Is nothing like before? COVID-19–evoked changes to tourism destination social media communication

Abstract

The outbreak of COVID-19 has boosted the significance of social media for tourism destinations. Therefore, this study investigates how and to what extent the pandemic has changed tourism destination communication on social media and consumers' social media engagement. Using data collected from 1136 Facebook posts by 85 Austrian tourism destinations, the authors compared three different phases of the COVID-19 pandemic (i.e. pre-COVID, lockdown, post-lockdown). Results show that 1) the use of linguistic text features of social media content has changed (i.e. more expressions of uncertainty, confidence, emotionality, more first-person storytellers, greater text length, and less specificity); 2) consumers’ social media engagement, measured as likes, shares, and comments, has increased; and 3) textual features explain changes in social media engagement rates only to a limited degree. Based on these findings, the study provides recommendations for destinations on how to design effective social media content during crises.

1. Introduction

The global COVID-19 pandemic initiated a disruptive change to the entire tourism system, since lockdowns and travel restrictions have widely brought tourism to a standstill (Kaczmarek et al., 2021; Rasoolimanesh et al., 2021; Wong et al., 2021; Zenker et al., 2021). The crisis also resulted in a fundamental change of tourism destination communication, such that marketers have turned even more intensely to social media communication (Čorak et al., 2020; Moorman & McCarthy, 2021). When subject to social distancing, closed borders, and isolation mandates, they began relying heavily on social networks (Frith, 2021); at the same time, COVID-19 enhanced consumers’ use of digital media to replace other communication platforms (e.g. exhibitions, travel agencies) restricted by the pandemic. Beyond communication effects, COVID-19 has also decimated sales of destination products and services, propelling marketers to increase their focus on brand communication as a way to stay in touch with customers (Platon, 2020). Furthermore, COVID-19 lockdowns and economic losses reduce destination marketing budgets, so marketers seek more efficient, affordable tools, such as social media (Itani & Hollebeek, 2021; Moorman & McCarthy, 2021; Sparks et al., 2013; Wong et al., 2020; Xiang & Gretzel, 2010). In such exceptional situations, digital and social media represent stable platforms for engaging consumers and opportunities to address all of these developments.

Current studies (e.g. Wut et al., 2021; Zenker & Kock, 2020) have already established the need for research into destination communication and the effects of COVID-19, especially its intense transition into digital and social media spheres (Čorak et al., 2020; Moorman & McCarthy, 2021). By offering time and location independence, actuality, and wide reach, social media are considered robust marketing communication tools, particularly during crises (Ham & Kim, 2019; Möller et al., 2018; Wut et al., 2021; Yang et al., 2021). The current COVID-19 pandemic demonstrates both potential and risks in social media communication (Kim & Kim, 2020; Mayer et al., 2021; Zhai et al., 2020; Zhou et al., 2021). Even in the unprecedented situation created by the virus, social media support ongoing dialogues between consumers and brands, which enhances brands’ reach and benefits their (future) economic performance (Kim & Kim, 2020; Nezakati et al., 2015; Xiang and Gretzel, 2010). However, social media are also used to disseminate misinformation (Zhou et al., 2021) or to attribute a crisis cause to specific destinations (Zhai et al., 2020). In spring 2020, the Austrian tourism destination Ischgl was for instance presented in the media as a “super-spreader” destination (Mayer et al., 2021).

While the importance of social media as part of the destination marketing mix is undisputed (Ryden et al., 2019), research lacks studies regarding the role of social media text content. For destination marketers it is essential to know what social media content is perceived as rich and thus drives consumer engagement. Recently, researchers established that times of crisis call for marketing communication conveying ambiguity, trust, resilience, but also a new hope (Ertimur & Coskuner-Balli, 2021; Mangiò et al., 2021; Robinson & Veresiu, 2021). However, studies to date have remained mute on how social media marketers can address the uncertainty of a situation while simultaneously conveying confidence and providing relevant information to consumers (Itani & Hollebeek, 2021; Moorman & McCarthy, 2021). The resulting complexity associated with creating and designing effective social media communication for tourism destinations during crises suggests the need for dedicated research efforts.

To address this research gap, the present study follows the call of current research (e.g. Ketter & Avraham, 2021; Mayer et al., 2021) and establishes how the COVID-19 pandemic has shaped social media content posted by tourism destinations. While several studies have compared how types of content (e.g. text, images) impact consumer response (e.g. Guidry et al., 2020; Li & Xie, 2020; Yi et al., 2019), research lacks linguistic analyses of social media posts (Lee & Kahle, 2016). Thus, we specifically investigate COVID-19-induced changes in text content of destination social media communication and analyze whether and how consumers respond to these changes, in terms of their engagement rates. Two main research questions guide the investigations:

• 1.How has the COVID-19 pandemic influenced the text features in destination social media content (i.e. uncertainty, confidence, emotionality, specificity, storyteller, and text length)?
• 2.How has the COVID-19 pandemic influenced consumer responses to and engagement with destination social media content (i.e. likes, comments, and shares)?

The present research thus investigates crisis-induced changes in linguistic features of social media posts and consumers’ responses to these changes. Such an approach expands theoretical knowledge of destination communication during crises and fosters a greater understanding of social media content richness and social media engagement (Berger & Milkman, 2012; Eggers et al., 2017; Leung et al., 2019). Consequently, the present study further contributes to destination brand communication by guiding marketers in the design of online text content, during and beyond crises like COVID-19.

