Destination Branding by Residents: The Role of Perceived Responsibility in Positive and Negative Word-of-Mouth

ABSTRACT The importance of residents’ communication about their home region as tourist destination is increasingly acknowledged in the place branding process. However, the extent to which residents feel responsible for communicating Destination Images (DIs), and how these attributions affect word-of-mouth (WOM) behavior has remained unclear. This paper addresses this topic by reporting a quantitative study (N = 913) among residents of the Dutch province of Fryslân. The results show that responsibility attributions for communicating positive regional images predict, next to Place Attachment and DI, positive and negative WOM behavior (pWOM and nWOM). Particularly attributing responsibility to citizens themselves is a significant predictor of pWOM and nWOM. Further, findings point to the role of age in variation in responsibility attributions and to previous holiday experience as predictor of nWOM. The implications for citizen engagement in destination branding and regional tourism management are discussed.


Introduction
Residents are increasingly included in regional marketing and place branding (Klijn, Eshuis, & Braun, 2012;Sartori, Mottironi, & Corigliano, 2012). Being acknowledged as important stakeholders (Kavaratzis, 2012), residents' and visitors' communication through informal networks, enhanced by social media, has taken at least part of the power to create Destination Images (DIs) away from destination marketing organizations (DMO's).
Destination marketing tends to construct holistic DIs through umbrella brands, often calling upon a supposedly homogeneous regional identity among residents of a destination (Jeuring, 2016). In this vein, the region as a meaningful, yet contested spatial category (Hurenkamp, Tonkens, & Duyvendak, 2011;Paasi, 2003;Terlouw, 2012), for the construction of Place Attachment (PA) and regional engagement among residents is expected to translate into positive destination word-of-mouth (WOM). However, the ways residents can complement and sustain these images are not automatically lining up with destination branding attempts, as DIs and PA differ between individual people, contexts and geographical levels. Therefore, supporting holistic brands among residents might demand a level of involvement that exceeds individual interests. As such, resident participation in destination branding can be seen as a form of citizenship (Bianchi & Stephenson, 2013;Zenker & Rütter, 2014) with duties and responsibilities at the benefit of the wider community.
Understanding responsibility attributions might be relevant for successfully engaging residents as place branding stakeholders and opinion makers. While PA and DI are known to be affecting various behaviors such as destination choice, loyalty and WOM, little is yet known about responsibility attributions for generating positive regional images among "bottom-up" stakeholders, such as residents. These links need to be explored in order to better understand the factors that affect the conditions for and limitations of citizen participation in destination branding.
Therefore, our paper aims to gain more insights into the extent to which residents are inclined to talk about their home region as tourist destination and who they hold responsible for generating positive regional DIs. Do attachment to their province of residence and perceived images of this province as tourism destination affect such WOM intentions? And do residents attribute responsibilities for sustaining positive regional DIs to themselves, to tourism entrepreneurs or to regional governments?
These questions are particularly relevant when considering that citizens have different, simultaneous relations with the places they inhabit. Not only are they residents, but they can also be tourists (Franklin & Crang, 2001), visiting various places within their region and spending holidays at locations that are geographically proximate to their place of residence (Müller, 2006;Singh & Krakover, 2015). This perspective complicates the traditional binaries of host-guest and local-tourist and demands tourism research that takes into account the various ways people experience and talk about places they find important.
To answer the research questions, a panel survey (N = 913) was employed in the Dutch province of Fryslân. Within the Netherlands, Fryslân is known for its strong regional identity (Betten, 2013;Duijvendak, 2008). This regional identity is rooted in an accumulation of political, cultural and natural dynamics. For example, a major (but declining) part of Frisian residents speak Frisian, the second language of The Netherlands. Fryslân is up to today an important source of self-identification for many people in the province, particularly within a context of the Dutch nation state.
Frisian destination branding and tourism marketing rely for an important part on a narrative of a "Frisian Identity", along which the province is promoted (Jeuring, 2016). Particularly, Jeuring shows how assumptions are made about a homogenizing regional identity that translates into a sense of commitment among residents to the administrative space of Fryslân or to Fryslân as tourism destination. For example, residents are envisioned to be hospitable ambassadors of the province, embodying "Frisianness" in the way they receive visitors or reinforcing the regional tourism brand by positive WOM (pWOM). However, thus far no evidence exists whether this sentiment prevails among residents.
In the following, an overview of relevant literature is given on stakeholder involvement in destination branding, WOM and its antecedents, and citizenship. Next, methods and materials are outlined, followed by the results of the study. Finally, the findings are interpreted, implications for destination marketing practices are given and topics for future research are suggested.

