Handshakes and hashtags: how changing social interactions make us feel awkward

ABSTRACT In the wake of the COVID-19 pandemic, social distancing measures were implemented across the globe. These measures demanded replacing taken-for-granted social practices such as shaking hands with new interaction rituals. Based on our personal observations, this collective process of learning new interactions often resulted in feelings of awkwardness. Awkwardness, in this sense, is more than an individual emotion; it is also a cultural marker helping us understand how interactions, interaction rituals and social norms are constituted. Therefore, we aim to obtain a better understanding of both what people perceive as failed interactions during the COVID-19 pandemic, and how they perceive these awkward moments. We do so by looking at how awkwardness is discussed in social and news media during the first wave of the pandemic. Combining a topic modelling of tweets and a thematic text analysis of news articles, we show the main topics representing awkwardness in relation to COVID-19, and how this links to new forms of face-to-face and mediated interactions. Moreover, we demonstrate that experiences of awkwardness often relate to the necessity of bodily and situational co-presence, creating a stronger sense of intimacy, synchronicity and sequency.


Introduction
On 11 March 2020, COVID-19 was declared a global pandemic by the World Health Organization.The infection control measures abruptly changed social interactions throughout the world.This external shock created new routines and transformed old ones (Mondada et al. 2020a).First, it forced people to work and meet online, collapsing work and private spaces and replacing face-to-face interactions by mediated interactions.While telework is not new, it became the dominant way to organize work for many (office) workers (Parker et al. 2020).These new routines had to be internalized almost overnight, pushing workers to learn and engage in new practices in the digital realm.Second, social distancing means the rules of offline interactions became restricted.Handshaking, hugging, or kissing generally were no longer socially accepted.Instead, 'new' rules of physical interaction had been introduced, such as the 'elbow bump' (Mondada et al. 2020b).As a result, people began to act in new ways without yet knowing or having internalized a new code of conduct, i.e. the social distancing norm has been a developing interactional process (Drury and Stokoe 2022, 985).Ergo, the COVID-19 pandemic defamiliarised the familiar at a global level.
Following these developments in socially distant and social distancing situations, we focus on how people discuss their experiences with such COVID-19 induced interactions on Twitter and in news media articles.We do so by exploring the perceived awkwardness of these interactions.Awkwardness is historically referred to as a situation heading in the wrong direction (Kotsko 2010).Yet, despite awkwardness being an important signifier of failed interactions, particularly in popular culture and social media (Kotsko 2010), we know very little about awkwardness in everyday life.Therefore, we examine what people perceive as failed interactions during COVID-19, and how they perceive these awkward moments.
We look at how awkwardness in social interactions is discussed on social media and in news media in the United States and the United Kingdom during the first wave of the COVID-19 pandemic.First, using topic modelling on Twitter data, we map the main types and situations of social interactions considered awkward.Second, using a text analysis of news media articles, we aim to understand what makes such social interactions awkward.Our focus on social media and (online) news media is both practical in terms of data availability as well as a sign of the times, as many people retreated to digital forms of interaction to stay informed about current developments and even 'curate [d] the affective self through social media' that helped them cope and express themselves in challenging times (Marshall 2022, 19).
With this paper, we make three contributions.First, most studies regarding the effects of the COVID-19 pandemic have focused on the economic (Glied and Levy 2020), psychological (Sood 2020), or societal consequences (Duvall 2022).Less attention has been devoted to how we engage in mundane, everyday social interactions, both online and offline -a notable exception being the project Human Sociality in the Age of COVID-19. 1  Second, in existing research on mediatization in everyday life, technology-mediated interactions are usually seen as of 'lower quality' than face-to-face interactions (see also Fernback 2007).For example, studies show that online arenas are less likely to induce meaningful interactions (Hine 2013) and more disinhibition encouraging unsocial behaviour (Kilvington 2021).This paper builds on these studies to explore which aspects of the mediatization of everyday life affect the success (or lack thereof) of social interactions during a global pandemic.Third, despite its ubiquity in mediatized popular culture, few attempts have been made to empirically study awkwardness, outside the context of television.This paper aims to further unravel what awkwardness is, how it manifests in today's (post-)pandemic society and what this tells us about (changing) social norms and rituals.This helps us to better understand the socially constructed nature of how social interactions and sociality are organized (Mondada et al. 2020a).

