Published vs. Postgraduate Writing in Applied Linguistics: The Case of Lexical Bundles

Introduction
Lexical  bundles,  also  known  as  clusters  and  chunks  (Hyland,  2008a,  2008b),  were  first
introduced  and  defined  by  Biber,  Johansson,  Leech,  Conrad,  and  Finegan  (1999).  They
referred  to  lexical  bundles  as  recurrent  expressions,  regardless  of  their  idiomaticity,  and
regardless  of  their  structural  status.  More  importantly,  they  considered  frequency  as  the
defining characteristic of bundles; in order for a word combination (e.g. on the other hand, at
the same time, it is necessary to, etc.) to count as a bundle, it must occur at least twenty times
in  a  corpus  made  of  one  million  words  with  the  additional  requirement  that  this  rate  of
occurrence be realized in at least five different texts to guard against idiosyncratic uses.
Lexical  bundles  are  identified  on  the  basis  of  frequency  and  breadth  of  use  (Cortes,
2002,  2004).  Fixedness  in  form  (e.g.,  on  the  basis  of  not  on  a  basis  of)  and  non-idiomatic
meaning  are  other  properties  of  bundles.  Among  other  registers,  lexical  bundles  have  been
found  to  be  an  important  part  of  academic  discourse  too  (Biber  et  al,  1999).Such  word
sequences have been classified structurally (Biber et al, 1999; Biber, Conrad & Cortes, 2004;
Biber,  2006;  Jalali,  Eslami  Rasekh  &  Tavangar  Rizi,  2008,  2009)  as  well  as  functionally
(Cortes, 2002, 2006; Biber & Barbieri, 2007; Hyland, 2008a, 2008b; Jalali, 2009, 2013; Jalali
& Ghayoumi, 2010). Lexical bundles can serve a wide range of discursive functions such as
organization  of  discourse,  expression  of  stance,  and  reference  to  textual  or  external  entities
(Biber  &  Barbieri,  2007;  Jalali,  2013).  Some  studies  conducted  in  this  regard  are  briefly
reviewed here.
Since  1999,  a  number  of  studies  have  been  specifically  launched  to  explore  possible
differences and\or similarities in the use of bundles between a few disciplinary fields (Cortes,
2002, 2004; Hyland, 2008a, 2008b), registers, such as conversation, fiction, news, academic
prose,  classroom  teaching  and  non-conversational  speech  (Biber  et  al,  1999;  Biber  et  al,
2004, Biber & Barbieri, 2007), genres (Hyland, 2008b; Jalali, 2013), and different degrees of
writing expertise (Cortes, 2002, 2004;  Jalali, 2009; Jalali et al., 2008, 2009).
Overall,  these  studies  have  indicated  that  lexical  bundles  are  strong  discipline,  genre,
and register discriminators. This means that apart from some overlaps, each discipline, genre,
or register draws on its own set of bundles to organize its discourse, express stance, and refer
to  different  parts  of  the  evolving  text  or  elements  outside  the  text.  The  findings  have  also
stressed that many lexical bundles favored by experts in a given disciplinary area may not be
used  by  novices  who  could  be  students  or  developing  writers  with  varying  degrees  of
language proficiency and disciplinary expertise.
Interestingly,  there  is  also  usually  a  correlation  between  the  structural  type  of  bundles
and the  function  they  serve  in  the  discourse  (Biber  et  al,  2004);  for  example,  anticipatory  it
bundles (e.g. it should be noted, it can be seen), the subject of the present study, are usually 
 
