Keywords

1 Introduction

Tourism is a highly hedonic activity that people like to think about and prepare for in advance, and will later talk about their trip experiences [6]. During the trip planning process, tourists tend to make trip-related decisions beforehand, taking into account the opinions and reviews of the tourists who have already visited the potential destinations. To systematically understand the trip planning processes of individual tourists, Mathieson & Wall [11] and Mansfeld [8] investigate the tourism-related product purchasing, information searching and decision-making. The view of the tourism information search process has been expanded into a broader communication scale [19], in which tourism information search behaviour satisfies the following needs: functional, hedonic, innovative, aesthetic, and sign, and they are realised through the following information sources: social, personal, marketing, and editorial. Due to the fact that information searches play an important role during the trip decision-making processes [6] and following the development of the Web 2.0 and social media, existing studies have explored trip decision-making and tourists’ information needs in the context of using online systems [2, 14]. Different types of online tourism information (online texts, videos, photo slides, and panoramas) have been investigated in terms of their influences on tourists’ desires to visit a particular destination [9], and it has been found that textual descriptions lead to more positive attitudes towards making a visit. Therefore, we aim to focus on the different types of textual tourism information that are available on existing online websites, and then to classify them, examine their characteristics, and explore their effects on tourists’ desires to visit a potential destination.

With regard to the different types of textual tourism information, current online services and platforms provide enormous opportunities to allow publication from various information sources, including social, personal, marketing and editorial [19]. From the perspective of social sources, Xiang & Gretzel [20] investigate the extent to which social media appear in search engine results in the context of tourism searches. However, the literature has not explored comprehensively the textual tourism information across current online services. Further, it lacks an analysis of the characteristics of each type of textual tourism information, although it has drawn attention to studies which examine how travel blogs affect tourists’ desires to visit a particular destination. Chen et al. [5] examine the characteristics (such as reliability, interestingness, novelty, etc.) of travel blog content which have an influence on tourists’ decision-making. However, since travel blogs are not the only form of textual tourism information which is available online, this study aims to explore all relevant such information, with a specific focus on how they influence tourists’ perceptions while reading the same, and their desires to visit destinations thereafter.

We have conducted two studies to contribute to the knowledge of: (1) identifying the different types of textual tourism information and understanding their characteristics; and (2) exploring the perceived characteristics of the identified textual tourism information and their different influences on tourists’ desires to visit a destination.

2 Background and Literature Review

2.1 Tourism Information Websites

Tourism is an ‘Information-Intense Industry’ [16] and online resources provide many platforms where tourism information can easily be both distributed and consumed. Websites stimulate the creating and sharing of tourism information (Pen et al. 2007) and serve as an important information source in tourists’ decision-making processes.

Web 1.0 allows users (tourists, marketers, official managers, etc.) to publish tourism-related information online, but it is only a ‘readable’ system and information is only updated once in a while. Web 2.0 allows users to interact with the websites, such that it is a dynamic ‘writable’ system which encourages free participation, collaboration, and information sharing. ‘Social media’, created by Web 2.0, consists of Internet-based applications that carry consumer-generated content. Xiang & Gretzel [20] propose a typology of online ‘social media’ websites, based on search engine results: 40% virtual community sites (e.g., LonelyPlanet), 27% consumer review sites (e.g., TripAdvisor), 15% blogs (e.g., TravelPod), 9% social networking sites (e.g., Twitter, Facebook) and only 7% media sharing sites (e.g., Youtube, Flickr, Arounder). In this research, the aim is to explore online tourism information from websites that are in the category of either Web 1.0 or Web 2.0, or social media.

2.2 Online Tourism Information

Since tourists generally have limited knowledge of any detailed information about potential destinations, they tend to seek out and be significantly influenced by online tourism information [22]. Online information, whether it is generated by peer tourists or by official marketers, is important for tourists’ information searches and decision-making processes [13].

Existing researches have explored the different types of online tourism information that are widely used by tourists, both in terms of how their characteristics are perceived by the users, and their overall effects on tourists’ desires to visit a particular destination. Marlow & Dabbish [9] compare four types of tourism information: texts, photo slides, videos, and panoramas. They find that across the four types of information, vividness and interactivity are the two characteristics that most influence users’ attitudes towards visiting an unfamiliar destination, and they also report that readers of text and panorama descriptions of destinations tend to have significant positive attitudes about visiting such places. Further, they explain that text descriptions are more likely to help readers to form vivid mental images in their minds, while panorama descriptions are more likely to allow interaction between the users and any relevant information.

