De Cock Granger 2021

Text&Talk 2021; 41(5-6): 691–713
Sylvie De Cock* and Sylviane Granger

Stance in press releases versus business
news: a lexical bundle approach
https://doi.org/10.1515/text-2020-0040
Received April 2, 2020; accepted July 19, 2021; published online August 11, 2021
Abstract: Press releases represent a hybrid business genre, which combines an

informational and a promotional communicative purpose. The objective of the
study is to assess the extent to which this duality is reflected in the language used,
and more particularly in the expression of stance, by comparing corporate press
releases with another business genre that is essentially informational, namely
business news reporting. The focus is on lexical bundles, as they have been found
to be a major conveyor of attitudinal and epistemic stance. Relying on the pattern-
matching approach to language, 3-word lexical bundles are extracted from a
1-million-word corpus of press releases (BeRel) and set against those found in a
similar-sized corpus of business news (BeNews). An examination of the key bun-
dles (keyword analysis) in each corpus reveals that the bundles that are distinctive
to press releases differ significantly from those found in BeNews, particularly in the
expression of modal, evaluative and personal stance.
Keywords: business English; keyword; lexical bundle; press release; stance
1 Introduction
Research based on computerized corpora that feature different registers has
shown that, far from being homogeneous, languages are highly heterogeneous,
with markedly different configurations of linguistic features. Nowhere is this
more apparent than in the Longman Grammar of Spoken and Written English
(LGSWE) (Biber et al. 1999), which provides a description of English grammar in
*Corresponding author: Sylvie De Cock, Centre for English Corpus Linguistics, University of
Louvain (UCLouvain), Collège Erasme, 1 Place Blaise Pascal, B-1348, Louvain-la-Neuve, Belgium,
E-mail: sylvie.decock@uclouvain.be. https://orcid.org/0000-0001-9572-4287
Sylviane Granger, Centre for English Corpus Linguistics, University of Louvain (UCLouvain),
Louvain-la-Neuve, Belgium, E-mail: sylviane.granger@uclouvain.be. https://orcid.org/0000-
0002-0047-2142
692 De Cock and Granger
four registers: conversation, fiction, news reportage and academic prose. The
corpus-based analysis shows that “each register has distinctive patterns, asso-
ciated with its particular communicative priorities and circumstances” (Biber
et al. 1999: 24). The authors of the LGSWE were aware, however, that there was
considerable variation within each register and called for “future investigations
of the sub-varieties”, such as editorials and reviews in the case of newspaper
writing (Biber et al. 1999: 17). With the help of dedicated software tools, the
automatic extraction of recurrent sequences of words, often referred to as the
n-gram method, brings out units of discourse that typify registers. This method
was used by the authors of the LGSWE to identify “lexical bundles”, i.e. “bundles
of words that show a statistical tendency to co-occur” (Biber et al. 1999: 989),
such as do you know what I mean in conversation or it has been shown that in
academic writing. As stated by Barbieri (2018: 251), lexical bundles “can reveal a
great deal about the unique linguistic characteristics and communicative func-
tions shaping registers”.
Focused on business English, this study aims to assess the extent to which
lexical bundles can reveal typical features of the sub-variety of press releases as
compared with another business English variety, that of business news reporting.
In line with Bhatia’s (1993) conceptualization of genres, press releases represent a
complex genre that serves a number of communicative purposes (see Section 2.1),
one of which is the promotion of the company that issues them. In view of this
promotional character, press releases are a tool of what Leech (1966: 64) calls
“prestige advertising”, i.e. “advertising which promotes a commercial enterprise”.
Leech’s (1966) book English in advertising concentrates mainly on “advertising
which promotes a product” (p. 64), and prestige advertising is only very briefly
discussed. Our study sets out to complete the picture provided in Leech (1966) by
exploring some of the recurrent linguistic characteristics of press releases as
compared with news reporting.
The paper is organized as follows: Section 2 sets the scene for our empirical
investigation; the focus is on corporate press releases and the linguistic analysis
centres on lexical bundles. The data and methodology used are described in
Section 3. Section 4 presents the findings: the degree of ‘bundleness’ of press
releases compared to business news reports and the key bundles typical of each
genre. The final section discusses the main results, and highlights some of the
lessons learned from the study and future avenues of research which could be
explored.
Stance in press releases versus business news 693
2 Literature review
2.1 Press releases
Texts connected with the domain of business broadly fall into one of the following
two categories: texts used to talk or write about business; and texts used to do
business and communicate with a range of internal and external stakeholders such
as employees, customers, suppliers, investors or the media (Nelson 2000). While
the former cover business-related topics and do not emanate from commercial
organizations, the latter are produced or issued by people who “communicate
using talk or writing in commercial organizations in order to get their work done”
(Bargiela-Chiappini 2007: 3). Business texts represent a wide variety of business
genres, which are generally identified on the basis of communicative purpose(s)
and situation(s) (Bhatia 1993; Koester 2010). Business genres that talk/write about
business include news articles about the world of business, research papers,
and business studies lectures (e.g. Crawford Camiciottoli 2007; Goossens 2013).
Examples of business genres used within the framework of organizations’ business
activities are internal emails, business meetings, earnings calls, press releases,
CEO letters and service encounters (e.g. Handford 2010; Huang and Rose 2018).
The main focus of this paper is on the written business genre of press releases,
which are issued by companies and are an instrument of public relations (PR).
Catenaccio (2008: 11) defines press releases as
relatively short texts resembling news stories and containing what is considered by the issuer
to be newsworthy information; they are generally sent to the journalistic community (but the
intended primary readership has been recently shifting to the general public) with the
purpose of having them picked up by the press and turned into actual news stories, thus
generating publicity, in the conviction that third-party endorsement is the best way to pro-
mote a company’s image and reputation.
Catenaccio’s definition highlights some of the major characteristics of press