Face-to-face and mediated social interactions in times of COVID-19
In his seminal work on social interactions, The Presentation of the Self in Everyday Life, Goffman (1959, 15) defines face-to-face interactions as 'the reciprocal influence of individuals upon one another's actions when in each another's immediate physical presence'.In such (bodily) co-presence, an individual gives an expression, but also gives off other signs that will be interpreted by other participants in the social occasion.In social situations, sociality must be curated, managed and performed by people in order to ensure playing the 'correct' role and making the 'right' impression.Social interactions are imbued with interaction rituals.They require a mutual focus of attention, creating a feeling of intersubjectivity (Collins 2005).Successful rituals evoke shared action and (symbolic) awareness: a positive feedback loop that strengthens group solidarity and leads to increased emotional energy.Failed rituals result in low levels of collective effervescence, no group solidarity, no heightened energy levels (Collins 2005, 51), and possibly in awkward emotions evoked by social interactions gone wrong.
Goffman's and Collins' work was drafted in a time when interactions occurred mainly face-to-face.Though such interactions have been declining in generalwith an increasing share of our time now spent online -the COVID-19 pandemic accelerated this process.At the onset of the pandemic, Collins (2020) deliberated how sharp disruptions of normal conditions of social interaction affected the validity of the theories formulated above.He noted that non-face-to-face forms of communication have the potential to supplement, but not to supplant, face-toface communication.As emphasized, bodily co-presence is important in successful interaction rituals, because it facilitates mutual focus, shared emotions and rhythmic entrainment -think about the emotions felt in attending a football match.Collins (2020) observed that COVID-19 pandemic hampered successful interaction rituals in two ways.
First, with most offices now requiring telework, many businesses have tried to compensate for the loss of face-to-face interactions by introducing mediated alternatives, such as virtual 'water cooler-sessions' or drinks.Anyone who has been part of such endeavours will probably confirm Collins' observation (Collins 2020) that the degree of involvement in such sessions is often limited.Seizing up the situation even in offline situations is a perilous moment, let alone when having to do this in a remote setting.The conditions for successful interaction rituals require synchrony (Collins 2005), which is hard to achieve with lagging video conversations and distorted voices (Ilomäki, Ruusuvuori & Laitinen 2021).Second, due to the almost worldwide obligation to wear face masks and keeping at least 1.5-metre distance, bodily co-presence became more complicated (Julia et al. 2020).Micro-characteristics of interaction, like appropriate speech pausing or mimicking emotional states, are fundamental solidarity builders during interactions (Koudenberg et al. 2017).Their dissipation gave rise to misunderstanding and potentially awkward situations.

Awkwardness as failed interaction rituals
With the reshuffling of rituals ushered in by the COVID-19 pandemic, looking at awkwardness potentially reveals which often given-for-granted dimensions govern rituals of interaction in moments of 'normality'.We, therefore, map awkward interactions stemming from forcefully socially distant (mediated) and social distancing situations and investigate how they are discussed in (news/social) media.Yet, first, we will dissect the concept of awkwardness.Kotsko (2010) discerns three types of awkwardness.First, everyday awkwardness refers to awkward individuals: those that struggle to navigate social norms.Though this might seem as if awkwardness is an inherently individual defect in line with Goffman's (1959Goffman's ( , 1967) ) perception of embarrassment, 2 Kotsko emphasizes that the emotions caused by these mishaps affect both the sender and receiver.Second, instead of referring to individual misconduct, cultural awkwardness concerns the 'general malaise that accompanies a relatively weak norm' (Kotsko 2010, 17).For example, an unsuccessful interaction such as failing to notice that one is still muted in a Zoom meeting is obviously an individual failure -leading to low levels of collective effervescence -but also the negligence of the audience not (subtly) informing the sender of their mishap.Third, radical awkwardness is an awkwardness that arises when a shared norm governing a given situation is absent altogether.In a world in which norms changed rapidly, with new norms not yet equally accepted in all social groups, it is this kind of awkwardness that seems to be ubiquitous.For example, the pandemic has demonstrated how radically different beliefs about the coronavirus have led to different ways of interacting: from strict isolation to a conspicuous hugging.
The weak and lacking social norms during the first wave of the pandemic seem to be fertile ground for awkward moments to occur.In this paper, we therefore examine what people perceive as failed interactions during COVID-19, and how they perceive these awkward moments.