Citizen engagement in destination branding
In recent years, destination management and destination branding are being challenged considerably by changing means of attributing values to products and places. For one thing, this is due to the decentralization of knowledge ownership, blurring the construction and contestation of place meanings (Paasi, 2012;Zimmerbauer & Paasi, 2013). Particularly, DMO's and tourism entrepreneurs aim to engage with their (potential) customers on a personal level by attempting to grapple with the informal circulation (Ateljevic & Doorne, 2004) of destination imaginaries (Salazar, 2012). At the same time, informal networks (e.g. travel blogs, review websites) increasingly are acknowledged as trusted information sources for, and ways of communication about touristic expectations, experiences and evaluations (Carson, 2008;Pan, MacLaurin, & Crotts, 2007). In that vein, engagement of "local" and "bottom-up" stakeholders is sought after in order to account for successful destination development and branding (Eshuis, Klijn, & Braun, 2014;Rehmet & Dinnie, 2013).
Place branding however has been criticized to be a top-down practice, reflecting the interests of a selected group of powerful stakeholders, such as politicians or boards of directors (Eshuis et al., 2014;Hankinson, 2007). These particular interests might not always align with the interests and ideas of other stakeholders and a more inclusive approach to tourism development is called for (Malek & Costa, 2014), which particularly pertains to practices of destination branding (Braun, Kavaratzis, & Zenker, 2013). For example, important benefits of the involvement of citizens and other local stakeholders that are mentioned by Klijn et al. (2012) include a more clear brand concept (or brand identity, i.e. the communicated meanings attributed to a place) and an increase in attraction of specific target groups. Arguably even more important, involving citizens in various planning practices is key in attempts "to build ties with local stakeholders in order to encourage them to become actively involved in changing conditions that affect the quality of their lives" (Malek & Costa, 2014, p. 1).
Involving citizens in destination marketing has only recently become an explicit topic on the research agenda of tourism scholars (Rehmet & Dinnie, 2013). Thus far, research has mostly built on research in the field of organizational behavior and product marketing, for example aligning with the idea of employees and consumers as brand ambassadors (Xiong, King, & Piehler, 2013). Similarly, ambassador networks (Andersson & Ekman, 2009) have been acknowledged as important ways of supporting institutionalized place marketing efforts. Particular benefits of ambassador networks include an effective and trustworthy means of communication, but also the enhanced competitiveness of a place or destination through the mere existence of an ambassador network (Andersson & Ekman, 2009). As such, the perceptions and activities of internal stakeholders, but also communication and collaboration with them should be an essential part of destination marketing (Choo, Park, & Petrick, 2011). Therefore, citizen engagement is a key factor in delivering externally communicated promises. Nonetheless, King states that "internal brand management [ … ] is yet to be thoroughly explored in the context of [tourism and hospitality]" (King, 2010, p. 2).
Building on citizen engagement in place branding implies assumptions on which roles inhabitants play in relation to their place of residence. After all, tourism development and place branding should start with the question "for who's benefit"? As outlined by Braun et al. (2013) for example, city residents have four different yet simultaneously played roles in place marketing. They are the audience receiving messages of place marketing campaigns, but at the same time they are part of the communicated place brand; who they are and what they do is inherently connected to how destinations are experienced by visitors. Third, residents are place ambassadors, who "live the brand" (Aronczyk, 2008) and finally, they play a role as a citizens who provide legitimization to any meaning attributed to public places. This approach points out how on a local level traditional tourism binaries such as host-guest and tourist-resident do not apply anymore in the blurred complexity of a "glocalized" world (Ritzer, 2003).
Grappling with the various roles of local residents in relation to the place they live, work, recreate and receive visitors calls for in-depth knowledge on the motivation of people to engage in various ways of circulating place brands and imaginaries of tourism destinations. An important aspect in this is the way residents talk about the places they live in. The next section therefore digs deeper into the holy grail of bottomup branding, WOM.