releases discussed by researchers who have investigated this genre, namely their
“multipurpose nature” (Bremner 2014: 260) and the fact that they are “meant for
different audiences” (Jacobs 1999: 22). Press releases aim to provide information
about the company (e.g. new products and services, financial results, employ-
ment), persuade journalists that the news is worthy of media coverage, and pro-
mote the company in a favourable light. As mentioned above, because of this
promotional dimension, press releases have an important part to play in what
Leech (1966: 64) calls “prestige advertising”. The multipurpose nature of press
releases has led to their characterisation as a complex hybrid genre (Catenaccio
2008) with both informational and promotional aims. As regards readership, press
releases have traditionally been directly targeted at journalists (to initiate press
coverage) and indirectly targeted at the general public (via any press coverage
based on the releases). However, with the availability of press releases on com-
panies’ corporate websites (Bremner 2018; Strobbe and Jacobs 2005), “press
release writers now have an opportunity to entirely bypass journalistic interven-
tion in reaching the public” (Catenaccio 2008: 15), especially potential and existing
investors (McLaren and Gurǎu 2005).
Genre-oriented research on press releases has examined their move structure
in some detail (Catenaccio 2008; McLaren and Gurǎu 2005) and has identified a
fairly stable standard format (announcement, elaboration, comments, contact
details, editor’s note – McLaren and Gurǎu 2005). In addition, a great deal of
attention has been devoted to the process involved in the production of press
releases by PR officers (Bremner 2014) and to the repackaging of the information
contained in the releases by journalists (Sissons 2012; Van Hout and Macgilchrist
2010) based, among other approaches, on ethnographic and cognitive-
psychological research methods. These studies have made it possible to empiri-
cally explore the complex interaction between PR practitioners, who “manage
[company] news” (cf. “news management”, Jacobs 2018: 181) and journalists, who
“make the news” in the form, for example, of news reports (cf. “newsmaking”,
Jacobs 2018: 181). In particular, they have cast some doubt on journalists’ claims
that they never use press releases and PR officers’ claims that “a lot of their copy
gets recycled verbatim” (Jacobs 2018: 181).
The key features of the language used in press releases discussed in the
literature include features mainly associated with (1) their resemblance to news
stories, which largely results from what Jacobs calls “preformulation” (2006: 201),
i.e. “a news style that requires little or no reworking on the part of the journalists
who receive [the press releases]”, and (2) the promotion of the company.
Jacobs’s work identified the following two linguistic preformulation features,
which can be seen not only to make the recycling of press releases into news
reports straightforward (which should encourage uptake) but also to make the
information in the texts appear more objective: third person self-reference and self-
quotation. These features are in line with the “style of objective reporting”
mentioned by Leech (1966: 65) in his brief discussion of prestige advertising.
Companies tend to refer to themselves in the third person singular (often using the
company’s name), which mirrors the way journalists refer to organizations in their
news reports. Quotes typically feature in the more ‘promotional comments’ move
of press releases, where the CEO and other high-ranking individuals from the very
organization issuing the release can be seen to comment on (aspects of) the
announcement. Such quotes are often referred to as self-quotes, because the
organization is quoting itself. Generally speaking, press releases could be said to
blur the boundary between ‘texts used to talk and write about business’ and ‘texts
used to do business’. While they are clearly used to ‘do business’, as they emanate
from companies and aim to serve their interests, they are preformulated to make it
easier for parts of the texts to be fed into news reports, i.e. into texts that are
typically used to ‘write about business’.
According to Leech (1966: 65), compared with consumer advertising (for
products and services), in prestige advertising “the more brazen forms of eulogy
are absent”. There is, however, widespread consensus that in press releases “the
information is presented in as favorable a light as possible from the corporate
viewpoint” (Pander Maat 2007: 61). The promotional character of press releases,
which stems from their “propagandistic purposes” (Pander Maat 2007: 61), has
been explored from a number of angles including the manner in which journalists
deal with promotional elements when they recycle press releases into news re-
ports. Pander Maat offers the most comprehensive and systematic empirical
analysis of promotional elements in press releases and in news publications based
on these releases. He has identified 13 kinds of promotional element in the press
releases and his is one of the rare studies that includes frequency information
about promotional elements. Some of the most frequent kinds are intensifying
adjectives (important), evaluative adjectives (excellent), property-specifying
adjectives (reliable), intensifying adverbs (considerably) and time adjuncts
(once again).
2.2 Lexical bundles
The idea of extracting and analyzing recurrent sequences of words from corpora
can be traced back to Altenberg’s (1990) study of recurrent word combinations. His
work was critical in identifying the forms and functions of routine formulae in
spoken English, a topic that had until then attracted little attention. However, the
strand of research initiated by Altenberg only truly took off with the introduction of
the construct of lexical bundle by Biber and colleagues (Biber 2009; Biber et al.
1999). Defined as the most frequently recurring sequences of words in a given
register, lexical bundles are similar to Altenberg’s recurrent word combinations
but differ from them in being operationalized on the basis of a specified set of
extraction criteria. The word sequences need to contain at least three words and
reach specified frequency and dispersion thresholds (for more details, see
Section 3.2).
It is important to bear in mind, however, that lexical bundles constitute raw
linguistic material that requires further processing. They only take on their sig-
nificance once they have been categorized in meaningful sets and interpreted.
Biber and colleagues suggested two main categorization schemes, which have
been widely used in lexical bundle research. The structural classification sub-
divides lexical bundles according to their syntactic makeup, distinguishing for
example between verb-based bundles (I don’t think so) and noun-based bundles
(the extent to which). This categorization highlights major differences across reg-
isters. For example, conversation turns out to be characterized by a dominance of
verb-based bundles, while the bundles typical of academic writing tend to be
noun-based (Biber 2009; Biber et al. 1999). The functional classification distin-
guishes between three functions: referential, discourse organizing and stance
(Biber et al. 2004: 384). Referential bundles make direct reference to physical or
abstract entities, or to the textual context itself (corporate law market). Discourse
organizers reflect relationships between prior and coming discourse (on the other
hand). Stance bundles express attitude or assessment of certainty (it is possible to).
Lexical bundles have spawned an impressive body of research. The bulk of it is
focused on academic writing and usually aims to raise novice writers’ awareness of
the patterns typical of academic writing and to help them produce more idiomatic
texts. A range of studies have explored differences in the use of lexical bundles
across disciplines and have uncovered a high degree of discipline-specificity
(Durrant 2017; Hyland 2008). Another very popular research strand compares the
use of bundles by native (L1) and non-native (L2) speakers. On the whole, L2 use of
lexical bundles is characterized by a mixture of over- and underuse, with
learner speech lacking in a range of multiword units that are typical of speech
and L2 argumentative essay writing characterized by an overuse of verb-based
bundles typical of speech (De Cock et al. 1998; Granger 2017; for an overview, see
Granger 2018).
We share Greaves and Warren’s (2010: 216) view that “academic genres have
attracted a disproportionate amount of interest compared with other genres” and
advocate the exploration of lexical bundles in non-academic texts such as texts by
professional communities (lawyers, journalists, businesspeople, etc.), other types
of online communities or the general public. Although the literature relying on this
type of text is still quite limited, some recent studies show its potential (e.g.
Alasmary 2019; Fuster-Márquez 2014; Gaspari 2013). A few studies have compared
different genres of one and the same register. Jablonkai (2009) compares two sets
of EU-related texts: official EU texts and online EU news texts. She finds that,
although the two corpora are made up of written texts and discuss EU topics, their
different communicative functions result in considerable variation in the use of
lexical bundles. Breeze (2013) focuses on four genres of legal English and finds
interesting differences, among other things, in the quantity and quality of verb
bundles. Case law proves to be characterized by the use of epistemic stance ex-
pressions (it seems to me), while in legislation and legal documents (e.g. contracts)
deontic categories (obligation/commission and permission) prevail (must be

accompanied by). Our study falls into this research perspective; it focuses on two
genres within the domain of business English: press releases and news reporting.
All these studies show that lexical bundles are a powerful window on the
patterned nature of registers and genres and the communicative functions they
serve. In our study special attention is paid to stance bundles, which seem
particularly relevant for the exploration of a hybrid genre such as press releases
that is both informational and evaluative.
3 Data and methodology