With regard to the written descriptions of a destination, there are different types of textual information, such as blog content, review content, official introduction content, etc. Chen et al. [5] focus on blog text content, and investigate which characteristics of the travel blogs attract tourists’ attention and influence their decisions. They find that the novelty, understandability, and interestingness are the perceived characteristics that have a positive influence on tourists’ desires to visit a destination.

In this research, we aim to explore all types of textual tourism information that are available online, and then to classify them, examine their characteristics and understand how they influence tourists’ decisions.

3 Study 1

3.1 Objectives

With the various types of tourism information that are available on websites to help tourists’ decision-making, Study 1 aims to identify the different types of online textual tourism information on a number of representative websites, and to examine their characteristics in terms of their linguistic features, the emotions presented, and the brand personalities of the destinations described in the information.

3.2 Method

We adopted a mixed-method approach that has been applied previously in collecting and analyzing social media data [3], which consists of online data crawling, text mining, and content analysis. Firstly, a manual content analysis was carried out through Affinity Diagram, to classify the available online textual tourism information. Secondly, text mining was employed to examine the characteristics of the identified types of online textual tourism information in a large scale using two processes: automatic online data crawling and WordStat data analysis.

3.3 Procedure

We selected one non-social website: Wikitravel (WT), and six social media websites: TripAdvisor (TA), Virtualtourist (VT), LonelyPlanet (LP), TravelPod (TP), Facebook (FB), and Arounder (AD), based on Xiang & Gretzel’s [20] work. In this study, we first collected small samples of tourism-related information to conduct a manual content analysis for the purposes of identifying the different types of textual tourism information. We collected different types of textual information about the same destination (London), and the same tourist attraction (the Tower of London). For the manual content analysis, we searched for tourism information across the aforementioned websites, browsed the webpages, observed their information formats and summarised the textual content forms of the websites.

For text mining, we chose to expand the scale and crawl 313,117 words’ text from the same websites. We developed and utilised a framework to collect data from WWW sites with unique domains. Web crawler is an automatic process of parsing HTML webpages and then extracting relevant information [18]. HTML skeleton patterns in each of the target websites were identified for the purposes of writing the parsing code fragment. The crawling framework was installed on a dual core CPU with 4 GB RAM included. The HTML parser was written in Eclipse Java and we deployed a multi-threaded version of the web crawler to make it scalable. We applied an existing core algorithm for fast reading the HTML skeleton on an Eclipse platform and embedded our own regular expressions in the system. For example, any white spaces between paragraphs on the TripAdvisor website should be ignored and skipped. For textual pre-processing, we removed any hash tags, eliminated any non-relevant pictures, user profile information and advertisements. This pre-processing was also machine coded.

3.4 Analysis

For the manual content analysis, we applied Affinity Diagram [15] to classify the online textual tourism information. For the text mining, WordStat 2010 was used to analyse the crawled 313,177 words text data separately into the groups of the classifications. To investigate the characteristics of the different types of textual information from the perspective of their linguistic features, WordStat is able to count the frequency of each word and calculate the percentage of its appearance. In this study, we only focused on the frequencies of the pronouns, namely the first person (e.g., I, our, my), the second person (e.g., you), and the third person (e.g., he, it). To investigate the emotions and personalities presented by the textual information, external dictionaries were installed in WordStat. We used the Regressive Imagery Dictionary [10, 17] to examine the emotions identified in the information, and the Brand Personality Dictionary [1, 12] to examine the brand personalities that were evident in the different types of textual information. Brand personality contains five traits with which to describe a brand or a product: sincerity (displaying feelings of kindness and thoughtfulness), sophistication (signifying something which is elegant and prestigious), excitement (indicating a carefree, spirited and youthful attitude), competence (suggesting success, accomplishment and leadership), and ruggedness (something which is rough, tough, outdoorsy and athletic).