Word-of-mouth
Destination marketing, and particularly destination branding activities, aim to construct positive meanings of places, by communicating a selection of physical, emotional and functional attributes of a place (Klijn et al., 2012). Braun et al. (2013) discern between three ways of place communication. Primary communication occurs through the physical features of a place. Secondary communication pertains to official marketing and public relations. Tertiary communication is the way in which residents of a place talk about their place, thus including WOM behavior.
Summing up assumptions of various scholars, "consumers have far more confidence in the views of friends and acquaintances than in a message that emanates from advertising or corporate spokespeople" (Andersson & Ekman, 2009, p. 43). Thus, WOM is acknowledged as an important means of circulating evaluations of intangible tourism offerings, hereby strongly influencing the success of tourism destinations. WOM has been defined in various ways, with most definitions pertaining to independent, face-to-face communication about products, services or companies between consumers (Chen, Dwyer, & Firth, 2014). Motivations to engage in WOM are rooted in needs for self-enhancement and self-affirmation, but links have also been found with social comparison, social bonding or an intention to help others (Alexandrov, Lilly, & Babakus, 2013).
WOM can take on different forms. For example, conventional WOM is face-to-face, but with the rise of the Web 2.0, electronic WOM (eWOM) has been studied too (Chu & Kim, 2011). Further, Chen et al. (2014) describe how WOM varies according to the number of senders and receivers, differentiating between one-to-one, one-to-many and many-tomany WOM. A more content-based typology is given by Naylor and Kleiser (2000), who studied negative WOM (nWOM) and pWOM. Alexandrov et al. (2013) found evidence for different motivations to engage in nWOM and pWOM. Importantly for tourism contexts, this latter distinction can shed light on the amplification of individual experiences into public opinions, for example taking shape through review websites for accommodation, restaurants, events or whole destinations (Carson, 2008;Pan et al., 2007).

WOM and PA
PA plays a central role in understanding people-place relations. Different conceptualizations of PA have been used (Lewicka, 2011). Moreover, the literature seems to lack consensus on a definition for PA and the concept tends to overlap (Lewicka, 2011) with, for example, Sense of Place (Jorgensen & Stedman, 2001) and Place Identity (Rijnks & Strijker, 2013). Similarly, PA has been stated to cover various sub dimensions. For example, Williams and Vaske (2003) employ two dimensions, discerning between affective and functional attachments. Alternatively, other scholars use three or four dimensions (Jorgensen & Stedman, 2001;Kyle, Graefe, & Manning, 2005). Chen et al. (2014) employ six dimensions, which largely overlap with the conceptualization of Jorgensen and Stedman. These are Place Identity (place as a basis for self-determination), Place Dependence (a perceived functional attachment), Social Bonding (social connections within a place), Affective Attachment (an emotional relation between person and place) and two interactional dimensions relating to people's memories about, and expectations of a place.
Various scholars provide evidence for a strong link between PA and WOM behavior (Alexandrov et al., 2013;Chen et al., 2014;Simpson & Siguaw, 2008). Indicators of spatial attachment on geographical levels varying from household to continent have been found to predict WOM intention and behavior. Explanations for this link lie in motivations of people to inform others about their experiences with a place, a sense of pride about a place or a wish to support a place by, for example, encourage other people to visit a place (Choo et al., 2011). So, PA appears to be enacted by social communications about a place. Therefore and in line with previous research, we hypothesize that: H1: Higher scores on PA concepts result in higher pWOM and in lower nWOM intentions.

WOM and DI
Next to PA, DI has been studied as WOM predictor in hospitality and tourism contexts (Agapito, Mendes, & Valle, 2010;Qu, Kim, & Im, 2011). Attitudes toward and perceptions about destinations appear important predictors of destination choice, return intentions but also loyalty toward a destination (Rodríguez Molina, Frías-Jamilena, & Castañeda-García, 2012). Moreover, given the self-enhancement and social positioning functions of WOM, talking positively or negatively about a destination is a way of expressing people's DI and of enacting self-identities. As such, WOM is "place branding in action" (Johansson, 2012, p. 3615), emphasizing the societal importance of tourism destinations in terms of performing everyday life identities and enacting a sense of belonging.
Conceptually, DI has a cognitive (functional) and an affective (emotional) component (Agapito et al., 2010). Additionally, a general evaluation of a destination is conceptualized as the overall image (OI) (Bigné Alcañiz, Sánchez García, & Sanz Blas, 2009). In some studies affective and cognitive attributes are dimensions of an OI, while in others OI is a more holistic appreciation of a destination (Rodríguez Molina et al., 2012), which forms a separate, third dimension of DI (Ahmed, 1991in Rodríguez Molina et al., 2012. Previous research has pointed to a large overlap between concepts of DI and image of large-scale spatial units such as countries and regions (Mossberg & Kleppe, 2005). Similarly, the province of Fryslân as a meaningful region contains numerous social and geographical variations, making it difficult to measure in terms of cognitive image. Moreover, as geographical units become larger or less familiar, their image becomes more holistic (Rijnks & Strijker, 2013) and people base their evaluations on affective associations, since people's image cannot account for all the (functional) intraregional differences. For this reason, regional identification is mainly affective (and less cognitive) (Pan, 2011). Also, in other studies it is argued that while affective DI is conceptually part of an overall perception of a destination, an OI pertains to an evaluation that is greater than the sum of its parts (Bigné Alcañiz et al., 2009) and can therefore add a significant predictive value in explaining people's behavior, such as WOM. So, DI is measured here in holistic terms of OI and Affective Image (AI) and we hypothesize that: H2: Higher scores on Affective Image and on Overall Image predict higher intention of pWOM and lower intention of nWOM.

