39Te minologija | 2012 | 19
Mul i-wo d pa e ns in he co pus
o In o ma ion and Communica ion
Technology. Te minological bundles
in he gen e – ‘Tex books in ICT’
MAREK WEBER
Bialys okUni e si yo Technology
KEYWORDS: co pus linguis ics, In o ma ion and Communica ion Technology, lexical bundles,
unc ional classi ica ion o lexical bundles, e minological bundles, lexical sc u iny
DATA FOR THE STUDY, THE CORPUS OF INFORMATION
AND COMMUNICATION TECHNOLOGY
Da a o he s udy consis s o an elec onic co pus o he b oad domain
o In o ma ion and Communica ion Technology (ICT) an umb ella e m
ha includes all echnologies de eloped o he pu poses o p ocessing
and he ans e o da a. The ex s we e selec ed o ep esen a c oss-
sec ion o he ield o ICT. The ex s we e di ided in o se en ca ego ies
which can be ega ded as gen es wi hin he domain o ICT:
Gen e 1: P o essional a icles in ICT;
Gen e 2: Academic a icles in ICT;
Gen e 3: Technical documen a ion o so wa e applica ions in ICT;
Gen e 4: Technical documen a ion o ICT ha dwa e;
Gen e 5: Tex books in ICT;
Gen e 6: Technical documen a ion o p og amming languages and p o-
g amming en i onmen s;
Gen e 7: Technical documen a ion o ne wo k echnologies.
The co pus con ains a collec ion o 973 ex s o aling almos 24 million
wo ds. The choice o da a o a co pus is one o he mos impo an con-
ce ns o he esea che as he co pus should be ep esen a i e o he ana-
lyzed domain o use. As Douglas Bibe sugges s he quan i y o ex s is
c ucial o esea ch in which he schola concen a es on he ex as he
p ima y uni o sc u iny. A su icien numbe o ex s should be included
in each gen e o accoun o a ia ion be ween ca ego ies and au ho s
(Bibe , 2006).
40 Ma ek Webe Mul i-wo dpa e nsin heco puso In o ma ion
 andCommunica ionTechnology.Te minological
 bundlesin hegen e–‘Tex booksinICT’
able 1. Composi ion o he In o ma ion and Communica ion echnology co pus
gen e Numbe
o ex s
Numbe o okens
( unning wo ds)
P o essional a icles 391 225551
Academic a icles 95 1272001
Technical documen a ion o so wa e applica ions
in ic
92 2472839
Technical documen a ion o ICT ha dwa e 100 2471772
Tex books 99 15315237
Technical documen a ion o p og amming
languages and p og amming en i onmen s
90 1282175
Technical documen a ion o ne wo k echnologies 106 725418
TOTAL 973 23764993
LEXICAL BUNDLE: THE TERM
Recu en sequences o wo ds which appea oge he mo e equen ly
han expec ed by chance a e an impo an pa o linguis ic ou pu . The
e m lexicalbundle was i s used by Douglas Bibe , S ig Johansson, Geo -
ey Leech, Susan Con ad and Edwa d Finegan in he now classic Longman
g amma o spokenandw i enEnglish o e e o mul i-wo d combina ions
iden i ied on he basis o he sole c i e ion o equency (Bibe e al. 1999)
while Mike Sco calls hem wo dclus e s in he manual o he so wa e
applica ion Wo dSmi hTools e sion 4 (1996) and Wo dSmi hTools e sion
5 (2010). The only conside a ion in iden i ying lexical bundles is hei
equency. “Clus e s a e wo ds which a e ound epea edly oge he in each
o he s’ company, in sequence” (Sco 2010: 337). Sco poin s ou ha wo d
clus e s may in ol e seman ic p osody i.e. he endency o ce ain lexemes
o co-occu wi h ce ain o he lexemes (e.g. he endency o cause o come
wi h nega i e e ec s such as acciden , ouble, e c.) (Sco 2010: 337).
