatic nature of a single PDA interface (Hayes, et al., 2016a). It remains unclear, however, whether the greater diversity in iconography and verbiage provided by Reactions will influence the communicative faithfulness of these cues in comparison to the Like.

On one hand, it seems possible that Reactions will be interpreted more faithfully than a generic Like. The inclusion of multiple response options could allow users to select the Reaction cue that most accurately communicates their desired emotional sentiment. For example, users can convey overtly negative and even indignant sentiments using the Anger button (Larsson, 2018). Likewise, users can convey a somber tone by selecting the Sad cue rather than being limited to Liking a sad post in an ironically supportive manner. While past research implies that users understand the literal and highly emotive meanings of Reactions cues as responses to mass media content (Badache and Boughanem, 2017; Larsson, 2018), it is less clear whether these cues will be interpreted faithfully within an interpersonal setting. If so, the introduction of Reactions could be said to lessen the phatic and ambiguous nature of PDAs.

On the contrary, Reactions might not be seen as more faithful than Likes. Despite their expanded vernacular, Reactions remain lightweight cues that are sent using an easy hover-and-click action. This ease of use is compounded with the fact that Facebook posts report the number of Reactions received in a bulk manner (i.e., the user sees which Reactions they received, along with a total count of all Reactions). Although further investigation can reveal individual Reaction counts (and who provided them), it is possible that users fail to distinguish the differences between individual Reactions, or between Reactions and Likes. A research question will therefore examine:

RQ1: Do Reactions and Likes differ in their perceived level of cue faithfulness?

Automaticity of use. A main utility of PDAs lies within users’ ability to click on a cue without exerting a great deal of thought. Hence, the provision of Reactions and Likes might also vary with regard to users’ expense of cognitive effort. Hayes, et al.’s (2016a) participants often described PDAs — particularly the Facebook Like — as important yet “nearly aimless, automated communicative action(s)” [1]. Sumner, et al. (2018) similarly referred to Likes as “the Facebook equivalent of a head nod” [2]. Indeed, Carr, et al. (2016) found people accounted for the potentially automatic nature of provision of PDAs when interpreting a PDA’s meaning and value. Similar to the above discussion of communicative faithfulness, these findings are were limited to social media interfaces offering only a single PDA option.

Facebook’s Reactions might challenge the extent to which PDAs, as social cues, are able to be provided in an automatic manner. The Facebook interface treats the Like as the default response cue, which can still be provided with a single click. Reactions require an additional step in which users must hover over the Like button before selecting which cue to click. Within Tian and colleagues’ (2017) sample, the Like button still reflected approximately 79 percent of response cues provided. This implies that the Like might still be used relatively automatically, with the other Reactions occurring in less frequent — and potentially more effortful — ways. It is unknown whether these small differences will require enough additional cognitive effort to alter the perceived automaticity of selecting a clickable (or hover and click) button as a form of communication. As such:

RQ2: Do Reactions and Likes differ in the perceived automaticity of their provision?




Participants and procedures

Data were collected from a convenience sample of 255 American undergraduate students who had accessed Facebook in the previous week, and received extra course credit for completing an online survey. Participants were an average of 20.17 (SD = 1.85) years of age, with more women (70.9 percent) than men (28.6 percent) in the sample. Gender and age presented no significant between-group differences in study variables, and were thus excluded from further analyses.

Utilizing an online survey platform, respondents were asked a series of questions about: 1) their own use of Facebook Likes and Reactions as response tools; 2) their perceptions of how others use Likes and Reactions; and 3) their demographic characteristics. Facebook was the only platform examined because it contained both a traditional PDA (the Like) and an expanded lexicon of Reactions. When relevant, the survey integrated visual representations from Facebook’s interface (i.e., cues) to enhance clarity.


Perceived cue faithfulness was measured using six items assessed on a seven-point Likert-type scale (1 = Strongly Disagree, 7 = Strongly Agree). Example items included, “People who use the emoji literally were saddened by or cried over my post,” “People who use the emoji literally were angered by or made mad over my post.” Each item reflects the extent to which participants believe that people use the Like or a particular Reaction cue literally. Higher values indicated more faithful interpretation of a particular response cue. An aggregate faithfulness of Reactions use was also calculated, using the means of the five Reaction responses.