Citizenship and responsibility
The concept of citizenship originally pertained to notions of belonging and rights within nation states (Bianchi & Stephenson, 2014). This has sparked strong relations with people's identity constructions in terms of the places where they reside (Misener & Mason, 2006). However, current understandings also include relations between individuals and their community, for example in terms of awareness, participation, loyalty and responsibility (Morais & Ogden, 2011).
Responsibility attributions are an important issue in current (Western) societies, as governments are decentralizing their tasks and societies are increasingly regulated by a complex interplay of stakeholders, by other referred to as a "Big Society" (Flinders & Moon, 2011;Kisby, 2010). As such, perceived responsibility of residents has received attention in various contexts. Important results are gained in the field of ecological behavior (Kaiser, Ranney, Hartig, & Bowler, 1999) and risk communication (Jeuring & Becken, 2013;Mulilis & Duval, 1997), particularly with respect to shared responsibilities between various stakeholders. Typically, stakeholders to whom responsibility can be attributed are individual residents, entrepreneurs and governmental authorities (Lalwani & Duval, 2000). Similarly, as sustainable tourism development builds on the involvement of multiple stakeholders, consensus about the attribution of responsibilities for various tasks in this process is essential.
Understanding resident involvement in destination branding might benefit from notions of citizenship and responsibility. As Choo et al. (2011) point out, residents are internal customers, and major local tourism benefits pertain to the ways residents can enjoy their local places through tourism and recreation (Canavan, 2013). Also, ideas of bottom-up brand support prevail for some time now in destination branding practice. Local engagement via tourism is linked with brand identification and in turn has been found to positively affect WOM behavior (Chen et al., 2014).
In this vein, local residents behaving responsibly toward their region implies a hospitable, positive attitude, including a general aim for creating and sustaining positive imaginaries about a region. Yet, Rehmet and Dinnie (2013), in a study on motivations of Berlin residents to participate in the "be Berlin" campaign, found that residents were little triggered by commitment and pride of their city, while expected personal benefits did motivate. So, assuming community feelings among residents as a basis for supporting holistic brands might overlook the importance of responsibility attributions as condition for citizenship behavior in the context of destination branding.
Addressing this issue and linking ideas of regional citizenship and resident engagement with destination branding, the aim is to get an insight in the predictive value of responsibility attributions for sustaining a positive image of Fryslân on WOM among residents of the province: H3: Attributions of responsibility predict nWOM/pWOM intentions, when controlled for PA and DI.
To tie up the theoretical basis for the study, the conceptual model as depicted in Figure 1 shows the concepts included and hypothesized relationships between them.

