Article Text

Download PDFPDF

Bubbles and lockdown in Aotearoa New Zealand: the language of self-isolation in #Covid19NZ tweets
  1. Jessie Burnette1,
  2. Maebh Long2
  1. 1 English and Linguistics, University of Waikato, Hamilton, Aotearoa New Zealand
  2. 2 English, University of Waikato, Hamilton, Aotearoa New Zealand
  1. Correspondence to Jessie Burnette, Arts, Law, Psychology and Social Sciences, University of Waikato, Hamilton 3240, Waikato, New Zealand; jessie.burnette{at}


In March 2020, as cases of COVID-19 were found in Aotearoa New Zealand, the government moved to eliminate community transmission of the virus through self-isolation. During this month, as the population discussed if, when and how households would be asked to stay at home, terms such as lockdown—the state of (national) closure—and bubble—the household isolating together—became common parts of everyday conversation.

In this article, we blend quantitative and qualitative research methodologies from corpus linguistics, literary studies and the medical humanities to compare the affective range of the terms lockdown and bubble as they were used in tweets containing the hashtag #Covid19NZ. Both lockdown and bubble are metaphors of containment that provided different ways of understanding and engaging with government stay-at-home measures by highlighting and minimising different aspects of the event. We found that while the strong, prison connotations of lockdown were reflected in discussions of the measure as a tough form of control exercised from above, the lighter associations of the term bubble led to the perception of this measure as more malleable and conducive to exertion of individual control. Yet, although the seemingly restrictive range of lockdown made it a useful term for the expression of negative affect, the term was actually more frequently used with neutral or unclear affect to share information. Conversely, while bubble tweets expressed more positive sentiment, humour and support towards government stay-at-home measures, this rendered the term surprisingly restrictive in its potential uses: its lightness makes it an effective way to limit the expression of antilockdown sentiment. As Kiwi Twitter users faced the uncertainty of the first COVID-19 lockdown, the pre-existing connotations of the metaphors used to frame stay-at-home measures also helped frame their own experiences of these measures.

  • COVID-19
  • metaphor
  • medical humanities
  • literary studies
  • internet

Data availability statement

Data are available on reasonable request. The data used in this paper have come from the CovidNZT Twitter Corpus and can be made available on reasonable request.

This article is made freely available for personal use in accordance with BMJ’s website terms and conditions for the duration of the covid-19 pandemic or until otherwise determined by BMJ. You may use, download and print the article for any lawful, non-commercial purpose (including text and data mining) provided that all copyright notices and trade marks are retained.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

“We are thrown into a world unsettling and unfamiliar” (Twitter user 21 2020b)

[W]e are all now preparing to go into self-isolation as a nation. […] I would rather make this decision now, and save those lives, and be in lockdown for a shorter period, than delay, see New Zealanders lose loved ones and their contact with each other for an even longer period. I hope you are all with me on that. (Ardern 2020b)

The above announcement, given by Prime Minister Jacinda Ardern on 23 March 2020, evoked complex, mixed reactions in many. The population of Aotearoa New Zealand had watched the pandemic strike countries across the world, but now the newly formed ‘Team of 5 Million’ had to contend with a lockdown of their own. Individual or household sequestration within a nation that had closed its borders brought sudden concerns about income, employment and mental health as well as suspicions of the powers granted to the government following the declaration of a state of emergency. As physical contact diminished, social media provided an alternate platform for users to find and maintain relationships with others. Information about lockdown measures was rapidly shared by ‘citizen journalists’ (Zappavigna 2021, 171), everyday users who reiterated announcements and reported on the events they witnessed while in lockdown, at times reframing events with the inclusion of their own opinions or emotional reactions. Through social media platforms, users were able to act as witnesses to the events happening directly around them, which may have otherwise gone unnoticed within the broader narratives of the pandemic.

On Twitter, many Kiwi users used the hashtag #Covid19NZ to generate ‘searchable talk’, defined by Michele Zappavigna as marking discourse used to connect with others through particular ideas and concerns (Kiwi 2021, 1). Hashtags, still a fairly new creation, are short-lived ‘language about language’ that perform a function of relation while constantly shifting alongside the interests and concerns of Twitter users. Through the use of the #Covid19NZ hashtag, users were able to generate points of connection with other users, mostly Kiwis, who were also sharing their thoughts and feelings about the pandemic. Gathering data through hashtags offers a unique and unmoderated view into the ways in which everyday people engage with the issues of the moment. In this article, we blend quantitative and qualitative methods from linguistics, literary studies and the critical medical humanities to examine the different ways people on Twitter using the hashtag #Covid19NZ framed their experiences of stay-at-home measures in Aotearoa New Zealand during March 2020.

With the decision to enforce pandemic restrictions in Aotearoa New Zealand, there was, as in the rest of the world, a need to quickly determine clear terminology to describe these measures. At the height of these restrictions, the term lockdown became the most prominent term, used in conjunction with alert levels, which ranged from 1 (the most minimal restrictions) to 4 (complete stay-at-home orders). Lockdown was not a new term, having already been used internationally to describe various security and health measures prior to the COVID-19 pandemic, from disease-controlling measures to security measures taken during prison riots and active shooting events. Used internationally to describe COVID-19 measures as early as January 2020 (Leung 2020), globally lockdown denoted a range of government-mandated health measures, including some more resemblant of Aotearoa New Zealand’s lighter alert level 2 restrictions. However, within Aotearoa New Zealand, lockdown was predominantly used in reference to the intensive alert level 3 or 4 restrictions.

The term bubble, whose pandemic usage originated in Aotearoa New Zealand, featured alongside lockdown, and was used to refer to ‘an exclusive social network’, frequently centred on a single household who isolated together (Long et al. 2020). While bubble was later used internationally to describe loosening restrictions introduced during May in the UK and the USA (Cowburn 2020; Thompson 2020), with Belgium allowing households to create a ‘corona bubble’ by adding four guests to their household (Rankin 2020), in Aotearoa New Zealand bubble first emerged as a complementary concept to lockdown, providing further guidance and clarification for Kiwis unsure of how to navigate intense pandemic restrictions. Although introduced at the height of the first alert level 4 lockdown, bubble continued to be used and adapted across alert levels as measures were loosened. As the country moved down alert levels, Kiwis expanded and merged their bubbles to include their broader social networks. Although bubble has previously been used as a conceptual domain for metaphors in several different contexts—for example, housing bubbles, or the protective measures taken for David Vetter, ‘the boy in the bubble’ (Immune Deficiency Foundation, 2022)—some of which, we argue below, have influenced public perception of the term, this particular metaphorical use was new. As bubble and lockdown became increasingly standardised parts of everyday conversation, their prior associations influenced how they were perceived and used by the public. Aotearoa New Zealand’s March 2020 lockdown has been largely viewed as highly successful, preventing the large numbers of COVID-19 deaths seen in many other countries during 2020 and allowing time for the development of a vaccine and subsequent vaccination of the community. Clear communication of health measures, including the tactical use of linguistic strategies such as effective metaphorical framing, has been largely recognised as a key factor in the country’s 2020 success.

