A Corpus-Based Analysis of
Tourism Discourse in the Digital Era
Kenryo Fu[1]*
|
*Correspondence: fujianliang@gmail.com |
|
1
Professional College of Arts and Tourism, Japan |
Abstract
Although
destination image has long been studied, little is known about how it is
linguistically constructed across genres of English-language tourism discourse.
This comparative corpus-based study examines three small
specialized corpora for Takarazuka City, Japan: an
Official Tourism Website Corpus (OTWC; 8,908 tokens), a User-Generated Tourism
Corpus of online reviews (UGTC; 6,054 tokens), and a Travel Journalism Corpus
from Time’s World’s Greatest Places 2025 (TJC; 25,130 tokens), the
last serving as a destination-general benchmark. Using frequency- and
dispersion-based word lists, semantic categorization informed by the
cognitive-affective model, 4-gram clusters, and log-likelihood keyness with effect sizes in AntConc
4.2.0, the study shows that the OTWC projects an institutional image centered
on cultural heritage and the Takarazuka Revue; the UGTC constructs a perceived
image through first-person reference, evaluative adjectives, and practical
concerns such as accommodation and access; and the TJC constructs an
editorially mediated image through novelty markers and quoted voices introduced
by says. The comparison reveals a mismatch between projected and perceived
images, especially the greater salience of manga in reviews and performing arts
on the official website. It demonstrates how small comparative corpora link
genre-specific lexical choices to destination image and offers implications for
English-language destination communication.
Keywords:
corpus-based
discourse analysis; specialized corpora; keyness;
log-likelihood; destination image; tourism discourse; online reviews; travel
journalism
1. Introduction
Corpus linguistics has become an established
methodology for investigating linguistic variation across registers, genres,
and discourse types through the systematic analysis of authentic language data (Biber et al., 1998; McEnery & Hardie, 2012). Advances in digital communication have expanded the possibility of
specialized domain-specific corpus compilation, and further enabled researchers
to examine how language is used in various contexts with both quantitative and
qualitative methods. Comparative corpus analysis, in particular, provides a
robust methodological framework for systematically identifying
discourse-specific lexical, semantic, and phraseological patterns across
different corpora. Despite the growing application of corpus linguistic methods
to domain-specific discourse, comparatively little attention has been paid to
how tourism discourse varies across different digital genres and how such
variation contributes to the linguistic construction of destination image. From
a register perspective (Biber & Conrad, 2019), official tourism websites, online reviews, and travel journalism differ in
key situational characteristics, including the relationship between writer and
reader, the communicative purpose, and the degree of institutional control over
text production. Such situational differences are expected to be reflected in
systematic lexical differences, which makes tourism discourse a productive site
for applied corpus research: the same destination is described by different
social actors for different purposes.
With the rapid development of digital
communication, online tourism platforms have become one of the main
communicative channels between tourism destinations and their potential
visitors. Among these online platforms, official tourism websites, which present
carefully curated information about local attractions, culture, history,
events, and tourism services, have attracted considerable scholarly attention
over the past few decades (Malenkina &
Ivanov, 2018; Pagano, 2023; Pagano & De Cantis, 2018). Through these strategically designed language and other multimodal resources,
these official websites aim not only to provide practical travel information
but also to establish and promote a distinctive and attractive image of a
destination. Such official online communication has been shown to influence
tourists’ perceptions, expectations, and destination choices in various ways, thereby playing a
significant role in destination image formation and travel decision-making (Li & Du, 2020; Pagano, 2023). However, while the role of
official tourism websites has been widely acknowledged, limited research
attention has been paid to how destination images are constructed across
different forms of digital tourism discourse (Pagano, 2023, p. 17).
In addition to official tourism websites,
user-generated content (UGC), particularly online reviews, has become an
increasingly influential factor in destination image construction in the
digital tourism environment. Unlike official websites, which present officially
designed information by institutions, UGC reflects visitors’ or tourists’
authentic experiences, personal evaluations, and recommendations, thereby
providing alternative perspectives on destination image. Previous studies have
shown that potential tourists heavily rely on user-generated reviews or UGC
when making travel and accommodation decisions (Fu, 2024; Ye et al., 2011). At the same time, tourism
practitioners, including hotel managers and hoteliers, actively analyze large
volumes of online review data to better understand visitors’ experiences,
preferences, and expectations, and to improve the quality of tourism services (Fu, 2024; Zainol
et al., 2022). Consequently,
a combined approach integrating the above official tourism websites and
user-generated tourism discourse is required to provide a more comprehensive
understanding of destination image construction, while comparison with travel
journalism offers a third, editorially mediated perspective that is neither
wholly institutional nor wholly personal.
Travel journalism or travel writing occupies a
distinctive position from official tourism websites and UGC discourses. In
contrast to official tourism websites and UGC, travel journalism is produced by
professional editors and journalists and is expected to maintain a degree of
journalistic objectivity while simultaneously promoting destinations (Brett, 2023; Lekant & Palau-Sampio, 2022). Time magazine’s World’s
Greatest Places feature, which has been published annually since its launch
in 2018, is such a representative example of
professional travel writing in international travel journalism[i].
Although the Time magazine feature is not devoted to one single tourism
destination, travel journalism discourse provides an important benchmark for
examining how destinations are represented in travel media in general.
Using Takarazuka City, Hyogo Prefecture, Japan, as
a case study, the present study employs comparative corpus analysis to examine
linguistic variation across three types of English-language tourism discourse:
official tourism websites, UGC, and travel journalism. Through a comparison of
three specialized corpora, the study identifies distinctive lexical, semantic,
and discourse patterns associated with each discourse type and explores how
these patterns contribute to the construction of destination image. Methodologically,
the study demonstrates the usefulness of multi-corpus comparison for
investigating tourism discourse and offers corpus linguistic insights into
destination image construction in the digital era. In doing so, the study aims
to contribute to applied corpus linguistics in three ways: it offers a
replicable procedure for comparing small, purpose-built specialized corpora (Koester, 2010); it combines
frequency, dispersion, and keyness measures with explicit reporting of
significance thresholds and effect sizes; and it links genre-specific lexical
choices to an established tourism model of destination image, thereby
connecting corpus-based discourse analysis with tourism research.
2. Literature Review
The term discourse, as used in expressions
such as tourism discourse and the discourse of socialism, has
multiple meanings in the discipline of linguistics. Structurally, discourse
refers to “any piece of connected language, written or spoken, which contains
more than one sentence” (Thornborrow
& Wareing, 1998, p. 240). Functionally, however, discourse is understood as
“language in use” (Brown
& Yule, 1983, p. 1). Following the
structural and functional perspectives in the previous studies, the present
study defines tourism discourse as the language used in tourism-related
communicative contexts and represented by the three domain-specific corpora
compiled in this study.
This section then reviews the basic concept of
destination image in the tourism context in the previous studies. Although
definitions of destination image vary slightly in wording across studies and
disciplines, researchers generally adopt Crompton’s (1979, p. 18) widely cited definition of destination image as “the
sum of beliefs, ideas, and impressions that a person has of a destination.”
Following this definition, destination image is commonly regarded as consisting
of two interrelated components: the cognitive factors, referring to visitors’
knowledge and beliefs about a destination, and the affective factors, referring
to their feelings and emotional responses toward a destination (Baloglu & McCleary, 1999). Baloglu and McCleary (1999) further argue that
destination image is mainly shaped by two major forces: stimulus factors, i.e.,
what tourists know about a destination, such
as information sources; and personal factors, i.e., how tourists feel about a
destination, including psychological and social characteristics. Baloglu and McCleary’s (1999) framework provides an important theoretical basis for examining how different
forms of tourism discourse contribute to destination image formation. This
perspective provides the theoretical motivation for comparing official tourism
websites, user-generated tourism discourse, and travel journalism discourse in
the present study.