2. Literature review and hypotheses

2.1. Destination social media communication

Social media communication (SMC) has “emerged as a game changer” (Ryden et al., 2019, p. 108) for tourism destinations. Social media enable synchronous and asynchronous multimedia communication, immediate responses and reactions (e.g. comments, likes), high consumer controllability and personalized communication, and are thus considered rich media (Carlson & Zmud, 1999; Dennis & Kinney, 1998; Klein, 2003). Hence in recent years, a vast body of literature identified the benefits of SMC such as creating and restoring destination image (Moya & Jain, 2013), communicating destinations' brand personality via social media posts (Lalicic et al., 2020), sharing information and promoting destination products and services (Buhalis & Sinatra, 2019; Demmers et al., 2020; Kim & Kim, 2020), building consumer-brand relationships (Jabreel et al., 2017), and collecting and analyzing user-generated content (Marine-Roiga & Clavé, 2015). Önder et al. (2020) show that big data from Facebook (i.e. Facebook likes) can even be used as an indicator of tourism demand. Prior studies have further investigated the impact of social media on consumers’ information search and travel behavior (e.g. Lim et al., 2012; Xiang & Gretzel, 2010), compared the effectiveness of firm-generated and user-generated content (e.g. Kaosiri et al., 2019; Kumar et al., 2016), categorized social media messages (e.g. Kwok & Yu, 2013; Leung et al., 2017), and identified factors that foster its persuasiveness, such as credibility and identification (e.g. Atwood & Morosan, 2015; Leung & Jiang, 2018).

Yet alongside these benefits, social media also inherit potential drawbacks. Brand communication is no longer confined to destination marketers as social media have empowered consumers to emerge as active brand co-creators (Lim et al., 2012; Oliveira & Panyik, 2015; Sparks et al., 2013). Tourists increasingly share negative travel experiences and dissatisfaction with destinations via online networks, which can result in so-called “social media storms” (Ryden et al., 2019). Another challenge is the consistency of destination-generated and user-generated content. Scholl-Grissemann et al. (2020) find that users' opinion of a destination becomes less positive, if the content provided by a tourism destination differs substantially from the content provided by other users, because edited or unrealistic posts cause unrealistic travel expectations. Cao et al. (2020) further caution that redundant, low-value information negatively impacts consumer response, and Önder et al. (2020) note that information overload can result in information anxiety or avoidance. Additionally, deficits in personalizing posts to target specific groups and adapting content to current circumstances can lead to consumer annoyance as well as information disutility, and ultimately to ‘unfollowing’ a social media account (Wang et al., 2020).

Since this research aims to understand how (crisis-induced change in) social media content impacts consumer response, Daft and Lengel's (1986, 1987) theory of media richness serves as an overarching theoretical framework. The theory proposes that communication varies in its ability to transmit information, which in turn shapes communication effectiveness and persuasiveness. Previous marketing research predicts that both channels (e.g. social media) and content (e.g. visuals, text) impact communication richness (Ishii et al., 2019; Moffett et al., 2021). Importantly, current research (e.g. Moffett et al., 2021) highlights that media richness can be particularly useful if communication involves ambiguous messages and uncertain situations, like during the COVID-19 pandemic.

In a social media context, Lee et al. (2021) note that communication richness is relevant in that it drives consumer engagement. This is in line with Klein (2003) stating that content richness positively affects response behavior and the evaluation of online platforms. Cao et al. (2021) find that social media richness determines consumer engagement behaviors, that is consumption, contribution, and creation of social media content. To date, several studies have examined the links between social media content, communication richness, and social media engagement. A focus of current studies is the comparison of posts using text and images. Guidry et al. (2020), for example, demonstrate that posts including images compared to posts using only text significantly increase response behavior. Li and Xie (2020) show that high-quality, professional pictures foster social media engagement, and Yi et al. (2019) find that video content drives the number of social media shares. Considerably less is known about linguistic-based features of text content and their impact on consumer response, and even less so in a tourism destination and crisis context. Therefore, the present study contributes to media richness theory by investigating linguistic features of tourism destinations’ social media content in the wake of the COVID-19 crisis.

2.2. Tourism crises and (social) media communication

Tourism destinations are associated with a large number of external impact factors (i.e. politics, economics, environment), and, thus, are vulnerable to change and crises (Backer & Ritchie, 2017; Çakar, 2018; Itani & Hollebeek, 2021; Möller et al., 2018; Ritchie & Jiang, 2019). Tourism crises denote unexpected events initiating a disruptive change to the entire tourism system, its core, assumptions and operations, which in turn results in loss of bookings, arrivals and stays (Mohanty et al., 2020; Pauchant & Mitroff, 1992). Since tourism has regularly experienced crises in the past (e.g. terror attacks, tsunami), research is well established in this field (Ritchie, 2004; Schroeder et al., 2013; Zenker & Kock, 2020). Crises can be classified into human-induced and nature-induced crises, or more specifically into economic, informational, physical, human resource, reputational crises, psychopathic acts, and natural disasters (Mitroff & Anagnos, 2001; Zhai et al., 2020). Previous research further argues that the perception of a crisis and its type is shaped by media discourses (Mayer et al., 2021; Ritchie & Jiang, 2019), since these discourses affect opinions, behavior and agendas (McLennan et al., 2017). Notably, different kinds of tourism crises can also be intertwined. Mayer et al. (2021) find that media discourses about COVID-19 triggered discourses about more sustainable (mountain) tourism in the form of, for example, less artificial snow-making or more respectful relations between hosts and staff.