Automaticity of response were measured for both the Like and individual Reactions, each via a modified four-item subscale of the Self-Report Habit Index (Verplanken and Orbell, 2003). A seven-point Likert-type scale (1 = Strongly Disagree, 7 = Strongly Agree) assessed automaticity of use; with higher scores representing more automatic (and thus less conscious and deliberate) delivery. Each pair of scales was administered twice, first to assess the lack of deliberateness of others’ provision of Likes and Reactions, and a second time later in the survey to assess respondents’ own level of deliberateness in provision. Sample items included, “[I/Facebook users] click [Like/Reaction] without having to consciously remember,” and “[I/Other] people [Like/React to] Facebook content before [I/they] realize [I’m/they’re] doing it.” All four scales demonstrated high reliabilities: αOthers-Likes = .86, αOthers-Reactions = .91, αSelf-Likes = .93, αSelf-Reactions = .96.




RQ1 asked whether Reactions are interpreted more literally than Likes. As an initial step, six independent-samples t-tests were conducted to compare the mean of each cue’s perceived faithfulness against the scale midpoint. The critical alpha was set at a more stringent .01 level to help control for Type 1 error related to running multiple comparisons. Results (see Table 1) indicated that the Like, Love, Haha, Wow, and Sad cues were all perceived more faithfully than ironically upon receipt (see Table 1). Angry did not differ from the scale midpoint; implying that it is not distinguishable as being faithful in nature. As such, it appears that Likes and Reactions are perceived in generally faithful or literal manner.

A series of paired t-tests were then conducted to compare within-respondent differences in the mean faithfulness of use values for each PDA among the 15 possible combinations. The critical alpha was again set at a more stringent .01 level. The tests revealed multiple significant differences in perceived cue faithfulness (See Table 1). To answer RQ1 broadly, the Like cue was seen as significantly more faithful in nature than the aggregate of all Reaction cues; t(254) = 8.61, p < .001, Cohen’s d = 1.08. On a more nuanced level, three trends emerged. First, the Like and Love cues did not differ from each other with regard to perceived faithfulness. Second, both the Like and Love cues differed from all other Reaction cues. Third, the remaining cues — Haha, Wow, Sad, and Angry — did not differ from each other.


Table 1: Descriptive statistics and paired t-test comparisons of faithfulness of interpretation of Facebook PDA options.
Notes: N = 255. Bolded values reveal the means (and standard deviations) for cue faithfulness of a particular PDA, with associated p-values revealing significant differences between that mean and the scale midpoint (i.e., 4.00).
*p < .05, **p < .01. Given the use of 15 paired t-tests, the more conservative p < .01 will be employed as the critical alpha.
PDA Reactions
Like5.25** (1.29) 
Love2.03* (252)5.06** (1.52) 
Haha8.29** (253)6.94** (251)4.30** (1.65) 
Wow7.97** (251)6.69** (249)-.74 (250)4.38** (1.47) 
Sad7.68** (254)6.33** (252)-.25 (253).41 (251)4.34** (1.68) 
Angry8.27** (252)7.03** (250).28 (251)1.05 (249).82 (252)4.27* (1.64) 
8.61** (254)8.00** (252)-2.39 (253)-1.80 (251)-2.27 (254)-3.18** (252)4.47** (1.24)


RQ2 inquired whether Reactions were provided less automatically than Likes. It was tested in regard to both the respondents’ perceptions of their own use, as well as their perceptions of how automatically other people use these cues. Paired t-tests compared (a) automaticity of others’ use of Reactions and Likes and (b) automaticity of respondents’ own uses of Reactions and Likes (See Table 2). Respondents perceived others as using Reactions (M = 3.30, SD = 1.48) less automatically than the Like (M = 4.74, SD = 1.38), t(254) = 14.13, p < .001, d = 1.77. Similarly, respondents reported themselves as using Reactions (M = 2.39, SD = 1.37) less automatically than the Like (M = 3.54, SD = 1.72), t(253) = 12.23, p < .001, d = 1.54. These tests indicate Reactions are perceived as more deliberate, less automatic communicative behaviors than Likes.


Table 2: Descriptive statistics and paired comparisons of automaticity of interpretation of Facebook Likes and Reactions for respondents and Friends.
Notes: N = 255; p < .001.
Likes3.54 (1.72)4.74 (1.38)11.61 (253)
Reactions2.39 (1.37)3.30 (1.48)9.50 (254)
t(df)12.23 (253)14.13 (254) 





The present research probed whether Facebook Reactions function in similar cognitive and communicative capacities as the Like, and might thus be considered as interchangeable in future scholarship. Guided by adaptive structuration theory, a survey of Facebook users facilitated the contrasting of characteristics associated with the perceived automaticity and communicative faithfulness of Likes and Reactions. Ultimately, results suggest that Likes and Reactions may be communicatively similar, yet involve different levels of cognition. Complicating matters, data additionally suggest a potential shift in how Likes themselves are used, perhaps due to the increased lexicon and number of options provided by Reactions.