Destination image has long been recognized as one
of the central concepts in tourism research because of its dual role in benefiting both tourism
destinations and tourists. As Li and Du (2020, p. 13) have pointed out,
destination image benefits both the supply side (tourism destinations) and the
demand side (tourists) in the tourism industry. For tourism destinations, a
favorable destination image enhances their competitiveness against their rivals,
whereas for tourists, it increases travel expectations, and enriches the
overall travel experience. At the same time, the ways in which destination
images are constructed, communicated, and transformed have evolved considerably
with the emergence of diverse forms of digital tourism discourse. In this
regard, Gartner (1994) has argued that destination images are formed through multiple
information sources, including induced, organic, and autonomous agents. However, relatively little emphasis has been
placed on the digital and multimodal introduction of destinations on dynamic
digital platforms that are continuously updated.
Among the various information resources that
contribute to tourism destination image formation, linguistic resources are of
particular importance. As Dann (1996, p. 2) has observed, “the
language of tourism attempts to persuade, lure, woo, and seduce millions of
human beings and, in so doing, convert them from potential into actual clients.” Dann further argues that linguistic features,
including strategically selected keywords, can “fire the imagination” of the
potential tourists and thereby serve a powerful persuasive function (1996, p. 174). This perspective highlights the importance of
examining lexical choices and recurring linguistic patterns in order to
understand how destination images are constructed and promoted through tourism
discourse.
Among various linguistic
elements that exert influence on destination image formation, the prominence of
first-person pronouns in travel journalism has particularly been observed by Brett (2023). Brett’s travel subcorpus of the Sassari Newspaper Article Corpus (SNAC)
was compiled from articles selected from the travel section of the online
version of the British newspaper The Guardian, covering a wide range of
travel topics, including weekend breaks, city breaks, skiing, golf, cruises,
gap years, and green travel (Brett, 2023, p.
4)[ii]. By
comparing the travel subcorpus with other subcorpora of SNAC, including Crime,
Football, and Banking, Brett found that
the first-person pronouns occurred far more frequently in their travel
subcorpus—the equivalent of travel journalism in the present study—than those
in the other subcorpora. He argues that this distinctive use of first-person
reference reflects the experiential and personal nature of travel journalism,
as detailed in (1).
(1) It is interesting to note how the most common form of reference in Travel
is first person, whereas in Tourism it is second person. Such findings would
tend to confirm a more literary nature of Travel Journalism, where the author
makes frequent reference to him- or herself throughout the journey, and the
texts take on some of the characteristics of the diary. (Brett, 2023, p.
7)[iii]
The travel journalism examined
by Brett (2023) is characterized by its experiential and personal nature, as professional
newspaper article writers or journalists narrate their own travel experiences
predominantly from a first-person perspective. By contrast, the Travel
Journalism Corpus (TJC) compiled and utilized in the present study consists of
edited features that construct destination image through multiple voices.
Although the two datasets have the same name of travel journalism, they
represent different styles and purposes. From a discourse perspective, Brett’s
travel journalism shares important linguistic characteristics with another
genre, i.e., user-generated tourism discourse, particularly its reliance on
first-person narration and personal experience.
Corpus-based research on the language of tourism
has itself developed considerably. Tourism discourse has been described as a
specialized discourse with its own lexical and textual conventions (Dann, 1996; Gotti, 2006), and corpus studies have
documented, for instance, the heavy reliance of promotional tourism texts on
evaluative adjectives (Pierini, 2009) and the persuasive
strategies of destination promotion (Manca, 2016). In parallel, research on online consumer reviews has
shown that reviewers construct credibility and evaluation through personal
narrative, first-person reference, and explicit evaluative lexis (Vásquez, 2014), a finding closely related
to the wider study of evaluation in text (Hunston & Thompson, 2000). Methodologically, keyword
analysis has become a central technique for identifying what is distinctive
about one corpus relative to another (Scott & Tribble, 2006), although its results depend
on the choice of reference corpus and of the keyness statistic (Baker, 2004; Gabrielatos,
2018).
These strands provide the methodological background for
the multi-corpus design adopted here.
Previous studies have considerably advanced our
understanding of destination image by examining official tourism websites,
user-generated tourism discourse, or travel journalism discourse from different
perspectives. However, these three forms of tourism discourse have rarely been
investigated within a single analytical framework. In particular, little
attention has been paid to how they differ in their lexical and semantic
characteristics and how these differences contribute to the construction of
destination image. To address these gaps, the present study adopts a
corpus-based case study approach to examine the destination image of Takarazuka
City across three different discourses. More specifically, the study addresses
the following three research questions (RQs).
(2) RQ1 What lexical and semantic features characterize the three specialized
corpora of official tourism website texts (OTWC), user-generated reviews
(UGTC), and travel journalism (TJC)?
(3) RQ2 How does the user-generated tourism corpus differ from the official
tourism website corpus in the lexical construction of the destination image of
Takarazuka City?
(4) RQ3 What does a comparison with a destination-general travel journalism
corpus reveal about the genre-specific features of the two Takarazuka corpora,
and what are the implications for Takarazuka’s English-language tourism
communication?
3. Methodology and Corpora
The paper presents a case study of Takarazuka
City’s destination image through the official tourism website of the Takarazuka
International Tourism Association[iv]
(TITA), Takarazuka-related UGC at Tripadvisor[v]
and Trip[vi],
and travel journalism discourse of Time magazine[vii].
The selection of Takarazuka City as the case study was based on the following
three considerations. First, despite its rich cultural and tourism resources,
including the internationally known Takarazuka Revue and the legacy of the
manga artist Osamu Tezuka, often called the “God of Manga,” the city has
received relatively little attention in corpus-based and destination image
studies. Second, Takarazuka City offers diverse tourism attractions, such as
performing arts, hot springs, historical sites, and manga-related culture,
which make it an appropriate case for examining how different tourism
discourses represent a multifaceted destination image. Finally, reasonably
scaled linguistic data are available from the official tourism website and UGC,
which enables a combined approach of quantitative and qualitative analyses and
motivates a systematic corpus-based comparison across different discourse
types.
Three specialized domain-specific corpora were
compiled by the author based on three different resources: the official tourism
website of the TITA, Takarazuka-related UGC at Tripadvisor and Trip, and Time
magazine’s online World’s Greatest Places 2025 feature. These corpora
represent three distinct forms of tourism discourse: official tourism
communication, user-generated tourism discourse, and travel journalism.
Together, they provide the basis for a corpus-based comparison of how the
destination image of Takarazuka is constructed across different communicative
contexts.
To ensure comparability, all three corpora were
compiled using the same
corpus-building procedures and subsequently processed using AntConc (version
4.2.0; Anthony, 2022)[viii] for
lexical and semantic analyses. Based on Stefanowitsch (2006) and revised by the author,
the following strategies were used for extracting examples and data. All texts
were collected manually and saved as plain-text (UTF-8) files, one file per web
page (OTWC), per review (UGTC), or per article (TJC). Reviews written in
languages other than English and machine-translated reviews were excluded.
AntConc’s default token definition was used; all items were case-folded, and no
lemmatization was applied, so that, for example, museum and museums
were counted as separate types. The only exception is the manual merging of theater
and theatre described in the note to Table 5. Given their size, the
three corpora are best described as small specialized corpora (Koester, 2010), which are well suited to
capturing the typical features of a narrowly defined discourse but call for caution when generalizing beyond it.
