!"#$ & '()*(+,-./0 ./ 1 +.23*(4 +.

'350
!"#$%&' )*+&&,-
. /
0 1$'2# 345*
6
0 728#$% 7*9%:;#$ 3'8$';%#2
.
428 72"# <=**#$ >"#%84*
.

1
0niveisity of Beigen;
2
Noiwegian School of Economics
7?%&$4@&
In this article we describe the KIAP corpus, a corpus of 450 research articles covering three disciplines
(economics, linguistics and medicine) and three languages (English, French and Norwegian) located at
the University of Bergen/Uni Research. We describe the rationale for article selection, the
functionalities of the corpus as well as important results coming out of the use of the corpus in the
KIAP project. Moreover, we discuss the use of the corpus as a methodological tool for the generation
and revision of hypotheses on linguistic phenomena characteristic of person manifestation in scientific
discourse. Finally, we present some key features of Norwegian medical articles that are relevant in the
discussion of the phenomenon of standardisation of scientific discourse.
!#ABC$8%: ieseaich aiticle; coipus; Noiwegian; uisciplines; languages
/ D$'2@'E4* @C2&4@&F
Kjeisti Flttum, Piofessoi
Bepaitment of Foieign Languages, BF-bygget, Su2u Beigen, Noiway
Tel.: +47 SS S8 22 82
E-mail: kjeisti.flottumif.uib.no
!" $%&''()* +" ,-.%* /" /" ,01203456 -61 /"7" 89541-%

1S8

67 "/,'.85+,-./
English acauemic language has foi seveial uecaues ieceiveu much attention fiom applieu
linguists anu foi the past uecaue oi two also fiom coipus linguists. Besciiptions of genies anu
typical linguistic anu uiscuisive featuies of these genies have been publisheu in ueuicateu
jouinals (e.g. !"#$%&' ") *$&+,&-./0, 1%-2$%&-."%&' !"#$%&' ") 344'.25 6.%+#.0-./0, 7%+'.08 )"$
942/.)./ *#$4"020, !"#$%&' ") 7%+'.08 )"$ 3/&52,./ *#$4"020) anu books (e.g. Swales 199u;
Beikenkottei & Buckin 199S; Bylanu 1998, 2uuu). Some of this ieseaich took uisciplinaiy
vaiiation into consiueiation (Bylanu 2uuu), while othei stuuies involveu compaiisons of English
language texts anu similai texts in anothei language (Nauianen 199S; vassileva 2uuu).
Bowevei, few stuuies uiu both (foi latei coipus-baseu stuuies uoing this, see e.g. Sanueison
2uu8; Lafuente-Nilln, Nui-Bueas, Lois-Sanz & vzquez-0ita 2u1u). The oveiaiching
ieseaich question exploieu in the KIAP pioject,
1
on the othei hanu, uiu ielate to both the
uiscipline factoi anu the language factoi. Acknowleuging that text is multivoiceu, we askeu to
what extent acauemic voices, as manifesteu in ieseaich aiticles, aie language-specific oi
uiscipline-specific, anu how the vaiious voices aie iealiseu. The pioject took its point of
uepaituie in featuies iepiesenting tiaces of SELF (the authois) as well as those of 0TBERS
(othei ieseaicheis anu ieaueis), in the shape of peisonal pionouns, metauiscouise, epistemic
mouifieis, aigumentative connectives, bibliogiaphical iefeiences, negation anu auveisatives. In
oiuei to examine these questions it was necessaiy to establish a uigital coipus of ieseaich
aiticles, ieflecting oui inteiest in the inteiaction of uiscipline anu national languages. The KIAP
coipus was establisheu in close collaboiation between the AKSIS uepaitment at the 0nifob
ieseaich oiganisation (now 0ni computing at 0ni Reseaich) anu the 0niveisity of Beigen. In the
piesent papei we woulu like to ieflect on some aspects ielateu to the establishing anu use of this
complex coipus.
The issues we biought into the pioject aie continuously being uiscusseu in touay's
ieseaich woilu. The oiientation towaius stanuaiuisation in acauemic wiiting becomes moie anu
moie piessing, anu uiveisification with uisciplinaiy anu national language tiauition piactices
may be incieasingly uifficult to maintain (Sanueison 2uu8). The uominant position of the
English language in acauemic uiscouise is inuisputable (Nauianen, Piez-Llantaua & Swales
2u1u), anu the INRAB stiuctuie (Gntiouuction, <ethous, Hesults 4nu 3iscussion; see e.g. Swales
199u), oiiginating in expeiimental science, is close to attaining the same geneial status in the
ieseaich aiticle, the most piestigious genie foi knowleuge uissemination. Bas this alieauy leu to
a loss of specific cultuial anu uisciplinaiy iuentities in acauemic wiiting piactices. To what
extent will such a potential loss have an impact foi ieseaich, highei euucation anu the piactices
of the stuuents. This situation constitutes a gieat uilemma foi many ieseaicheis whose fiist
language is not English anu whose wiiting tiauition is not founueu on the Anglo-Ameiican one.
Nost of us woulu like to uefenu oui national anu uisciplinaiy tiauitions, but at the same time we
want to publish in inteinationally ienowneu outlets piomoting stanuaiuisation in both language
anu foim.
It was against this backgiounu that we staiteu builuing the KIAP coipus. In this papei we
piesent the pioject anu its main finuings (section 2), befoie we ieflect on the coipus as a
methouological tool foi the geneiation, uevelopment anu ievision of hypotheses anu ieseaich
questions (section S). The KIAP pioject incluueu texts in Noiwegian, a small acauemic language
which has ieceiveu little attention compaieu to English acauemic language. We woulu theiefoie
like to focus specifically on oui finuings foi Noiwegian heie anu contiast these with oui finuings