used to act as metadiscourse elements (Hyland, 2000, 2008a, 2008b) or expressions of stance
(Biber, 2006). Biber et al. (1999) have shown that it clauses followed by either to (as in it is
important to note that this relationship may always be true) or that (as in it is clear that this
policy  is  unlikely  to  lead  to  fruitful  results)  are  common  in  academic  writing  and  their
relatively  frequent  presence  has  been  substantiated  in  a  range  of  academic  genres  (Hewings
& Hewings, 2002). 
The study of this structural group of lexical bundles can be important for two reasons.
First,  there  is  some  evidence  to  suggest  that  for  many  non-natives,  this  structure  can  pose
serious degrees of difficulty, mostly because of the absence of an anticipatory  it structure in
some languages (Jacobs, 1995, Hewings & Hewings, 2002). Second, due to the importance of
this  structure  as  a  metadiscursive  element  or  a  stance  expression,  it  can  be  important  to
identify the range of interpersonal meanings conveyed by such word clusters as it bundles are
usually good means by which writers can express their opinions, evaluate the subject matter,
and engage with readers (Hewings & Hewings, 2002).
According  to  Hewings  and  Hewings  (2002),  lexical  bundles  starting  with  an
anticipatory  it  have  four  metadiscoursal  or  interpersonal  roles:  hedges  (showing  speaker  or
writer's  tentativeness  and  uncertainty  about  the  following  proposition),  attitude  markers
(expressing writer's attitude toward the content), emphatics (stressing writer's certainty  about
the  force,  and  the  credibility  of  the  propositional  meaning),  and  attribution  (convincing  the
reader through a general or specific reference). The review of the literature showed that very
few  studies  have  focused  on  the  use  of  anticipatory  it  bundles  within  some  key  genres  of
academy (see Hewings & Hewings, 2002; Hyland, 2008a). Especially important is that there
is the scarcity of studies addressing specific phraseological practices in different disciplinary
areas, especially with an aim to describe and explain possible differences and/or similarities
between experts and novices in their use of these word combinations in their respective high-stakes genres.
The purpose of this study was to compare the use of one structural class of bundles in
some key written academic genres of one disciplinary area of applied linguistics through the
use  of  two  corpora  of  academic  writing.  The  assumption  was  that  exploring  possible
variations in the use of such word combinations across genres could be a good contribution to
a  better  understanding  of  phraseological  preferences  and  practices  in  different  discourse
communities. 
 

More  specifically,  the  study  probed  the  use  of  anticipatory  it  lexical  bundles  in  two
genres of applied linguistics. Applied linguistics was selected as the discipline of interest for
two reasons: (1) it has not been subject to rigorous analysis in terms of such clusters and (2),
raising awareness  of  genre  features through  such  studies  can become  part  of  its  disciplinary
content.  Accordingly,  two  corpora  of  research  articles  and  postgraduate  writing  in  applied
linguistics were employed to find the extent to which these two academic genres in a single
disciplinary area are similar to or different from each other. At the same time, by comparing
the  two  genres  of  applied  linguistics,  this  study  attempted  to  show  the  extent  to  which
students' use of anticipatory it bundles could be compared to that of published writers. 
 
The study
Research questions
This study, therefore, addressed the following questions:
1. What are the most frequent four-word anticipatory it lexical bundles in applied linguistics
published and postgraduate writing?
2. To what extent is there evidence to support similarity or contrast in the range, frequency,
and function of anticipatory it lexical bundles across the two genres?  
 
Corpora
Two  corpora  were  used  in  this  study.  The  first  corpus  included  published  writing  in  the
discipline of applied linguistics, and the second one represented students' unpublished writing
at  post  graduate  level.  The  second  corpus  consisted  of  master  theses  and  doctoral
dissertations written  by  some  EFL  students  within the  discipline  of  applied  linguistics,  with
relevance  to  English  language  teaching  and  translation.  Each  of  these  corpora  will  be
described more below. The first corpus had been originally prepared by Jalali (2009) for his
study on variations in the use of lexical bundles within a single discipline: applied linguistics.

The basis for the selection of journal articles was mostly previous corpus-based studies
done  on  the  scientific  discipline  of  applied  linguistics  (e.g.  Ruiying  &  Allison,  2003),  the
advice  given  by  experts  in  the  field,  and  access  to  the  electronic  files  of  papers.  Table  1
represents  the  journals,  the  numbers  of  texts,  and  the  number  of  words  in  this  corpus.  The
second  corpus,  also  collected  by  Jalali  (2009),  included  master  theses  and  doctoral
dissertations written by some postgraduate EFL students during 2004-2009 time period.