(5) Strategies for extracting examples and data using AntConc (Version 4.2.0)
a. Manual searching
b. Searching for target vocabulary using the KWIC tab
c. Extracting target words using the Word tab (Min. Freq = 1, Min.
Range = 1)
d. Extracting 4-gram clusters using the N-Gram tab (N-Gram Size = 4,
Open Slots = 0, Min. Freq = 1, Min. Range = 1)
e. Extracting target words using the Keyword tab (Min. Freq = 1, Min.
Range = 1; keyness statistic: Log-likelihood; effect-size measure: Dice;
significance threshold: p < .0001 (LL ≥ 15.13))
The descriptive statistics of the three corpora
compiled by the author are elaborated, as shown in Table 1. First, the Official
Tourism Website Corpus (OTWC) consists of 67 English-language texts collected
from the official website of the TITA. These files cover the “our top 12 picks”
and “Suggestions” sections of the official website. Second, the User-Generated
Tourism Corpus (UGTC) comprises 58 English-language user reviews of “The Best
Hotels in Takarazuka 2025,” and “Top Attractions in Takarazuka” collected from
Tripadvisor, and 18 English-language user reviews of tourism facilities and
attractions in the city collected from the Trip website. Restricting the UGTC
to 2025 would result in a dataset too small to support meaningful corpus-based
analysis. For this reason, reviews published between 2011 and 2025 were
included to obtain a sufficiently substantial body of user-generated discourse
concerning Takarazuka[ix].
Third, the Travel Journalism Corpus (TJC) consists of 100 online articles
featured in Time magazine’s World’s Greatest Places 2025. Unlike
the OTWC and the UGTC, which are both Takarazuka-oriented, the TJC covers a
broader range of destinations and is intended to serve as a destination-general
benchmark against which the features of the two Takarazuka corpora can be
evaluated. Table 1 summarizes the descriptive statistics of the three corpora,
including their data collection periods, numbers of texts (files or reviews),
tokens, types, Type-Token Ratio (TTR), and Standardized Type-Token Ratio
(STTR). These statistics provide an overview of the size and composition of
each corpus and serve as the basis for the subsequent lexical analyses.
Table 1. Descriptive Statistics of the OTWC, the
UGTC, and the TJC
|
|
OTWC |
UGTC |
TJC |
|
Time
Period |
2025 |
2011-2025 |
2025 |
|
Files
/ Reviews |
67 |
76 |
100 |
|
Tokens |
8,908 |
6,054 |
25,130 |
|
Types |
2,272 |
1,386 |
6,414 |
|
TTR
(%) (Type-Token
Ratio) |
25.51 |
22.89 |
25.52 |
|
STTR[x] (500
tokens, %) |
53.32 |
51.55 |
65.34 |
Note: Tabulated and compiled by the author from the OTWC, the UGTC, and the TJC
using AntConc[xi]
4. Results and Discussion
4.1
The
Official Tourism Website Corpus (OTWC)
To identify the prominent lexical features
underlying the officially projected destination image of Takarazuka City, OTWC
was analyzed using AntConc. The word list was finalized for analysis according
to the following three procedures and screening criteria. First, a raw word
list was generated from the OTWC using AntConc, with the parameters Min.
Frequency = 1 and Min. Range = 1. Second, the 50 most frequent types were
extracted from the raw word list because 50 words were deemed enough to provide
a representative lexical overview of the corpus while maintaining a manageable
dataset for qualitative semantic analysis. Third, function words (stop words)
and semantically neutral or
non-topical items were removed from the list so that only high-frequency
content words[xii],
namely, nouns, verbs, adjectives, and adverbs based on Biber et al. (1999, p. 62), were retained for
subsequent semantic analysis, as presented in Table 2. Because a cut-off at the
50 most frequent types yields different numbers of content words in each corpus
(13 in the OTWC, 15 in the UGTC, and 12 in the TJC), the
finalized lists are used for qualitative profiling rather than for direct
numerical comparison.
Table 2. The Finalized Word List from the OTWC
|
No. |
Type |
Rank |
Raw Frequency |
Raw Range |
Normalized Frequency (pmw) |
Normalized Range |
|
1 |
takarazuka |
7 |
144 |
42 |
16,165 |
0.627 |
|
2 |
temple |
21 |
47 |
9 |
5,276 |
0.134 |
|
3 |
revue |
26 |
37 |
14 |
4,154 |
0.209 |
|
4 |
enjoy |
28 |
34 |
20 |
3,817 |
0.299 |
|
5 |
hotel |
30 |
32 |
10 |
3,592 |
0.149 |
|
6 |
shrine |
30 |
32 |
7 |
3,592 |
0.104 |
|
7 |
theater |
30 |
32 |
9 |
3,592 |
0.134 |
|
8 |
area |
38 |
26 |
12 |
2,919 |
0.179 |
|
9 |
japanese |
41 |
23 |
15 |
2,582 |
0.224 |
|
10 |
grand |
43 |
22 |
7 |
2,470 |
0.104 |
|
11 |
hall |
43 |
22 |
9 |
2,470 |
0.134 |
|
12 |
time |
46 |
20 |
13 |
2,245 |
0.194 |
|
13 |
museum |
50 |
18 |
8 |
2,021 |
0.119 |
Note: Tabulated and compiled by the author from the OTWC using AntConc
Table 2 presents the final set of 13 high-frequency
content words from the OTWC, together with their distributional statistics, including raw frequency, raw
range, normalized frequency, and normalized range. Normalized frequency, as
noted in McEnery and Hardie (2012, pp. 49-50), refers to the number of occurrences of each lexical
item per million words (pmw), as calculated using Equation (6)[xiii].
Normalized range represents the proportion of texts in the corpus in which a
lexical item occurs, and is calculated by dividing its raw range by the total
number of texts, as shown in Equation (7)[xiv].
Normalized frequency and normalized range facilitate comparative analyses
across corpora of different sizes by indicating both the relative frequency and
the textual distribution of each lexical item. These values therefore provide a
more comprehensive account of the lexical features of the OTWC than raw
frequency or raw range alone. Two caveats apply. First, given the small size of
the corpora, frequencies per million words involve considerable extrapolation;
they are reported here for comparability with previous studies, but frequencies
per 10,000 words would express the same proportions on a more realistic scale
(e.g., 16,165 pmw = 161.65 per 10,000 words). Second, normalized range is a
simple measure of dispersion; more refined measures such as Gries’s (2008) DP could be considered in future
work.
(6) 
(7) 
The 13 high-frequency lexical items presented in
Table 2 were further classified into seven semantic categories by the author
according to their functions in the tourism context. The classification was
carried out manually on the basis of the concordance lines for each item, so
that items were assigned according to their predominant use in context rather
than their dictionary meaning. As shown in Table 3-1, these categories include
destination identity, nation identity, cultural heritage, performing arts, tourism
infrastructure, and general contextual terms (six cognitive factors), and
promotional or evaluative language (one affective factor). These semantic
categories reveal that the official tourism website constructs the destination
image of Takarazuka primarily through its distinctive cultural heritage
(displayed by three lexical items, temple, shrine, and museum)
and performing arts resources (displayed by four lexical items, revue,
theater, grand, and hall). In particular, lexical items such as temple,
shrine, and museum highlight the city’s
historical and cultural attractions, whereas revue, theater, grand,
and hall collectively emphasize the symbolic importance of the
Takarazuka Revue as a unique tourism asset[xv]. These
semantic categories mainly contribute to the cognitive dimension of destination
image (Baloglu & McCleary, 1999), while the frequent use of
promotional expressions such as enjoy reflects the affective aspect of
destination image (Baloglu &
McCleary, 1999) and the persuasive function of tourism discourse (Dann, 1996). These findings answer RQ1 by demonstrating that official tourism
discourse is characterized by destination-specific lexical choices that reflect
cultural heritage and the performing arts in constructing the destination image of Takarazuka.