Major tasks in corporate marketing include not only brand advertising but also crisis communication (Kim, 2013). Fall and Massey (2006) describe crisis communication as the process of actively providing and sharing information in response to stakeholder questions emerging in a crisis. Depending on the stage of a crisis, these communication efforts are further categorized into crisis preparation, crisis response, and crisis recovery (Ketter & Avraham, 2021). Brand communication during crises has several functions, such as responding to current questions, limiting damage, and inspiring confidence (Ritchie, 2004; Ritchie et al., 2004; Wong et al., 2021), but also expressing compassion and care (Ertimur & Coskuner-Balli, 2021) in order to avoid an ignorant and egoistic brand image (Mangiò et al., 2021).

Crises change not only marketing content, but also functionality of communication platforms. For instance, COVID-19-induced lockdowns and regulations restricted several communication channels (e.g. travel agencies, exhibitions) resulting in increased spending on online brand marketing to remain relevant to consumers (Moorman & McCarthy, 2021; Platon, 2020; Wong et al., 2020). Gretzel et al. (2020) claim that during the COVID-19 crisis, travelers increasingly share vacation memories in digital networks and use destination websites to plan future holidays. Social media function well as a communication tool during crises (Ham & Kim, 2019; Wut et al., 2021), because they are time- and location-independent, can rapidly disseminate messages to a wide community, and address the concerns and needs of pertinent target groups (Hua-Hsin, 2008; Möller et al., 2018; Wut et al., 2021). The key question emerging for destination marketers is how to adapt social media content (e.g. in terms of textual features) to foster positive consumer responses during a tourism crisis.

2.3. Social media text content during crises

Textual social media content constitutes discourses combining and framing linguistic elements, which are interpreted in the context of current circumstances (Keller, 2011). A crisis creates new demand for appropriate content (Platon, 2020), which needs to be reflected in social media texts. Linguistic expressions reflecting uncertainty, confidence, emotionality and informativeness can alter consumer response (Blankenship & Craig, 2007; Escalas, 2004; Tandoc & Lee, 2020), because they convey compassion and care (Ertimur & Coskuner-Balli, 2021). In a similar vein, linguistic features, such as the storyteller (third person vs. first person) or the number of words, can have an impact on consumer attention and response (Ryu et al., 2018; Schreiner et al., 2019; Tulloch, 2014).

This research specifically focuses on crisis-induced changes in social media text content, that is uncertainty, confidence, emotionality, and specificity. Crises cause uncertainty, because familiar operations need to be restructured (Pauchant & Mitroff, 1992). As a health crisis, the COVID-19 pandemic demands new forms of social coexistence, which also result in increased uncertainty (Grace & Tham, 2020; Itani & Hollebeek, 2021; Tandoc & Lee, 2020). Social media posts of tourist destinations should reflect this uncertainty in specific linguistic expressions. Crisis communication research recommends actively addressing uncertainties (Grace & Tham, 2020), though the unique features of crises make it difficult to forecast and publish reliable information. To counteract uncertainty, brand communication during a crisis needs to express confidence, for instance by addressing crisis-related topics such as solidarity, health, and safety (Platon, 2020). Confidence implies a state of security, in which feelings and emotions are matched by actions (Rashid et al., 2020). Blankenship and Craig (2007) explain that confidence can be expressed and generated by linguistic features. Rastegar et al. (2021) highlight the impact of trust on prospective tourists' intention to travel to a destination with high COVID-19 fatality rates. Specifically, the authors claim that trust and safety play key roles in destination management organizations' crisis communication. Therefore, linguistic expressions of confidence are expected to increase in SMC during the COVID-19 pandemic. By initiating fear of the disease, social changes, and uncertainty about the future, the COVID-19 pandemic has further heightened collective emotionality (Moran & Green, 2020; Wen-Ying & Budenz, 2020; Wong et al., 2020). In addition to negative outcomes, the pandemic can evoke positive emotionality, such as optimism, cohesion, and confidence (Lu et al., 2020). Emotions affect consumers’ attention, recall, and involvement (Lu et al., 2020), and thus are relevant to social media posts. In a marketing context, emotions can be created by the emotional tonality of the language, which in turn depends on textual features (Escalas, 2004). It can be expected that heightened collective emotionality is reflected in a greater use of emotional expressions during the COVID-19 pandemic. Apart from linguistic expressions of uncertainty, confidence and emotionality, text content varies in its expression of informativeness. Since information gathering is a primary motive for using social media, informativeness is a relevant content feature (Aydin, 2020; Demmers et al., 2020; Kanuri et al., 2018). The level of informativeness in turn depends on content specificity. According to Peters et al. (2013), specificity makes content more useful and drives response behavior. Specific content based on available resources as well as tangible and intangible characteristics of a tourism destination (e.g. traditional events, geographical attractions, customs, dishes) also contributes to brand differentiation and facilitates authenticity (Chiu et al., 2012; Mei et al., 2020). However, the unstable circumstances created by COVID-19 prevent the publication of destination-specific content, such that specificity decreases in SMC during a crisis. Summing up, the present study proposes four hypotheses:

H1a

Destination social media posts show higher levels of uncertainty during COVID-19 than before COVID-19.