3.1 Corpora
Our study is based on a recently compiled one-million-word corpus of corporate

press releases (BeRel, Centre for English Corpus Linguistics, UCLouvain),
compared with a one-million-word corpus of business news reports (BeNews,
Centre for English Corpus Linguistics, UCLouvain, Goossens 2014). Table 1, which
draws on Biber et al. (1999: 5), summarizes the main situational differences be-
tween the two genres being studied.
The BeNews corpus, which is used as a reference corpus, was compiled be-
tween 2006 and 2009 and contains 1,864 news reports from the business sections
of mainstream and specialist British and North American newspapers and maga-
zines (e.g. the Financial Times, the International Herald Tribune, Time Magazine,
The Economist). The texts deal with a wide variety of topics such as financial
markets, global trade, inflation, company news and taxation. The BeNews corpus
includes only the main body of the news reports: the texts were stripped of
headlines, leads, dates, journalists’ names, pictures, tables and figures.
Our analytical focus in this paper is the BeRel corpus. It is made up of press
releases collected from the official corporate websites of the top 200 American-based
companies in the 2019 Fortune 500 list (http://fortune.com/fortune500/list/). It was
Table : Main situational differences between press releases and business news reports.
Mode Main communicative Target audience

purpose(s)
Press releases Written Inform, persuade, promote Journalists, investors,

general public
Business news reports Written Inform, evaluate General public
Table : Description of the corpora used in the study.
BeRel BeNews
Tokens ,, ,,

Number of texts , ,
Average length per text  words  words
decided to start from the top of the list, as higher-ranking companies, unlike lower-
ranking companies, have a well-established tradition of providing access to press
releases in dedicated press rooms (Callison 2003). BeRel comprises approximately
5,000 words (between 10 and 15 press releases) per company. As the corpus is
intended to be representative of corporate press releases in general, the 2,559
releases included were issued by a wide range of companies operating in different
sectors (e.g. finance, IT, consumer goods, engineering, pharmaceuticals, construc-
tion, transportation) and cover a variety of company-related topics (e.g. financial
results, product launches, new management, job creation). The corpus is made up
of the main body of the releases (announcement, elaboration and comments, cf.
Section 2.1) without any headlines, leads, dates and places of release, pictures,
tables or figures. Boilerplate elements such as editors’ notes and contact details were
also eliminated from the corpus, for two main reasons: not only are they seen as
peripheral components of press releases (Sissons 2012), but they also tend to be
identical in the various releases published by the same company. The average length
of a release in BeRel is shorter than that of a news report in BeNews (400 words vs.
570 words, cf. Table 2).
It is important to point out that the overwhelming bulk of the research on press
releases carried out to date is essentially qualitative in nature and that empirical
investigations of actual texts make use of rather limited sets of data, the size of
which tends to be given in terms of the number of texts analyzed and not the total
number of words in the samples. Skorczynska Sznajder (2016) is one of the very few
studies that actually mentions the number of tokens in the corpora used, namely
c. 120,000 words in each of the three corpora analyzed. It is often difficult,
therefore, to assess the degree of representativeness of the data used in many
analyses of press releases.
To our knowledge, the one-million-word BeRel corpus is the largest comput-
erized corpus used in research concentrating on press releases in English. Corpus
size is of critical importance in a study that explores lexical bundles. According to
Cortes (2008: 46), it is “highly advisable to work with at least one million words to
identify lexical bundles in a corpus and to draw reliable comparisons when using
more than one corpus”.
3.2 Methodology
There are several methodological approaches to lexical bundles. We adopted

Biber’s (2009: 281) “radical corpus-driven approach”, which has the following
three characteristics: (1) it is based on the actual word forms used in the corpus, not
lemmas; (2) it is based on sequences of words irrespective of their grammatical/
syntactic structure; and (3) it focuses on frequent, recurrent combinations of
words. As regards bundle length, although it is standard to focus on 4-word
bundles, we opted for 3-word bundles as they are much more numerous and we
wanted to base our analysis of stance (cf. Sections 4.2.2 and 4.2.3) on a rich
diversity of bundles. According to Biber (2009: 282) “there are almost 10 times as
many 3-word bundles as 4-word bundles”. In addition, as pointed out by Simpson-
Vlach and Ellis (2010: 509), many of the 3-word bundles have high formulaic value.
Similarly, Gaspari (2013), who based his analysis solely on 4-word bundles, con-
cludes that it would be desirable to widen the analysis to sequences of 3 words (as
well as 5/6-word bundles).
The frequency cut-off point varies considerably in the lexical bundle literature.
Biber (2006: 134) recognizes that “[t]he actual cut-off used to identify lexical
bundles is somewhat arbitrary”. We opted for a minimum frequency of 20 occur-
rences per million words (pmw), which has been found to be high enough to ensure
that the selected sequences are unlikely to be due to chance (Bestgen 2018: 220).
Range was also taken into consideration when retrieving the 3-word bundles from
the BeRel corpus; following Biber et al. (1999), the bundles under study had to
recur in at least 5 different press releases. To extract the bundles and identify those
that are key in each corpus we used the Wordlist and Keywords functions in
WordSmith Tools 6 (Scott 2012; henceforth WST), supplemented with Concord and
Collocate Display to contextualize the bundles.
An examination of lists of automatically extracted lexical bundles reveals that
some bundles are closely connected. For example, the sequences the end of
(405 tokens, BeNews) and end of the (147 tokens, BeNews) actually share some of
their occurrences and can thus be seen to overlap when they occur as part of the
longer sequence the end of the (122 tokens, BeNews). As rightly pointed out by Chen
and Baker (2010: 33), “overlapping word sequences could inflate the results of
quantitative analysis” if they are not dealt with. Very few studies of lexical bundles
based on corpora containing more than 100,000 words adopt a method that
tackles overlapping bundles, however, as this involves highly complex and
time-consuming manual work. This approach was not used here either, as our
study is essentially exploratory and the main focus is not on quantitative findings.
4 Data analysis
4.1 Degree of ‘bundleness’
In view of the stable standard core moves of press releases (announcement,