Communicative faithfulness (RQ1)

According to past qualitative research and mixed methodological research, Likes are often provided and interpreted ironically relative to the cue’s verbiage (Hayes, et al., 2016a; Sumner, et al., 2018). Users have reported clicking Like to signal interpersonal sentiments that ranged from literal appreciation of a post to support of the person who posted it. While it may seem odd to Like someone’s post about the death of a loved ones, social media users seem to have developed social norms that allowed this ironically supportive meaning to be gleaned from the context of a Like. That said, scholarly research regarding interpersonal uses of PDAs has lagged behind the introduction of Reactions during 2016.

Somewhat surprisingly, results of RQ1 revealed Likes and Reactions were generally perceived as being more faithful than ironic in meaning (i.e., each cue’s mean faithfulness score was significantly above the neutral scale mid-point, with the exception of Angry). This similarity suggests Reaction and Like may be communicatively compatible, yet also implies that the multi-cue interface may have undercut the previously ironic and ambivalent nature of a single-cue PDA system. Consistent with Korzenny’s (1978) electronic propinquity theory, it may be the broadening of iconographic feedback cues (i.e., increased bandwidth) decreased users’ need to assume or ‘read into’ Likes and Reactions as phatic cues. Rather, people might use the broader bandwidth to assume more literal meaning from a given cue. This finding might challenge researchers’ ability to compare the multi-cue Facebook PDA system with single cue systems such as Twitter and Reddit; a possibility that should be tested in future multi-platform research.

Further parsing RQ1, respondents perceived Likes as even more faithful in meaning than the seemingly more specific Reactions cues; with the exception of the Love cue which was similarly high in faithfulness. There are multiple ways to interpret this finding. First, Reactions — though they denote specific verbiage — might serve as nonverbal cues similar to Walther and D’Addario’s (2001) attribution of emoticons. Emoticons, like most nonverbal communication, can convey meaning while remaining a little ambiguous. A person might not know why someone smiled at them, but this nonverbal gesture is generally taken as prosocial. The same might be true of a smiley face emoticon. So, while Likes and Reactions were interpreted relatively faithfully by participants in the present study, this literal interpretation might only extend to facilitating phatic social interaction rather than communicating specific information (Malinowski, 1972; Miller, 2008). For instance, users might interpret the Love cue as a faithful indicator that someone “loved” something about the post, without reading too much into what exactly was “loved” and for what purposes. In other words, the general social and emotional sentiment might be faithfully interpreted without nuanced sensemaking. This speculative reading of results is somewhat consistent with the original conceptualization of PDAs, yet reflects the ebb and flow of how users employ a given technology within adaptive structuration theory (DeSanctis and Poole, 1994).

Second, it seems likely that users allow a little more ambivalence when it comes to interpreting Reactions that are emotionally vague or negatively charged, yet are more willing to perceive positive sentiments as extremely literal in meaning. The Angry cue (M = 4.27, SD = 1.64) was rated as the least faithful and did not differ from the scale midpoint. Conversely, Like (M = 5.25, SD = 1.29) and Love (M = 5.06, SD = 1.42) were rated as significantly more faithful than the rest of the Reactions cues (Haha, Wow, Sad, and Angry). In comparison, Haha, Wow, Sad, and Angry failed to display statistically different faithfulness levels from each other. As noted above, liking and loving might occur in different forms (e.g., romantic, familiar, friendly, etc.), yet are generally seen as positive and prosocial emotional constructs. Users might be prone to read these overtly positive cues in a literal manner. Haha and Wow reflect more ambivalent potential (Phillips and Milner, 2017): You can laugh with someone or at someone’s expense, and can be wowed by something that is amazing or horrific in nature. This ambivalence might extend to the Sad and Angry cues. While these terms carry negative social connotations, it is possible to be sad that a friend is in pain, or angry that a friend was treated poorly. In the case of the Angry cue, users might be less willing to interpret an antisocial cue in a literally negative manner; especially given the potential for the cue to incite anger and hatred (Matamoros-Fernández, 2018). These interpretations are once again speculative, which future research should seek to clarify.