Methodology
A panel survey was conducted among residents of the province Fryslân. Managed by the Frisian research institute Partoer, the panel contained at the time of surveying (September 2015) 1286 people with an age of 18 or older. The panel is consulted every couple of months on various societal issues, ranging from multilingualism to landscape preferences, from health care to tourism. In total, 913 respondents participated (71% response rate), of which 51% were female and 49% were male. The average age of respondents was just over 54 years (SD = 13.67). The panel should be understood as a convenience sample, without being representative for, for example, the population of the province of Fryslân. As this study's main aim was to test theoretical relationships and testing hypotheses, our concern was less with generalizing to the Frisian population. Therefore, the use of the panel was held to be appropriate for the purpose of this study. Nevertheless, the results must be interpreted with this limitation in mind.
The items and concepts used for this paper were measured as part of a survey containing a larger number of items about intraregional leisure and tourism. Here, only items and concepts relevant for this study are reported (Table 1). Internal consistency (Table 1) was measured using Spearman's rho for two-item scales and Cronbach's alpha for multipleitem scales (Eisinga, Te Grotenhuis, & Pelzer, 2013;Vaske, 2008). Data were analyzed using IBM SPSS Statistics 20.
Negative and positive WOM. pWOM and nWOM (i.e. talking positively and negatively with others about Fryslân as tourism destination) were measured with items adapted from Alexandrov et al. (2013). Both scales performed well on internal consistency.
Place attachment. PA was measured with a shortened 12-item version of the scale developed by Chen (2012). Chen's scale intends to measure six dimensions (i.e. Place Identity, Place Dependence, Social Bonding, Affective Attachment, Interactional Potential and Interactional Past, see also Section 2.3). However, factor analysis on the scores of our respondents on the 12 items (with Varimax rotation) did reveal only two dimensions. The first dimension consisted of five items and reflected perceptions of social belonging attributed to the region of Fryslân. The second dimension was formed by four other items, reflecting a perceived continuity and development of self-identity, facilitated by living in Fryslân. Each subscale had sufficient internal reliability (Table 1). DI. In line with Hosany, Ekinci, and Uysal (2007), one item to measure Overall DI was used (1-10 scale, with higher scores indicating more positive impression): "What is your overall impression of Fryslân as holiday destination?" A semantic differential scale (7point scale) was used to assess Affective DI. The ten items were partly adapted from previous research of Hung and Petrick (2011) and Kastenholz (2010), complemented by selfconstructed items (Table 1). Factor analysis (with Varimax rotation) revealed two dimensions. Three items were deleted due to low or ambiguous factor loadings. The first consisted of five items and pertained to perceptions of being comfortable and feeling "in place" at the destination. The second dimension consisted of two items, but was excluded from the analysis, due to low internal consistency. Thus, AI was measured with a summated five-item scale.
Responsibility attributions. Three one-item scales measured attributions of responsibility for sustaining a positive regional image for Fryslân. The items were constructed based on the approach of Lalwani and Duval (2000), discerning between attributions of responsibilities to citizens, entrepreneurs and regional government.
Next to these concepts, socio-demographics and previous intraregional holiday behavior were measured. This include age, gender and individual income (low, <€31.000; medium €31.000-35.000; high, €35.000<). Previous holiday behavior was measured with a dichotomous variable, by asking if respondents had spent their main yearly holiday in Fryslân in the last five years (yes or no).

Results
Descriptive statistics are presented in Table 1. Overall, respondents rather likely engage in pWOM about Fryslân and are unlikely to give nWOM. Further, respondents score above scale means on the PA dimensions. Particularly, the Self-Continuity dimension appears important in the way respondents feel connected to Fryslân. Similarly, DI scores are high, with an average of 4.46 (SD = .96) on AI and 7.90 on OI (SD = 1.28). Interestingly, responsibility for sustaining a positive regional image was mostly attributed to citizens (M = 3.87) and entrepreneurs (M = 3.75), and less (although still relatively high) to regional government (M = 2.84). Correlation analysis showed that less responsibility attributed to citizens relates to more responsibility attributed to regional government (not in table).
The following two sections further assess responsibility attributions and WOM intentions as bottom-up place branding aspects, by comparing scores between different groups of respondents according to demographics and previous intraregional holiday behavior. Next, the results zoom in on findings about the predictive value of responsibility attributions on pWOM and nWOM, when controlling for PA and DI scores.

Regional image responsibility attributions
In order to get some more insights in variations of responsibility attributions for sustaining a positive image, average scores on responsibility attributions were calculated and compared for age, gender, previous intraregional holiday behavior and income (Table 2). No significant differences in attributing responsibility for sustaining a positive regional image were found between various age groups. Next, when comparing male and female respondents, male respondents attributed more responsibility to entrepreneurs than did their female counterparts, but no differences were found for citizens and regional government. Similarly, residents who had spent at least one main holiday in Fryslân in the last five years attributed more responsibility to citizens than respondents who had not, but these groups did not differ in attributing responsibility to entrepreneurs and regional government. Finally, discerning between various income groups did not reveal significant results.