elaboration and comments), it would not be unreasonable to expect the BeRel
corpus to display a higher degree of ‘bundleness’, i.e. a higher density of lexical
bundle types and tokens, than the BeNews corpus.
The numbers in Table 3 appear to lend support to our hypothesis: BeRel
contains more bundle types, and especially tokens, than BeNews (the results are
statistically significant at p < 0.001 according to chi-square and log-likelihood
tests). These findings and the frequency counts for the top bundles in BeRel (as well
as: 603 and chief executive officer: 486) and in BeNews (one of the: 431 and the end
of: 405) suggest that the bundle types in BeRel display a higher degree of recur-
rence than the bundle types in BeNews, hence the lower bundle type/token ratio.
This tendency is confirmed as we move further down the list of bundles. The 3-word
bundles in BeRel are particularly well dispersed across the corpus: they can be
found in 40 different releases on average, with the top 15 bundles featuring in 210
to 512 texts. The degree of ‘bundleness’ in BeRel has no doubt been boosted by the
large number of different texts displaying a similar move pattern in the corpus.
An examination of the bundle types that occur in only one of the corpora
(exclusive bundles) or in both (shared bundles) reveals that, while the majority of
bundles are shared (61% of the bundle types), there is a larger number of exclusive
bundles in BeRel than in BeNews (cf. Table 4). As is clear from Table 5, and in line
with our observations above, the exclusive bundles in BeRel recur with much
higher frequencies than the exclusive bundles in BeNews.
Table : -word bundle types and tokens in BeRel and BeNews.
BeRel BeNews
Bundle types , ,

Bundle tokens , ,
Bundle type/token ratio . .
Table : Exclusive and shared -word bundle types and tokens in BeRel and BeNews.
Exclusive to BeRel Shared Exclusive to BeNews
Types  , 

Tokens , , ,
Table : Top  exclusive and shared -word bundles in BeRel and BeNews.
Exclusive BeRel Shared Exclusive BeNews
vice president and  as well as per cent to 

today announced that  one of the Bank of England 
for more information  the end of the credit crunch 
announced today that  in the United States is likely to 
learn more about  in the U. said in a 
we are excited  in addition to cent in the 
to learn more  as part of of the last year 
today announced the  and chief executive the financial crisis 
are excited to  is expected to in a statement 
principal amount of  around the world Bank of Scotland 
the securities act  part of the Said it would 
diversity and inclusion  will continue to Royal Bank of 
senior notes due  the number of cent of the 
we are proud  of the year that the government 
aggregate principal announcement  in the past the collapse of 
Some of the most frequent exclusive bundles in BeRel can clearly be connected
with the announcement move and self-quotes from the organizations’ high-
ranking officers (in the comments move), which are typical of press releases. The
BeNews exclusive bundles in the table reflect the reporting dimension of the ar-
ticles as well as the wide range of topics, not exclusively related to company news
(financial markets, the financial crisis, government policies). The top shared
bundles include bundles that can be found in many different written genres:
discourse-structuring bundles and referential bundles denoting time, place and
quantity.
4.2 Key bundles
One particularly efficient method for identifying the distinctive linguistic features
of a given register or genre is the keyword method. Scott (1997: 236) defines a
keyword as “a word which occurs with unusual frequency in a given text” and adds
that “[t]his does not mean high frequency but unusual frequency, by comparison
with a reference corpus of some kind”. As pointed out by Skorczynska Sznajder
(2016: 51), the identification of keywords “can enable the detection of recurrent
patterns of meaning, which would otherwise be difficult to access and recognize,
especially in a horizontal reading of texts in a corpus”. The notion of keyness is not
limited to single words: it also applies to word sequences. Both keywords and key
phrases are highly revealing of the distinctiveness of a given register or genre in
terms of the language used and the communicative functions it serves: they “may
be shown to be indicative of the writer’s position and identity, as well as of the
discourse community, with its values and beliefs about the subject matter and the
genre that characterize it” (Bondi 2010: 7).
Using the keyword function in WordSmith Tools, we identified the key bundles
in press releases relative to news reports and then reversed the procedure and
identified the key bundles in news reports with press releases as the reference
corpus.1 The extracted key bundles will be referred to as BeRel bundles and
BeNews bundles respectively. We limited the analysis to bundles that have a
minimum frequency of 20 and that reach a significant log likelihood keyness value
(p < 0.01). In what follows we give a general overview of the structural categori-
zation of the key bundles (Section 4.2.1) before zooming in on three categories of
stance bundle (Sections 4.2.2 and 4.2.3).
4.2.1 Structural categorization
We distinguish three main structural categories: VP-based bundles which contain

a verb component (appears to have, to cut costs); NP-based bundles which include
any noun phrase (+ post-modifiers) (general corporate purposes, a variety of); and
PP-based bundles which start with a preposition followed by a noun or noun
fragment (against the dollar, in return for). The category ‘Other’ contains bundles
that do not fit into any of these three categories (many of the).
The first point to note in Table 6 is that BeRel has far more key bundles than
BeNews, which suggests that corporate press releases are a more distinctive reg-
ister than business news reports. As regards categorization, NP-based bundles
dominate in the two registers: they account for almost half of the key bundles. This
is not very surprising, as domain-specific registers feature a wide range of
specialized multiword terms. The vast majority of the bundles are topic-
dependent, but they are quite different in the two corpora. In BeRel they largely
1 Although keyword analysis often relies on a reference corpus that is larger than the foreground
corpus, this is not a necessary condition (cf., for example, studies by Bondi 2010 and Malavasi and
Mazzi 2010).
Table : Structural categorization of key bundles.
Bundle type BeRel bundles BeNews bundles
NP-based  (%)  (%)