Bibe , Johansson, Leech, Con ad and Finegan de ine lexicalbundles as
ecu en exp essions ega dless o hei s uc u al p ope ies, i.e. hey a e
sequences o wo d o ms ha equen ly co-occu in discou se and poin
ou ha sho e bundles a e o en inco po a ed in o mo e han one lon-
ge lexical bundle (Bibe e al. 1999). They also obse e ha in mos
cases lexical bundles do no o m s uc u al uni s and mos o hem a e
41Te minologija | 2012 | 19
no exp essions ha language use s would ecognize as idioms o o he
ixed lexical exp essions. Bibe , Johansson, Leech, Con ad and Finegan
es ablish he ollowing cu o h esholds o mul i-wo d sequences o
quali y as lexical bundles: hey mus occu in a leas 10 imes pe million
wo ds and a leas in i e ex s (Bibe e al. 1999).
Bibe no es ha a su p ising esul o a equency d i en app oach is ha
lexical bundles ha e wo unexpec ed cha ac e is ics (Bibe 2006: 134). Fi s -
ly, mos o hem a e no idioma ic in meaning bu he meanings a e usu-
ally anspa en om indi idual wo d- o ms. Ken Hyland also poin s o he
ac ha mos bundles a e seman ically anspa en and “ o mally egula ,
p o iding he building blocks o cohe en discou se” (Hyland 2008: 6).
Secondly, mos bundles a e no comple e g amma ical s uc u es. Bibe ,
Johansson, Leech, Con ad and Finegan obse ed ha app oxima ely 15 %
o he lexical bundles in con e sa ion o med comple e s uc u al uni s
while only app oxima ely 5 % o he lexical bundles in academic p ose
could be conside ed as comple e ph ases o clauses (Bibe e al. 1999: 1000).
A aluable inding in o he syn ac ic na u e o lexical bundles by Bibe is
ha hey a e lexical uni s ha equen ly cu ac oss g amma ical s uc u es,
o example hey can b idge wo ph ases o clauses in such a way ha he
las wo ds o a bundle a e he beginning pa s o a second syn ac ic s uc-
u e (Bibe 2006: 135). Hyland also emphasizes ha lexical bundles being
iden i ied solely on he basis o hei equency usually span s uc u al uni s
(Hyland 2008: 6).
A ange o co pus s udies ha e been de o ed o he analysis o ecu en
s ings o unin e up ed wo d- o ms and hey demons a e how impo an
hey a e in a ious ypes o discou se as well as hey show conside able
a ia ion o lexical bundles in di e en gen es and egis e s (e.g. Bibe
2006; Bibe , Con ad, Co es 2004; Hyland 2008; Sco , T ibble 2006;
Gozdz-Roszkowski 2011).
METHODOLOGY USED IN THE STUDY
The decision was made o concen a e on he analysis o 4-wo d bundles
because on he one hand he equencies o 4-wo d bundles a e subs an-
ially highe han 5-wo d sequences and on he o he hand hey equen -
ly con ain 3-wo d s ings and enable o iden i y mo e appa en pa e s and
s uc u es han 3-wo d exp essions. Lis s o 4-wo d bundles in each o he
42 Ma ek Webe Mul i-wo dpa e nsin heco puso In o ma ion
 andCommunica ionTechnology.Te minological
 bundlesin hegen e–‘Tex booksinICT’
se en ICT gen es we e gene a ed using he Wo dSmi hTools . 5.0 – he
so wa e package o sea ching pa e ns in co po a.
The ollowing wo cu -o poin s we e se in his s udy: a minimum
equency o occu ence o 20 imes pe million wo ds and ano he c i-
e ion ha a bundle should occu in a leas i e ex s. The p ocess o
applying he cu -o c i e ia equi ed app op ia e calcula ions in each o
he se en iles con aining he lis o bundles in a gi en gen e. The cal-
cula ions a e desc ibed in he ollowing s eps:
1. C ea ing an addi ional column named “F equency pe million” in
he sp eadshee ob ained om he Wo dSmi hTools so wa e.