Several studies have considered aspects of either COVID-19 and metaphor or COVID-19 and social media, such as trends in public perception and discussion of certain aspects of COVID-19 on Twitter (Kwon et al. 2020; Wicke and Bolognesi 2020; Saleh et al. 2021); the depictions of certain groups during the pandemic (Jimenez-Sotomayor, Gomez-Moreno, and Soto-Perez-de-Celis 2020); the origins and uses of different metaphors to describe COVID-19 (Abdel-Raheem 2021); pandemic social media humour (Mpofu 2021) and the ways in which politicians and media around the world have used metaphors to frame COVID-19 and their responses (Esfandiary 2021, 191–198). However, the ways the public adopted, rejected or played with different metaphorical framings of stay-at-home measures on social media has been sorely under-researched.

In Aotearoa New Zealand, as people adjusted to an abrupt influx of new concepts and realities, metaphors of containment proved particularly useful for Kiwis discussing government stay-at-home measures in March 2020. One way, most commonly found in tweets using lockdown to describe the situation, was to frame stay-at-home measures as strong restrictions which must be endured or overcome in order to return to secure ‘normalcy of life in the future’ (Twitter user 1 2020). Another way of engaging with the unpredictability of the ‘new normal’, linked predominantly to tweets using bubble, was to embrace the ‘uncertainty, volatility and flux’ by framing stay-at-home measures as an opportunity to re-evaluate old ways of being and search for alternative ways forward (Twitter user 21 2020a). In using these terms, Twitter users were able to shape their experiences of the COVID-19 pandemic and stay-at-home measures in quite different ways.

In what follows, we analyse the affective range of lockdown, noting the ways it was used to understand and critique government stay-at-home measures by highlighting and minimising different aspects of the experience. We then compare it with bubble, a term whose affective connotations are very different to that of lockdown. We found that while lockdown was predominantly used to discuss a uniform, restrictive measure which needed to be endured or overcome as a community in order to return to the previous status quo, in #Covid19NZ tweets bubbles were often presented as semi-malleable domestic situations which could be individually customised to provide opportunities for creativity and change. Where bubbles may have been perceived as allowing for higher levels of individual control, lockdown seemed to be understood more as control exercised from above. For users, lockdown signified an event which could be affected by the greater collective, but only to a lesser extent by the individual. Metaphors have long been recognised as having a powerful influence in defining our perception of events and, in turn, our lived social realities, and this remained true as Kiwis used the metaphors of lockdown and bubble, both metaphors of containment, to conceptualise and respond to COVID-19 stay-at-home measures.

“I am fascinated by metaphors, and thinking about the ‘bubble’ of lockdown” (Twitter user 20 2020)

In the time of COVID-19 we are contained, to varying degrees as risk escalates and abates, within our countries, our communities and our households, as well as by a heightened awareness of our bodies. We contain our breath within masks, which highlights our vulnerability in the face of an invisible threat that resists containment as well as our inescapable interdependence on each other. Within such an uncertain and unsettled time, metaphors play an important role in helping individuals and communities ‘conceptualise the less clearly delineated in terms of the more clearly delineated’ (Lakoff and Johnson 2003, 59). As stated by Lakoff and Johnson, metaphors that become culturally prominent play ‘an essential role in characterising the structure of […] experience’, affecting both our perception of events and, in turn, our lived social realities (2003, 118). That our perceptions of illnesses are largely influenced by the metaphors associated with them is, by this point, well-known. As explored by Susan Sontag in her foundational works Illness as Metaphor and Aids and its Metaphors, epidemic diseases are often viewed through the lens of military metaphors, which frequently frame illness as a war and diseases as threats arising from the Other. Studies of the metaphors of the COVID-19 pandemic have found that ‘Health workers are “servicemen” on the “frontline” “battling” “an invisible enemy”’ (Rose and Dolezal 2020), and that ‘war framing is used to talk about certain topics, such as the virus treatment, but not others, such as the effects of social distancing’ (Wicke and Bolognesi 2020, 1); that the progression of the outbreak has been mapped as ‘movement in the form of “waves” of the disease’ (Craig 2020, 1029) and that ‘fire metaphors can be particularly appropriate and versatile in communication about the COVID-19 pandemic’ (Semino 2021, 56). Most importantly, as shown by Elena Semino, we see that ‘a well-informed and context-sensitive approach to metaphor selection can be an important part of public health messaging’ (2021, 56). The Aotearoa New Zealand government’s presentation of lockdown ‘as a collective and meaningful cause’ has been considered by researchers (Beattie and Priestley 2021, 7), and some of its negative connotations have been touched on in relation to bubble (Trnka and Davies 2020, 172), but the function of lockdown as a metaphor has received insufficient attention. There has been some research on the effectiveness and limitations of bubble as a metaphor that depicts ‘households as sites of responsibility and care’ (Trnka and Davies 2020, 167); its function as a spatial metaphor in shaping an elimination narrative in response to COVID-19 (Kearns 2021, 327; Appleton 2020) and its meaning in relation to colonialism and capitalism (Manderson and Veracini 2020). However, none of this scholarship has engaged with social media or large quantities of naturally occurring data from a broad range of subjects. Furthermore, the public reception and comparative use of these metaphors in naturally occurring language has not, to the best of our knowledge, been examined—a gap we address in this article.