1
For a complete overview of the KIAP (Kulturell identitet i akademisk prosa/Cultural
identity in academic prose) project, see Flttum, Dahl and Kinn (2006).
"#$% & '()*(+,-./0 ./ 1 +.23*(4 +.'350

1S9
foi English anu Fiench (section 4). In the final section (S), we attempt to assess the value of
ielatively small, but specializeu coipoia like the KIAP coipus, befoie we outline some
peispectives foi potential futuie uses of the coipus.
97 !"#$ & 3'.:(+,; +.'350 1/8 <(/('1* )-/8-/<0
The pioject "Cultuial Iuentity in Acauemic Piose", abbieviateu by the Noiwegian acionym KIAP,
is locateu at the 0niveisity of Beigen, but has piofiteu fiom stiong inteinational collaboiation. It
was financeu by the Reseaich Council of Noiway uuiing the yeais 2uu2-2uu6. New ieseaich is
still being unueitaken within the pioject, which (so fai) has iesulteu in 8 mastei theses anu 2
PhB uisseitations in Beigen. Thiough inteinational collaboiation, the pioject is also iepiesenteu
in PhB uisseitations outsiue of Noiway (e.g. Pouuat 2uu6; Rinck 2uu6). This ieseaich
enviionment has stiongly contiibuteu to the uynamics of the pioject. Seveial national anu
inteinational woikshops anu symposia have been linkeu to the pioject, anu in teims of wiitten
uissemination of the finuings, both the ueuicateu publication seiies 3:&52,.0: *$"0&; 9:$.)-2$ )$&
<13* anu numeious inteinational publications may be mentioneu (some can be founu in the
Refeiences section of the cuiient papei).
0ui point of uepaituie foi the pioject was that we consiueieu acauemic uiscouise to be
cleaily ihetoiical, a position which we still veiy much auheie to. 0ui oveiall objective has been
to give substance to the commonly expiesseu contestation of the conception of acauemic
uiscouise as neutial, objective anu non-inteiactional. Seveial stuuies have obseiveu the cleai
piesence of the authoi(s)' voice in acauemic wiitings, such as the use of the fiist peison pionoun
combineu with an aigumentative veib - 1=>2 /'&., -8&- ? (e.g. Beige 2uuS; Bieivega 2uuS;
Bylanu 1998; vassileva 2uuu) In this peispective, we auopteu the view that the ieseaich aiticle
is ihetoiical in the sense that it iepiesents a uiscouise cieateu in oiuei to inuuce coopeiative
attituues anu actions (see Pielli 1989), on the one hanu, anu to give the possibility foi the
authois to position themselves in oi in ielation to a paiticulai uiscouise community, on the
othei. This position implies the eviuent fact that the ieseaich aiticle is auuiesseu, i.e. uiiecteu at
someone (Beige 2uuS). Even if the manifestation of auuiessivity may be uiffeient in uiffeient
tiauitions, this obseivation points to the simple but impoitant fact that the ieseaich aiticle
constitutes a piece of communication. The piimaiy ihetoiical aim of a ieseaich aiticle is to
cieate effects which convince the auuience to such a uegiee that the aiticle becomes an
integiateu pait of a paiticulai fielu's liteiatuie, which iepiesents the fielu's knowleuge base.
In oiuei to unueitake the ielevant analyses, we neeueu a ielatively laige coipus. In
collaboiation with AKSIS (see above), we staiteu the collection of ieseaich aiticles fiom
iefeieeu jouinals. Since we wanteu to puisue a uoubly contiastive peispective involving both
uiscipline anu language, we collecteu aiticles wiitten in thiee languages, viz. English, Fiench anu
Noiwegian, anu within thiee uisciplines, viz. economics, linguistics anu meuicine. This enableu
us to compaie, e.g., economics aiticles wiitten in the thiee languages as well as, e.g., Noiwegian
aiticles within the thiee uisciplines. The coipus uesign thus maue it possible to investigate the
impoitance of both the language factoi anu the uiscipline factoi, as well as theii inteiaction. The
complete electionic coipus consists of 4Su aiticles (about S,uuu,uuu woius) publisheu in
iefeieeu jouinals, in the peiiou 1992-2uuS. The coipus is uiviueu in 9 subcoipoia with Su
aiticles in each: English economics (2%+2/"%), English linguistics (2%+'.%+), English meuicine
(2%+,25); Fiench economics ()$2/"%), Fiench linguistics ()$'.%+@A Fiench meuicine ()$,25);
Noiwegian economics (%"2/"%), Noiwegian linguistics (%"'.%+), Noiwegian meuicine (%",25).
AKSIS' contiibution was ciucial in setting up anu ueveloping a seaich function which maue it
possible to peifoim seaiches accoiuing to uiffeient ciiteiia, such as language, uiscipline,
inuiviuual aiticle, aiticle section, etc. The seaich output also incluueu fiequency calculations
which became the basis foi quantitative analyses (foi a uetaileu uesciiption of the KIAP coipus
as well as the iesults of the analyses, see Flttum et al. 2uu6).
!" $%&''()* +" ,-.%* /" /" ,01203456 -61 /"7" 89541-%