Data and methods
First, to map which COVID-19 related topics people considered awkward, we analysed English language tweets 3 related to awkwardness.The outcomes of this first step informed our search strategy for our second step of analyses that looked at the English language 4 news media articles that describe particular to COVID-19 awkwardnessinducing situations in greater detail.

Social media data
Twitter (now X) is one of the largest online social networks that allows users to post and interact via messages known as 'tweets'.We used a combination of search terms, combining the word-stem of 'awkward*' and the following COVID-19 related terms: 'covid*', 'corona*', or 'rona'.To access these data, we used a Twint script which is frequently used in social media research (Bonsón, Perea, and Bednárová 2019).We collected English language Twitter data from 1 January 2020 until 31 December 2020, covering the early stages of the pandemic in which people were already social distancing by their own initiative before official measures took place, as well as the months in which people felt less confident about -or used to -new forms of communication.Our analysis starts from 1 March as this is approximately the date during which awkwardness started to feature more prominently in the discourse and remained consistent thereafter (Figure 1).
The entire dataset consists of 24,188 tweets; the dataset from 1 January 2020 until 31 December 2020 includes 23,802 tweets.Both encompass the tweet text, date of the tweet, username, number of replies and favourites.Eighty-eight per cent of the tweets were from unique users.The text data were cleaned in Python by stemming, lowercasing and removing stop words (e.g.and, or, that), corpus-specific words, included in >80% of the text (e.g.awkward, coronavirus, etc.), and rare words that appeared in less than 200 tweets.We use Twitter data cumulatively to look for themes, rather than analysing concrete users, which should be similar to anonymous public observation.Additionally, to protect the users from any unwanted attention, we anonymized all the tweets and private Twitter accounts mentioned in this research.
Topic modelling (TM) -an inductive unsupervised machine learning techniquewas used to discover themes associated with awkwardness in the context of COVID-19 on Twitter.We used the Latent Dirichlet Allocation (LDA) topic modelling technique that explores latent structures in texts by clustering words that occur in documents together more frequently than one would expect by chance (DiMaggio, Nag, and Blei 2013, 578).Moreover, LDA arranges these latent structures, or themes, proportionally.This helps to identify which themes are dominant.While tweets can have multiple themes at once, for the purpose of this study, we only look at the dominant themes of the tweet, specifically, the theme with the highest proportion in the document.To conduct TM analysis, we used mallet tool for Mac (McCallum 2002).By running the topic models with increasing number of topics and qualitatively assessing the coherence of topics' output and how logical are the analytical categories, we settled on a topic model with 30 topics (see Online appendix A). 5 As the aim of this study is to analyse awkwardness in relation to social interactions, we selected broad search terms representing the context in which these social interactions occur to ensure that the topics are exhaustive.Therefore, TM output also included topics related to awkwardness in the context of COVID-19 but not related to social interactions (e.g. the way governments handled COVID-19 responses, awkwardness in relation to vaccines, deaths worldwide).We have discarded the topics that are not related to social interactions (the complete TM is available on request).The number of topics related to social interactions was 10 (see Online appendix B).In other words, awkward social interactions represent one-third of the COVID-19 discourse for topics that were considered awkward. 6