Word-of-mouth
The same groups were compared on their scores on pWOM and nWOM. While the overall tendency was fairly homogenous, a number of significant differences were found (Table 3). First, it appeared that younger respondents (between 18 and 39 years old) were significantly less likely to engage in pWOM than older respondents (although scores were still relatively high). At the same time, the youngest age group was also most inclined to give nWOM, compared to respondents of 50 years and older. Note however that all respondents were not likely to engage in nWOM, with the highest average score of 1.60 on a one to five scale. A significantly higher intention for pWOM was also found for respondents who previously had spent a holiday in Fryslân. No significant results were found when discerning between male and female respondents, or between various income groups.

Predicting WOM by PA, DI and responsibility attributions
Hierarchical regression analyses were used to test the hypothesized relations. As we discerned between nWOM and pWOM, the models were tested separately for these two

Predictors of pWOM
Variables were added in four blocks (Table 4). In the first step, the two PA dimensions (Social Belonging and Self-Continuity) appear to be significant predictors (F(2, 783) = 128.38, p < .001), explaining 25% of variance in pWOM. Second, Affective DI and Overall DI were added, with a significant increase in explained variance (R 2 change = .16). The third block of variables consisted of the three responsibility attribution variables. Attributing responsibility to citizens appeared to be a significant positive predictor of pWOM, thus indicating that perceiving citizens to be responsible for communicating positive images of Fryslân results in higher intention to indeed engage in pWOM. On the other hand, responsibility attributions to regional government were negatively related to pWOM, signifying that attributing more responsibility to regional governments is related to lower intention among respondents to engage in pWOM themselves. Responsibility attributions to entrepreneurs did not contribute to explaining variance of pWOM. This model explained 44% of variance in pWOM (F(7, 778) = 88.55, p < .001). Finally, the three personal characteristics that showed significant relations with pWOM and nWOM (Table 4, past holiday experience, age and gender) were added to the model. Age was a significant positive predictor of pWOM, indicating that older people are more likely to talk positively about Fryslân as holiday destination, when controlling for the other variables in the model. This is in line with the previously reported findings (Table 4). The full model explained 46% of variance in pWOM (F(10, 775) = 67.76, p < .001).

Predictors of nWOM
In the first step, the two PA variables were added. Both emerged as significant predictors (F (2, 783) = 99.09, p < .001), explaining 20% of variance in nWOM. As to be expected, negative signs for the beta weights were found, indicating that stronger attachment in terms of social belonging and Self-Continuity will less likely result in people talking negatively about their region. Adding Overall DI and Affective DI resulted in a significant increase of explained variance (R 2 change = .07). Again, negative beta weights were found. In this second model, however, the social belonging factor did not contribute anymore as significant predictor, indicating that DI accounts for a major part of the variance explained by social belonging. Adding the three responsibility predictors to the model did not result in a significant increase of explained variance. This model explained 27% of variance of nWOM (F(7, 778) = 42.63, p < .001, Table 5). Similar to the pWOM model, a fourth block with personal characteristic variables (past holiday experience, age and gender) was added. A small but significant increase in explained variance was found (R 2 = .29, F(10, 775) = 32.83, p < .001). Again, age was a significant predictor, however the flipped sign indicates (in line with the previously reported ANOVA's (Table 3)) that younger people are more likely to spread nWOM about Fryslân than older people. Further, previous holiday experience significantly adds to explaining nWOM variance. Interpreting the negative weight of this dummy variable (yes = 0, no = 1) leads to the impression that having spent a holiday in Fryslân in the last five years result in an inclination to engage in nWOM. This somewhat counterintuitive result will be addressed in the Discussion section. Finally, in the full model responsibility attributions to citizens was a significant negative predictor of nWOM, indicating that a perception of citizens to be less responsible for positive regional images of Fryslân results in a stronger inclination to engage in nWOM.