2. Inse ing he unning wo ds alue in o a ee Excel sp eadshee cell
wi hin a gi en ca ego y.
3. Filling in he i s cell o he “F equency pe million” column wi h
he ollowing o mula:
4. Copying he o mula o o he cells by clicking on he igh bo om
co ne o he illed-in cell and d agging in downwa ds o o he cells.
5. Highligh ing he “Numbe o ex s” column.
6. Choosing he “Da a” menu and selec ing he ield “So &Fil e ” in
he so ing ool.
7. Choosing he descending so ing o de .
8. Dele ing all ows o he able, o which he alues o he “Numbe
o ex s” ield is lowe han 5.
9. Highligh ing he “F equency pe million” column.
10. Choosing he “Da a” menu and selec ing he ield “So & Fil e ”
in he so ing ool.
11. Choosing he descending so ing o de .
12. Dele ing all ows o he able, o which he alues in he “F e-
quency pe million” ield a e lowe han 20.
1886 di e en bundles we e ound in he co pus a e applying he
abo e cu -o c i e ia. The o al numbe o bundles iden i ied in he
en i e da a amoun ed o 197 553.
43Te minologija | 2012 | 19
able 2. Dis ibu ion o lexical bundles in IC gen es
gen e The o al numbe
o bundles a e
applying cu -o
c i e ia
Numbe o di e -
en bundles a e
applying cu -o
c i e ia
% o unning
wo ds in bundles
P o essional a icles 1057 131 1,8
Academic a icles 5548 120 1,7
So wa e applica ions 44669 403 7,2
Ha dwa e 55296 672 8,9
Tex books 72483 119 1,9
P og amming lan-
guages and p og am-
ming en i onmen s
10984 180 3,4
Ne wo k echnologies 7516 261 4,1
To al 197553 1886
The pe cen age o unning wo ds in bundles was ob ained by mul iply-
ing he numbe o o al cases o a gi en gen e by 4 (4-wo d bundles a e
analyzed) and di iding by he numbe o unning wo ds in a gi en gen e
and hen mul iplying by 100 % acco ding o he o mula:
In ou iew wo pa ame e s: he ange o di e en bundles and he pe -
cen age o unning wo ds in bundles can be ega ded as indica o s o he
deg ee o which a gi en gen e is o mulaic and epe i i e in compa ison
o o he gen es. In o he wo ds he deg ee o o mulaici y and epe i i e-
ness o a gen e can be measu ed by he ange o di e en bundles employed
in a gen e and he pe cen age o unning wo ds in bundles.
The ICT gen es can be di ided in o h ee g oups acco ding o he c i-
e ion o o mulaici y and epe i i eness.
Gen e 3 – ‘ echnical documen a ion o so wa e applica ions in ICT’
and gen e 4 – ‘ echnical documen a ion o ICT ha dwa e’ a e ma ked
by high deg ees o o mulaici y and epe i i eness as hey display com-
pa able pa e ns by employing he bigges scope o di e en bundles
(403 and 672 espec i ely) and he highes pe cen age o unning wo ds
in bundles (7,2 % and 8,9 % espec i ely).

44 Ma ek Webe Mul i-wo dpa e nsin heco puso In o ma ion
 andCommunica ionTechnology.Te minological
 bundlesin hegen e–‘Tex booksinICT’
Figu e 1: Gen es wi h high deg ees o o mulaici y and epe i i eness
In con as , gen e 1 – ‘p o essional a icles in ICT’, gen e 2 – ‘aca-
demic a icles in ICT’ and gen e 5 – ‘ ex books in ICT’ a e cha ac e ized
by ela i ely low deg ees o o mulaici y and epe i i eness as hey employ
he lowes scope o di e en bundles (131, 120 and 119 espec i ely) and
he lowes pe cen age o unning wo ds in bundles (1,8 %, 1,7 % and
1,9 % espec i ely).