Data and methods

The data used for this study come from the CovidNZT Corpus (COVID-19 NZ Twitter Corpus), which consists of tweets using the hashtag #Covid19NZ and its variations (eg, #covid19nz). From this larger corpus, we have made a subcorpus using the data only from March 2020, a time when both the COVID-19 pandemic and stay-at-home measures were still completely new to Aotearoa New Zealand. There are 10 549 tweets within our full corpus of March 2020 data. When viewing the larger March corpus using AntConc software, lockdown 1 and lock down* generated 1963 concordance hits in comparison with bubble* 183; isolation* 395; quarantine* 170 and rāhui* (the te reo Māori—Aotearoa New Zealand’s Indigenous language—word for a temporary, protective prohibition on access to a particular area or resource implemented by the tribal authorities of the region).2 The lower numbers for bubble are due in part to the term’s later addition to Kiwis’ pandemic vocabulary. Despite the larger recurrence of isolation than bubble, our comparison draws on the latter term as it illustrates a markedly different metaphorical range to lockdown. For parity, we have used 272 tweets containing lockdown or bubble for our in-depth analysis of these two terms. Of these, 154 tweets contain lockdown and 118 tweets contain bubble.

When selecting bubble and lockdown tweets, the following were excluded: retweets, single word tweets and tweets where the term did not occur within a complete sentence (eg, a tweet saying “we’re in #Covid19NZ #lockdown” would be included, but “I want to leave the house. #Covid19NZ #lockdownnz” would be excluded). Where terms were used multiple times within the same tweet, the tweet was still coded once to avoid privileging any one user’s perception of the term. This only left 118 individual tweets which contained bubble; all of these tweets were coded. As lockdown was far more frequent, 154 lockdown tweets were randomly selected for analysis in order to create a more manageable dataset.

The ethics of using and presenting social media data in research is still a contended topic. According to Twitter’s privacy policy, public content (including tweets) is available for research use, however, as found by Casey Fiesler and Nicholas Proferes, many Twitter users fail to realise this (Fiesler and Proferes 2018, 6). To address this uncertainty, we have anonymised the tweets shown and quoted within this paper. Furthermore, all of the tweets shown or quoted were still publicly available in January 2022 and had not been deleted.

While Twitter offers us a unique opportunity to glean data on public uses and perceptions of terms, it is not without limitations. The demographic of Twitter users tends to be skewed towards younger users, as well as speakers of English. Furthermore, while we have followed the hashtag #Covidnz19 when gathering our data, this excludes those who discussed stay-at-home measures without using a hashtag, and does not guarantee that all of the tweeters were New Zealanders as not all users provide their geographic location. Because of this, our findings cannot be deemed exhaustive; further studies using data from other sources would benefit research in this area.

It was not appropriate or possible to involve patients or the public in the design, or conduct, or reporting, or dissemination plans of our research.

This research considers both prominent and more uncommon trends in uses of lockdown and bubble. To further support our findings in relation to more uncommon trends involving stance opposing government measures and discussion of inequitable outcomes for Māori (the Indigenous people of Aotearoa New Zealand), an additional 800 tweets were randomly selected from our March 2020 corpus for comparison.

In this article, we have manually coded all of the variables. The initial variables coded for were stance towards government stay-at-home measures and basic emotional sentiment. There were three categories of stance coded for: ‘supportive of current government measures’; ‘opposed to current government measures’ and ‘supportive of government measures but calling for stronger measures’. Where stance was either unstated or otherwise could not be clearly determined, this was coded as ‘unclear stance’. Emotional sentiment has been coded in polarity terms—positive/negative—and has been determined by considering lexical decisions, multimodal features such as emojis and links, and semantic prosody. As with stance towards measures, where emotional sentiment was unstated or otherwise could not be clearly determined, this was coded as ‘unclear sentiment’.

As we will discuss, sentiment and stance are expressed in significantly different ways dependent on whether the term lockdown or bubble is used when discussing stay-at-home measures. To better understand these differences in usage, we have coded a further five variables: the use of humour or language play; the depiction of stay-at-home measures as holding opportunity for the individual or collective society; pronoun usage; what we describe as the perceived hardness or softness of the measure and whether the stay-at-home measure was described as applying to the individual (I, me, you) or a collective, broader community (we, New Zealand, the city). We coded these particular features as they offer a strong indication of user perceptions of events, offering insights into the ways that the metaphors of bubble and lockdown shape Twitter users’ experiences of stay-at-home measures. Analysing this large number of variables provides a nuanced and in-depth view of how Twitter users perceived and used these terms. In the following sections, we analyse the findings from both quantitative and qualitative perspectives.

‘New Zealand goes into lockdown’ (Twitter user 2 2020)

Prior to the outbreak of COVID-19, the term lockdown was perhaps most strongly associated with the following (Oxford English Dictionary 2022) definitions:

  1. The confinement of prisoners to their cells for an extended period of time, usually as a security measure following a disturbance; the time at which such confinement begins. Also in the context of a psychiatric hospital or other secure unit. Frequently in on lockdown, in lockdown.

  2. More generally: a state of isolation, containment or restricted access, usually instituted for security purposes or as a public health measure; the imposition of this state.

Governing this term were senses of confinement, restriction and disturbance, exacerbated by the negative stigmas associated with prisoners, psychiatric patients and fears of intrusive state control. For many, lockdown holds connotations of punishment: when applied to the home, the term may suggest house arrest (Young 2021, 4). Alternatively, these same restrictions could be understood as safeguards, securing us as we are moved through unstable, dangerous times. For many in Aotearoa New Zealand, the term was most frequently heard in association with school shootings in America. With the arrival of COVID-19 in Aotearoa New Zealand and the country’s move up alert levels, lockdown grew to become the most commonly used term to refer to government mandated stay-at-home measures. Even as the term developed a more specific meaning within the country during the time of COVID-19, its prior associations continued to linger.

For some, the major threat arising from the pandemic was not the loss of the life, but the loss of freedom. Much academic analysis of states of emergency or exception focuses on the ways in which ‘under the mantle of humanitarian protection, governments, including democracies… [usher] in authoritarian rule’ (Trnka 2020, 11). While considering the formation of modern disciplinary systems, Michel Foucault suggests that the advent of plague ‘brings about the political dream of an exhaustive, unobstructed power’ through increased surveillance, partitioning of the public and militarisation, leading to a moment where ‘political power is exercised to the full’ (Foucault 2003, 47). While this ‘integration of medicine and politics, which is one of the essential characteristics of modern biopolitics’ (Agamben 1998, 143), could be viewed as a form of exhaustive humanitarian protection of individuals, it is often framed by Foucault in negative terms: in the name of protection from plague, governments are increasingly enabled to transition to military rule (2003, 47). These ideas are further supported by the works of Giorgio Agamben, which warn that ‘the state of exception tends increasingly to appear as the dominant paradigm of government in contemporary politics’ (Agamben 2005, 2). Declarations of states of exception are often viewed with suspicion, and with good reason—these measures carry with them the ghosts of Guantanamo Bay, Rwanda and Nazi concentration camps, among numerous other examples of legalised violence. The potential for governments to abuse states of emergency in response to COVID-19 was not lost on Agamben himself, who published an antilockdown manifesto in late February 2020 (Agamben 2020), or on Twitter users, several of whom compared Aotearoa New Zealand’s stay-at-home measures with ‘martial law’ (Twitter user 3 2020) and ‘the Stasi’ (Twitter user 4 2020).