H1b

Destination social media posts show higher levels of confidence during COVID-19 than before COVID-19.

H1c

Destination social media posts show higher levels of emotionality during COVID-19 than before COVID-19.

H1d

Destination social media posts show lower levels of specificity during COVID-19 than before COVID-19.

In addition, the storyteller format reflects the textual design of a social media post. Marketing communication generally features either third-person or first-person narrators, who might include a personification of the brand, customers, or employees (Ryu et al., 2018; Tulloch, 2014). Third-person, external storytellers (e.g. the destination, the region) use third-person pronouns and are not actively involved in the post as character. First-person narrators instead use first-person pronouns (e.g. I, our destination) and have an active, internal role in the story, such that they act simultaneously as storyteller and (main) character (Graesser et al., 1997; Tulloch, 2014). First-person narratives generally convey higher emotionality, intimacy, and trustworthiness (Davis et al., 2004; De Graaf et al., 2012). Although previous research lacks any explicit evidence about whether and to what extent crises might change the use of different storytellers, it can be expected that the increase in collective emotionality during COVID-19 (Moran & Green, 2020; Wen-Ying & Budenz, 2020; Wong et al., 2020) has encouraged the use of first-person voice as a storytelling form. Finally, text content in social media posts varies in word count. Social media are fast-moving, and posts usually combine various media (e.g. text, images, emojis, links), such that marketers tend to use shorter texts (Schreiner et al., 2019). Yet, crises require extended communication efforts to provide information, address uncertainties by responding to stakeholders’ questions, and increase confidence (Palttala & Vos, 2012; Ritchie et al., 2004; Wong et al., 2021). The present study assumes that these changes in content needs are reflected in increased word counts of destination social media posts since the outbreak of COVID-19. Thus, the following hypotheses are proposed:

H1e

Destination social media posts show more first-person storytellers during COVID-19 than before COVID-19.

H1f

Destination social media posts are longer during COVID-19 than before COVID-19.

Social media content can be consumed both actively/reactionary (e.g. likes, comments, shares, creating a review) and passively/non-reactionary (e.g. reading, browsing). Since active social media engagement enhances a brand's reach and can therefore determine economic performance (Kim & Kim, 2020; Nezakati et al., 2015; Sparks et al., 2013), measuring consumer reactions to posted content is a relevant task for destination marketing. Social media engagement can be described as a behavioral response driven by emotional reactions to a post (Zaidi et al., 2020). The level of behavioral engagement can be assessed by the number of likes, comments and shares, as well as engagement rates (Demmers et al., 2020; Lalicic et al., 2020; Leung, 2019; Solem and Pedersen, 2016). These measures might not translate directly into higher sales, booking rates, or spending, but they reflect consumers' willingness to engage with a brand and their attitudes toward the destination content (Demmers et al., 2020). Likes, comments, and shares are all responsive behaviors, but they differ in the level of psychological and behavioral involvement they reflect. Clicking ‘like’ is the least demanding reaction; commenting on and sharing a social media post require more effort and thus signal higher levels of involvement (Solvoll and Larrson, 2020).

With reduced opportunities to spend time together, the COVID-19 pandemic let people search for other ways to stay socially connected, including interacting through social media (Nguyen et al., 2020). According to Gretzel et al. (2020), travelers increasingly share travel memories in social networks. In testing whether and how COVID-19 has affected social media engagement in a tourism destination context, beyond the increased consumption of digital and social media in general and the growing importance of SMC for brand management (Moorman & McCarthy, 2021; Wong et al., 2020), the present study predicts that social media engagement (measured as likes, comments, and shares) has increased since the outbreak of COVID-19:

H2

Consumer engagement rates with destination social media posts, measured as (a) likes, (b) comments, and (c) shares, are higher during COVID-19 than before COVID-19.

3. Empirical research

3.1. Research design

The Alpine destination Austria represents the context of the present study. With 5.6 million arrivals, 79 million overnight stays and a 5.5% contribution to the GDP (Statistics Austria, 2021) Austria is a globally popular tourism destination. In the wake of COVID-19, Austria gained increased media attention around the world because numerous skiing vacationers returning from the Austrian tourism destination Ischgl “reportedly became infected with the virus during their stay and then spread it in their home regions” (Mayer et al., 2021, p. 1).

To test the hypotheses, social media posts by Austrian tourism destinations are compared in three stages: before the COVID-19 pandemic (pre-COVID), during the first lockdown (lockdown) and during the phase after the first lockdown (post-lockdown). Using a social media–derived secondary data set enables us to investigate COVID-19-induced changes in marketing content created and published by tourism destinations. With social media engagement rates (likes, comments, shares) as dependent variables the study further captures how consumer behavior on social media changed as the COVID-19 crisis unfolded.