VP-based  (.%)  (%)
PP-based  (.%)  (%)
Other  (%)  (%)
Total  (%)  (%)
reflect the self-centered company perspective typical of the genre. A whole series of
bundles refer to the companies’ senior figures, who are quite systematically quoted
in press releases (chief executive officer, board of directors), and to companies’
financial deals and strategies (customary closing conditions, net proceeds from).
BeNews features many key NP-based bundles related to financial institutions
(Bank of England) and to stock market news (per cent of). Temporal bundles are
particularly prevalent among the BeNews key bundles, with as many as 40 bundles
containing the words day, week, month or year (final three months), which is hardly
surprising given that time is one of the key elements of news reporting (Bell 1991).
PP-based bundles account for a slightly smaller proportion of BeRel bundles
than BeNews (14 vs. 19%). PP-based BeRel bundles are mainly topic-related
(connected with companies’ stakeholders and finances: for our customers, of
common stock), with the exception of a few complex prepositions (in support of).
Here too, temporal bundles prove to be characteristic of BeNews (in recent months).
The preponderance of topic-related bundles boosts Biber et al.’s referential cate-
gory, which dominates in the two key bundle lists. However, the VP-based category
accounts for a sizeable proportion of the key bundles, with a slightly higher pro-
portion in BeRel than BeNews (34 vs. 28%), which counterbalances the lower
proportion of PP- and NP-based bundles. Unlike the other two categories, which
prove mainly to have a referential function, the VP category plays a key role in the
expression of stance. It allows speakers and writers to modulate their message in
light of the addressee and the situation thanks to a range of choices they can make
in terms of tense and aspect, voice and modality, as well as of types of subject and
complement. We have chosen to focus more particularly on stance bundles since
stance is an important element in both press releases and news reports, given their
main shared and specific communicative purposes: to inform (both genres), to
persuade and promote (press releases) or to evaluate (news reports). In the
following two sections we report our findings on the types of stance that emerge
from our data as particularly distinctive: modal, personal and evaluative.
4.2.2 Modal stance
The study of modal stance is often limited to core modals and semi-modals, but in a
discourse-based approach to modality it is advisable also to include the “large
number of other relatively fixed expressions with meanings similar to the modal
auxiliaries; for example, want to, be able to, be obliged to, be likely to, be willing to”
(Biber et al. 1999: 484).
Table 7 gives the number of modal key bundles in each corpus, distinguishing
between those that contain a modal auxiliary (will serve as) or a semi-modal
(is going to) from those that contain a modal expression (are likely to).
Table : Modal and non-modal key bundles in BeRel and BeNews.
Bundle type BeRel BeNews
Non-modal  (%)  (%)

Core modals and semi- modals  (%)  (%)
Modal expressions  (%)  (%)
Total (%)  (%)
The table shows that, while the non-modal VP-based bundles dominate in the
two corpora, this dominance is particularly striking in BeRel (89 vs. 68%). BeNews
contains many more modal key bundles than BeRel (32 vs. 11%). Several of these
bundles contain modal expressions (e.g. seem to, be thought to, be unlikely to),
which shows the benefit of extending the notion of modality beyond the traditional
categories of core and semi-modals.
A detailed analysis of all the modal markers is beyond the scope of our study
but a few points are worth noting. Overall, the table shows how different the modal
bundles are in the two genres. Not only are they more frequent in BeNews than in
BeRel but they are also quite different. In BeRel the main modal is the core auxiliary
will, which may be a pure future (see example 1) but in many cases expresses the
company’s strong commitment (see example 2). Volitional will is described by
Leech (2004: 87) as follows: “[o]ccurring mainly with first-person subjects (except
in indirect speech), will in this sense can convey a promise, a threat, an offer or a
shared decision. The volitional element is reinforced by a feeling that in the act of
speaking, a decision is made, and that the fulfilment of the intention is guaran-
teed”. While the two main meanings of will – futurity and volition – can be clearly
distinguished in some contexts, Leech (2004: 56) rightly observes that in general
they “are so closely intermingled that it is difficult to separate them”. The second
most frequent auxiliary is shall, which has a much more limited use as it is only
found in the disclaimers included at the end of some press releases (see example 3).
(1) The Company will host a conference call on Thursday, May 18, 2017 (…) [BeRel]
(2) Northwestern Mutual Future Ventures Fund II will continue to advance the
company’s investment strategy (…) [BeRel]
(3) This press release shall not constitute an offer to sell or the solicitation of an
offer to buy any securities, nor shall there be any sale of these securities (…)
[BeRel]
In BeNews the most frequent modal is the core auxiliary would, typically used in a
that-clause after a speech verb (often in the simple past) to report a company’s or a
government’s intentions, promises and claims. This is evidenced by the two most
frequent key bundles – said it would and that it would – and confirmed by the
Collocate function of WST, which gives say as the main lexical collocate of would.
But other uses of would are also found, often in hedged statements (example 4).
The presence of have to as the second most frequent modal and the total absence of
must are in line with the declining use of must to express obligation noted by Leech
et al. (2009: 87–89).
Probably the most interesting finding that emerges from our analysis is the
number and variety of modal expressions used in BeNews and their total absence
in BeRel. All of them signal epistemic stance expressed by either adjectives (in
particular, likely and unlikely), copulas (appear and seem) or passive verbs such as
be expected to. The bundle is likely to actually has the highest keyness value of all
the modal bundles in BeNews (see example 5). Most of these modal expressions
would not have been investigated in a corpus-based study relying solely on the
traditional core modals and semi-modals.
(4) It would be hard to find a large industry in America which is adding jobs. For
that matter, it would be hard to find a large American company making net
additions to its payroll. [BeNews]
(5) The MPC might cut by another half-point, but this is likely to be coupled with the
unprecedented step of “printing money”. [BeNews]
4.2.3 Personal and evaluative stance
Biber et al.’s (1999: 1001–1024) analysis of 3- and 4-word bundles in conversation

and academic writing shows that conversation is characterized by personal stance,
with frequent use of the first person pronoun I and second person pronoun you,
while academic writing displays a more impersonal stance, characterized by the
use of anticipatory it-structures and bundles containing passive verbs. In her study
of epistemological positioning in English news discourse, Bednarek (2006: 652)
stresses the importance of identifying the source of a proposition as more objective
or more subjective. She illustrates the former with impersonal structures marked
with it (it is certain) and the latter with phrases marked with the pronoun I (I’m
certain that) (p. 654). In the remainder of this section we distinguish between
bundles with the 1st person pronouns I and we (I look forward to, we continue to)
and the 3rd person pronouns it and there (it could be, there was no) in order to
ascertain the extent to which the different communicative functions of press re-
leases and news reporting are manifested in the pronoun-headed bundles.
Table 8 shows that in BeRel we, and to a lesser extent I, are commonly used in
formulaic chunks, which almost invariably occur as part of company self-quotes.
Interestingly, the majority of them express not only personal stance but also
evaluative stance. This is due to the presence of a large number of we are + adjec-
tive chunks featuring highly positive adjectives: pleased, proud, committed,
excited, honored, thrilled, delighted, confident (see examples 6 and 7).
(6) We are excited to be developing the infrastructure necessary to make this option
available (…),” said Bob Sulet, CBRE’s president and chief executive officer.
[BeRel]
(7) “(…) We are proud of our performance on key environmental, social and
governance issues (…),” Tim Casey, senior vice president of LNG, commented
(…) [BeRel]
In his study of promotional language in press releases, Pander Maat (2007: 73) has
found that quotes tend to contain more promotional elements than non-quoted
material. As a result, he suggests that “quotes may be a device to smuggle some
promotion into the news report”, as news reports typically include quotes and
“journalists have less freedom in editing quotes than editing other sentences”
(Pander Maat 2007: 73). The distinctive dominance of positively laden adjectives in
press releases also emerges clearly from a comparison of the top 10 key adjectives
in the two corpora (see Table 9). While the adjectives in BeRel are overwhelmingly
positive, the top key adjectives in BeNews express a range of different meanings,
few of which are clearly positive.
Table : First and third person pronoun key bundles in BeRel and BeNews.
Person pronoun BeRel BeNews
I  
we  
it  
there  
Total  
Table : Top  key adjectives in BeRel and BeNews.
Rank Press Releases Business News
 excited () last ()