However, as shown by Susanna Trnka, many of these analyses suggest that all states of emergency are ‘inherently unjustifiable’, which fails to account for the preferences of everyday citizens and the ways in which ‘collective responsibility, care and blame are envisioned and enacted, not only by governments but by (advanced liberal) citizenries’ (2020, 11). When examining Aotearoa New Zealand’s response to COVID-19, surveys of citizens’ attitudes were ‘dominated by expressions of support for the emergency measures and trust in government decision-making’ (Trnka 2020, 13). While acknowledging the impingement on citizens’ rights, many expressed ownership over their decisions to comply with government measures, making ‘concerted efforts to shift the affective tenor of the “lockdown” to a more positive valence’ (Trnka 2020, 12), whether it was by placing teddy bears in windows for children to see during government-permitted walks, performing free concerts online, or volunteering to deliver groceries for vulnerable members of the community.

Despite lockdown’s associations with prisons and institutions of state control, in #Covid19NZ tweets the term lockdown is the most emotionally neutral of the terms used for stay-at-home measures. As figure 1 shows, in our corpus lockdown was used in an almost equal numbers of positive and negative tweets—34.4% expressed negative sentiment while 31% expressed positive sentiment. Similarly, while tweets containing lockdown with clear stance were largely in favour of government measures (28%), there were a greater number of tweets showing unclear stance (54%), which illustrates the term’s use in impartial statements.

Figure 1

Sentiment and stance across terms (figure by author).

When compared with other terms for stay-at-home measures, lockdown also falls in the middle when examined in relation to stance: it is the third most common term used in tweets opposed to government measures, for government measures and calling for stronger government measures. Because lockdown became increasingly standardised by public figures when discussing level 3 and 4 stay-at-home measures, the seeming neutrality of the term may simply be reflective of it gaining a new principal referential meaning: the official term for Aotearoa New Zealand’s self-isolation.

In many tweets, the lockdown was something to report on, something that could be described but not necessarily altered, to be entered into and endured as a collective in order to return to the previous status quo. This apparent neutrality also seems to come with the view of lockdown as a hard, uniform object which collectively effects entire communities (here, often the country as a whole) and becomes simply a reality to be endured. Within this national uniformity, however, lockdown also enabled a range of thinking about risk. That is, if a prison lockdown suggests the containment of a threat, potentially caused by those incarcerated, within the place of incarceration, a school lockdown, while tainted by the association with lax gun control, implies the protection of the vulnerable from a threat that has breached the institutional perimeter. While both meanings allude to control through separation and containment, in the former sense the threat comes from within, while in the latter the threat comes from without. This range within the term lockdown enabled Kiwis to conceptualise a threat arriving from outside—outside the country, outside the home—as well as threat that comes from within—within the community, stemming from each other. By blending the hard associations of prison lockdowns with the protective elements of a school lockdown and tempering both with the message that the level 4 lockdown was ‘a collective and meaningful cause’ (Beattie and Priestley 2021, 7), a necessary response to a threat rather than the direct wielding of biopower, health officials offered the public a clear sense of what a lockdown entailed while garnering large-scale support.

Within Aotearoa New Zealand, announcements of lockdown were commonly paired by Prime Minister Ardern with the ideas that the government was ‘going hard, and going early’ (Ardern 2020a), and that we, as a collective, were ‘in this together and must unite against COVID-19’ (Ardern 2020b). This strong referential meaning is reflected in the absence of pronoun use in the majority of lockdown tweets. Unlike bubbles, as we will see below, lockdowns were not perceived as the possessions of individuals but as uniform objects entered into by broader communities and better described using articles such as the. When a pronoun was used in conjunction with lockdown, we received the highest usage. Even without a pronoun, descriptions of collective experience were shared by many Twitter users, especially those who were supportive of government measures, and were present in 44.2% of tweets containing lockdown. In tweets of this nature, some users worked to further identify themselves as a part of a united Kiwi collective. This was commonly reflected both linguistically (as in the use of te reo Māori loan words and other features of New Zealand English) and through the use of multimodal features such as images and GIFS. It is these rather stoic understandings of lockdown that were overwhelmingly represented in our March 2020 corpus—however, the less common perceptions of lockdowns, some of which we detail below, offer important insights into more varied perspectives.

Biopower: ‘imprisoning entire population in homes is inhumane’ (Twitter user 5 2020)

As countries around the world began using states of emergency to instigate health measures—cities closed, citizens were ordered to stay at home, families began making their own provisions, workplaces avoided meetings—exertions of biopower were made increasingly visible. These enforced stay-at-home measures constituted a highly visible form of biopolitics, reinvigorating debate over citizen’s rights and the extent to which governments can—or should—exert control over the bodies of individuals (Agamben 1998, 122). As such, despite the framing by government officials of self-isolation as ‘a collective and meaningful cause’, and ‘the COVID-19 virus as an obstacle that required collective buy-in to overcome’, not all Kiwis wished to ‘buy-in’ (Beattie and Priestley 2021, 7). While only four lockdown tweets in our corpus expressed full opposition to government measures, on conducting a check of 800 further March 2020 tweets, the themes discussed below were repeated in the lone 12 tweets—1.5%—which were found to oppose government stay-at-home measures. The rarity of tweets opposing these measures reinforces previous reports on the largely unified public support for government mandated stay-at-home measures in March–April 2020 which was so necessary for their success (Hafner and Sun 2021; Jamieson 2020). For three of these, the imposition on personal freedoms was positioned as an intolerable threat to the future of the country. Fears of increasing restrictions from an untrustworthy government conspiring to limit freedoms to control the citizenry, or enact ‘Military rule’ (Twitter user 6 2020), were explicitly cited in two of these tweets and implied in the third. In the third tweet, as in other tweets from the broader dataset, COVID-19 was constructed as ‘a virus that is no worse than the seasonal flu’ (Twitter user 7 2020), with lockdown measures presented as both illogical and threatening. Here, measures were compared with the response of Sweden who only ‘restricted people over 70 to their homes because they are the ones most likely to need a vent’, as opposed to ‘imprisoning an entire population’ (Twitter user 5 2020).