News media data
To gain better insights into the mechanisms behind awkward interactions during the COVID-19 pandemic, we investigated the use of awkwardness in coronavirus-related English language news media, again during the period ranging from 1 January 2020 to 31 December 2020 in the US and the UK.We used the Nexis Uni database and included the 20 of the most read newspapers in the UK (YouGov 2020) and in the US (Infoplease 2020) (see Online appendix C and D).Building upon the observations of Collins (2020) and our own Twitter analysis, we focused on mediated interactions and everyday interactions affected by the social distancing measures.The queries used for finding news items related to awkwardness and COVID-19 measures were 'social distancing' and 'awkward'.'Awkward' and 'Zoom' or 'Skype' were used to include news articles on mediated interactions through the most prevalent videotelephony platforms.This sampling procedure produced a sample of 50 units (32 social-distancing, 18 mediated interactions).We used thematic text analysis to identify convergence points within their narratives of awkwardness that could allow for a deep understanding of this emotion and its object (Braun and Clarke 2006).

Mapping awkward social interactions ('what')
Our TM Twitter research provided a list of topics considered awkward during the coronavirus pandemic (see Online appendix B).In line with Collins (2020), it also shows that these topics are usually driven by two (meta)situations of distance: socially distant, mediated forms of interactions and social distancing measures leading to uncertainties on how to interact in a situation.

Social distancing
Most of the topics concern social distancing-related issues: the awkwardness of day-today interactions in close spatial proximity to others.According to Twitter users, the social interactions that were considered most awkward during COVID-19 were performances of 'old' interactions while socially distancing.Such topic as 'New/old life comparisons' (40%) featured many instances of individuals comparing themselves before and (hypothetically) after the COVID-19 pandemic.'If you guys thought I was awkward before . . .damn has COVID made me forget how to socialize', wrote one user.Often, in this topic, the juxtaposition of the new reality over previously given-for-granted ways of interacting is the very source of awkwardness.Collins (2020) observed that artificial, digital emulations of social encounters hampered interaction by not allowing sufficient involvement among participants.These findings seem to expand on Collins' observations by indicating that such emulations of 'normal' interactions are prone to failure (awkwardness) also in offline (social distancing) environments.
Recurring issues summarized under the topic of 'Situation scripts' (13%) also concerned a series of daily, often given-for-granted interactions whose scripts needed to be re-written in the light of social distancing rules, as indicated by the keywords 'door', 'walking' and 'toilet'.Taking about the very relatable case of elevators, one user suggests that 'Elevators have always been an awkward space.But how do you maintain social distancing in an elevator during COVID-19?''.The topic of 'Family/celebrations' (10%) concerned situations such as family reunions or celebrations in general.Awkwardness here is, for example, related to having to reject invitations.These too are caused by uncertainty concerning interaction routines (e.g.'Corona has made me socially awkward when I arrive to gatherings cause idk how people feel about touching or saying hi Latino-ly'), which also shows the strong cultural connotations of types of interaction.'Etiquette' (10%) brings forth the awkwardness felt in relation to physical greetings when meeting other people, showing that new practices of greeting, albeit broadly known, are nonetheless still strange and somewhat laughable to participants (Mondada et al. 2020b).For instance, one user described 'The romantic tension between me and every new person I'm introduced to who momentarily forgets Covid, reaches out their hand to shake mine, remembers Covid then‚ embarrassed, lets their hand fall gently to their side.Their face?Masked.Their eyes?Yearning, awkward'.In the situations above, from 'Situation scripts' to 'Etiquette', we might see awkwardness as emerging from a lack of intersubjectivity (Collins, 2004), which in turn is caused by the inability to perform rituals of interaction like shaking hands, sitting closely during a romantic date.
In addition to these broader categories of social interactions, Twitter users also referred to more specific aspects or sites of social interactions.The use of face masks, for example, has emerged as topic 'Masks' (7%).This led to shifting power dynamics due to the increased importance of good individual citizenship, e.g. one user stated that 'It can be awkward and uncomfortable to remind others of the importance of masking to prevent the spread of #COVID19‚ but it is important'.Behaviour in public spaces has also surfaced in 'Displaying COVID-19-like Symptoms in Public' (5%); discussing the awkwardness of presenting coronavirus in public, such as coughing.For example, one user reports that 'It's so awkward being an allergy sufferer during this era of Covid.*tries not to sneeze in public*' Finally, 'Delivery/rides' (3%) related awkwardness to situations involving delivery services, and 'Hair/looks' (3%), grouped tweets relating to self-care practices which, in absence of hairdressers, nail salons and other services, had to be carried out at home, or were not taken care of at all.Thus, the awkwardness arising in situations of social distancing interactions was predominately correlated to lacking norms and rituals which individuals could rely on to mould their behaviour.These awkward instances indicate a situation of what Kotsko (2010) calls radical awkwardness.Here, appropriate forms of behaviour are shrouded in doubt and uncertainty, and are developing interactional processes (Drury and Stokoe 2022).This leads to discomfort as individuals wait to see if their behaviour will be judged appropriately by their peers, demonstrating how 'moral concerns about actual or potential violations to these rules became relevant in and through social interaction during this period' (Ekberg et al. 2021, 684).