The emaining wo gen es: gen e 6 – ‘ echnical documen a ion o p o-
g amming languages and p og amming en i onmen s’ and gen e 7 – ‘ ech-
nical documen a ion o ne wo k echnologies’ o m he hi d g oup wi h
he alues o bo h pa ame e s in he middle o he ange (numbe s o
di e en bundles: 180 and 261 espec i ely; pe cen age o unning wo ds
in bundles 3,4 % and 4,1 % espec i ely).
Howe e , i is wo h bea ing in mind ha as S anislaw Gozdz-Roszkow-
ski igh ly poin s ou di ec compa isons can only be sa ely made be ween
gen es wi h simila wo d coun s. In o de o accoun o ha conside a ion
he pa ame e pe cen ageo  unningwo dsinbundles is compu ed o each
gen e in such a way as o e lec he wo d coun s o each o he subco -
po a ep esen ing espec i e gen es (Gozdz-Roszkowski 2011: 111).
FUNCTIONAL CLASSIFICATION OF BUNDLES
A amewo k o he unc ional analysis o he bundles ob ained in his
co pus was es ablished om Bibe ’s (Bibe 2006; Bibe e al. 2004),
Hyland’s (2008) and Gozdz-Roszkowski’s (2011) axonomies.
Bibe ’s (2006) classi ica ion esul ed om he sc u iny o a b oad co pus
o spoken and w i en egis e s which co e ed among o he s such ypes
o discou se as: casual con e sa ions, class sessions apes, class oom each-
ing, o ice hou s, s udy g oups, on-campus se ice encoun e s, ex books,
cou se packs, ins i u ional ex s (e.g. uni e si y ca alogs, b ochu es).
45Te minologija | 2012 | 19
The co pus in Hyland’s (2008) s udy (size 3,5 million wo ds) com-
p ises esea ch a icles, PhD disse a ions and MA/MSc heses om ou
disciplines: elec ical enginee ing and mic obiology om he applied and
pu e sciences, and business s udies and applied linguis ics om he social
sciences.
Gozdz-Roszkowski (2011) analyzed a co pus o Ame ican Law con ain-
ing o e 5,5 million wo ds and ep esen ing se en gen es wi hin Ame i-
can legal cul u e and educa ion: academic a icles, b ie s, con ac s, leg-
isla ion, opinions, p o essional a icles and ex books.
D awing upon he abo emen ioned axonomies lexical bundles in ICT
we e unc ionally di ided in o h ee b oad ca ego ies wi h espec o hei
meanings in he ex s: esea ch-cen e ed, ex -cen e ed and pa ic-
ipan -cen e ed. Bundles g ouped in he i s ca ego y help w i e s o
o ganize hei ac i i ies and expe iences in he domain o ICT. Tex -
cen e ed bundles a e employed o indica e he o ganiza ion o he ex
and i s meaning. Finally pa icipan -cen e ed bundles a e used o
signal di e en a i udes o assessmen s and hey a e ocused on he w i e
o eade o he ex .
Figu e 2: Func ional axonomy o lexical bundles
Each o he h ee majo unc ional ca ego ies o lexical bundles was
u he subdi ided in o a numbe o subca ego ies.