Specifically, the authors randomly sampled 1136 Facebook posts published in German on the official Facebook accounts of 85 Austrian tourism destinations between December 12, 2016, and November 16, 2020. The sample includes solely organic informational and promotional posts relating to the destinations (e.g. hiking routes, COVID-19 regulations). Next to personalized messages, sponsored content and sweepstakes, organic posts created by destinations are a way to communicate with tourists via social media (Lee et al., 2021) in order to affect their attitudes, intentions, and behavior (Leung et al., 2019; Molina et al., 2020; Platon, 2020). These destination-generated postings usually combine various media elements (e.g. text, images, video, links) and seek to share information (e.g. destination products and services, promotions) (Hautz et al., 2014; Molina et al., 2020; Shen & Bissell, 2013). In most countries the pandemic led to the alternation of two states, lockdown and gradual opening/post-lockdown, alongside specific restrictions and travel options (Ketter & Avraham, 2021). Hence, data was classified into pre-COVID (December 12, 2016, until March 15, 2020), lockdown (March 16, 2020, until April 13, 2020), and post-lockdown phases (April 14, 2020, until November 16, 2020) in line with the classification determined by the Austrian federal government. Regarding the hypotheses, the latter two phases represent the intra-COVID phase. Distinguishing lockdown and post-lockdown states allows us to capture changes that might have occurred in different phases of the pandemic.

For the statistical data analysis, the present study relied on SPSS and LIWC (Linguistic Inquiry Word Count). The publication date of the social media posts (pre-COVID, lockdown, post-lockdown) provides the independent variable, and social media engagement rates (likes, comments, and shares) are the dependent variables, which can capture any changes in consumer response behavior on social media. We divide the likes, comments, and shares for each Facebook post by the number of Facebook followers for that specific destination, similar to the measures used in previous social media studies (e.g. Demmers et al., 2020; Leung, 2019). To assess the impact of COVID-19 on destination social media content, changes in the linguistic expressions (uncertainty, confidence, emotionality, specificity) as well as in textual features (storyteller, text length) were measured. For uncertainty, the percentage of words in a post that represent uncertainty (e.g. maybe, unsure, possibly) was considered. In LIWC, the word category ‘uncertainty’ includes 463 words, with an internal consistency of 0.81 (LIWC, 2020; Meier et al., 2018). For confidence, a LIWC factor variable derived from previous linguistic studies, with scores ranging from 0 to 100, was used. The higher the score, the greater the linguistic expression of confidence (Meier et al., 2018). For emotionality, the study used another factor variable provided by LIWC, on a scale ranging from 0 to 100, such that higher scores indicate more intense textual expressions of emotionality (Pennebaker et al., 2015). Specificity was coded manually, using a seven-point rating scale (1 = not destination specific at all, 7 = highly destination specific), by two coders who searched for destination-specific cues in each social media post (e.g. regional events, destination-specific geographical attractions, destination-specific brand names). For the type of storyteller, the study again used manual coding but integrated the LIWC, by analyzing the content according to the presence of third-versus first-person pronouns. Text length was assessed by a word count of the German version of LIWC using a dictionary of 18,711 words (Meier et al., 2018).

3.2. Results

3.2.1. Effects of COVID-19 on social media text content

Analysis of variance (ANOVA) tests the impact of publication date (n_pre-COVID = 517, n_lockdown = 58, n_{post-lockdown} = 561) on linguistic expressions and textual features of the destinations’ social media posts (Table 1 and Fig. 1 ). The results confirm that uncertainty, confidence, specificity and text length differ significantly across the three phases of pre-COVID, lockdown and post-lockdown (all ps < .05). Contrary to expectations, linguistic expression of emotionality does not differ significantly across all three stages (p = .10), although post-hoc tests show significant differences between pre-COVID and post-lockdown. In greater detail, post-hoc tests (LSD) reveal that: post-lockdown social media posts convey higher levels of uncertainty than pre-COVID posts (p < .01); linguistic expressions of confidence increased during lockdown (p = .02) and post-lockdown (p = .05); emotionality of language increased in the post-lockdown phase (p = .04); the level of specificity in destination social media posts declined during lockdown (p = .03) and post-lockdown (p < .01); word count increased during lockdown (p < .01) and post-lockdown (p < .01). No significant differences emerged between lockdown and post-lockdown for any linguistic expression category (all ps > .10). A chi-square test shows significant differences in the use of first-person storytellers across the three phases of pre-COVID, lockdown and post-lockdown (p < .01). Specifically, z-tests for two independent proportions show significantly greater use of first-person narrators in destination social media posts in the lockdown (p < .01) and post-lockdown phases (p < .01) compared with the pre-COVID phase, and significantly greater use of first-person narrators in the lockdown than in the post-lockdown phase (p < .01). Given these results, hypotheses 1a-f are supported, although in some cases only in the post-lockdown phase.

Table 1.

Effects of COVID-19 on destination social media posts.

	Independent Variable: COVID-19						Sig.
	Pre-Covid		Lockdown		Post-Lockdown
	M	(SD)	M	(SD)	M	(SD)
Social Media Post Textual Features:
Uncertainty	1.52	(2.86)	1.85	(2.41)	2.06	(2.42)	.003
Confidence	81.08	(18.61)	86.74	(14.09)	83.06	(15.01)	.019
Emotionality	72.68	(35.67)	77.39	(31.60)	77.00	(32.82)	.101
Specificity	3.53	(1.90)	2.97	(2.06)	2.88	(1.70)	<.001
Storyteller (First-Person)	43%		71%		50%		<.001
Word Count	34.71	(24.04)	57.28	(49.60)	51.06	(27.73)	<.001
Social Media Engagement Rate (%):
Likes	.55	(.68)	1.28	(1.61)	1.10	(4.13)	.005
Comments	.04	(.08)	.06	(.09)	.06	(.15)	.062
Shares	.05	(.10)	.10	(.13)	.12	(.20)	<.001

Fig. 1.

Changes in linguistic expressions and textual features of destination social media posts.