Socially distant interactions
The socially distant form of interaction -i.e.mediated interactions -also appeared repeatedly in the TM.Mediated interactions formed its own topic and related mostly to 'Video calls' (7%).Here, key terms such as 'etiquette' or 'call' relate awkwardness to lacking clarity on appropriate behaviour (situational scripts) in certain online scenarios.Awkwardness here is solved in the eventual cultural consolidation of such routines.In one example, one user suggested that 'In the new world post COVID-19, online socialization will be new paradigm.Initially you may feel awkward but you may get used to this sooner or later.We may hear now, lets meet online with BYOD (Bring your own drink)'.Socially distant interactions also featured within some of the topics described above.The 'New/old life comparisons' (40%) topic included students reflecting on online education: 'Such a weird thing to end a class on-line.No final exam, no final goodbyes, no handshakes, just an awkward, "Thanks" folks.Be safe and stay home. . .So very, very weird.I hate coronavirus'.'Etiquette' (10%) showed that in-person interactions were not the only source for awkward greeting situations, being a source of embarrassment also for mediated interactions: 'Leaving calls has got to be the most awkward thing.Ever since COVID I don't know if I should shake your hand'.'Family/celebrations' (10%) exemplifies the interactions with family that traditionally took place in face-to-face context and how being socially distant exacerbated the awkwardness: 'Zoom call w fam was awkward bc my older brother and I are both in quarantine & his gf nearly cried talking about how COVID has uprooted their lives while my entire extended family nodded solemnly from my mom's living room'.
These findings position awkwardness as emerging from a lack of internalized norms and rituals to deal with this sudden new reality and ways of interaction.This situation resembles Kotsko's (2010) concept of cultural awkwardness, a general state of weak norms of interaction producing a widespread discomfort during interactions.Online interactions, similar to greetings in offline interactions (Mondada et al. 2020a), seem to be somewhat ritualized yet not sufficiently internalized in that participants are able to interact without thinking about the interaction itself.This gap arguably triggers the experienced awkwardness.Here, the concept of awkwardness allows us to observe this discontinuity as regarding not only moments in which the sequence is intrinsic to the interaction, such as the opening of videocalls (Ilomäki & Ruusuvuori 2020), but also latent in other moments of social interaction.

COVID-19 as a conversation topic
Finally, one topic goes beyond the two categories above and concerns how even talking about COVID-19 itself engenders feelings of awkwardness.The topic 'Small talk' (19%) includes Tweets like 'I've got the flu and my Polish tutor asked me (via zoom) if it was "the Covid" and I laughed and said "hey, maybe" and she kinda laughed and then I coughed and we sat in awkward silence for a good 4 seconds and continued the lesson'.Small talk about COVID-19 was not only awkward at times on its own accord, but also it was amplified by social distancing and exacerbated socially distant interactions, such as in this example: 'In an attempt to kill an awkward silence, my dad just told the guy trying to fix our internet a CORONA VIRUS joke.He made the guy half-chuckle and then the awkward silence resumed.Dad, why are you this way?'.Digital tools contributed their share to this topic, as indicated by the key terms 'call' and 'phone' of this topic.As a Twitter user puts it, 'if COVID-19 scientists were stuck in an awkward [Zoom] breakout room they would find a cure within 5 minutes'.These examples show that COVID-19 as a conversation topic might be the awkward individual among conversation topics.Twitter users often referred to the actions of individuals (cf.'dad' in the example above) struggling to navigate social norms, which explicitly connects to Kotsko's (2010) everyday awkwardness.These tweets and small talk about the awkwardness of COVID-19 can be seen as collective attempts at sense-making.This includes sharing their experiences online hoping for peer feedback (e.g.likes, retweets) to help calibrate their behaviours (Han & Kuipers 2021); a set of practices Marshall (2022) coined 'covidiquette'.