3.5 Results and Discussions

Through the manual content analysis, four types of online textual tourism information were identified: blogs (i.e., personal blogs), reviews (i.e., review comments), messages (i.e., commercials and notifications), and articles (i.e., official introductions and comprehensive descriptions). This classification is consistent with the conclusion by Vogt & Fesenmaier [19], that tourism information is derived fundamentally from four sources: social (reflecting the reviews), personal (reflecting the blogs), marketing (reflecting the messages), and editorial (reflecting the articles).

This study shows that the four types of textual tourism information have different characteristics with regard to one particular destination in terms of: (1) the linguistic features - the narrators’ viewpoints, namely the first person, second person and third person; (2) the emotions expressed via the information - the intangible environments that the information creates; and (3) the brand personality - a set of human characteristics that are attributed to a brand name, in this case - London. The following presents and discusses the characteristics detected by WordStat across the four types of textual tourism information (see results in Table 1):

Table 1. Study 1 results
Full size table

  • Blogs tend to describe tourism information using first person narratives, displaying a strong sense of affection and positive emotional effect, and presenting a destination with sincerity and excitement. Due to the nature of blog content, which normally consists of tourists’ personal travel stories, experiences, and journals, they often present a trip in a first person narrative. Blogs, written in the first person, place the readers read as the narrators so they will associate closely with the travel stories. Readers will immerse themselves spontaneously in the blog contents, will feel what the narrators have experienced and will be affected easily. Meanwhile, blogs usually contain a mixture of objective facts (destination facilities, environments and services) and subjective opinions (tourists’ personal experiences and feelings) regarding a particular destination. Therefore, readers will find blogs to be very affective, and in this particular case, the destination of London was presented with a sense of sincerity and excitement.

  • Reviews tend to describe tourism information using first and third person narratives, often displaying aggression and anxiety, and presenting a destination with sophistication and excitement. Review comments contain the travel experiences and opinions of a substantial number of tourists who have been to a particular destination. Meanwhile, the nature of writing review comments means they are not limited to making compliments, but rather expressing genuine opinions based on real experiences, whether good or bad. Therefore, the content of any review comments may vary dramatically due to the various perspectives of different users, and may even show polarisation regarding one destination. It is reflected in this result that review comments reveal emotions of aggression and anxiety to readers, and in this case the destination of London was presented with a sense of sophistication and excitement.

  • Messages tend to describe tourism information using first and second person narratives, displaying a sense of affection and positive emotional effect, and presenting destinations with sophistication and excitement. Due to the nature of commercials and notifications on social media, the goals of managers sending this type of information are to promote the tourism industry of a particular destination and to attract more tourists. It is logical that messages use first and second person points-of-view (POV) to shorten the distances between readers (potential customers) and destinations.

  • Articles tend to describe tourism information using third person narratives, presenting a destination with sincerity, but the emotions cannot be detected. It reflects the nature articles that they will normally introduce a tourism destination from a distance, from a third person POV with very few personal opinions, with the majority of the information content being factual, accurate and reliable. Therefore, the results from this study show that articles display no emotions, and the destination of London was presented with a sense of sincerity.

4 Study 2

4.1 Objectives

Based on the results of Study 1, four types of online textual tourism information were identified, each of which exhibit different characteristics to readers. To examine how these four types of textual information have an influence on tourists’ decision-making, Study 2 aims to empirically explore how the users perceive both their characteristics and their impacts on tourists’ desires to visit a destination.

4.2 Hypotheses

Existing studies have identified many information quality variables from tourists’ perspectives, in which the variables have an influence on their reading experiences and further impact on their behavioural intentions to visit a destination [5, 7, 9]. Chen et al. [5] find that the perceived reliability and understandability of tourism blog content will influence the blog usage enjoyment, and will further influence tourists’ intentions to visit the destination described. Since the information search behaviour of a tourist usually has a specific purpose of finding sufficient information to allow decisions to be made, the functional need (i.e., the usefulness of the information in terms of tourism decision-making) is a fundamental factor in the examination of tourism information [19]. Meanwhile, Marlow & Dabbish [9] find that the perceived vividness of a piece of information has a positive influence on tourists’ desires to visit a destination through the concept of psychological distance. Therefore, in Study 2, we aim to investigate how different types of information content (based on the results of Study 1: blogs, reviews, messages, and articles) influence tourists’ reading experience and their desires to visit a destination. The following hypotheses are offered:

  • H1a. The form of information content influences the perceived reliability of the tourism information.