3.2.2. Effects of COVID-19 on consumer social media engagement

ANOVA also reveals how the COVID-19 pandemic (pre-COVID vs. lockdown vs. post-lockdown) changed consumer response behavior, in terms of likes, comments, and shares of the tourist destinations' social media posts (Table 1 and Fig. 2 ). Findings indicate significant increases in liking (p < .01) and sharing (p < .01) destinations’ Facebook posts, but only marginally significant changes in commenting behavior (p = .06). Compared to pre-COVID, post-hoc tests confirm a significant increase in consumer engagement in terms of liking during post-lockdown (p < .01), in commenting during post-lockdown (p = .02), and in sharing behavior during lockdown (p = .03) and post-lockdown (p < .01). No significant differences emerged between lockdown and post-lockdown for any of the three social media engagement measures (all ps > .10). Hence, the results support hypotheses 2a-c.

Fig. 2.

Changes in social media engagement rates.

Considering that the findings affirm that the COVID-19 pandemic changed both destination social media content and consumer engagement, the statistical analysis also examines whether the specific linguistic expressions and textual features (uncertainty, confidence, emotionality, specificity, storyteller) represent the mechanisms by which social media engagement rates (likes, comments, shares) change, as well as whether these effects depend on the lockdown and post-lockdown states in the wake of COVID-19. Thus, the authors use the PROCESS macro (Hayes, 2013, model 1, bootstrapping procedure, 5000 samples) to estimate how the pandemic (pre-COVID vs. lockdown vs. post-lockdown) moderates the effect of the five linguistic expressions and textual features on likes, comments, and shares of social media posts. Each linguistic expression (uncertainty, confidence, emotionality, specificity) and textual feature (storyteller) represents an independent variable, COVID-19 (pre-COVID vs. lockdown vs. post-lockdown) is included as the moderator, and each social media engagement measure (likes, comments, shares) serves as separate dependent variable. As the moderator is multi-categorical, the authors estimate two models each, one comparing lockdown with pre-COVID, and one comparing post-lockdown with pre-COVID. Furthermore, the analysis controls for the potential influences of other textual features; the word count (Berger & Milkman, 2012); the presence of links, hashtags, and emojis (McShane et al., 2020); and the type of content (image vs. video) (Yi et al., 2019) as covariates.

Results indicate very few interaction effects. The significant interaction effect of language emotionality indicates that language emotionality has a significant negative effect on commenting (B = −0.0007, p < .01) and sharing (B = −0.0007, p < .01) but only during the post-lockdown phase. The significant interaction effect of specificity indicates that destination specificity has a significant negative effect on sharing during lockdown (B = −0.0286, p < .01) and post-lockdown (B = −0.0098, p = .01). Therefore, during the pandemic, emotionality and specificity in social media posts can decrease consumers’ commenting and/or sharing behavior—effects that were not present before COVID-19.

4. Contribution to theory and practice

The present study empirically investigates differences in destinations' social media content and consumer responses across three phases (pre-COVID, lockdown and post-lockdown). Thus, it follows the call of previous studies to extend research on online destination communication during crises (e.g. Wut et al., 2021; Zenker & Kock, 2020), consumer response to advertising content during crises (e.g. Kim, 2013), and individual elements such as richness of text shaping social media posts (e.g. Lee et al., 2021; Wang et al., 2020; Zaidi et al., 2020). Specifically, findings reveal that since the outbreak of COVID-19, the use of linguistic expressions and textual features of social media content has changed (more expressions of uncertainty, confidence, emotionality, more first-person storytellers, greater text length, and less specificity, RQ1), and that consumers’ social media engagement, measured as likes, shares, and comments, has increased (RQ2). In addition, the study finds that only language emotionality and specificity have significant negative effects on social media engagement during the COVID-19 pandemic. The following sections discuss the theoretical contribution and managerial implications of these findings.

4.1. Contribution to theory

Studies on social media related to crises have recently increased (Wut et al., 2021). However, they mainly reflect two research fields: the impact of social media on tourists' risk perceptions, and the use of social media for post-crisis brand recovery (e.g. Horster & Gottschalk, 2012; Schroeder et al., 2013). Mayer et al. (2021) point to a lack of studies into the role of crisis-related media communication to stakeholders, and Ketter and Avraham (2021) note that SMC during a pandemic is still an emerging research field. To extend current knowledge on how SMC is used and how it changes during tourism crises (Möller et al., 2018; Nigel et al., 2020; Ritchie & Jiang, 2019; Schroeder et al., 2018), this study clarifies which changes occur and how tourism crises affect tourists’ responses on social media. Specifically, results contribute to theory in three ways.

First, this research contributes to media richness theory in the context of social media marketing. Extant research widely agrees on the significant impact of communication richness on social media engagement, and previous studies have compared the impact of content and image types on social media richness (e.g. Guidry et al., 2020; Li & Xie, 2020; Yi et al., 2019). Yet, research to date has neglected text content as an integral part of social media posts (Lee & Kahle, 2016). The present study explicitly links social media communication of tourism destinations with linguistic expressions and other textual features of social media posts in the context of a tourism crisis. It illuminates which linguistic expressions and textual features are subject to change during a tourism crisis, and how these changes affect consumer engagement on social media. Specifically, results indicate that content richness in terms of specificity of social media texts declines in a tourism crisis, but this decline has a positive impact on consumer social media engagement. While this is admittedly only a starting point for further investigations, it can be concluded that social media text content significantly contributes to communication richness, which extends the understanding of the role of individual (text) content elements in social media communications (Lee et al., 2021; Lee & Kahle, 2016; Wang et al., 2020; Zaidi et al., 2020).