Table 3-1. Classification of the Finalized Word
List of the OTWC
|
No. |
Factor Type |
Attribute |
Lexical Items |
|
1 |
Cognitive |
Destination Identity |
takarazuka |
|
2 |
Cognitive |
Nation Identity |
japanese |
|
3 |
Cognitive |
Cultural Heritage |
temple, shrine, museum |
|
4 |
Cognitive |
Performing Arts |
revue, theater, grand, hall |
|
5 |
Cognitive |
Tourism Infrastructure |
hotel |
|
6 |
Cognitive |
General Contextual Terms |
area, time |
|
7 |
Affective |
Promotional / Evaluative Language |
enjoy |
Note: Tabulated and compiled by the author from the OTWC using AntConc
Further, a 4-gram cluster analysis was conducted on
the OTWC. A 4-gram cluster refers
to a recurrent sequence of four consecutive words and four-word sequences are
the length most frequently investigated in research on lexical bundles (Biber et al., 2004; Partington & Morley, 2004). The results reveal that
eight of the ten most frequent 4-gram clusters in the OTWC are related to the
performing arts, which is consistent with the lexical patterns presented in
Tables 2 and 3-1. The most frequent 4-gram cluster is the Takarazuka Grand
Theater (Rank = 1, Frequency = 16, Range = 5). The remaining two clusters, is
said to have (Rank = 6, Frequency = 5, Range = 2) and dinner is served
in (Rank = 10, Frequency = 4, Range = 2), fall into the categories of
general contextual terms and tourism infrastructure, respectively. These
findings further demonstrate that recurrent lexical bundles in the OTWC
reinforce the prominence of performing arts in the linguistic construction of
Takarazuka’s destination image. Because several of these clusters occur in only
two to five texts, they are best interpreted as recurrent phrasing on
particular pages rather than as corpus-wide lexical bundles, which are usually
required to be dispersed across at least five texts (Biber et al., 2004).
4.2
The
User-Generated Tourism Corpus (UGTC)
4.2.1 The Finalized Word List of
the UGTC
Following the same analytical procedure, the UGTC
was analyzed to identify the lexical features that characterize visitors’
perceived destination image of Takarazuka City. The finalized word list of the
UGTC is presented in Table 4. The 15 high-frequency content words from the UGTC
were subsequently divided into the seven semantic categories (Table 3-2). An
overall comparison of the OTWC and the UGTC reveals three major differences in
three semantic categories: cultural heritage, performing arts (both cognitive
factors), and promotional / evaluative language (an affective factor), which
are further elaborated in the subsequent sections.
Table 4. The Finalized Word List of the UGTC
|
No. |
Type |
Rank |
Raw Frequency |
Raw Range |
Normalized Frequency (pmw) |
Normalized Range |
|
1 |
hotel |
16 |
45 |
21 |
7,433 |
0.276 |
|
2 |
very |
16 |
45 |
27 |
7,433 |
0.355 |
|
3 |
nice |
27 |
29 |
17 |
4,790 |
0.224 |
|
4 |
takarazuka |
27 |
29 |
20 |
4,790 |
0.263 |
|
5 |
good |
31 |
27 |
18 |
4,460 |
0.237 |
|
6 |
great |
34 |
26 |
20 |
4,295 |
0.263 |
|
7 |
room |
35 |
25 |
17 |
4,130 |
0.224 |
|
8 |
place |
36 |
24 |
19 |
3,964 |
0.25 |
|
9 |
station |
37 |
23 |
19 |
3,799 |
0.25 |
|
10 |
beautiful |
46 |
19 |
15 |
3,138 |
0.197 |
|
11 |
japanese |
46 |
19 |
17 |
3,138 |
0.224 |
|
12 |
manga |
49 |
18 |
8 |
2,973 |
0.105 |
|
13 |
museum |
49 |
18 |
9 |
2,973 |
0.118 |
|
14 |
temple |
49 |
18 |
6 |
2,973 |
0.079 |
|
15 |
walk |
49 |
18 |
13 |
2,973 |
0.171 |
Note: Tabulated and compiled by the author from the UGTC using AntConc
One striking lexical difference between the OTWC
and the UGTC concerns the lexical item manga. Manga, an important
tourism resource in Takarazuka through the legacy of Osamu Tezuka and the
Takarazuka Osamu Tezuka Manga Museum, occurs 18 times in eight texts (NormFreq
= 2,973 pmw[xvi];
NormRange = 0.105[xvii]) in the
UGTC but only eight times in three texts (NormFreq = 898 pmw; NormRange =
0.045) in the OTWC. Osamu Tezuka is widely recognized as the “God of Manga”
(Encyclopedia Britannica[xviii]).
A further analysis identified two examples (8) and (9) referring to Osamu
Tezuka as the “God of Manga” in the UGTC, whereas no corresponding
examples were found in the OTWC. Although the absolute frequencies are small,
the difference is statistically significant (LL = 8.77, p < .01), with manga
occurring approximately 3.3 times as frequently in the UGTC as in the OTWC.
This discrepancy suggests that visitors perceive manga as a more salient
destination attribute than is reflected in the official tourism website, which
indicates a gap between the officially projected destination image and the
perceived destination image by tourists. More broadly, the finding implies that
manga culture, despite its strong association with Takarazuka City, remains
underrepresented in the city’s official English-language tourism communication.
Table 3-2. Classification of the Finalized Word
Lists of the OTWC and the UGTC
|
No. |
Factor Type |
Attribute |
OTWC |
UGTC |
|
1 |
Cognitive |
Destination identity |
takarazuka |
takarazuka |
|
2 |
Cognitive |
Nation identity |
japanese |
japanese |
|
3 |
Cognitive |
Cultural Heritage |
temple, shrine, museum |
manga, museum, temple |
|
4 |
Cognitive |
Performing Arts |
revue, theater, grand, hall |
|
|
5 |
Cognitive |
Tourism Infrastructure |
hotel |
hotel, room, station |
|
6 |
Cognitive |
General Contextual Terms |
area, time |
place, walk |
|
7 |
Affective |
Promotional / Evaluative Language |
enjoy |
very, nice, good, great beautiful |
Note: Tabulated and compiled by the author from the OTWC and UGTC using AntConc
(8) The entire museum is dedicated to Osamu Tezuka, “the god of manga”
and they’ve added a lot of decorations and architecture for his various works.
(UGTC, 2014, italicized by the author)
(9) Our tour group stopped for a short restroom break here on the way to Awaji
Island, so was not able to fully explore what was available at this rest area.
Overall, seemed very clean and ultra-modern. Noticed a lot of Harry Potter
promotional merchandise (a tie-in with Universal Studios Osaka?) and the
characters of the famed “God of Manga,” Osamu Tezuka. Not sure what to
think about the Dr. Blackjack toilet paper that was on sale. (UGTC, 2024)
A second lexical difference is observed in the
category of performing arts between the OTWC and the UGTC. While lexical items
related to the Takarazuka Revue (revue, theater, grand, and hall)
constitute one of the most prominent semantic categories in destination image
formation in the OTWC, no corresponding high-frequency lexical items appear in
the UGTC. As shown in Table 5, the lexical evidence suggests that the
performing arts, although strongly emphasized in the official tourism website,
are considerably less salient in visitors’ online reviews. For example, the
normalized frequency of grand in the OTWC (NormFreq = 2,470) is 7.48
times that of grand in the UGTC (NormFreq = 330); the normalized range
of grand in the OTWC (NormRange = 0.104) is 4 times as high as that in
the UGTC (NormRange = 0.026). Log-likelihood tests confirm that all four items
are significantly less frequent in the UGTC than in the OTWC: revue (LL
= 9.18, p < .01), theater (LL = 7.63, p < .01), grand (LL =
12.67, p < .001), and hall (LL = 22.82, p < .0001). These findings
indicate a discrepancy between the officially projected destination image and
the perceived destination image with respect to one of Takarazuka’s most
distinctive tourism resources.