Data analysis tools
Computer programs
Two  computer  programs  were  used  in  this  study:  Antconc3.2.1w  (Anthony,  2007),  and
Wordsmith  (Scott,  2008).  The  former  was  used  for  the  identification  of  lexical  bundles  and
concordancing while the latter was only employed to find the number of texts within  which
each bundle had been used. These two are described more below.
Antconc3.2.1.w  is  a  free  concordance  program  designed  and  developed  by  Anthony
(2007)  (see  Fig.1).  This  study  used  it  to  identify  anticipatory  "it"  lexical  bundles  and  find
their  frequency.    It  has  useful  tools  such  as  concordance,  concordance  plot,  file  view,  N-grams, collocates, word list, and keyword list that are used to analyze texts of different kinds
and  lengths.  The  concordancer  also  makes  it  possible  to  see  each  of  the  clusters  in  actual
textual context within which it has originally been used.
Among  all  these  tools,  there  is  one  by  which  it  is  possible  to  identify  word
combinations,  clusters,  or  lexical  bundles  of  different  lengths  and  frequencies  in  small  or
large  corpora.  All  lexical  bundles  in  corpora  of  different  sizes  with  their  actual  frequencies
can be found and displayed by inserting a set of commonly key words with which the bundles
collocate,  such  as  prepositions  (e.g.,  at,  of,  on,  etc),  modals  (e.g.,  can,  should,  could,  may,
etc),  etc,  and  deciding  on  the  minimum  optimal  frequency  (e.g.  twenty  in  a  corpus  of  one
million words) and the required number of words in clusters (i.e. three, four, five, or six). 
However, As Antconc3.2.1.w could not count and display the number of different texts,
WordSmith  tools5  (Scott,  2008)  was  applied  for  the  identification  of  lexical  bundles  in
different texts. This program is similar in many ways to Antconc3.2.1.w, but it does show the 
 
number of texts in which bundles have been used. So when all candidate lexical bundles were
identified  by  the  first  computer  program,  each  of  them  was  again  searched  on  Wordsmith
tools5  to  find  the  number  of  texts  with  which  they  have  been  used.  Only  those  four-  word
combinations could count as lexical bundles that had been used ten times and in at least five
different  texts  no  matter  how  frequent  they  were  (Biber  et  al,  1999).  This  was  to  guard
against  idiosyncratic  and  repetitive  uses  of  the  same  bundle  in  the  same  text  by  the  same
writer.

Functional analysis of bundles
The focus of  this study  was on 4-word  it bundles  because previous research  has shown  that
they  are  far  more  common  than  5-word  strings  and  offer  a  wider  range  of  structures  and
functions than 3-word bundles (Cortes, 2004). Bundles are essentially extended collocations
defined by their frequency of occurrence and breadth of use, but the actual frequency cut offs
are  somewhat  arbitrary.  This  study  took  a  conservative  approach  by  setting  a  minimum
frequency of 10 times per million words and occurrence in at least 10% of texts, i.e. the word
combinations has to appear in five or more texts to be regarded as a lexical bundle.

The  data  were  analyzed  in  three  steps.  First,  all  anticipatory  it  lexical  bundles  were
identified  in  the  two  corpora  along  with  their  actual  frequencies  and  the  number  of  texts  in
which they had been used. Second, by using a functional typology of it-clauses developed by
Hewings  and  Hewings  (2002)  (see  table  3)  and  the  AntConc  3.2.1  concordancer  (Anthony,
2007) and  Wordsmith  tool5  (Scott, 2008)  for  conducting  the quantitative  analysis  of  lexical
bundles, an attempt was made to probe the context in which bundles had been used to decide
on  the  most  predominant  functions.  This  was  done  by  both  authors  until  reaching  an
agreement  of  100%  on  all  cases.  In  the  third  stage,  the  results  were  compared  to  determine
the  extent  to  which  research  articles  of  applied  linguistics  were  different  and/or  similar  to
postgraduate writing in terms of range, frequency, and function of anticipatory  it bundles. It
must be noted that while there are already some functional classifications of lexical bundles
(e.g.  Cortes,  2002;  Biber  et  al,  2004;  Hyland,  2008a,  2008b),  Hewings  and  Hewings'
functional taxonomy of it-clauses (2002) was used in this study since it specifically dealt with
the interpersonal functions of this structural group.

Results
Lexical bundles in applied linguistics published writing
Table 4 shows anticipatory "it" lexical bundles in the corpus of published writing in applied
linguistics along with the frequency and the number of texts in which they had been used. A

total of seventeen different it-bundles were drawn from this corpus. The overall use of these
bundles was 449, mounting to 0/036% of the whole corpus.  In terms of function, this corpus
capitalized maximally on attitude markers (43.20%) and minimally on the attribution markers
(3.80%)  (see  table  5).  Some  of  the  most  frequent  it-bundles  were:  it  is  important  to  (88
times),  it  should  be  noted  (40  times),  it  is  possible  that  (38  times),  and  it  is  difficult  to  (36
times). A large number of anticipatory "it" lexical bundles in this corpus had also the pattern
of it +V+ adjective + that/to. It also seemed that the use of such bundles by published writers
in  applied  linguistics  helped  writers  to  encode  different  interpersonal  meanings.  The  following
examples from this corpus can show the use of some of such bundles by published writers:
(1)  As  a  result  of  these  experiences,  it  is  possible  that  these  students  retrospectively
constructed the mainstream basic writing section as being “for American students” and
assumed  that  such  an  environment  would  have  been  more  stressful  for  them  than  the
multilingual one.