Analysing awkward social interactions ('how')
So far, we have highlighted situations or topics Twitter users perceived as particularly prone to awkwardness.But what exactly makes these situations awkward?The more indepth news media articles provide valuable insights: intimacy, synchronicity and sequency.
First, socially distant and social distancing interactions were seen as being not or less able to establish a sense of intimacy -unable to generate sufficient emotional energy (Collins 2005).Lacking synchronicity (e.g. by non-physical co-existence) resulted in imitated interpersonal intimacy and as such more awkward interactions.Getting into the same rhythms -provoking rhythmic entrainment -seemed to be much more difficult in, for example, online dates, making it feel 'a bit like a job interview the first time [. . .] [a] dodgy connection can really ruin what is already a pretty awkward date' (Jehring 2020).In a similar vein, a lack of intimacy made it much more difficult to establish trust and solidarity -normally outcomes of intersubjectivity in successful interaction rituals, but notoriously hard to forge through online interactions (Hine 2013) -in therapy sessions: 'Trust me, it would've been a lot easier [to go through regular psychological therapy], and way less awkward, than sitting in silence with a therapist in front of you for 45 minutes' (Hancock 2020).
The articles also showed how such lack of synchronicity trickled down to face-to-face interactions, where a sense of 'incompleteness' -especially a perceived lack of intimacylingered.This potentially made every ritual or form of communication awkward.Obviously, this lack of intimacy plays out in 'intimacy oriented' interactions such as dating (see Chilcott 2020), yet other forms of interaction -especially rituals -too suffer from social distancing measures.Take Gallagher (2020), who lamented how the social distancing measures made comforting rituals impossible.'[Attending a funeral during lockdown] was very awkward, all I could say to my grandad was hello and goodbye even though he was really upset and his wife had just died.I don't have a way with words and you can't do what comes naturally and just hug them'.
This insufficiency also occurs in the spoilt sequences of interactions.For example, a participant in a romantic meeting reported how intimate -and quite awkward -conversations such as coronavirus status, quarantine credentials and exclusivity need to be held before engaging in the first kiss, a previously spontaneous demonstration of affection (Raza 2020).Arguably, this makes it much harder to evoke sufficient emotional energy (Collins, 2005) to move towards the next steps in strengthening interpersonal solidarity.In a different setting, Clark (2020) highlights 'the awkward experience of sending out pre-party group emails to strategise about the bathroom'.Here, awkwardness tended to pass 'as everyone settled in, six feet apart, wine glasses in hand, we gradually eased out of the awkwardness and remembered what it was like to eat and drink with loved ones on a warm summer night' (Clark 2020).These findings confirm previous research on how the sequential organization of the opening of the video calls among family members is used as a resource to sustain relation intimacy (Gan, Greiffenhagen & Licoppe 2020, 379).