Table 5. Comparison of Performing Arts-Related
Lexical Items in the OTWC and the UGTC
|
Corpus |
Type |
Rank |
Raw Frequency |
Raw Range |
Normalized Frequency (pmw) |
Normalized Range |
|
OTWC |
revue |
26 |
37 |
14 |
4,154 |
0.209 |
|
theater |
30 |
32 |
9 |
3,592 |
0.134 |
|
|
grand |
43 |
22 |
7 |
2,470 |
0.104 |
|
|
hall |
43 |
22 |
9 |
2,470 |
0.134 |
|
|
UGTC |
revue |
110 |
9 |
7 |
1,487 |
0.092 |
|
theater |
127 |
8 |
8 |
1,321 |
0.105 |
|
|
grand |
387 |
2 |
2 |
330 |
0.026 |
|
|
hall |
— |
— |
— |
— |
— |
Note: Tabulated and compiled
by the author from the OTWC and the UGTC using AntConc. The UGTC contains one
occurrence of the American spelling “theater” and seven occurrences of the
British spelling “theatre.” To ensure consistency in the lexical analysis, these
spelling variants were manually combined by the author, and their frequencies
and distributional measures (Rank, Raw Frequency, Raw Range, Normalized
Frequency, and Normalized Range) were calculated as a single lexical item. By
contrast, no instances of the British spelling theatre were found in the
OTWC.
One possible explanation for the gap mentioned
above is the limited accessibility of theater attendance. Although the
Takarazuka Revue is one of the city's most distinctive tourism resources,
references to the performing arts occur far less frequently in the UGTC than in
the OTWC. Unlike museums, temples, or other attractions that visitors can
access almost throughout the day, attending a theatrical performance is often
constrained by fixed performance schedules, ticket availability, and advance reservations, as
noted by Nozu et al. (2023). As a consequence, many visitors may experience other more accessible
attractions without necessarily attending a Takarazuka Revue performance,
resulting in the relatively limited representation of performing arts in
user-generated tourism discourse. This explanation remains tentative, as it
cannot be tested with the present corpus data alone.
A third lexical difference
concerns promotional and evaluative language between the OTWC and the UGTC. In
contrast to the OTWC, which primarily provides factual and descriptive
information, the UGTC contains a considerably larger number of evaluative lexical
items such as nice, good, great, and beautiful.
These expressions reflect visitors’ personal evaluations and emotional
responses, corresponding to the affective aspect of destination image (Baloglu & McCleary, 1999). Overall, the comparison demonstrates that official tourism discourse
emphasizes destination promotion and branding, whereas user-generated tourism
discourse foregrounds visitors’ authentic experiences and evaluations.
The differences between the OTWC and the UGTC are
further demonstrated through a comparison executed in the larger unit of 4-gram
clusters. Unlike the OTWC, none of the ten most frequent 4-gram clusters in the
UGTC are related to the performing arts. Overall, six of the ten most frequent
clusters fall into the tourism infrastructure category, whereas the remaining
four belong to the general contextual terms category. The most frequent 4-gram
cluster is from the train station (Rank = 1, Frequency = 5, Range = 5),
which belongs to the tourism infrastructure category and represents a cognitive
factor. These findings suggest that visitors to Takarazuka focus more on
practical aspects of travel, particularly tourism infrastructure such as hotels
and transportation, than on destination-specific cultural resources.
4.2.2 Keyness Analysis in the UGTC
Keyness analysis was conducted using AntConc to
identify lexical items that are significantly overrepresented in the UGTC (target corpus)
relative to the OTWC (reference corpus). As noted by Partington et al. (2013, p. 18), a word list and a keyword
list serve different analytical purposes. A word list (as presented in Table 4)
provides information about the frequency of individual lexical items within the
target corpus, identifying words that occur relatively frequently in that
corpus. In contrast, a keyword list (as presented in Table 6) identifies
lexical items that occur with statistically significantly higher frequencies in
the target corpus than would be expected on the basis of their frequencies in a
reference corpus. The finalized keyword list of the UGTC (Target Corpus: the
UGTC; Reference Corpus: the OTWC) is presented in Table 6. Using the log-likelihood (LL) statistic (Dunning, 1993; Rayson & Garside, 2000), all eight keywords yielded
the LL values exceeding the threshold of 15.13 (p < .0001), indicating
highly significant lexical differences between the two corpora. To complement
the statistical significance indicated by LL, the keyness (effect) values
reported by AntConc, calculated using the Dice measure selected in AntConc’s
keyword settings, are also presented as indicators of the practical magnitude
of lexical uniqueness. As effect size is a descriptive measure rather than a
statistical significance test, it is interpreted as supplementary evidence in
conjunction with the LL measures. Because keyness is calculated for every type
in the corpus, many comparisons are performed; the stringent threshold of p
< .0001 was therefore adopted partly to reduce the risk of false positives
arising from multiple testing (Gabrielatos, 2018). Only
positive keywords, i.e., items overrepresented in the target corpus, are
reported in Table 6.
Table 6. Keyness Analysis (Target Corpus: the
UGTC; Reference Corpus: the OTWC)
|
No. |
Type |
Keyness (Likelihood) |
Keyness (Effect) |
|
1 |
i |
161.84 |
0.029 |
|
2 |
my |
36.23 |
0.007 |
|
3 |
nice |
44.83 |
0.01 |
|
4 |
great |
24.89 |
0.009 |
|
5 |
good |
23.84 |
0.009 |
|
6 |
stay |
20.72 |
0.005 |
|
7 |
beautiful |
17.30 |
0.006 |
|
8 |
room |
17.02 |
0.008 |
Note: 3.84, p < .05; 6.63,
p < .01; 10.83, p < .001; 15.13, p < .0001[xx]
First, the keyness analysis demonstrates the
overrepresentation of first-person pronouns (I and my) in the
UGTC relative to the OTWC. As shown in Table 6, among all keywords, the lexical
item I exhibits the highest LL value (LL = 161.84, p < .0001) as well as the largest
effect size (effect size = 0.029). Unlike the OTWC, which adopts an
institutional and relatively impersonal style in promoting the destination, the
UGTC constructs destination image primarily through visitors’ first-hand
experiences and personal perspectives. This finding is broadly consistent with Brett (2023, pp. 6–7), who reports that first-person pronouns are among the
most prominent forms of personal reference in the travel subcorpus of the SNAC[xxi],
compiled in his study. In the present study, the frequent use of first-person
pronouns enhances the authenticity and experiential nature of user-generated
tourism discourse, thereby contributing to the construction of a more personal
and experience-based destination image. This interpretation is consistent with
research on online consumer reviews, in which first-person narrative is a
central resource for constructing credibility (Vásquez, 2014).
The keyness analysis further
shows that the UGTC is characterized by lexical items associated with
subjective evaluation (nice, great, good, and beautiful)
and personal experience (stay), i.e., affective factors, and tourism
infrastructure-related terms (room), i.e., cognitive factors. These
lexical items reflect both the affective and cognitive components of
destination image (Baloglu &
McCleary, 1999), which indicates that
user-generated tourism discourse constructs destination image primarily through
personal experience, subjective evaluation, and practical travel concerns. By
contrast, the official tourism website constructs destination image mainly by
presenting tourism resources and destination attributes from a promotional but
impersonal perspective.