 available () biggest ()
 innovative () British ()
 leading () economic ()
 proud () bad ()
 global () worst ()
 unique () likely ()
 effective () big ()
 committed () European ()
 strategic () foreign ()
In BeNews key bundles, personal stance is totally absent. While the pronoun it
may refer to some specific entity, it can also be the impersonal pronoun it used in
anticipatory it structures, for example. Unfortunately, apart from a few bundles
which are unambiguously personal (it plans to) or impersonal (it is hard), the
majority of the it bundles are too short for it to be possible to distinguish between
the two uses (it is not, it could be, it was not). However, it is possible to disambiguate
the bundles by using the concordancing facility of WordSmith Tools. For example,
searching for the sequence it is followed by the word that in a window of 10 words
to the right of the search string allowed us to identify a wide variety of anticipatory
it clauses in BeNews. This method brings out the diversity of stance expressions
used by news writers and is therefore a highly valuable complement to lexical
bundle extraction: the lexical bundle approach identifies what is fixed in
Table : Examples of anticipatory it bundles in BeNews.
it is abundantly clear that

it is almost certain that
it is appalling that
it is broadly true that
it is widely agreed that
it is hardly surprising that
it is comforting to know that
it is equally wrong to assume that
it is not a good sign that
it is particularly alarming that
it is not far-fetched to believe that
language, while the more qualitative, fine-grained concordance-based analysis

brings out the rich variety of means used by news writers to modulate their
statements (it is broadly true that) and to express evaluation (it is appalling that)
and epistemological positioning in an impersonal manner (it is almost certain that)
(see Table 10 for a few representative examples).
5 Discussion and conclusion

Our study fully confirms Biber and Barbieri’s (2007: 265) statement that “each
register employs a distinct set of lexical bundles, associated with the typical
communicative purposes of that register”. It does so on the basis of two business
English genres – press releases and news reporting. It shows the full power of a
corpus-driven approach to language, in particular when it is combined with the
keyword procedure, which, as observed by Groom (2010: 73), “offers researchers a
powerful means of gaining access to the meanings and values of specialized
discourse communities”.
Our analysis shows that the two genres are characterized by the use of quite
distinct lexical bundle types which reflect their respective communicative pur-
poses: mainly informational in the case of news reporting and both informational
and persuasive in that of press releases. Despite the fact that press releases are to
some extent preformulated to resemble news reports, there are marked differ-
ences in the way the actual information is presented. The impersonal framing in
news reports (e.g. there bundles, anticipatory it bundles) stands in stark contrast
to the more personal framing in news releases (e.g. we and I bundles). The
impersonal framing is in line with journalists’ reporting role, as they are expected
both to distance themselves from, and to provide an external evaluation of, the
news they are reporting on. The more personal self-centered approach of press
releases combined with the routine foregrounding of firm intentions/promises
(e.g. will bundles) and the focus on positive elements (e.g. bundles with posi-
tively laden adjectives) tie in well with their promotional role (among their
stakeholders, potential investors and the general public). The persuasive func-
tion of press releases (to persuade journalists to cover the news) is reflected in the
quotes from high-ranking figures invariably included in the releases (e.g. we are
excited) and the emphasized ‘superlativeness’ (e.g. innovative or unique) to
highlight the ‘news values’ of the events (Bell 1991). Our findings also confirm
previous research based on much smaller sets of data regarding the move
structure of press releases. The structure is remarkably stable and is manifested
in highly routinized bundles (e.g. for the announcement move: today announced
that/the).
As also shown by other studies, the lexical bundle approach proves to be a

very powerful discovery procedure, which brings to light formulaic units at the
interface between discourse, syntax and lexis that would not easily be acces-
sible via other means. Admittedly, it only highlights fully fixed combinations of
contiguous words. However, our study has demonstrated that these can be used
as prompts to explore the range of variants they display, using some of the
many other functions offered by linguistic analysis software tools, in particular
concordancing and automatic collocate extraction. Although the findings
discussed in this paper are a worthwhile starting point, it is essential that the
leads provided by our lexical bundle approach be pursued in more in-depth
investigations so as to paint a fuller picture of the discursive features and
practices that characterize press releases (compared with news reports). In
addition, future research is needed to determine the extent to which the
quantitative tendencies observed still apply when overlapping bundles (cf.
Section 3.2) are taken into account.
Lexical bundle research has contributed to demonstrating the ubiquity and
importance of formulaic language in a range of registers and genres. So far,
however, it has had very little impact on the teaching and learning of languages in
spite of many calls for integration (see for example Mhedhbi 2014). Lexical bundles
are a major marker of stance and they therefore deserve to be focused on alongside
purely grammatical or lexical aspects in courses for both general and specific
purposes. As noted by Leech (2001), to achieve this aim, we need corpora that
closely match the future communicative needs of learners. Business English is
often presented as a homogeneous variety with little regard for the many different
genres it comprises. This may lead to misguided advice. For example, Roberts
(n.d.: 91), while recognizing the difference in communicative purpose between
news writing and public relations writing such as press releases, provides the
following general advice: “Write the body of the press release using news writing
techniques and style”.2 As our findings demonstrate, press releases call for specific
writing strategies in keeping with their dual – information and promotion –
objective. The extraction and analysis of lexical bundles is one way of uncovering
these strategies, but further research is necessary to form a fuller picture. There is
scope for further studies using the same approach but based on different corpora
and relying on different parameters (in terms of bundle size, frequency threshold,
etc.). As they span the fields of grammar, lexis, discourse and pragmatics, lexical
bundles call for a highly diversified body of research if they are to be interpreted in
meaningful ways. As evidenced by our study, the rich, multi-faceted research
2 https://ohiostate.pressbooks.pub/stratcommwriting/chapter/press-release-structure-and-
format/.
carried out by Geoffrey Leech in areas as diverse as corpus linguistics, grammar,