Discussion
The findings of the study give raise to a number of issues to be discussed further. Also, the results provide some interesting insights that can help policy-makers concerned with destination marketing and branding campaigns, particularly with respect to stakeholder collaboration and implementing citizen participation in the planning process of regional tourism development. First, respondents see the responsibility for sustaining a positive image of Fryslân as tourist destination to be shared mostly between themselves and tourism entrepreneurs, while regional governments are attributed less responsibility for this destination marketing task. However, our regression analyses prove that citizen responsibility is the most significant type of responsibility attribution. Respondents who feel that citizens are responsible for communicating DIs are more inclined to pWOM, while attributing responsibility to regional government appears to inhibit pWOM intention. Respondents who think that citizens are not responsible for communicating DIs are more inclined to nWOM. These findings indicate that having a sense of self-responsibility is important for citizens to engage in behavior supportive of a regional "greater good", at least in terms of communicating positive DIs.
Second and somewhat surprisingly, previous holiday experiences in Fryslân result in inclinations to engage in nWOM. As previous holiday experience does not add significantly to pWOM, it may be possible that this finding has to do with the type of holiday experience: negative experiences may have more impact than positive experiences. However, we did not measure this. In other research, satisfaction has been found to mediate between destination loyalty and predictors such as PA and DI (Prayag & Ryan, 2012). So, negative touristic experiences should be prevented as much as possible. Third, another interesting finding pertains to the relation between age and WOM, being positive with pWOM and negative with nWOM. The increased nWOM tendency among younger respondents is an important issue from a tourism management perspective: how to engage younger people in terms of regional attachment, perceived self-responsibility and their role as place ambassadors? The phase in their life course however might limit the options to get younger people locally engaged. The interests of young adults might lie in exploring the world and experience new places (Lepp & Gibson, 2008) instead of bonding with home. Then, nWOM might be more a way to express self-identity than anything else. Indeed, Alexandrov et al. (2013) found that nWOM can be triggered by a need for self-affirmation and social comparison. DMO's should take into account such variations between age groups and find ways to connect with the particular needs and perspectives of young residents in particular in order to be able to get residents across the whole age spectrum involved in destination branding.
Fourth, the focus of this study on responsibility attributions is helpful for DMO's and other tourism management stakeholders in terms of finding ways to connect and engage with tourism entrepreneurs and local residents in place branding processes. Feeling responsible for positive destination WOM appears to be related to an intrinsic sense of belonging (Self-Continuity). This is good news for a region such as Fryslân, which is known to spark a strong sense of regional identity among many of its residents (Duijvendak, 2008;Jeuring, 2016). Such senses of belonging however can play out in different ways. It can lead to people wanting to protect their belongings from interference from the outside (e.g. incoming tourists). On the other hand, it can indeed result in promoting their region to others. Therefore, it is important to acknowledge that supporting holistic brands by residents might demand a level of involvement that exceeds individual interests. The way individual residents make meaning of a region is likely to differ from the ideas of destination marketers or regional authorities (Vainikka, 2012) and affects the way individual perceptions translate into pWOM or nWOM. A challenge thus pertains to get an insight in these dynamics and connect holistic marketing narratives with the perceptions and narratives of residents. This might imply communicating holistic imaginaries that are not just paradisiac and recreate the tourism idyll, but rather ones that reveal the grounded and lived experiences of local residents and their visions on the contemporary and future advantages of spending time in their region, either as tourist or as inhabitant.
Thus, for successful bottom-up support for place branding, DMO's and regional governments should start to take into account principles of citizenship in their destination marketing policies. For example, Eshuis et al. (2014) emphasize the importance of creating dialogues between DMO's, entrepreneurs and residents. In other words, residents need to be facilitated in finding a way to translate their sense of responsibility for their region into actual behavior. This enhances chances of residents getting a sense of ownership of communicated destination brands. At the same time, it allows DMO's to get an insight in the personal needs, interests and narratives of residents.