4.3
The Travel Journalism Corpus (TJC)
4.3.1 The Finalized Word List of the TJC
Following the same analytical procedure, the TJC
was analyzed to identify the salient lexical features underlying the
editorially mediated destination image presented in international travel
journalism. Unlike the OTWC and the UGTC, which are destination-specific
corpora focusing on the specific destination of Takarazuka City, the TJC, which
is destination-general in nature, comprises 100 texts describing 49 places
to visit and 51 places to stay featured in Time magazine’s World’s
Greatest Places 2025. The TJC serves as an external benchmark for the OTWC
and UGTC, both of which focus on Takarazuka. As a destination-general corpus
covering multiple tourist destinations, the TJC provides a broader,
cross-destination perspective on tourism discourse, enabling comparisons of
lexical diversity and linguistic characteristics across the three corpora in
this study. In general, the finalized word list demonstrates that the
high-frequency content words (Table 7) from the TJC are more diverse and are
distributed unevenly across the seven categories shown in Table 3-3.
Table 7. The Finalized Word List of the TJC
|
No. |
Type |
Rank |
Raw Frequency |
Raw Range |
Normalized Frequency (pmw) |
Normalized Range |
|
1 |
new |
20 |
94 |
55 |
3,741 |
0.55 |
|
2 |
hotel |
26 |
65 |
30 |
2,587 |
0.3 |
|
3 |
art |
31 |
55 |
30 |
2,189 |
0.3 |
|
4 |
world |
32 |
54 |
38 |
2,149 |
0.38 |
|
5 |
first |
34 |
52 |
40 |
2,069 |
0.4 |
|
6 |
guests |
40 |
49 |
36 |
1,950 |
0.36 |
|
7 |
local |
42 |
48 |
34 |
1,910 |
0.34 |
|
8 |
museum |
43 |
46 |
20 |
1,830 |
0.2 |
|
9 |
says |
43 |
46 |
32 |
1,830 |
0.32 |
|
10 |
opened |
45 |
45 |
35 |
1,791 |
0.35 |
|
11 |
resort |
48 |
43 |
19 |
1,711 |
0.19 |
|
12 |
country |
49 |
40 |
29 |
1,592 |
0.29 |
Note: Tabulated and compiled by the author from the TJC using AntConc
One notable characteristic of the TJC is the
balanced use of promotional and evaluative language, representing an
intermediate pattern between the OTWC and the UGTC. Unlike the OTWC, which
mainly employs promotional language to market the destination, or the UGTC,
which predominantly contains visitors’ personal evaluations, the TJC combines
descriptive, evaluative, and journalistic language to present destination image
in an informative yet engaging manner. Specifically, the distinctive feature of
the TJC is the occurrence of lexical items such as new, first, local, and
opened, none of which appear among the high-frequency lexical items in
either the OTWC or the UGTC. Strictly speaking, these items are not evaluative
in the same sense as nice or beautiful. They function rather as
markers of newsworthiness, foregrounding novelty and locality, and carry an
implicit positive appraisal; it is in this sense that they are grouped with
promotional and evaluative language in Table 3-3. These words foreground novelty,
authenticity, and newsworthiness, which are characteristic features of
editorial tourism discourse. As illustrated in (10) – (13), new is used
to introduce a recently developed resort in (10), first highlights a
pioneering dome in (11), local emphasizes authentic cultural experiences
with the local wildlife conservationist in (12), and opened reports a
newly established museum in (13). Together, these lexical choices distinguish
travel journalism from both official tourism and user-generated tourism
discourses. They also suggest that incorporating greater emphasis on novelty
and authenticity could further enhance the English-language tourism
communication of Takarazuka.
Table 3-3. Classification of the Finalized Word
Lists of the Three Corpora
|
No. |
Factor Type |
Attribute |
OTWC |
UGTC |
TJC |
|
1 |
Cognitive |
Destination identity |
takarazuka |
takarazuka |
world, country |
|
2 |
Cognitive |
Nation identity |
japanese |
japanese |
— |
|
3 |
Cognitive |
Cultural Heritage |
temple, shrine, museum |
manga, museum, temple |
museum |
|
4 |
Cognitive |
Performing Arts |
revue, theater, grand, hall |
— |
art |
|
5 |
Cognitive |
Tourism Infrastructure |
hotel |
hotel, room, station |
hotel, resort, |
|
6 |
Cognitive |
General Contextual Terms |
area, time |
place, walk |
guests, says |
|
7 |
Affective |
Promotional / Evaluative Language |
enjoy |
very, nice, good, great beautiful |
new, first, local, opened |
Note: Tabulated and compiled by the author from the OTWC, the UGTC, and the TJC
using AntConc
(10) “Since Madagascar is [largely] untouched, it’s a new destination for
a lot of people,” says Philippe Kjellgren, founder and owner of the new
luxury resort, Voaara Madagascar, on Sainte-Marie’s northwestern coast, where
you can spot humpback whales migrating from June to September. (Voaara)
(11) Aiming to be the NBA’s first carbon-neutral arena includes solar
panels and water preservation efforts like low-flow faucets, drought-tolerant
landscaping, and water-efficient irrigation.
(Intuit Dome)
(12) Other tour highlights include night safaris, a tiger-painting workshop, and
drinks with a local wildlife conservationist. (Oberoi Vindhyavilas Wildlife Resort,
Bandhavgarh)
(13) The two-story museum opened in October in Uji, about 30 minutes
south of Kyoto, on the site of the company’s former Uji Ogura plant, where the
Hanafuda trading cards that put the company on Japan’s map were once printed.
(Nintendo Museum)
4.3.2 Keyness Analysis in the TJC
Keyword analysis was also conducted by comparing
the TJC (target corpus) with the OTWC (reference corpus). Because the TJC
comprises travel journalism covering 100 destinations worldwide rather than a
single destination, its lexical profile is more diverse and less
destination-specific. Consequently, the comparison between the TJC and OTWC
yielded far fewer statistically significant keywords (Table 8) than the
comparison between the UGTC and OTWC (Table 6). This result is attributable to
the broader topical coverage and greater lexical diversity of the TJC, which is
not restricted to a single destination. It should be noted, however, that the
TJC differs from the OTWC in both genre and topical scope, so the keywords
obtained from this comparison cannot be attributed to genre alone; the result
is therefore treated as exploratory.
Specifically, one notable keyword in the TJC is says
(LL = 27.936, p < .0001; effect = 0.004), a reporting verb that
introduces direct speech and contributes to the authenticity of travel
journalism. In contrast to Brett (2023), who argued that direct speech is
“relatively rare in the Travel” corpus and “is practically non-existent” in the
Language of Tourism corpus (p. 6), a concordance analysis of the TJC in this
study revealed that all 46 occurrences of says
in 32 texts (see Table 7) introduce direct quotations from local residents,
business owners, experts, or other stakeholders, as presented in (14), (15),
and (16). The frequent use of direct speech contributes to authenticity and
immediacy by presenting destination information through multiple voices rather
than through the journalist’s narration alone.