  • H1b. The form of information content influences the perceived understandability of the tourism information.

  • H1c. The form of information content influences the perceived vividness of the tourism information.

  • H1d. The form of information content influences the perceived usefulness of the tourism information.

  • H2. The form of information content influences tourists’ desire to visit a destination.

4.3 Method

An empirical scenario-based experiment was carried out, in which each participant was asked to read all four types of textual tourism information. After they finished reading each type of information, they were required to complete a questionnaire regarding the characteristics of the given information and their desires to visit the destination described therein. Therefore, this is a within subject study with non-parametric data, and it has one testing variable with four levels (four types of textual tourism information). This experiment contained three parts. The first part presented the scenario to the participants (see Table 2). The second part displayed four pieces of tourism information to the participants, based on the textual information identified in Study 1. The four pieces of information were randomly shown to the participants, and after each piece of information had been read, closed-end questions (see Table 2) were posed to investigate participants’ opinions and desires towards visiting the destination featured in the information. Finally, the third part entailed a follow-up interview (see Table 2), to obtain the in-depth knowledge of why the participants formed certain impressions and made particular decisions after reading the different types of information.

Table 2. Study 2 material
Full size table

4.4 Participants

Sixteen participants (8 males, 8 females, aged 18 to 39) were recruited through emails and posters for this within-subject experiment. All participants were Chinese, who had neither been to Mexico, Peru, or Brazil, nor had any prior knowledge about the detailed tourism information in these countries.

4.5 Material

The four pieces of tourism information used in Study 2 were designed based on the four types of textual tourism information identified from Study 1: blogs, reviews, messages, and articles. In this study, we selected four destinations in Mexico, Peru, and Brazil, for which similar tourism content was chosen: natural landscapes and heritages. The actual information content for the chosen destinations was downloaded from TravelPod (blogs), TripAdvisor (reviews), WikiTravel (articles), and a travel agent website (messages). The questionnaires were designed in English and later translated to Chinese for the participants, who were more comfortable with their first language.

4.6 Data Collection and Analysis

The questionnaire data was collected by 7-point Likert questions, each time a participant finished reading a piece of information. A Freidman’s test was employed to analyse the one variable, within-subject, and non-parametric ordinal data. The follow-up interview was carried out after each participant finished reading all four pieces of information and had completed the questionnaires. Due to the nature of the qualitative data, interviews were audio-recorded, fully transcribed by the first author, and analysed based on an Emergent Themes Analysis [21].

4.7 Results and Discussions

The results from Study 2 (see Table 3) indicate that there are significant differences in how users perceived the four types of textual tourism information identified in Study 1, and how the four types of information had different influences on tourists’ desires to visit a destination.

Table 3. Study 2 results
Full size table

Reliability.

AFreidman’s test shows that the perceived reliability is significantly different across the four types of textual tourism information (\( {\text{X}}_{( 3)}^{ 2} = 1 3. 3 3 6,p = .00 4 \)), where pairwise tests reveal that the textual tourism information in the forms of articles is perceived the most accurate, reliable, and closest to the facts. It reflects the finding from Study 1 that tourism information in the form of articles presents a destination with sincerity. Due to the nature of articles, they tend to narrate the facts of a destination, in terms of history, location, culture, population, etc., so it is quite reasonable that articles are the most reliable type of information source. Interestingly, the form of review comments and personal blogs are also perceived to be very reliable, and are significantly more reliable than messages (i.e., commercials and travel agent notifications). It can be explained by Cheong et al.’s [4] findings that users have more trust in product information created by other consumers than in information generated by manufacturers, and this trustworthiness attributed to user-generated content (including blogs and peer comments) remains, regardless of whether the particular views are positive or negative.

Understandability.

A Freidman’s test shows that the perceived understandability is significantly different across the four types of textual tourism information (\( {\text{X}}_{( 3)}^{ 2} = 1 9. 4 6 5,p = .00 0 \)), where pairwise tests reveal that the textual tourism information in the forms of reviews and blogs is perceived to be the easiest to understand and to follow with little effort. It can be explained by the comments of one participant, who described how “Blogs and review comments are written by other travellers, as tourists with the same perspectives of myself. The language is easy and straightforward, and it feels like I am experiencing their trips through the words”. This comment reflects the finding from Study 1 in which blogs and reviews tend to use first person narratives, so that readers feel closer to the information, as if they are the narrators throughout the whole reading. Therefore, it is easier to understand the content in blogs and reviews.