(2) It may be that students in the sciences, all PhD students in our case, focused more on the
explicit  goals  of  the  courses,  which  answer  an  urgent  need  to  publish;  others  seemed  rather
more open to acknowledging more personal gains.
(3)  It  is  important  to  emphasize  in  this  section  that  although  the  majority  of  the  words  that
remind  us  of  a  non-Spanish  spelling  are  grouped  among  those  which  form  their  plural  by
adding  the  suffix  -s,  we  have  found  two  examples  of  zero  plural  morpheme:  Bluetooth  and
réflex.
(4)  By  way  of  final  comment,  it  is  interesting  to  note  that  the  results  of  the  study  are
compatible  with  a  view  of  language  learning  that  distinguishes  the  acquisitional  processes
involved  in  the  development  of  implicit  L2  knowledge  from  the  general  deductive  learning
strategies involved in the development of explicit knowledge.

Lexical bundles in applied linguistics postgraduate writing
As shown in table 6, there were again seventeen different anticipatory "it" lexical bundles in
the corpus of postgraduate writing: it was found that, it is important to, and it should be noted
were  some  of  the  more  frequent  lexical  bundles  used  by  postgraduate  students.  The  overall
frequency  of  all  it-bundles  in  this  corpus  was  354,  covering  0.038%  of  the  whole  corpus.
However, the overall  frequency of  "it" lexical bundles  in this  corpus was lower  than that  of
applied  linguistics  research  articles  (0/036%).  Interestingly,  however,  there  were  some
bundles  in  this  corpus  (i.e.  it  should  be  mentioned,  it  was  revealed  that,  and  it  is  assumed
that)  that  were  only  used  by  postgraduate  students,  not  published  writers,  in  applied
linguistics.

Functional analysis also showed that postgraduate students, like published writers, were
able to employ lexical bundles in the discourse to serve a wide variety of different functions
(see  Table  7).  As  can  be  seen,  among  the  five  categories,  34.45%  of  all  it-bundles  were
devoted to those sequences expressing epistemic meanings. Emphatics were the second group
of  bundles  in  terms  of  the  occurrence,  covering  around  26%  of  all  it-bundles,  with  attitude
markers (19.77%) and hedges (16.37) were the next. And finally, the category of attribution
was found to be the least used, with a portion of 3.38%.  The following examples can show
some of these different uses:
5) In general, it seems that the newspapers through the language used and, more specifically,
through  the  sequence  of  discursive  features  that  include  transitivity,  thematization, 
 
lexicalization, and modality encode and reinforce asymmetries between EU and Iran in their
representation, in the context of west-dominated international politics.
6)  Also,  according  to  the  suggestions,  it  is  possible  to  speculate  the  meaning  of  unknown
words when 95 percent of the words in the text are familiar to the reader.
7) Thus, when authors use expressions such as my purpose for you in this chapter is to, it is
important to note that, perhaps, and surprisingly, they are using metadiscourse.
8) Thus, although  cognates seem to  be better remembered  than non-cognates, it  is not  clear
that  this  is  due  to  their  sharing  a  memory  representation,  as  there  is  a  great  deal  of  debate
over how bilingual memory is organized.
 
Comparisons
Comparisons in terms of variety and frequency of bundles
Probably, the most  surprising finding  of this  study was related  to the  similarity between  the
two corpora  under  investigation in  terms  of the  range  of  it-bundles  employed.  Although  the
number of texts used in the corpus of applied linguistics articles was six times more than that
of  postgraduate  writing,  these  two  corpora  were  very  similar  in  terms  of  variety.  Out  of
seventeen (17) bundles used in applied linguistics research articles, fifty-three percent (53%)
were used  in the  other corpus  too. Table  8  shows shared  it-bundles in  the two  corpora.  The
results  obtained  also  showed  that  the  frequency  of  it-bundles  was  almost  the  same  in  the
corpus of  applied  linguistics  published  writing  and  the  corpus  of  postgraduate  writing  (368,
and 386, respectively), as shown in table 9.