(14) “It’s been a long journey,” Van Andel says, “but a worthwhile and
rewarding one.” (Peter Island Resort)
(15) “Being the older sibling, I’d sacrifice it for my brother; now I can make
enough for everyone to share,” says the Eleven Madison Park alum who
helms the kitchen at the 12-seat chef’s counter Papa’s Bombay. (Papa’s Bombay)
(16) “Travelers to the region can cross borders easily, spending less money on
visas, time on checkpoints, and exploring more of the conservation area on a
single trip,” says Barbara Murasiranwa-Hughes, regional representative
for the Tourism Business Council of Zimbabwe. (Mbano Manor Hotel)
Table 8. Keyness Analysis (Target Corpus: the TJC;
Reference Corpus: the OTWC)
|
No. |
Type |
Keyness (Likelihood) |
Keyness (Effect) |
|
1 |
says |
27.936 |
0.004 |
|
2 |
new |
27.797 |
0.007 |
Note: 3.84, p < .05; 6.63, p < .01; 10.83, p < .001; 15.13, p < .0001[xxii]
4.4
Across
Corpora Analysis
4.4.1 Type-Token Ratio (TTR) and Standardized
Type-Token Ratios (STTR) across the Three Corpora
The TTR and STTR values provide an indication of
lexical diversity across the three corpora. Because the raw TTR decreases as
corpus size increases (McEnery & Hardie, 2012, p. 50), the raw TTR values in Table
1 are not directly comparable: the TJC is roughly three to four times larger
than the OTWC and the UGTC. The almost identical raw TTRs of the OTWC (25.51%)
and the TJC (25.52%) therefore conceal a real difference, because the TJC
achieves the same ratio over a much larger number of tokens. The STTR, calculated over consecutive 500-token
segments, controls for this effect and is therefore used as the primary measure
here. On this measure, the TJC (STTR = 65.34%) is clearly the most lexically
diverse corpus, followed by the OTWC (53.32%) and the UGTC (51.55%). The
relatively low value for the UGTC is consistent with the repeated use of a
small set of evaluative and experience-based items that characterizes
user-generated reviews, whereas the high value for the TJC reflects its
coverage of 100 different destinations. Even so, the STTR values are
interpreted descriptively: the UGTC yields only about 12 segments of 500
tokens, compared with about 18 for the OTWC and 50 for the TJC, so the small
difference between the OTWC and the UGTC (1.77 percentage points) should not be
over-interpreted.
4.4.2 Convergence and Divergence across Corpora
Beyond the individual corpus analyses, a
cross-corpus comparison reveals that the three discourse types construct
destination image through different combinations of content, stance, and voice.
The OTWC foregrounds cultural heritage and performing arts in Takarazuka City
through a promotional voice, thereby projecting an officially constructed
destination image. By contrast, the UGTC constructs a perceived destination
image through first-person experience, subjective evaluation, and practical
visitor concerns, which reflects an experiential and highly personal
communicative stance. The TJC, meanwhile, combines destination description with
novelty, locality, and direct quotations, resulting in an editorially mediated,
multi-voiced representation of destinations. Taken together, these findings
suggest that destination image is not simply projected by tourism organizations
but is co-constructed through the interaction of institutional, experiential,
and editorial discourses.
The cross-corpus comparison also identifies both
convergence and divergence in the lexical display of destination image. As
shown in Table 3-3, only two high-frequency lexical items, hotel and museum,
are shared across all three corpora, indicating that accommodation or tourism
infrastructure and cultural attractions constitute common cognitive attributes
of destination image regardless of discourse type. By contrast, the greater
prominence of manga in the UGTC relative to the OTWC and performing
arts-related lexical items (revue, theater, grand, and hall)
in the OTWC relative to the UGTC reveals a mismatch between visitors’ perceived
destination image and the officially projected destination image. From a
practical perspective, this divergence suggests that official tourism websites
could place greater emphasis on Takarazuka’s manga-related resources while
providing more accessible and user-friendly information on theater
performances. Such improvements may help narrow the gap between destination
promotion and visitors’ actual interests and experiences.
5. Conclusion, Limitations, and Future Prospects
This study has presented a corpus-based case study
of destination image construction in the digital era by comparing three
English-language tourism corpora compiled by the author: the Official Tourism
Website Corpus (OTWC), the User-Generated Tourism Corpus (UGTC), and the Travel
Journalism Corpus (TJC). Through word lists, semantic categories, and keyness
analyses, the study examined how destination image is linguistically
constructed across different types of tourism discourse.
Regarding RQ1, the analysis demonstrated that each
corpus exhibits distinctive lexical and semantic characteristics reflecting its
communicative purpose and style. The OTWC is characterized by
destination-specific lexical items related to cultural heritage, performing
arts, tourism infrastructure in Takarazuka City, and promotional language,
officially projecting the city primarily as a culturally distinctive
destination. In contrast, the UGTC contains a greater proportion of evaluative
expressions, first-person pronouns, and accommodation-related vocabulary,
reflecting visitors’ or tourists’ experiences and their perceptions of the
destination. The TJC is characterized by a more diverse lexical profile,
including markers of novelty and locality, and direct quotations introduced by
the reporting verb says, revealing distinctive features of travel
journalism in the online tourism environment.
Regarding RQ2, the comparison across corpora
demonstrated that the destination image of Takarazuka City constructed by the
UGTC is different from that formed by the OTWC. The OTWC projects destination
image from an institutional and promotional perspective, whereas the UGTC
constructs a perceived destination image through first-hand experiences and
subjective evaluation. The comparison further reveals both convergence and
divergence between the two discourse types. While cultural attractions and
tourism infrastructure such as museums and hotels are commonly
represented across the three corpora, notable discrepancies are also observed
in the prominence of manga-related discourse and performing arts, suggesting
that the officially projected destination image does not always fully
correspond to visitors’ perceptions.
Regarding RQ3, the comparison with the
international travel journalism corpus suggests several implications for
improving English-language tourism communication on official tourism websites.
In particular, incorporating local perspectives, direct quotations from
residents and tourism practitioners, and greater emphasis on novelty may
enhance the authenticity, credibility, and international appeal of official
tourism websites. These findings indicate that effective destination
communication may benefit from integrating institutional promotion with
visitors’ experiences and editorial storytelling.
Theoretically, this study contributes to tourism
discourse research and applied corpus linguistics by demonstrating that
destination image is not constructed solely through official tourism promotion
but is co-constructed through institutional, experiential, and editorial
discourses. Methodologically, it illustrates the value of combining word list
analysis, semantic categorization, and keyness analysis in examining projected
and perceived destination images from complementary perspectives. It also shows
that small, purpose-built specialized corpora, when analyzed with dispersion
information, explicit significance thresholds, and effect sizes, can reveal
interpretable differences between closely related genres of the same discourse
domain. Overall, the findings suggest that destination image in contemporary
tourism communication should be understood as a dynamic and negotiated outcome
emerging from the interaction of institutional, experiential, and editorial
voices rather than as a unidirectional product of official destination
marketing.
Several limitations should be acknowledged. First,
the study focuses on a single Japanese destination and therefore the findings
cannot be generalized to all tourism destinations. Second, although the three
corpora represent different discourse types, their sizes and sources are not
identical, which may influence lexical distribution despite the use of
normalized frequency and keyness analysis. Third, the UGTC spans 2011–2025,
whereas the OTWC and the TJC represent 2025, so diachronic change in visitors’
reviews and in the website itself cannot be ruled out.
Fourth, the semantic categorization was carried out manually by a single
analyst, and the 4-gram analysis relied on low frequency and range thresholds.
Fifth, the analysis is restricted to verbal text and does not consider images
or other multimodal resources. Future research could investigate multiple
destinations across different countries, incorporate additional discourse
genres and multimodal resources, test the manual categories with automatic semantic tagging (e.g., the USAS
tagger in Wmatrix; Rayson, 2008), compare each corpus with a
large general reference corpus of English, and employ more advanced corpus techniques such as collocation and
concordance-based analyses of evaluation to provide a more comprehensive
understanding of destination image construction in digital tourism
communication.