In the fourth tweet, however, distrust in lockdown was bound with a theme resonant in five other tweets with different stances toward government measures: economic inequity. For this user and the other five, the economic insecurity brought about by the pandemic foregrounded the financial precarity already present in their lives as they found themselves unemployed or having otherwise lost wages. These users, three of whom identified themselves as low-waged ‘essential’ workers who were unable to remain at home during lockdown, found themselves on the front line of the pandemic, contending both with the ‘chaos’ (Twitter user 8 2020) of a panicked public and their own peculiar position: as both essential and, by nature of their low wages and dangerous work, expendable, this group found themselves in the role of Foucault’s crows, ‘already sentenced to death and thereby [garnering] a strange mobility in life’ (Brown 2020, 7). Faced with the precarity of their circumstances in the face of economic uncertainty, our fourth opposing user viewed the lockdown with resentment, as a measure that unevenly benefited ‘the rich pricks’ at the expense of the everyday person (Twitter user 9 2020). Feelings of disenfranchisement and fear, along with the sense of ‘Us’ and ‘Them’ already present within an uneven society, were bolstered by COVID-19 and self-isolation. While for some lockdown measures exposed the ‘fragile and exploitative character’ of modern capitalism and neoliberal hegemony (Manderson and Veracini 2020), insufficient engagement with tangata whenua—the Indigenous people of the land—drew attention to further racial inequities.

If the severity of the pandemic across the world led most Kiwis to accept what they deemed to be justifiable and temporary, if extreme, government restrictions to the whole population, others felt that the ‘we’ of the lockdown excluded them. Insufficient inclusion of ‘the “other” Treaty partner’ in determining and communicating the public health response to COVID-19 signified for Rhys Jones ‘the broader lack of Māori participation in the pandemic response’ (Jones 2020). Inequitable outcomes for Māori were discussed in five tweets within our corpus. In these tweets, users were, on the whole, supportive of government mandated lockdown and health measures, but called for further Māori inclusion in government decisions or for tino rangatiratanga (sovereignty)—a self-determined, Māori-driven response. One user critiqued the lack of Māori inclusion in the health response with specific reference to the language used in government communications, expressing disapproval of the use of non-Indigenous term lockdown to describe stay-at-home measures rather than rāhui (figure 2; Twitter user 16 2020). Fears that Māori, who historically suffered disproportionately in previous pandemics, would again suffer from ‘racist outcomes’ (Twitter user 18 2020) were explicitly stated in three tweets, with the last tweet discussing how to support additionally vulnerable ‘immunocompromised whānau (families)’ (Twitter user 19 2020). For these Twitter users, the government COVID-19 response highlighted ‘perceived ruptures in the social contract’, leading them to ‘express colonial trauma and critique both the non-democratic distribution of science’s benefits, and the state’s role in supporting this’ (Sobo and Drazkiewicz 2021, 70). That one tweet specifically criticised the government’s use of the term lockdown, with its strong prison associations, rather than rāhui, takes on additional meaning when considering the overincarceration of Māori within Aotearoa New Zealand—as of March 2021, Māori made up 52.7% of the country’s prison population (New Zealand Government 2021) despite only accounting for 16.5% of the overall population (New Zealand Government 2020). It is again important to note that no tweets using rāhui or discussing inequitable outcomes for Māori in this corpus expressed complete opposition to government measures. Although it could seem counterintuitive, tweets containing rāhui were overwhelmingly positive and supportive of government measures—more so than any other term. This positivity may have represented attempts to reframe the lockdown in a positive light, more compatible with Māori cultural practises, so as to encourage whānau adherence to protective measures and fight for equitable outcomes for Māori while subtly acknowledging insufficient action by the government. While these five lockdown tweets represent a small percentage of the overall corpus, the concerns they express are again replicated in a further 21 tweets in the broader corpus and offer valuable insight into wider perspectives on lockdown.

Figure 2

Example tweets (sources: data from anonymised Twitter users 11, 15, 16, 17, 6; image used by Twitter user 11 from Frost and Lynch (1990).

Even as COVID-19, the virus ‘which does not discriminate’ (Hopgood 2020), posed a threat to individuals across society, it simultaneously highlighted already-present inequalities across both Aotearoa New Zealand and the world. The language use within the plague city, and the uneven ways in which biopower is exerted, is manifest in the connotations, as well as the ramifications, of lockdown. Regardless of whether negative engagement with self-isolation was due to fears of an insufficient or overzealous government response, the tension between government biopower and individual freedoms were often central to concerns. Despite the majority of tweets representing the country’s compliance with self-isolation rules, for those who wished to position stay-at-home measures as authoritarian or otherwise unjust, lockdown’s connotations of imprisonment and control made it an ideal term. As we will see below, bubble was perceived as a softer, more fragile measure and while it is also a non-Indigenous term popularised by governmental use, it did not have the same problematic connotations as lockdown and did not generate similar concerns within our data.

“Inside our bubble with our bubble mates” (Twitter user 17 2020a)

Within a week or two of entering lockdown in 2020, mentions of bubbles became common in houses around the country. The bubble metaphor, first used in response to COVID-19 by Prime Minister Ardern on 24 March 2020, was vital in encouraging its widespread adoption (Trnka and Davies 2020, 169). Originally coined by Tristram Ingham while considering the pandemic response for the disability sector, the term was soon being used in COVID-19 press conferences, households across Aotearoa New Zealand and around the world (Herald 2020). As stated by Prime Minister Ardern, ‘Bubble just felt like a natural way to describe something that really needed to be contained’ (Alexander 2021). That bubble became a prominent, culturally powerful metaphor is unsurprising, given both its accessibility and its modelling by the Prime Minister and others in positions of power.