Conclusion
We examined how awkwardness in social interactions is discussed in (social/news) media during the first wave of the COVID-19 pandemic.Using the concept of awkwardness, we foreground the sociological and mediated aspects of the 'new normal' and understand the widespread unease and psychological unsettledness of these times.This paper also proposes an empirical application of awkwardness as a cultural indicator to map and understand moments of cultural change.This under-studied (yet ubiquitous) emotion is further explored in a new mediated context (Goffman 1959(Goffman , 1967)), based on empirical research instead of philosophical observations (Kotsko 2010).
We first mapped the main social interactions considered awkward ('what') and connected these to the types of awkwardness developed by Kotsko (2010).We found that almost all topics relating to the awkwardness of social distancing, comparing the old and new realities, the need to rewrite situational scripts, dealing with new etiquettes of physical interaction and so on, fit well within the description of radical awkwardness.This is likely due to the abruptness with which these new, carefully coordinated forms of interactions were introduced in most societies.Mediated, socially distant topics were limited to 'new/old comparisons', 'etiquette', 'family and celebrations' and -of course -'videocalls'.Often awkwardness here arises from a lack of online socialization.The norms in these interactions have not (yet) gathered sufficient strength to convey emotional energy, leading to diminished rather than heightened emotional energy.Hence, cultural awkwardness seems to predominate in mediated interactions.Finally, our results highlight the everyday awkwardness of COVID-19 as a conversation topic ('small talk').Talking about the virus is referred to by Twitter users as awkward per se, often related to (trying to) make jokes about it.This echoes the work of Browne (2016), arguing how humour helps to overcome the everyday awkwardness of discussing taboo topics.
Second, we aimed to understand what makes such social interactions awkward ('how').The more detailed reports of awkwardness gathered from news media evince that these situations are perceived as awkward (appearing, therefore, as failed rituals) due to a lack of intimacy, synchronicity and sequence.These faults greatly undermine the emotional energy and interpersonal solidarity, confirming recent observations that mediated (or significantly altered) forms of interaction do a poorer job in facilitating (pro-)social behaviour (Kilvington 2021) and sustaining friendship (Lai and Fung 2020).Though online platforms -be it through voluntary participation in, e.g., social networks, or in this case, more involuntary 'migration' to video call services such as Zoom -are able to mirror much of offline social life, they still lack the affective elements, intensifying feelings of awkwardness (Marshall 2022).
Processes of digitalization and mediatization affect how we spend time with others online, but also how we spend time in the physical co-presence of others (Cahir and Lloyd 2015).Consequently, social norms and expectations on social interactions are overturned.The COVID-19 pandemic greatly accelerated this development, potentially catalysing an even greater sense of awkwardness.Unsuccessful or incomplete interactions go beyond what Kotsko (2010) calls individual awkwardness but also encompasses a sense of unease caused by the diffusion of all kinds of social norms (cultural awkwardness) or even a lack of clear norms at all (radical awkwardness).Shaking hands may again be allowed, yet this does not mean that everyone will equally feel the need to do so.Though measures have been relieved, they created a ripple effect that might impact social interactions for times to come.To build on one of the Tweets mentioned above, COVID-19 scientists may cure us of the virus, but they might be less able to cure us of the awkwardness caused by ambiguous and unsettled norms of interaction.
Nevertheless, our approach has some limitations.First, we recognize that Twitter might overrepresent younger, highly educated and affluent groups (Wojcik and Hughes, 2019).Second, albeit engaging with two national contexts (UK/US), we did not compare these contexts.With awkwardness being a social emotion, national differences are likely to exist correlated with broader cultural differences across societies.Third, topic modelling, while affording vast sweeps of information to be analysed by letting an algorithm categorize the data, does not allow for smaller differences within topics to emerge and be investigated.Finally, we devoted little attention to how people repair failed interaction rituals, for example, by replacing the handshake with the elbow bump.Future contributions are invited to make use of qualitative tools such as interviews or observations to add nuance and specificity to the dimensions and characteristics of awkwardness observed in this research, as well as to how people creatively repair breakages in interaction rituals and social relationships (see also Clegg 2012).

Notes
1. https://www.lorenzamondada.net/humansociality-covid192. Whereas embarrassment is a reactive form of acute self-consciousness about a faux pas that has already occurred and shyness involves anticipatory anxiety about imagined or expected blunders (Scott 2004, 123), awkwardness can be both as some social interactions are laden with latent awkwardness, such as first dates.3. We chose Twitter because it allowed researchers to download tweets without being registered.4.This decision has largely been driven by the fact that awkwardness itself is an English expression. 5.The standard default number of keywords in Mallet is set to 20. 6. Topic proportions add up to more than 100% because tweets often have more than one topic.In our analysis, we assign a single topic to a tweet where the topic proportion is higher than 50%.