In sum, while the Like and Reaction cues were generally seen as more faithful than ironic, the present study implies that users built in a little more ambiguity into their interpretations of neutral and negative cues.

Cognitive automaticity (RQ2)

Previous research (e.g., Hayes, et al., 2016a; Carr, et al., 2016) described Liking as being somewhat a somewhat automatic prosocial social media behavior. The results of RQ2 indicate Reactions are not perceived as being used automatically, as values for automaticity of use of Reactions fell significantly below the scale midpoint. Users reported using and perceiving Reactions as deliberate, thoughtful behaviors. Reactions were also significantly more deliberate than Likes in their use. Tian and colleagues’ (2017) participants used the Like substantially more than other Reactions cues: The Like comprised nearly 79 percent of all response cue provisions. Our results might help explain why the Like continues to reign in popularity. Perhaps users still prefer the more automatic cue over the more deliberate and thoughtful ones.

Reactions being perceived as intentional behaviors is unsurprising, given the additional (albeit minimal) cognitive and physical effort users expend to provide them. This result may be reflective of the user interface and design of Reactions cues. Rather than making a single click, Reactions users must pause and select options from a pop-up menu. As users assert more cognitive effort to find and use the appropriate Reaction, they will not be able to engage in “mindless” behavior associated with automaticity. That said, Reactions are still relatively new to Facebook users, so it could also be they have not used the feature enough for norms of use to stabilize.

Limitations and future research directions

The present study possesses limitations that can be used to help identify areas for future research. First, this study employed a convenience sample of undergraduate students, and is thus limited in generalizability. Study population and sampling concerns are important within adaptive structuration theory (DeSanctis and Poole, 1994), which argues that technology and social norms share bidirectional influence. As such, context matters. This is particularly true within the present study, due to the phatic and ambiguous nature of PDAs (Hayes, et al., 2016a) and the ambivalent nature of the Internet (Phillips and Milner, 2017). Different sub-groups of users might attribute different meanings to PDAs, which was not examined within the present study. This void is notable, since the bulk of PDA research within the United States has employed convenience samples comprised of mostly undergraduate students (e.g., Hayes, et al., 2016a; Sumner, et al., 2018; Zell and Moeller, 2018). Future research should therefore follow suit of Wohn and colleagues (2016) who recruited from national panels and reflected a wider variety of age, socioeconomic status, and education levels.

Relatedly, the present study’s sample was located within the United States, yet the study of PDAs and other lightweight nonverbal social media cues possess a rich international body of research. Facebook Reaction cues have been examined in relation to mass media engagement and larger cultural practices within Scandinavian countries (Larsson, 2018) and Belgium (Matamoros-Fernández, 2018), as well as across four different countries (United Kingdom, United States, France, and Germany (Tian, et al., 2017). Indeed, Lim (2015) notes that cultural factors such as being high context (i.e., a tendency to leave certain information unsaid while relying on contextual and nonverbal cues) or low context (i.e., a preference for direct statements) might play a role in how groups of users employ heavily nonverbal and emotional social media cues such as emojis and PDAs. The present study’s more interpersonal orientation should be further examined using similarly international sampling methods.

Finally, the present study is limited to one platform; Facebook. This decision was made to control for the potential error that would be introduced by comparing a single-PDA platform such as Twitter, to a multi-PDA platform like Facebook. It seems likely that the presence of multiple response cue options changes the way that users interact with any given cue. Future research, ideally of a cross-platform nature, is necessary to more fully understand how Facebook’s Like/Reaction interface might operate as similar or different than single-cue PDA systems, or other multi-cue PDA systems.




The present work conceptually and pragmatically contrasted the use and interpretation of Reactions and Likes, and in doing so suggests social media offering users multiple nonverbal response cues provide different communicative affordances than social media with single one-click cues. Adaptive structuration theory (DeSanctis and Poole, 1994) argues that technology and social structures and norms are intertwined, and a change in one can engender change in the other. Extant research implies that Likes are often ironic (Hayes, et al., 2016a; Sumner, et al., 2018), whereas the present study reveals them to be more faithful than ironic. Likewise, the present study’s results reveal Reactions are used less automatically and less literally than Likes; which implies that these cues might possess important differences in both use and interpretation. The unexpected change in how the Like is used and interpreted in this data could reveal that Reactions have engendered a change in the social norms of one-click cue delivery on Facebook. As such, the present study serves as a springboard for an entirely new question; do Facebook one-click cues — with the increased nonverbal communication afforded by Reactions — still fit the definition boundaries of paralinguistic digital affordances? Future research with more diverse sampling methods is needed to more fully interrogate the communicative and cognitive dimensions of Reactions and Likes within an interpersonal framework. End of article