Declaration regarding usage of
generative AI (Artificial Intelligence) tools
This manuscript made use of ChatGPT to assist in
searching for previous studies in related fields and in paraphrasing and
improving the clarity of text. No AI tools were used for the generation of
research hypotheses or the writing of conclusions. All suggestions generated by
ChatGPT were carefully reviewed, selectively adopted, and revised as necessary
by the listed author, who takes full responsibility for the final content of
the manuscript.
Conflict
of Interest
The authors declare no conflicts of interest
regarding this manuscript.
References
Anthony, L. (2022). AntConc
(Version 4.2.0) [Computer software]. Waseda
University. https://www.laurenceanthony.net/software
Baker,
P. (2004). Querying keywords: Questions of difference, frequency, and sense in.
keywords analysis. Journal of English Linguistics, 32(4), 346–359. https://doi.org/10.1177/0075424204269894
Baloglu, S., & McCleary, K. W. (1999). A model of destination image
formation. Annals of Tourism Research, 26(4), 868–897. https://doi.org/10.1016/S0160-7383(99)00030-4
Biber, D., & Conrad, S. (2019). Register, genre, and style (2nd
ed.). Cambridge University Press.
Biber,
D., Conrad, S., & Cortes, V. (2004). If you look at…: Lexical bundles in
university teaching and textbooks. Applied Linguistics, 25(3), 371–405. https://doi.org/10.1093/applin/25.3.371
Brett,
D. F. (2023). Is Travel Journalism more similar to Newspaper Language or the
Language of Tourism? A corpus-based study. Studies in Travel Writing, 26(2),
152-167. https://doi.org/10.1080/13645145.2023.2261635
Brown,
G., & Yule, G. (1983). Discourse Analysis. Cambridge University
Press.
Crompton, J. L. (1979). An assessment of Mexico as a vacation destination
and the influence of geographical location upon that image. Journal of
Travel Research, 17(4), 18–23. https://doi.org/10.1177/004728757901700404
Dann,
G.M.S. (1996). The Language of Tourism: A Sociolinguistic Perspective.
CABI.
Fu, K.
(2024). Text Analysis on Toyooka-based Hotel Customers’ Online Reviews. Journal
of Arts and Tourism, 3(1), 193-200. https://doi.org/10.60292/artsandtourism.3.0_194
Gartner, W. C. (1994). Image Formation Process. Journal of Travel &
Tourism Marketing, 2(2–3), 191–216. https://doi.org/10.1300/J073v02n02_12
Gotti,
M. (2006). The language of tourism as specialized discourse. In O. Palusci & S. Francesconi (Eds.), Translating tourism:
Linguistic/cultural representations (pp. 15–34). Editrice
Università degli Studi di Trento.
Gries,
S. Th. (2008). Dispersions and adjusted frequencies in corpora. International
Journal of Corpus Linguistics, 13(4), 403–437. https://doi.org/10.1075/ijcl.13.4.02gri
Koester, A. (2010). Building small specialised corpora. In A. O’Keeffe &
M. McCarthy (Eds.), The Routledge handbook of corpus linguistics (pp.
66–79). Routledge.
Lekant,
M., & Palau-Sampio, D. (2022). Reinventing travel journalism in the digital
age: Quality, specialization, technology, and a unique perspective. Comunicación y sociedad, 19. https://doi.org/10.32870/cys.v2022.8250
Li, W.,
& Du, W. (2020). Understanding destination Image in UGC: A lexicon
approach. Journal of Travel, Tourism and Recreation, 2(1), 13-28.
https://doi.org/10.22259/2642-908X.0201002
Malenkina, N., & Ivanov, S. (2018). A linguistic analysis of the
official tourism websites of the seventeen Spanish Autonomous Communities. Journal
of Destination Marketing & Management, 9, 204-233. https://doi.org/10.1016/j.jdmm.2018.01.007
Nozu,
N., Yao, Y., Kawamura, T., Fu, K., & Takahashi, N. (2023). A comparative
study on temporary linguistic landscapes in local art festivals: Toward
application to the Toyooka Theater Festival. Journal of Arts and Tourism, 2,
41-57. https://doi.org/10.60292/artsandtourism.2.0_41
Pagano,
N. (2023). Keywords in UNESCO websites: The role of image in destination
marketing. International Journal of Islands Research, 4 (1),
17-32. https://doi.org/10.21427/9GEX-1A64
Pagano,
N., & De Cantis, S. (2018). Tourism destination image, tourism discourse
and UNESCO sites: A contrastive analysis. In P. Calò & G. Ruggieri (Eds.), 10th
International Conference on Islands Tourism: Book of full papers (pp.
160–170). Observatory on Tourism for Islands Economy (OTIE).
Partington, A., Duguid, A., & Taylor, C. (2013). Patterns and
meanings in discourse: Theory and practice in corpus-assisted discourse studies
(CADS). John Benjamins Publishing Company. https://doi.org/10.1075/scl.55
Pierini, P. (2009). Adjectives in tourism English on the web: A
corpus-based study. Círculo de Lingüística
Aplicada a la Comunicación,
40, 93–116.
Rayson,
P. (2008). From key words to key semantic domains. International Journal of
Corpus Linguistics, 13(4), 519–549. https://doi.org/10.1075/ijcl.13.4.06ray
Rayson,
P., & Garside, R. (2000). Comparing corpora using frequency profiling. In Proceedings
of the Workshop on Comparing Corpora (pp. 1–6). Association for
Computational Linguistics. https://doi.org/10.3115/1117729.1117730
Vásquez, C. (2014). The discourse of online consumer reviews.
Bloomsbury.
Ye, Q.,
Law, R., Gu, B., & Chen, W. (2011). The influence of user-generated content
on traveler behavior: An empirical investigation on the effects of
e-word-of-mouth to hotel online bookings. Computers in Human Behavior, 27(2),
634-639. https://doi.org/10.1016/j.chb.2010.04.014
Websites and Resources
https://chatgpt.com/ (Retrieved July 26, 2026)
https://kanko-takarazuka.jp/english/ (Retrieved July 23, 2026)
https://kanko-takarazuka.jp/english/lp_introduction/ (Retrieved March 24, 2025)
https://time.com/ (Retrieved July 25, 2026)
https://time.com/5375033/2018-worlds-greatest-places-choice/ (Retrieved July 21, 2026)
https://time.com/collections/worlds-greatest-places-2025/ (Retrieved July 23, 2026)
https://ucrel.lancs.ac.uk/llwizard.html (Retrieved July 25, 2026)
https://www.britannica.com/search?query=%22god+of+manga%22 (Retrieved July 25, 2026)
https://www.city.takarazuka.hyogo.jp/1060683/index.html (Retrieved July 25, 2026)
https://www.laurenceanthony.net/software/antconc/ (Retrieved July 25, 2026)
https://www.theguardian.com/uk/travel (Retrieved July 2, 2026)
https://www.tripadvisor.com/ (Retrieved July 26, 2026)
https://www.trip.com/ (Retrieved July 26, 2026)
Author biodata
Kenryo Fu is an Associate Professor at Professional College
of Arts and Tourism in Hyogo, Japan. He holds a Ph.D. in Language,
Communication, and Culture from Kwansei Gakuin University. His research interests include English
linguistics, semantics, contrastive linguistics, corpus linguistics, and
linguistic landscapes, with particular attention to language use in tourism and
intercultural communication. He is the author of The
Present Perfect in English: From Semantic, Evolutionary, and Contrastive
Perspectives (Kaitakusha,
2021). His recent research focuses on linguistic features of tourism discourse,
online reviews, and language use in public spaces.
Email: fujianliang@gmail.com
ORCID:
0000-0001-6773-8788