Conclusion
This paper explored who are responsible for promoting the Dutch province Fryslân as tourist destination, according to the residents of this region. Also, it measured how such responsibility attributions affect, next to PA and DI, the extent to which these residents are engaged in WOM behavior about Fryslân as tourism destination. We hypothesized that PA (Hypothesis 1), DI (Hypothesis 2) and responsibility attributions for sustaining a positive image (Hypothesis 3) influence both positive and nWOM.
We found that residents of Fryslân are likely to engage in pWOM, and are unlikely to speak negatively about Fryslân. PA and DI are strong predictors for pWOM and nWOM, although DI is more important in explaining pWOM, while PA is more important in explaining nWOM. Responsibility attributions at least partly contribute significantly in explaining the variance in pWOM and nWOM.
Getting residents engaged in destination branding is among the key topics of current branding literature and practice. However, the extent to which people feel responsible and to who they attribute responsibility for communicating DIs has remained unclear. This study has started to explore these relationships. The study contributes to knowledge about the antecedents of WOM and the roles of residents in place branding in a number of ways. Particularly, the findings point to a small but significant role of responsibility attributions for communicating DIs in predicting both pWOM and nWOM. Moreover, by including responsibility perceptions for sustaining and communicating positive images of a region as tourism destination, it highlights how stakeholdersin this case residentsattribute responsibilities to themselves and others in the process of destination branding.
Some limitations apply to this study and the findings and implications should be interpreted by taking these into account. The data for this study were obtained through a survey among a convenience sample of residents from the province of Fryslân. The online panel of residents contains relatively few young people. Also, participation in the panel is voluntary, which might pre-select more actively engaged or outspoken respondents. Moreover, the results may reflect local circumstances that are typical for Fryslân as tourism destination and which might not be applicable to other regions (e.g. political situation, climate or broader societal issues). For example, PA scores in this study were high, compared to similar findings in other Dutch regions (Rijnks & Strijker, 2013). Whether this has to do with a possible positive engagement bias among respondents in the panel, or that it reflects a strong regional "Frisian" identity is difficult to say.
Despite these limitations, the results of the study align with other research on the importance of PA and DI for predicting WOM behavior (Alexandrov et al., 2013;Chen et al., 2014;Simpson & Siguaw, 2008). Yet, we note that scholars vary considerably in the way they operationalize these concepts, making it difficult to interpret and compare the results. This study found a two-dimensional solution of PA. Social Belonging, pertaining to a sense of "fitting in", is somewhat similar to what other studies call Social Bonding (Chen et al., 2014), Affective Attachment (Yuksel, Yuksel, & Bilim, 2010) and Environmental Fit (Droseltis & Vignoles, 2010). In turn, Self-Continuity incorporates a notion of Place Identity (Jorgensen & Stedman, 2001) but also has a temporal aspect, reflecting Chen et al.'s (2014) Interactional Past and Interactional Potential. So, the PA concept remains in need for a more robust and unified conceptualization (Lewicka, 2011). Similarly, results point to a conceptual difference between nWOM and pWOM.
While correlated, talking negatively or positively about a destination appears to be triggered in different ways. This is consistent with other studies (Alexandrov et al., 2013) and we recommend other studies on the role of WOM in place branding to not merely focus on WOM as a positive attribute.
Future research could dig deeper into the role of responsibility attributions in relation to place branding and stakeholder collaboration in regional tourism development. For example, pertaining to the scale issues that apply to the relations between responsibility attributions and the ways residents make meaning of their everyday environment both as citizen and as (possible) tourist. For example, it would be interesting to discern between responsibility attributions for promoting places on different spatial levels (e.g. village, city, visited places as tourist). Also, such responsibility attributions could be studied in relation to broader societal dynamics, such as political preference, as a certain attitude toward governmental bodies might affect attributions of responsibility to external stakeholders in various contexts. This is particularly relevant in times of change toward decentralized "Big Societies", where residents become responsible for all sorts of tasks previously covered by governmental authorities (Flinders & Moon, 2011).
Similarly, next to perceiving an important role for residents themselves, residents of Fryslân attribute much responsibility for promoting Fryslân as tourism destination to tourism entrepreneurs. Often, however, tourism entrepreneurs have small businesses and limited capability for branding. While current regional tourism policy in Fryslân explicitly aims to facilitate entrepreneurs with communication channels for tourism promotion, the potential of engaging residents in the promotion of local tourism businesses is thus far only partly explored. The question is whether authorities should play a role in this or not. The tendency to decentralize government responsibilities (Kisby, 2010) complicates the way place branding activities will be shared among stakeholders. Therefore, other studies could take on a similar approach to responsibility perceptions in a place branding context, from the perspective of tourism entrepreneurs, employees of DMO's or policy-makers. Relevant in this respect is to point to the importance of getting an insight in the effect of policy measures and place branding initiatives on responsibility perceptions. Therefore, future research could employ a longitudinal design, measuring attributions of responsibility at several moments in time, for example before, during and after branding and marketing campaigns.
Future research should also look into the role of previous touristic experiences of residents within their own province. The tourism market is competitive, and for many regions receiving tourists from far away will remain an utopia. Therefore, we argue that residents living nearby and within destinations should become a main target group for whom tourism is developed. When residents engage in touristic activities in their region of residence (Díaz Soria & Llurdés Coit, 2013;Jeuring, 2016), this can enhance regional identification and tourism may become an inclusive part of citizenship behavior. Not only in terms of creating positive experience as a basis of outward-oriented branding for incoming tourism (Zenker & Rütter, 2014), but also as an activity that contributes to a thriving, livable region where touristic and everyday life experiences are balanced and reinforcing each other (Canavan, 2013). A better message to communicate to regional tourism management is hard to find.