advertising, semantics, pragmatics and language teaching has been a major
source of inspiration for our research. We gratefully dedicate our study to his
memory.
References
Alasmary, Abdullah. 2019. Lexical bundles in contract law texts: A corpus-based exploration and
implications for legal education. International Journal of English Linguistics 9(2). 244–257.
Altenberg, Bengt. 1990. Speech as linear composition. In Caie Graham, Arnt Haastrup,
Lykke Jakobson, Jorgen Erik Nielson, Jorgen Sevaldsen, Henrik Specht & Arne Zettersten
(eds.), Proceedings from the fourth Nordic conference for English studies, 133–143.
Copenhagen: University of Copenhagen.
Barbieri, Federica. 2018. I don’t want to and don’t get me wrong: Lexical bundles as a window to
subjectivity and intersubjectivity in American blogs. In Joanna Jopaczyk & Jukka Tyrkkö (eds.),
Applications of pattern-driven methods in corpus linguistics, 251–275. Amsterdam:
Benjamins.
Bargiela-Chiappini, Francesca, Catherine Nickerson & Brigitte Planken. 2007. Business discourse.
Basingstoke: Palgrave Macmillan.
Bednarek, Monika. 2006. Epistemological positioning and evidentiality in English news
discourse: A text-driven approach. Text & Talk 26(6). 635–660.
Bell, Allan. 1991. The language of news media. Oxford: Blackwell.
Bestgen, Yves. 2018. Evaluating the frequency threshold for selecting lexical bundles by means of
an extension of the Fisher’s exact test. Corpora 13(2). 205–228.
Bhatia, Vijay Kumar. 1993. Analysing genre: Language use in professional settings. London:
Longman.
Biber, Douglas. 2006. University language. A corpus-based study of spoken and written registers.
Amsterdam: Benjamins.
Biber, Douglas. 2009. A corpus-driven approach to formulaic language in English: Multi-word
patterns in speech and writing. International Journal of Corpus Linguistics 14(3). 275–311.
Biber, Douglas & Federica Barbieri. 2007. Lexical bundles in university spoken and written
registers. English for Specific Purposes 26. 263–286.
Biber, Douglas, Susan Conrad & Viviana Cortes. 2004. If you look at…: Lexical bundles in
university teaching and textbooks. Applied Linguistics 25(3). 371–405.
Biber, Douglas, Stig Johansson, Geoffrey Leech, Susan Conrad & Edward Finegan. 1999. Longman
grammar of spoken and written English. London: Longman.
Bondi, Marina. 2010. Perspectives on keywords and keyness. An introduction. In Marina Bondi &
Mike Scott (eds.), Keyness in texts, 1–18. Amsterdam: Benjamins.
Bremner, Stephen. 2014. Genres and processes in the PR industry: Behind the scenes with an
intern writer. International Journal of Business Communication 51(3). 259–278.
Bremner, Stephen. 2018. Workplace writing. Beyond the text. London: Routledge.
Breeze, Ruth. 2013. Lexical bundles across four legal genres. International Journal of Corpus
Linguistics 18(2). 229–253.
Callison, Coy. 2003. Media relations and the Internet: how Fortune 500 company web sites assist
journalists in news gathering. Public Relations Review 29. 29–41.
Catenaccio, Paola. 2008. Press releases as a hybrid genre: Addressing the informational/
promotional conundrum. Pragmatics 18(1). 9–31.
Chen, Yu-Hua & Paul Baker. 2010. Lexical bundles in L1 and L2 academic writing. Language,
Learning and Technology 14(2). 30–49.
Cortes, Viviana. 2008. A comparative analysis of lexical bundles in academic history writing in
English and Spanish. Corpora 3(1). 43–57.
Crawford Camiciottoli, Belinda. 2007. The language of business lectures. Amsterdam: Benjamins.
De Cock, Sylvie, Sylviane Granger, Geoffrey Leech & Tony McEnery. 1998. An automated approach
to the phrasicon of EFL learners. In Sylviane Granger (ed.), Learner English on computer,
67–79. London: Addison Wesley Longman.
Durrant, Philip. 2017. Lexical bundles and disciplinary variation in university students’ writing:
Mapping the territories. Applied Linguistics 38(2). 165–193.
Fuster-Márquez, Miguel. 2014. Lexical bundles and phrase frames in the language of hotel
websites. English Text Construction 7(1). 84–121.
Gaspari, Federico. 2013. A phraseological comparison of international news agency reports
published online: Lexical bundles in the English-language output of ANSA, Adnkronos,
Reuters and UPI. VariEng Studies in Variation, Contacts and Change in English 13.
Goossens, Diane. 2013. Assessing corpus search methods in onomasiological investigations:
exploring quantity approximation in business discourse. In Hilde Hasselgård, Jarle Ebeling &
Signe Oksefjell Ebeling (eds.), Corpus perspectives on patterns of lexis, 271–292.
Amsterdam: John Benjamins.
Goossens, Diane. 2014. Quantity approximation in business language. A contrastive, corpus-
driven approach (Dutch, English, French). Unpublished PhD Thesis. Louvain-la-Neuve:
Université catholique de Louvain.
Granger, Sylviane. 2017. Academic phraseology: A key ingredient in successful L2 academic
literacy. In Ruth Vatvedt Fjeld, Kristin Hagen, Birgit Henriksen, Sofie Johannson, Sussi Olsen
& Julia Prentice (eds.), Academic language in a Nordic setting: Linguistic and educational
perspectives, Oslo Studies in Language, 9(3), 9–27. Oslo: University of Oslo.
Granger, Sylviane. 2018. Formulaic sequences in learner corpora: Collocations and lexical
bundles. In Anna Siyanova-Chanturia & Ana Pellicer-Sanchez (eds.), Understanding
formulaic language: A second language acquisition perspective, 228–247. New York:
Routledge.
Greaves, Chris & Martin Warren. 2010. What can a corpus tell us about multi-word units? In
Anne O’Keeffe & Michael McCarthy (eds.), The Routledge handbook of corpus linguistics,
212–226. London: Routledge.
Groom, Nicholas. 2010. Closed-class keywords and corpus-driven discourse analysis. In
Marina Bondi & Mike Scott (eds.), Keyness in texts, 59–78. Amsterdam: Benjamins.
Handford, Michael. 2010. The language of business meetings. Cambridge: Cambridge University
Press.
Huang, Ying & Kate Rose. 2018. You, our shareholders: metadiscourse in CEO letters from Chinese
and Western banks. Text & Talk 38(2). 167–190.
Hyland, Ken. 2008. As can be seen: Lexical bundles and disciplinary variation. English for Specific
Purposes 27. 4–21.
Jablonkai, Réka. 2009. “In the light of”: A corpus-based analysis of lexical bundles in two
EU-related registers. WoPaLP 3. 1–16.
Jacobs, Geert. 1999. Preformulating the news: An analysis of the metapragmatics of press
releases. Amsterdam: John Benjamins.
Jacobs, Geert. 2006. The dos and don’ts of writing press releases (and how learners act upon
them). In Paul Gillaerts & Philip Shaw (eds.), The map and the landscape: Norms and
practices in genre, 199–218. Bern: Peter Lang.
Jacobs, Geert. 2018. Organizations and corporate communication: linguistic ethnography in the
newsroom. In Colleen Cotter & Daniel Perrin (eds.), The Routledge handbook of language and
media, 178–189. London: Routledge.
Koester, Almut. 2010. Workplace discourse. London and New York: Continuum.
Leech, Geoffrey. 1966. English in advertising: A linguistic study of advertising in Great Britain.
London: Longman.
Leech, Geoffrey. 2001. The role of frequency in ELT: New corpus evidence brings a re-appraisal.
Foreign Language Teaching and Research 33(5). 328–339.
Leech, Geoffrey. 2004. Meaning and the English verb, 3rd edn. Harlow: Pearson Education.
Leech, Geoffrey, Marianne Hundt, Christian Mair & Nicholas Smith. 2009. Change in contemporary
English: A grammatical study. Cambridge: Cambridge University Press.
Malavasi, Donatella & Davide Mazzi. 2010. History v. marketing: Keywords as a clue to disciplinary
epistemology. In Marina Bondi & Mike Scott (eds.), Keyness in texts, 169–184. Amsterdam:
Benjamins.
McLaren, Yvonne & Cǎlin Gurǎu. 2005. Characterising the genre of the corporate press release.
LSP and Professional Communication 5(1). 10–30.
Mhedhbi, Malek. 2014. Lexical bundles and the construction of an academic voice in business
writing. Advances in Language and Literary Studies 5(6). 1–9.
Nelson, Mike. 2000. A corpus-based study of the lexis of business English and business English
teaching materials. Unpublished PhD thesis. Manchester: University of Manchester.
Pander Maat, Henk. 2007. How promotional language in press releases is dealt with by journalists:
Genre mixing or genre conflict. Journal of Business Communication 44(1). 59–95.
Roberts, Jasmine. n.d. Writing for strategic communication industries. Columbus: Ohio State
University Pressbooks. https://ohiostate.pressbooks.pub/stratcommwriting/.
Scott, Mike. 1997. PC analysis of key words – and key key words. System 25(2). 233–245.
Scott, Mike. 2012. WordSmith Tools version 6. Stroud: Lexical Analysis Software.
Simpson-Vlach, Rita & Nick C. Ellis. 2010. An academic formulas list: New methods in phraseology
research. Applied Linguistics 31(4). 487–512.
Sissons, Helen. 2012. Journalism and public relations: A tale of two discourses. Discourse &
Communication 6(3). 273–294.
Skorczynska Sznajder, Hanna Teresa. 2016. A comparative study of keywords in English-language
corporate press releases from European companies: insights into discursive practices.
Discourse and Interaction 9(1). 49–64.
Strobbe, Ilse & Geert Jacobs. 2005. E-releases: A view from linguistic pragmatics. Public Relations
Review 31. 289–291.
Van Hout, Tom & Felicitas Macgilchrist. 2010. Framing the news: An ethnographic view of business
newswriting. Text & Talk 30(2). 169–191.
Bionotes
Sylvie De Cock
Centre for English Corpus Linguistics, University of Louvain (UCLouvain), Louvain-la-Neuve,
Belgium
sylvie.decock@uclouvain.be
https://orcid.org/0000-0001-9572-4287
Sylvie De Cock is Professor of English language, linguistics and business communication at the
University of Louvain (UCLouvain). She conducts her research at the Centre for English Corpus
Linguistics (founded by Sylviane Granger). Her interests include corpus linguistics, phraseology,
English for specific purposes (business communication), learner corpus research (spoken and
written learner corpora) and pedagogical lexicography.
Sylviane Granger
Centre for English Corpus Linguistics, University of Louvain (UCLouvain), Louvain-la-Neuve,
Belgium
sylviane.granger@uclouvain.be
https://orcid.org/0000-0002-0047-2142
Sylviane Granger is Professor Emerita of English Language and Linguistics at the University of
Louvain (UCLouvain). In 1990 she launched the first large-scale learner corpus project, the
International Corpus of Learner English, and since then has played a key role in defining the
different facets of the field of learner corpus research. Her current research interests focus on the
analysis of phraseology in native and learner language and its integration into reference and
instructional materials.