Vividness.

A Freidman’s test shows that the perceived vividness is significantly different across the four types of textual tourism information (\( {\text{X}}_{( 3)}^{ 2} = 2 1. 6 1 3,p < .000 \)), where pairwise tests reveal that the textual tourism information in the forms of blogs is perceived to be the most likely to create images, sounds, or smells in users’ minds, and make the users feel like they are immersed in the environment/activity/location while reading the information. It can be explained by one participant, who stated that “Blogs tend to describe the process of a trip, how the authors get to the destination, how the trip begins, what they experience… in an order. So it is very easy and natural to create a sequence of images of their trips”.

Usefulness.

A Freidman’s test shows that the perceived usefulness is significantly different across the four types of textual tourism information (\( {\text{X}}_{( 3)}^{2} = 1 2. 7 80,p = .00 5 \)), where pairwise tests reveal that the textual tourism information in the forms of reviews is perceived to be the most useful while undertaking the tasks for making tourism decisions. Most of the participants expressed the view that all four types of information were useful, to some extent, at different stages of their trip planning processes. However, review comments played the most influential and useful role in the decision-making process, and as one participant explained, “No matter how good a destination is as a travel agent claims, or as an official article introduces, or even as a personal blog describes, all these types of information come from one or a very limited number of sources. I cannot make decisions based on this small amount of information. What I like about reviews is that I can get many up-to-date pieces of information from the viewpoints of different tourists, which are very straightforward, quick to access, and also comprehensive. I would still eliminate a destination after reading an article emphasising how good it is, if the review comments from most of the peer travellers were negative”.

Desire to visit.

A Freidman’s test shows that the tourists’ desire to visit a destination is significantly different across the four types of textual tourism information (\( {\text{X}}_{( 3)}^{2} = 1 3. 4 8 9,p = .00 4 \)), where pairwise tests reveal that the textual tourism information in the forms of reviews and blogs has the most significant influence on tourists’ desires to visit a destination. The desire to visit a destination has been shown to be positively associated with vivid mental imagery [9], and the perceived understandability of the information [5]. This study shows that blogs, which are the easiest to understand and have the most vivid type of textual tourism information, have a significant influence on tourists’ desires to visit a particular destination. Meanwhile, review comments, which are the easiest to understand and have the most useful type of textual tourism information, also have a significant influence on tourists’ desires to visit a destination. Reliability, as one of the most important characteristics of online tourism information, does not have any effect on blog usage enjoyment, nor on tourists’ desires to visit a destination [5]. It is consistent in Study 2 that articles, as the most reliable type of information, do not exert much influence on tourists’ desires to visit a destination. One participant explained that, “The reliability of information is very important, but it is more like an initial condition. I won’t even read it carefully if I believe the information is fake, dodgy, or exaggerating the truth. But it does not mean that I would like to go there if I believe the information is true”.

5 Conclusion

This research first investigated the online textual tourism information through a mixed-method approach that has previously been applied in collecting and analysing social media data [3]. Study 1 identified and classified four types of textual tourism information: blogs, reviews, messages, and articles, in which we found that their characteristics varied between in terms of the linguistic features, the emotions contained, and the brand personality presented in the case of one common destination. Study 2 empirically explored how tourists perceived the four types of textual tourism information, and we found that: (1) articles are perceived as the most reliable type of information; (2) reviews and blogs are perceived as the easiest to understand; (3) blogs are perceived as the best for creating image in users’ minds; and (4) reviews are perceived as the most useful resource when making tourism decisions. Finally, Study 2 empirically explored how the four types of textual tourism information influence on the tourists’ levels of desire to visit a destination. The results show that blogs and review comments have a significant influence on tourists’ desires to visit a destination.

The investigation of online textual tourism information in this paper provides knowledge of how the different types of textual tourism information differ in terms of the presentations of the information, how tourists perceive them, and how they are affected by them. Technology designers may consider combining different types of textual tourism information to users at different stages in the trip planning process, so that tourists feel more comfortable that their needs are being met while reading any such customised information.