Comparisons in terms of functions of bundles
In  terms  of  generic  differences  in  the  variety  of  bundles  used  in  each  major  functional
category, it was found that the variety of it-bundles serving as hedges and attitude markers in
applied linguistics published writing was more than that of postgraduate writing. While in the
case  of  emphatics  and  attributions,  there  was  a  slight  difference,  for  epistemic  meanings,  it
was the postgraduate writing that made a considerably heavier use. 
There were attitude markers (i.e. it is interesting to, it is important that, and it is hoped
that) that were only used by published writers in applied linguistics. Especially important was
the  higher  frequency  of  it  is  important  to  in  the  corpus  of  research  articles.  It  is  difficult  to
was  another  bundle  which  was  also  used  more  heavily  by  applied  linguistics  writers.
Interestingly  and  in  contrast  to  some  findings  of  the  previous  research  (e.g.  Hyland,  2008a,
2008b,  Cortes,  2004),  postgraduate  students,  who  might  not  have  established  themselves  as
members  of  their  disciplinary  communities,  were  found  to  be  confident  in  using  those
stretches  that  involved  making  emphasis.  This  showed  that  postgraduate  students  could
express their attitudinal meaning in a straightforward manner.

Discussion and conclusion
Postgraduate students' relatively frequent use of anticipatory it bundles in their writing could
be taken as a surprising result in this study as the previous research (e.g., Cortes, 2004) had
shown that students tended to rely less on bundles in the development of their discourses. The
analysis  of  the  corpus  of  postgraduate  writing  used  showed  that  the  number  of  different
lexical  bundles  used  by  students  in  their  writing  was  almost  as  many  as  those  used  by
published  writers.  It  seemed  that  students,  both  at  the  master's  and  doctoral  levels,  could
handle  the  use  of  anticipatory  it  lexical  bundles  for  a  wide  variety  of  discursive  functions.
However, while this relatively frequent use of it bundles with metadiscursive functions could
be indicative of writing expertise and disciplinary growth, it can also be argued that the heavy
use of such a wide variety of bundles may not always be a sign of proficient language use and
disciplinary  expertise.  Less  proficient  language  users  may  need  to  rely  more  on  formulaic
expressions  like  lexical  bundles.  Research  article  writers,  apart  from  lexical  bundles,  may
rely  on  other  resources  like  specialized  vocabulary,  diverse  word  choices,  conjunctions,
discourse markers, and manipulation of syntactic devices to develop their arguments.
Students' relatively frequent use of anticipatory it bundles could also be due to the fact
that  they  have  already  been  exposed  to  such  word-sequences  several  times  in  their  prior
readings of applied linguistics published literature. There is almost no doubt that postgraduate
students  have  repeatedly  observed  different  lexical  bundles  in  different  research  articles  to
which they have been exposed for doing and writing their own research. Furthermore, given
that anticipatory it lexical bundles are very pervasive in university written language (Biber at  al,
1999; Biber & Barbieri, 2007) and they may have a formulaic status (Wray, 2000), the use of
such word combinations may not confront students with a very difficult task. 
Also,  it  has  been  postulated  that  lexical  bundles  are  retrieved  and  stored  whole  from
memory  through  holistic  rather  than  analytical  processes  (Conklin  &  Schmitt,  2008)  and
therefore, postgraduate students may have little if any difficulty not only in understanding but
also in producing lexical bundles. There may be a processing advantage in the use of lexical
bundles  as  some  formulaic  sequences  have  been  shown  to  be  easier  to  use  (Conklin  &
Schmitt,  2008).  It  can  also  be  postulated  that  lexical  bundles  can  act  as  handy  short-cuts  or
frames  (Biber  &  Barbieri,  2007)  through  which  writers  can  scaffold  their  propositional
meanings with a relative ease. However, automatic acquisition of lexical bundles should not
be taken for granted as this study also showed that there were some lexical bundles in applied
linguistics  published  writing  on  which  students  did  not  draw  quite  frequently  or  were  not 
 
used at all. These word sequences are not idiomatic in meaning and hence they may be easy
to  understand,  but  they  may  not  seem  to  be  marked  and  perceptually  salient.  Consequently,
there  may  still  be  a  need  to  incorporate  them  in  L2  syllabus  or  EAP  (English  for  academic
purposes)  courses  for  an  increased  pedagogical  focus  on  lexical  bundles.  This  is  especially
important  for  those  students  who  need  to  understand  and  use  such  lexical  bundles  in  their
future target genres (Hyland, 2008b).