Resea ch-cen e ed bundles include he ollowing subca ego ies:
Quan i y bundles (e.g. oneo  hebigges ;ala genumbe o ;oneo  he
ollowing; henumbe o elemen s);
Time e e ence bundles (e.g. a  he imeo ;endo  heyea ; henex 
ewweeks;in hecomingweeks);
Place / di ec ion e e ence bundles (e.g. in heUni edS a es;o  he
mainwindow;bo omo  hewindow;in hes a usba );
46 Ma ek Webe Mul i-wo dpa e nsin heco puso In o ma ion
 andCommunica ionTechnology.Te minological
 bundlesin hegen e–‘Tex booksinICT’
P ocedu e bundles – used o desc ibe di e se unc ions pe aining o
ICT (e.g. heuseo  he; heuseo a);
Topic indica o bundles – pe aining o he a ea o esea ch (e.g.
p oceedingso  heIEEE;p og amminglanguagesandsys ems; hed opdown
menu; he ollowingcon igu a ionop ions);
Mul i- unc ional e e ence bundles – bundles which can be used as
ime / place / ex e e ence (e.g. heendo  he; hes a o  he; hele 
o  he;a  hebeginningo );
Desc ip ion bundles – speci y cha ac e is ics o he ollowing noun (e.g.
hecomplexi yo  he; hesu aceo  he; hescopeo  he; heheigh o  he).
Tex -cen e ed bundles include he ollowing subca ego ies:
Elabo a ion bundles – u he elabo a e on he analyzed opic and
cla i y i (e.g. a  hesame ime,aswellas he, hismeans ha  he,in his
case he);
T ansi ion bundles – p o ide addi i e o con as i e connec ions be-
ween po ions o ex s (e.g. on heo he hand,asopposed o he,inaddi-
ion o he,incon as  o he);
F aming a ibu es bundles – “si ua e a gumen s by speci ying limi -
ing condi ions” (Hyland 2008: 14) o making claims o a gumen s (e.g.
in e mso  he,in hecon ex o ,in hep esenceo , hecon en so  he);
Condi ions bundles – exp ess condi ions (e.g. i youwan  o,i youdo
no ,i youha ea,i youha ean);
Resul s bundles – indica e logical links be ween elemen s in e ms o
cause and esul ela ionships (e.g. so ha youcan,asa esul o ,asa
esul  he,asa unc iono );
S uc u e ma ke s bundles – a e used o poin o o he pa s o he
ex (e.g. asshownin igu e,asdiscussedinsec ion,isshowninexample,
la e in hischap e ).
Pa icipan -cen e ed bundles include he ollowing subca ego ies:
Engagemen bundles – add ess eade s di ec ly; “ac i ely add ess ead-
e s as pa icipan s in he un olding discou se” (Hyland 2008: 18) (e.g. do
oneo  he,selec  he ypeo ,is ecommended ha you,selec oneo  he);
S ance bundles – con ey emo ions, a i udes, alue judgmen s and as-
sessmen s; “p o ide a ame o he in e p e a ion o he ollowing p op-
osi ion” (Bibe , 2006: 139) (e.g. i isnecessa y o,isnogua an ee ha ,i is
impo an  o,i isno possible,i ispossible o);
47Te minologija | 2012 | 19
Modali y bundles – exp ess ha some hing is p obable, pe missible o
necessa y (e.g. mus be ul illeda,canbeused o,ascanbeseen,inwhich
youcan,mayno bedisplayed,maybe ep oducedo ,mus accep anyin e -
e ence,in e e ence ha maycause, ha maycauseundesi ed);
P edic ion bundles – exp ess he w i e ’s p edic ion o some u u e
ac ion (e.g. isexpec ed obe,youwillbep omp ed,willbedisplayedin, his
willopen he,willbeasked o).
Table 3 shows he numbe s o bundles belonging o he h ee majo
unc ional ca ego ies in each gen e.
able 3: Dis ibu ion o lexical bundles ac oss unc ional ca ego ies
Resea ch-
cen e ed
Tex -
cen e ed
Pa icipan -
cen e ed
O he s
P o essional a icles 45 23 13 45
Academic a icles 48 35 928
So wa e applica ions 123 55 122 83
Ha dwa e 207 51 138 187
Tex books 32 35 18 20
P og amming languages and
p og amming en i onmen s
39 39 26 62
Ne wo k echnologies 91 24 11 91
Figu e 3: Classi ica ion o esea ch-cen e ed bundles