Second, this study contributes to the body of literature which analyses how crises change the tourism industry. Beyond obvious effects (e.g. declining demand, loss of bookings), the precise changes remain hard to clarify (Bausch et al., 2020). By focusing on the impact of crises, specifically COVID-19, on social media content, this research follows a recommendation to test the impact of crises on destination communication (Zenker & Kock, 2020). Previous studies emphasize the importance of adapting brand content during crises (Mangiò et al., 2021), yet without detailing how these adaptions are actively generated. As long ago as 2004, Ritchie (2004) noted that instilling confidence is a key task of brand communication. Ketter and Avraham (2021) also argue that destinations are increasingly striving to spread hope and inspiration since the outbreak of the COVID-19 pandemic. Ertimur and Coskuner-Balli (2021) suggest focusing on themes of solidarity, trust, and resilience, but none of these authors details how text-based features of marketing communication activities need to be adapted. Our research shows how crisis-related themes such as uncertainty (e.g. “maybe,” “we don't know yet”), confidence (e.g. “we stick together,” “things are looking up”), emotionality (e.g. “we are sad to cancel the event”), and specificity are expressed and how these expressions changed in the wake of the COVID-19 pandemic. Results show, for example, an increase in linguistic expressions of confidence after the COVID-19 outbreak, in line with the general recommendations mentioned above. Therefore, this study can serve as a contribution that links conceptual themes of marketing communication in a tourism crisis to linguistic expressions and other textual features that can be used by destination marketers.

Third, this study contributes to the emerging research field of SMC and consumer social media behavior during crises (Ketter & Avraham, 2021). Previous research predicts that crises tend to increase consumption of social media (Moorman & McCarthy, 2021; Wong et al., 2020); but these studies do not refer to destinations, nor do they confirm whether greater consumption of social media content translates into increased consumer engagement. For example, if consumers use social media mainly to gather information, they might not increase their commenting or sharing behavior. If instead they follow tourism destinations' social media posts to gain inspiration for their future travel plans, they may exhibit enhanced social media engagement rates. As the study confirms empirically, the pandemic has evoked higher numbers of likes, comments on, and shares of tourism destinations' social media posts. This finding supports Nguyen et al. (2020) and Gretzel et al. (2020) who highlight that social media helped people to share memories and to stay socially connected during the crisis. Similarly, other research (e.g. Platon, 2020; Wong et al., 2020) suggests using social media in brand communication to remain relevant to consumers. Interestingly, no changes occurred between lockdown and post-lockdown for any of the three social media engagement measures. Apparently, in the absence of actual travel opportunities or in times of restricted travel options, social media content helps consumers to virtually connect with tourism destinations. In seeking a deeper understanding of SMC during crises, we also identified differences in the impact of linguistic expressions and textual features on consumer response compared with a non-crisis context. The results indicate that during the COVID-19-induced lockdown, the emotionality of social media posts reduces consumers' commenting and sharing behavior, in contrast with findings in non-crisis contexts (e.g. Kanuri et al., 2018; Kim & Youn, 2017; Peters et al., 2013) that highlight the positive effects of emotionality. It can be concluded that the unstable and insecure circumstances of crises call for more rational and less emotional language in brand communication. In addition, during phases of lockdown and post-lockdown, content specificity leads to fewer social media shares, an effect that was not present before the COVID-19 pandemic. Thus, we speculate that announcing and promoting specific products and services does not fit well with limited travel options and the general uncertainty present during crises. Although these findings need to be considered preliminary, the present investigations respond to previous studies pointing to a lack of knowledge on how and why consumers react to marketing communication during crises (e.g. Kim, 2013) and are thus an important building block for an increased understanding of how destination marketers' social media posts drive consumers’ social media engagement.

4.2. Managerial implications

Previous research on marketing communication during crises mainly addresses a strategic level underlining the need for detailed communication strategies (Ritchie, 2004; Ritchie et al., 2004) and providing best-practice-frameworks (Fall & Massey, 2006). The present research, however, delves into more operative levels (e.g. choice of textual features), which is needed to implement and adapt communication strategies. By guiding destination marketers in the textual design of online content, we identify managerial implications which go beyond the current pandemic. Specifically, tourism practitioners learn how text can actively transmit uncertainty, confidence, emotionality and specificity. Additionally, findings suggest avoiding language emotionality and destination-specific content (e.g. events, restaurant opening hours) in unstable times of crises, since they negatively affect consumer engagement. Table 2 provides an overview of specific linguistic expressions and textual features destination marketers can use, text examples, and a summary how these linguistic expressions and text features changed in SMC during the COVID-19 pandemic, and whether and how these changes affected consumers’ social media engagement.

Table 2.

Text design of destination social media posts.