If lockdown evokes Foucault and Agamben, bubble calls to Sloterdijk's work on air, atmosphere and spheres (Sloterdijk 2011). For Sloterdijk, biopolitics operates in a space of spheres: we construct atmospheric envelopes around ourselves, but our separation within these bubbles is never complete. For Sloterdijk, before there is being-in-the-world there is being-with: our bubbles are always intersubjective. We are both separate and connected, and we dwell in cities made of the foam of multiple, loosely touching bubbles. This is particularly clear in the time of an airborne pandemic virus: our bubbles are never complete spaces of isolation as our interdependency always permeates our attempts at environmental enclosure.

There is, also, of course, a nostalgic whimsy to bubbles: they bring back memories of bubble bath beards and the ever-coveted bubble wand. Childhood bubbles went hand in hand with soap and all spaces of safety and cleanliness; perhaps even bringing to mind the ‘bubble boy’ popularised by the light-hearted film (IMDb 2021). Their more recent use at weddings to replace polluting confetti or bird-endangering rice links them to adult celebrations of love and connection. Transparent and glistening, bubbles are fragile, self-contained spaces, both within and separate from the world they reflect back in their iridescent sheen. Above all, bubbles are temporary—thin, momentary barriers, there one moment and gone in the next.

It is likely that these connotations were in the minds of the government of Aotearoa New Zealand when they began using the term bubble to describe small groups of families, housemates and friends entering isolation together. Bubbles implied a separation that was light—mere child’s play—and most importantly, impermanent. Of course, less positively, the very fragility of bubbles, with their connotations of impermanence, highlights their danger as a form of protection during a pandemic. After all, bubbles are easily popped, again stressing our interdependence on each other—for better or worse. The see-through nature of bubbles can suggest openness and transparency, but this may also suggest increased surveillance by both the government and neighbours. While bubbles may make us think of protection, they may also bring to mind the insulated nature of privileged societal groups who remove themselves to gated communities, or of the loneliness of isolation. We may be reminded of economic and housing bubbles, both of which have particularly negative connotations in the wake of the 2007–2009 financial crisis and Aotearoa New Zealand’s ongoing housing crisis.

Bubbles are also intrinsically connected to air; in some ways bubbles have the ability to make air most visible. It is this invisible air, something we usually take for granted, which is now at the heart of our worries: it is air that carries COVID-19. Our coughs and sneezes spread the virus in invisible droplets so that sharing space and air with others has become more of an anxiety-inducing activity than a social one. It is the air that we attempt to contain by wearing masks as our breath becomes potentially hazardous. It is air which grew cleaner as we stayed home, reducing pollution and the deaths it causes under normal circumstances, but it is the inability to breathe air which leads to the need to ventilate those severely affected by COVID-19. The lack of air has led to an unfathomable number of deaths caused by the virus, and, as we exited the first 2020 lockdown, it was the violent denial of air which led to the death of George Floyd and the subsequent global Black Lives Matter protests.

For the seeming simplicity of a referent like bubble, the potential interpretations of the bubble metaphor are surprisingly varied. Moreover, it became clear that these varied connotations were not lost on those using this term. Our research shows that bubble was largely used in tweets indicating support for current government measures (55.1% for measures, 2.5% opposed to measures, with a further 7% calling for stronger measures and 36% containing unclear stance), and contained positive sentiment more frequently than lockdown (41% positive sentiment, 25% negative sentiment, with 35% unclear). Of the commonly used terms for stay-at-home measures (lockdown, bubble, isolation, quarantine and rāhui), bubble contained the second most positive sentiment and the second most support for government measures, only following the infrequently used rāhui in both categories. Bubble did not seem to carry many of the same negative connotations as lockdown—there were no tweets discussing inequitable outcomes for Māori. Furthermore, where rāhui only occurred once in lockdown tweets in order to criticise the use of lockdown as a term, it occurred alongside bubble three times, all in positive contexts.

However, even as bubbles generated mostly positive discussion of stay-at-home measures, this was tempered by their perceived fragility: as bubbles were entered into by individuals and small groups for protection it then became the responsibility of these individuals to defend and maintain them lest they pop. 44.9% of bubble users described bubbles as soft—for some, this meant that they were perceived as more comfortable; for others, more vulnerable and ineffective.

In contrast to the sense of uniform collective experience often engendered by the term lockdown, on Twitter, the personalised bubbles of individuals were more likely to be the focus of discussion. The perceived ownership of bubbles was most prominently seen in the pronouns used when discussing these measures: where lockdown was largely discussed using articles such as the, rather than pronouns, bubble occurred with pronouns just over 70% of the time. Most frequently, the term was used alongside the possessive pronouns your, our and my (figure 3).

Figure 3

Pronoun usage (figure by author).

This assertion of individual ownership of bubbles implies personal accountability, as well as an element of control—where lockdown was largely engaged with as a hard, uniform object, born of governmental biopower and governing the lives of Kiwis, bubbles could more easily take the shape and tone that their inhabitants desired. This is further supported by figure 4, which indicates that while lockdown was discussed in 44.2% of tweets as something entered into as a collective—most frequently as an entire country—only 20.3% of tweets discussed bubbles in this same way. That is, for most users, while Aotearoa New Zealand was in lockdown, individuals were in their bubbles.

Figure 4

Collective versus not collective (figure by author).

Amid a pandemic and the uncertainty of stay-at-home measures, the responsibility for creating and maintaining home bubbles may have presented an opportunity for some to retain a semblance of control and positivity. By exerting control of the bubble through creative ‘little gestures’ (Twitter user 22 2020), the uncertain, dramatic experience of pandemic and stay-at-home measures are ‘re-construed as an opportunity of change […] a precious experience [that] may promote the possibility of elaborating a sense of co-responsibility that sees us as protagonists, together with other people, in the reconstruction of a way of living’ (Cipolletta and Ortu 2021, 288).

As COVID-19 spread throughout the world, closely followed by various stay-at-home measures, so too did increasing levels of anxiety and depression (Nkire et al. 2021, 2). However, as found by Sabrina Cipolletta and Maria Cristina Ortu, those who responded to the uncertainty of the pandemic by reconstruing the experience as ‘an opportunity of change’ tended to engage with the situation in more positive ways (2021, 288). Creative and flexible approaches to the pandemic created the ‘opportunity to reconstruct a limited and frustrating reality in a way that allow[ed] the person to feel freer’ (Cipolletta and Ortu 2021, 288). Where this approach was rarely seen in tweets containing lockdown, occurring only 10% of the time, it was present in 25% of tweets which discussed bubble. As we will see below, the use of language play or humour and the framing of stay-at-home measures as an opportunity for change are both far more common in tweets discussing bubble.