About the authors

Erin M. Sumner (Ph.D., Arizona State University) is an Associate Professor of Human Communication at Trinity University in San Antonio, Texas. Her research primarily focuses on understanding the role of computer-mediated communication (e.g., social media, online dating sites, and other technologies) in social and personal relationships.
E-mail: ebryant [at] trinity [dot] edu

Rebecca A. Hayes (Ph.D., Michigan State University) is an Associate Professor of Communication at Illinois State University. Her research interests lie in the brand and personal implications of social media, and crisis communication.
E-mail: rahayes [at] ilstu [dot] edu

Caleb T. Carr (Ph.D., Michigan State University) is an Associate Professor of Communication at Illinois State University. His research primarily explores the use and role of technology on the convergence of organizational and interpersonal communication, decision-making, and the development and presentation of identity.
E-mail: ctcarr [at] ilstu [dot] edu

Donghee Yvette Wohn (Ph.D., Michigan State University) is an Assistant Professor at New Jersey Institute of Technology and director of the Social Interaction Lab ( Her research is in the area of Human Computer Interaction (HCI) where she studies the role of algorithms and social interactions in livestreaming, esports, gaming, and social media.
E-mail: donghee [dot] y [dot] wohn [at] njit [dot] edu



1. Hayes, et al., 2016a, p. 182.

2. Sumner, et al., 2018, p. 1,452.



M. Ahmadi, M.E. Schneider, R. Kadam, and D.Y. Wohn, 2016. “Designing paralinguistic digital affordances for social support,” CSCW ’16 companion: Proceedings of the 19th ACM Conference on Computer Supported Cooperative Work and Social Computing Companion, pp. 221–224.
doi:, accessed 9 January 2020.

I. Badache and M. Boughanem, 2017. “Emotional social signals for search ranking,” SIGIR ’17: Proceedings of the 40th International ACM SIGIR Conference on Research and Development in Information Retrieval, pp. 1,053–1,056.
doi:, accessed 9 January 2020.

N.N. Bazarova, Y.H. Choi, V. Schwanda Sosik, D. Cosley, and J. Whitlock, 2015. “Social sharing of emotions on Facebook: Channel differences, satisfaction, and replies,” CSCW ’15: Proceedings of the 18th ACM Conference on Computer Supported Cooperative Work & Social Computing, pp. 154–164.
doi:, accessed 9 January 2020.

M. Burke, C. Marlow, and T. Lento, 2010. “Social network activity and social well-being,” CHI ’10: Proceedings of the SIGCHI Conference on Human Factors in Computing Systems, pp. 1,9094–1,912.
doi:, accessed 9 January 2020.

C.T. Carr, R.A. Hayes, and E.M. Sumner, 2018. “Predicting a threshold of perceived Facebook post success via Likes and Reactions: A test of explanatory mechanisms,” Communication Research Reports, volume 35, number 2, pp. 141–151.
doi:, accessed 9 January 2020.

C.T. Carr, D.Y. Wohn, and R.A. Hayes, 2016. “👍 as social support: Relational closeness, automaticity, and interpreting social support from paralinguistic digital affordances in social media,” Computers in Human Behavior, volume 62, pp. 385–393.
doi:, accessed 9 January 2020.

J. Donner, 2007. “The rules of beeping: Exchanging messages via intentional ‘missed calls’ on mobile phones,” Journal of Computer-Mediated Communication, volume 13, number 1, pp. 1–22.
doi:, accessed 9 January 2020.

G. DeSanctis and M.S. Poole, 1994. “Capturing the complexity in advanced technology use: Adaptive structuration theory,” Organization Science, volume 5, number 2, pp. 121–147.
doi:, accessed 9 January 2020.

R.A. Hayes, E.D. Wesselmann, and C.T. Carr, 2018. “When nobody ‘likes’ you: Perceived ostracism through paralinguistic digital affordances within social media,” Social Media + Society (18 September).
doi:, accessed 9 January 2020.

R.A. Hayes, C.T. Carr, and D.Y. Wohn, 2016a. “One click, many meanings: Interpreting paralinguistic digital affordances in social media,” Journal of Broadcasting & Electronic Media, volume 60, number 1, pp. 171–187.
doi:, accessed 9 January 2020.