Text Design of Destination Social Media Posts
To convey …	… use expressions such as …		… during crises …
uncertainty	“maybe”, “possibly”, “we might close we don't know yet”		uncertainty-expressions increase during crises
confidence	“we stick together”, “things are looking up”, “together we can do it”, “let's look at each other”		confidence-expressions increase during crises
emotionality	“we are sad to cancel the event”, “we are so happy to open again”, “Unfortunately, today is the last day for the time being to welcome our guests”, “our hotels are closed but we provide you with beautiful and uplifting pictures from our region”		emotionality-expressions increase AND an increase in emotionality-expressions decreases social media comment and share engagement
specificity	“in the Schladming Dachstein region, you can explore 230 km slopes, 88 lifts and 80 winter huts”, “try our Brettljause and Schladminger beer”		specificity-expressions decrease AND a decrease in specificity-expressions increases social media share engagement
To use a …	… use words such as …	… representing …	… during crises …
first-person storytelling format	I, we, in our destination	hoteliers, famers, rangers, locals, guides	first-person storytellers increase

Previous research argues that brand content during crises needs to reflect the uncertainty of the situation while simultaneously conveying confidence (Itani & Hollebeek, 2021; Moorman & McCarthy, 2021; Platon, 2020). From a practical perspective, the present study enhances these findings by showing how uncertainty and confidence can actively be created and expressed by social media content. Results indicate that crises require less emotional communication content. Therefore, destination marketers should refrain from conveying too much emotionality via social media posts (e.g. “We are so sad and very sorry that we have to cancel the festival”). With regard to linguistic expressions of specificity, prior research suggested using specific content to achieve brand differentiation and authenticity (e.g. Mei et al., 2020; Ryu et al., 2018). In times of crises, however, present findings suggest avoiding destination-specific content, because it lowers customer responses in terms of comments and shares. Apparently, specific content (e.g. “Our weekend-package includes 2 overnight stays, breakfast and spa entrance. Visit us, enjoy and relax”) does not make sense for consumers, because it conflicts with their uncertain reality. Furthermore, specific information about tourism destinations (e.g. “In our region 80 idyllic alpine huts wait for your visit”) is less useful when consumers do not know when or how they will be able to travel again.

This study further supports the importance of destinations' social media presence during crises. Compared to pre-COVID, consumers' interactions with destination posts increased in terms of likes (by 133% during lockdown, 100% during post-lockdown), comments (by 50% during lockdown, 50% during post-lockdown) and shares (by 100% during lockdown, 140% during post-lockdown). To determine the practical implications of these increases, we compare the number of likes, comments and shares on the official Facebook account of Gasteinertal, a typical Austrian mountain destination, before COVID-19 and during lockdown. On average each posting published by the destination received 74% more likes (pre-COVID = 356.81, lockdown = 619.25), 170% more comments (pre-COVID = 9.37, lockdown = 25.25) and 77% more shares (pre-COVID = 25.12, lockdown = 44.50). When users contribute to a destination's brand-related online activities in these ways, they transform into brand ambassadors (Hautz et al., 2014; Muntinga et al., 2011); when they actively engage with social media content, they also co-create (tourism) experiences (Buhalis & Sinatra, 2019). Therefore, marketers should pay attention to the present finding regarding increased social media engagement during crises, because it represents an organic form of promotion.

5. Limitations and further research

Due to its disruptive impact, COVID-19 has created many new research fields. With the investigation of its impact on social media content and engagement rates, this study focuses on destination communication and identifies concrete changes in text content and consumer responses. Specifically, results show that crises significantly impact linguistic characteristics and textual features of destination social media posts and consumer engagement with these posts. Testing hypotheses with a social media secondary data set has high external validity as it captures consumers’ reactions to destination social media postings, in terms of likes, comments, and shares, in a real, everyday setting. However, to increase the internal validity of results, future studies could design experiments and manipulate linguistic expressions of uncertainty, confidence, (positive/negative) emotionality, or content specificity, and measure cognitive and affective consumer responses. Such controlled experiments could rule out confounding factors that are always present in a real-world crisis environment where factors other than social media text content may affect consumer engagement. In this vein, further research also needs to address additional impact factors, such as source of content, type of posting and presentation mode. Perceptions of and reactions to social media brand communications further depend on consumer characteristics (e.g. socio-demographics). Therefore, additional studies examining specific target groups in controlled lab studies are needed to gain an in-depth understanding of the phenomenon and the underlying processes.

In field studies, future research should investigate other content-based (e.g. emojis, hashtags, images) and perception-based (e.g. tourists’ mood or need for cognition) contingencies and mediating processes to deepen understanding of engagement processes during crises. In doing so, future studies may want to consider other relevant theories apart from media richness theory used in this study, such as media exposure or source credibility theory. In addition, further research could explicate whether the content changes and related activities are strategically planned or were simply driven by circumstances. Previous research on crisis communication calls for active expressions of confidence (e.g. Ritchie, 2004), and the present study confirms significantly more verbal expressions of confidence in posts published during COVID-19 (lockdown, post-lockdown). Yet it cannot be established conclusively that destination marketers purposefully chose linguistic expressions that signal confidence.

The present research captures crisis-driven changes in destination marketing content and customer social media engagement by comparing pre-COVID, lockdown, and post-lockdown phase. Given the fact that the first lockdown ordered by the Austrian government lasted only about a month (March 16, 2020, until April 13, 2020), the sample for this stage is significantly smaller compared to pre-COVID and post-lockdown (n_pre-COVID = 517, n_lockdown = 58, n_{post-lockdown} = 561). The study's sample further consists solely of social media posts by Austrian tourism destinations, which are predominantly mountain destinations. To understand the implications of this global pandemic, it would be interesting to extend the research to other international destinations. Such approaches could include more diverse destination types (e.g. city or beach destinations), which would increase the generalizability of findings.