The light-hearted associations of the term bubble have clearly affected the ways in which it was used by members of the public; 55.1% of bubble tweets were in support of government measures, with 41% of bubble tweets containing positive sentiment, while only 2.5%–3 tweets–expressed opposition to stay-at-home measures. In all of these tweets, the focus was still domestic—one asked why a bubble could go for a walk together but not bury a loved one; one sought information on the inclusion of a relative within their bubble and one expressed concerns for victims of domestic violence becoming trapped within their bubbles. As such, for the majority of users, while governmental biopower and pandemics seemed inextricable in a nation in lockdown, biopower seems to have no place in tweets about bubbles. Yet, the near absence of critical tweets using bubble should make us consider the value of the term to a government wanting—needing—to make a nation agree to radical measures. The affective range of bubble encouraged Twitter users to use the term in creative and comedic ways, subtly guiding other users into framing stay-at-home measures in a positive manner. When viewed in this way, the term bubble may be considered an effective form of government influence over public perception: the very name of this stay-at-home measure interrupts potential resistance. While the connotations of the term lockdown make this an ideal measure for protest, as noted in a report by Alexander Mitchell for Newshub, if you “try to say bubble angrily, you can’t do it” (2021). Biopower can work most effectively for governments when it is willingly adopted or barely perceived.

Language play and humour in your bubble during lockdown

The time of COVID-19 certainly was, and still is, a time unsettled by crises of disempowerment, mortality, authority and truth within Aotearoa New Zealand, as it was for much of the world. As anxieties about the pandemic grew, they set the scene for a rise in carnivalesque atmosphere—perhaps especially on social media, a space described by Ana Deumert as already encouraging the carnivalesque and increasingly relied on to maintain connection while physically distanced (2014, 23). The carnivalesque, initially theorised as a literary mode by Mikhail Bakhtin, names a response to crisis or uncertainty that takes the form of the reversal or subversion of power structures through humour, satire and the grotesque. The carnivalesque spirit of the internet can enable ‘loss of self and pure enjoyment’ (Deumert 2014, 34) and a playful transgression of the usual social order through revelry in ‘free-time and free-space’ (Lachmann, Eshelman, and Davis 1988, 132). It is this carnivalesque atmosphere which lends itself to the banter, ‘shitposting’ (Twitter user 11 2020)—posts solely intended to be disruptive through humour, absurdity or offensive content—and, especially, language play which are prevalent on the internet.

An excellent example of a carnivalesque post resounding with dark absurdity is shown in the tweets collated in figure 2. Here, an image of Sherilyn Fenn who, as Twin Peaks’ Audrey Horne is receiving a shocking phone call, is tagged with the lines “It’s Kurt Cobain from 1991. He says we’re stupid and contagious” (Twitter user 11 2020). The Nirvana frontman, whose death sent shockwaves through the mid-1990s, calls from the past to warn us, through a line from ‘Smells Like Teen Spirit’, of our move from the metaphorical contagion of chaotic teenage angst to the literal contagion of a pandemic. Yet, he does so by calling Twin Peaks, one of the greatest but also most uncanny and surreal television shows ever made, and which aired in 1990, before ‘Teen Spirit’ was released in 1991. Cobain thus calls the past to speak across the decades to our present. Cobain’s suicide, the murder on Twin Peaks, and the deaths of the pandemic are combined with temporal, tonal and genre disturbances into a carnivalesque tweet whose macabre humour comments on the disruptive, often grotesque nature of the pandemic.

The carnivalesque space of the internet enables users to engage in ‘ludic self-construction’, relating to themselves ‘and others in a playful manner’ (Deumert 2014, 23). Ludic identities, as defined by Jos de Mul, are those identities constructed based on multilinear plots (rather than clear, singular narratives), lack of closure, interactivity—and, frequently, playfulness and humour (Deumert 2014, 25). On Twitter, users share 280-character snapshots of themselves as they wish to be perceived. They can then build on these tweets and, indeed, their internet personas, through engagement with others via hashtags, retweets and comments: all of these features combine to make Twitter an ideal space for the construction of the ludic self. Interestingly, the willingness to approach the uncertain state of liminality caused by the pandemic with the flexibility, creativity and humour associated with the performance of ludic identities is associated with positive mental health outcomes (Cipolletta and Ortu 2021, 288). While both lockdown and bubble tweets used the carnivalesque space of the internet to share humour, this was done to different extents and with further variation in the types of humour associated with each term. Humour was present in 39% of tweets containing bubble, and many of these used ‘language play and the creative use of linguistic resources’ (Trye et al. 2020, 15) to create new words and lexical blends (30.7% of humorous bubble tweets) or to play on the polysemous nature of the term bubble (20.5% of humorous bubble tweets). Given the nature of Twitter as a multimodal platform, many users added additional flourishes of images, GIFs and emojis to add to their word play.

In collapsing the barrier between the public and private by using humour and engaging the ambient audience generated by hashtags, users are able to ‘create community from individual, commonplace experience[s]’ of stay-at-home measures (Zappavigna 2021, 167). The humour of bubble tweets served the discourse function of ‘a softening device […] aimed at defusing tension in a delicate and socially-charged situation’ (Trye et al. 2020, 15). This approach was commonly seen in bubble tweets containing language play, which were overwhelmingly positive in tone and supportive of stay-at-home measures, while simultaneously acknowledging the restrictions of those measures. As shown in figure 5, of 46 bubble tweets containing humour or language play, 72% contained clearly positive sentiment and 85% were supportive of government measures. Of greater interest, none of the humorous bubble tweets indicated opposition to stay-at-home measures or a desire for stronger measures. While the confining nature of the bubble was ever-present, these tweets tended to offer creative strategies to make the best of the situation and ‘lift your spirits’ (Twitter user 12 2020). Embracing the liminal potential of the online carnivalesque through the use of humour and language play seems to correlate with the search for opportunities of change amid stay-at-home measures, support for government measures and a positive outlook amid the crisis of COVID-19. However, while bubble tweets may have used carnivalesque aspects of social media to share light, playful humour, humorous lockdown tweets were more likely to engage with darker forms of humour closer to the traditional carnivalesque.

Figure 5

Bubble and lockdown humour (figure by author).