R.A. Hayes, C.T. Carr, and D.Y. Wohn, 2016b. “It’s the audience: Differences in social support across social media,” Social Media + Society (16 November).
doi:, accessed 9 January 2020.

F. Korzenny, 1978. “A theory of electronic propinquity: Mediated communication in organizations,” Communication Research, volume 5, number 1, pp. 3–24.
doi:, accessed 9 January 2020.

A.O. Larsson. 2018. “Diversifying Likes: Relating reactions to commenting and sharing on newspaper Facebook pages,” Journalism Practice, volume 12, number 3, pp. 326–343.
doi:, accessed 9 January 2020.

S.S. Lim, 2015. “On stickers and communicative fluidity in social media,” Social Media + Society (11 May).
doi:, accessed 9 January 2020.

E. Lowe-Calverley and R. Grieve, 2018. “Thumbs up: A thematic analysis of image-based posting and liking behaviour on social media,” Telematics and Informatics, volume 35, number 7, pp. 1,900–1,913.
doi:, accessed 9 January 2020.

A. Matamoros-Fernández, 2018. “Inciting anger through Facebook reactions in Belguim: The use of emoji and related vernaular expressions in racist discourse,” First Monday, volume 23, number 9, at, accessed 9 January 2020.
doi:, accessed 9 January 2020.

P.B. O’Sullivan and C.T. Carr, 2018. “Masspersonal communication: A model bridging the mass-interpersonal divide,” New Media & Society, volume 20, number 3, pp. 1,161–1,180.
doi:, accessed 9 January 2020.

M. Ozanne, A. Cueva Navas, A.S. Mattila, and H.B. Van Hoof, 2017. “An investigation Into Facebook ‘Liking’ behavior: An exploratory study,” Social Media + Society (10 May).
doi:, accessed 9 January 2020.

A.M. Rosenthal-von der Putten, M.R. Hastall, S. Köcher, C. Meske, T. Heinrich, F. Labrenz, and S. Ocklenburg, 2019. “‘Likes’ as social rewards: Their role in online social comparison and decisions to like other people’s selfies,” Computers in Human Behavior, volume 92, pp. 76–86.
doi:, accessed 9 January 2020.

L. Scissors, M. Burke, and S. Wengrovitz, 2016. “What’s in a Like? Attitudes and behaviors around receiving Likes on Facebook,” CSCW ’16: Proceedings of the 19th ACM Conference on Computer-Supported Cooperative Work & Social Computing, pp. 1,501–1,510.
doi:, accessed 9 January 2020.

E.M. Sumner, L. Ruge-Jones, and D. Alcorn, 2018. “A functional approach to the Facebook Like button: An exploration of meaning, interpersonal functionality, and potential alternative response buttons,” New Media & Society, volume 20, number 4, pp. 1,451–1,469.
doi:, accessed 9 January 2020.

Y. Tian, T. Galery, G. Dulcinati, E. Molimpakis, and C. Sun, 2017. “Facebook sentiment: Reactions and emojis,” Proceedings of the Fifth International Workshop on Natural Language Processing for Social Media, pp. 11–16.
doi:, accessed 9 January 2020.

B. Verplanken and S. Orbell, 2003. “Reflections on past behavior: A selfreport index of habit strength,” Journal of Applied Social Psychology, volume 33, number 6, pp. 1,313–1,330.
doi:, accessed 9 January 2020.

J.B. Walther and K.P. D’Addario, 2001. “The impacts of emoticons on message interpretation in computer-mediated communication,” Social Science Computer Review, volume 19, number 3, pp. 324–347.
doi:, accessed 9 January 2020.

D.Y. Wohn, C.T. Carr, and R.A. Hayes, 2016. “How affective is a ‘Like?’ The effect of paralinguistic digital affordances on perceived social support,” Cyberpsychology, Behavior, and Social Networking, volume 19, number 9, pp. 562–566.
doi:, accessed 9 January 2020.

A.L. Zell and L. Moeller, 2018. “Are you happy for me ... on Facebook? The potential importance of ‘likes’ and comments,” Computers in Human Behavior, volume 78, pp. 26–33.
doi:, accessed 9 January 2020.


Editorial history

Received 17 January 2019; revised 20 December 2019; accepted 7 January 2020.

Creative Commons License
This paper is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.

Assessing the cognitive and communicative properties of Facebook Reactions and Likes as lightweight feedback cues
by Erin M. Sumner, Rebecca A. Hayes, Caleb T. Carr, and Donghee Yvette Wohn.
First Monday, Volume 25, Number 2 - 3 February 2020