De Cock Granger 2021

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

De Cock Granger 2021

Uploaded by

Copyright:

Available Formats

Text&Talk 2021; 41(5-6): 691–713

Sylvie De Cock* and Sylviane Granger

Abstract: Press releases represent a hybrid business genre, which combines an

Keywords: business English; keyword; lexical bundle; press release; stance

Catenaccio’s definition highlights some of the major characteristics of press

2.2 Lexical bundles

deontic categories (obligation/commission and permission) prevail (must be

3 Data and methodology

Our study is based on a recently compiled one-million-word corpus of corporate

Mode Main communicative Target audience

Press releases Written Inform, persuade, promote Journalists, investors,

Table : Description of the corpora used in the study.

Tokens ,, ,,

There are several methodological approaches to lexical bundles. We adopted

In view of the stable standard core moves of press releases (announcement,

Table : -word bundle types and tokens in BeRel and BeNews.

Bundle types , ,

Exclusive to BeRel Shared Exclusive to BeNews

Types  , 

Exclusive BeRel Shared Exclusive BeNews

vice president and  as well as per cent to 

4.2 Key bundles

4.2.1 Structural categorization

We distinguish three main structural categories: VP-based bundles which contain

Table : Structural categorization of key bundles.

Bundle type BeRel bundles BeNews bundles

NP-based  (%)  (%)

4.2.2 Modal stance

Table : Modal and non-modal key bundles in BeRel and BeNews.

Bundle type BeRel BeNews

Non-modal  (%)  (%)

4.2.3 Personal and evaluative stance

Biber et al.’s (1999: 1001–1024) analysis of 3- and 4-word bundles in conversation

Person pronoun BeRel BeNews

Table : Top  key adjectives in BeRel and BeNews.

Rank Press Releases Business News

 excited () last ()

Table : Examples of anticipatory it bundles in BeNews.

it is abundantly clear that

language, while the more qualitative, ﬁne-grained concordance-based analysis

5 Discussion and conclusion

As also shown by other studies, the lexical bundle approach proves to be a

carried out by Geoffrey Leech in areas as diverse as corpus linguistics, grammar,

You might also like