Lockdown, which for our Twitter users appeared to be defined by its hard associations with imprisonment and control, produced far fewer tweets containing humour or language play than tweets containing bubble: 12% for the former compared with 39% of the latter. In addition, even humorous lockdown tweets tended to portray lockdown as a hard object; something challenging which would need to be overcome (50% of humorous lockdown tweets portray lockdown as hard as opposed to 8% of bubble tweets). At times, these tweets actively drew on or exaggerated the hard associations of lockdown as well as the distressing aspects of life during lockdown, with users joking about supply shortages, societal collapse and being ‘#LockedUp’ (Twitter user 13 2020). Where bubbles offered opportunities for creativity and individual control during a time of liminal instability (25% of bubble tweets discussed opportunity vs 10% of lockdown tweets), lockdown seemed to be the term that represented loss of individual control and the cause of much of that instability. In comparison with the largely positive tone of bubble tweets, many tweets containing lockdown, even where humour was used, were angry or fearful. This was especially prominent in tweets that were opposed to government measures or calling for stronger measures.

Conclusion: ‘there’s lots of anxiety’ (Twitter user 14 2020)

The March 2020 lockdown required Kiwis to come to terms with stay-at-home measures. Despite the anxiety surrounding both COVID-19 and self-isolation, individuals’ language choices helped them to frame their experiences in different ways. The lightness and whimsy of bubble encouraged many Twitter users to describe their domestic spaces as soft, positive places of play, humour and opportunity—indeed, as indicated by the lack of opposition to stay-at-home measures in these tweets, it may have proven too difficult for disgruntled Kiwis to express anger at bubbles. In contrast, the perceived hardness of lockdown often generated more negative reactions: while the majority of tweets contained neutral or ambiguous stance, the presence of critical and conspiracy tweets suggests that it was far easier to be antilockdown than antibubble. Where bubbles were perceived as the possessions of individuals, and bubble tweets often shared humour, personal experiences and suggestions for how to enjoy the opportunities presented by stay-at-home measures, lockdown was most often the term used for citizen reportage or reiterated announcements that would affect the community. Perceived as a hard measure that entire communities either entered or were put into, lockdown tended to have a more negative valence.

As we continue in 2022 to enter uncharted territory with the COVID-19 pandemic, our words matter. Our words help to situate us, to stabilise or destabilise us, to frame our lives and our memories. In considering the terms lockdown and bubble, we are forced to reckon with our own subjectivity; the ways in which we can be manipulated, and the ways in which we can retain a sense of our own agency while in a seeming free-fall. Ironically, the COVID-19 pandemic and its ‘new normal’, largely characterised by absence, have made explicitly visible a number of truths about the prepandemic ‘normal’. In considering the language of this ‘new normal’, we can begin to reconsider how we frame the world around us, and conceive new, beneficial, ways of engaging with the current pandemic and stay-at-home measures in the future.

List of tweets

Twitter user 1 (@anonymised). “Saw the #covid10nz news.” Twitter post, 23 March 2020.

Twitter user 2 (@anonymised). “New Zealand goes into.” Twitter post, 23 March 2020.

Twitter user 3 (@anonymised). “With people, being told.” Twitter post, 26 March 2020.

Twitter user 4 (@anonymised). “This week has been.” Twitter post, 31 March 2020.

Twitter user 5 (@anonymised). “Watch Sweden.” Twitter post, 27 March 2020.

Twitter user 6 (@anonymised). “What if I told you.” Twitter post, 26 March 2020.

Twitter user 7 (@anonymised). “We’re day 5.” Twitter post, 31 March 2020.

Twitter user 8 (@anonymised). “So a day and.” Twitter post, 23 March 2020.

Twitter user 9 (@anonymised). “So, most calling for.” Twitter post, 22 March 2020.

Twitter user 10 (@anonymised). “As we adjust.” Twitter post, 25 March 2020.

Twitter user 11 (@anonymised). “Shitposting: it’s the only.” Twitter post, 16 March 2020.

Twitter user 12 (@anonymised). “From my bubble to yours.” Twitter post, 25 March 2020.

Twitter user 13 (@anonymised). “#LockedUp.” Twitter post, 26 March 2020.

Twitter user 14 (@anonymised). “It’s true there’s lots.” Twitter post, 21 March 2020.

Twitter user 15 (@anonymised). “Hubby is now home.” Twitter post, 25 March 2020.

Twitter user 16 (@anonymised). “It’s not a lockdown.” Twitter post, 27 March 2020.

Twitter user 17 (@anonymised). “As we adjust to being at home.” Twitter post, 25 March 2020a.

Twitter user 17 (@anonymised). “#bubblerule Celebrate every day.” Twitter post, 26 March 2020b.

Twitter user 18 (@anonymised). “Many of us have.” Twitter post, 18 March 2020.

Twitter user 19 (@anonymised). “Some great stories this morning.” Twitter post, 29 March 2020.

Twitter user 20 (@anonymised). “I am fascinated by metaphors.” Twitter post, 26 March 2020.

Twitter user 21 (@anonymised). “Surprisingly fitting conclusions from.” Twitter post, 31 March 2020a.

Twitter user 21 (@anonymised). “We are thrown into a world.” Twitter post, 31 March 2020b.

Twitter user 22 (@anonymised). “Wonderfully weird deja vu.” Twitter post, 29 March 2020.

Data availability statement

Data are available on reasonable request. The data used in this paper have come from the CovidNZT Twitter Corpus and can be made available on reasonable request.

Ethics statements

Patient consent for publication

Ethics approval

Not applicable.


We thank David Trye for extracting and cleaning the Twitter data used in this paper and Andreea S Calude for her valuable insights.


1. Asterisks signal that zero or more characters can follow the term (eg, lockdowns).

2. We initially coded an additional 276 tweets including these terms: 29 containing rāhui, 140 containing isolation and 106 containing quarantine. While these provide helpful context when considering the stance, sentiment and popularity of the terms lockdown and bubble, the differences between the uses of lockdown and bubble have ultimately proven the most interesting for the purpose of this study. While isolation, quarantine and rāhui are briefly referred to in places, they are not the focus of this article.



  • Twitter @JessiePBurnette

  • Contributors JB and ML made substantial contributions to the planning and drafting of this work; JB developed and performed the data coding; both authors discussed the results, gave final approval for this version to be published and agreed to assume responsibility for the overall content.

  • Funding This work is partially supported by funding from the Marsden Fund of the Royal Society of New Zealand.

  • Competing interests None declared.

  • Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.

  • Provenance and peer review Not commissioned; externally peer reviewed.