14u
It is impoitant to emphasise that foi the KIAP pioject, specific linguistic anu uiscuisive
issues, ielateu to the genie of the ieseaich aiticle, foimeu oui point of uepaituie, anu also that
coipus stuuies weie seen as an impoitant contiibutoi to oui ,2-8"5"'"+./&' &44$"&/8B Bowevei,
we uiu not stait by ueveloping the coipus anu then foimulating peitinent ieseaich questions
which woulu fit the coipus. We uiu it the othei way aiounu, staiting with investigating what
woulu be ielevant questions in ielation to the puipose of oui stuuy. Then, in oiuei to have
sufficient uata, we pioceeueu to a systematic collection of ieseaich aiticles, which then weie
stiuctuieu in a uigital coipus. This coipus thus became an integial pait of oui methou. As
iegaius coipus linguistics, Stig }ohansson's uefinition of the object of coipus linguistics fits well
with oui own position:

"|.j the object of coipus linguistics is %"- the stuuy of coipoia. It is iathei the stuuy of language
thiough coipoia, |.j." (}ohansson 199S:19)

A key issue in the pioject was to iuentify possible cultuial iuentities as manifesteu in
linguistic tiaces of acauemic voices in the genie of the ieseaich aiticle. Within this peispective,
oui main focus was on peison manifestation - what we chose to call the SELF anu the 0TBER
uimensions of acauemic voices. This focus was specifieu thiough thiee ieseaich issues, ielateu
to (1) the manifestation of the authois (the SELF uimension) in the texts, (2) the piesence of the
voices of ieaueis anu othei ieseaicheis (the 0TBER uimension) anu (S) the piesentation of the
authois' own ieseaich. The focus on peison manifestation also explains oui choice of theoietical
fiamewoik, which has been baseu on a bioau inteipeisonal anu polyphonic peispective (see
Nlke et al. 2uu4; Flttum 2uuSa), with genie theoiy as an oveiaiching appioach (see foi
example Beige 2uuS; Rastiei 2uu1; Swales 199u).
In oiuei to auuiess the thiee ieseaich issues, we selecteu uiffeient linguistic featuies
which may iealise the SELF anu the 0TBER uimensions of peison manifestation - in uiffeient
ways. The main featuies stuuieu aie the following:

- fiist peison anu inuefinite subject pionouns
- veibs combineu with these pionouns
- maikeis of epistemic mouality
- aigumentative connectives
- metatextual expiessions
- the constiuction '2- #0='2- ,2 + infinitive (anu its counteipait in Fiench anu Noiwegian)
- polyphonic constiuctions (polemic negation anu concession)
- bibliogiaphical iefeiences

0ui main hypothesis thioughout the pioject was that uiscipline is moie impoitant than
language in the iuentification anu piouuction of cultuial iuentities; i.e. that theie aie moie
similaiities between, foi example, a Noiwegian meuical authoi anu a Fiench meuical authoi
than between a Noiwegian meuical authoi anu a Noiwegian linguist authoi. As iegaius
uisciplinaiy uiffeiences, a few exploiatoiy analyses unueitaken at the beginning of the pioject
"#$% & '()*(+,-./0 ./ 1 +.23*(4 +.'350

141
alloweu us to uiaw up thiee pieliminaiy caiicatuie piofiles: the meuical ieseaichei as iathei
absent, hiuing behinu the text (as in C8.0 0-#5D 8&0 08">% -8&- ?), the economist as somewhat
piesent but in a mouest way (as in E2 0-$#/-#$2 -82 4&42$ &0 )"''">0 ? ) anu the linguist as
cleaily anu polemically piesent in the text (as in 1% /"%-$&0- -" FA >2 &$+#2 -8&- ?). 0ui
investigations confiimeu the geneial hypothesis in most iespects: uiscipline tiumps language.
Authois of ieseaich aiticles tenu to wiite moie like theii uisciplinaiy colleagues wiiting in othei
languages than like theii language-community co-membeis wiiting in othei uisciplines - with
iespect to most of the linguistic featuies stuuieu in the pioject (see Flttum et al. 2uu6).
=7 >?( !"#$ +.'350 10 1 2(,?.8.*.<-+1* ,..* ).' ,?( 8(@(*.32(/, .)
?A3.,?(0(0 1/8 '(0(1'+? B5(0,-./0
In this context, we woulu like to emphasise the heuiistic value of a coipus such as KIAP as a
methouological tool. The coipus helpeu us not only to geneiate new hypotheses, but also to
iefocus anu aujust oui initial hypotheses. The quantitative analyses (concoiuances, collocations,
fiequency analyses) as well as the semantic-piagmatic analyses of the coipus ievealeu seveial
inteiesting phenomena. In auuition, the piactical (albeit time-consuming) auvantage of having
paiticipateu in the collection anu couing of the coipus mateiial pioviueu us with invaluable
insight into the makeup of the texts, sometimes giving iise to unexpecteu insights. In this section
we piesent an oveiview of ieseaich topics (a-u) that emeigeu uuiing the coipus analysis
(Flttum 2uuSb).

G&@ H$", -82 +2%2$&' 8D4"-820.0 ") &#-8"$.&' 4$202%/2 ,&%.)20-25 ID -82 #02 ") ).$0- 42$0"%
4$"%"#%0 -" -82 8D4"-820.0 ") &#-8"$.&' 4$202%/2 ,&%.)20-25 ID -82 .%52).%.-2 4$"%"#% !" .% H$2%/8
$202&$/8 &$-./'20 (see Flttum 2uuS, 2uu4a)

We still consiuei fiist peison pionouns to be a ielevant anu obvious explicit maikei of
authoiial piesence in the ieseaich aiticle. Bowevei, the uiffeiences between the use of these
pionouns in English anu Noiwegian on the one hanu anu in Fiench on the othei, tuineu out to
be so impoitant that we soon iealiseu that it was necessaiy to exploie othei linguistic
constiuctions in the Fiench ieseaich aiticles that might potentially caiiy a 'peisonal' value. The
analyses showeu that the inuefinite pionoun "% (coiiesponuing to English "%2) is highly
fiequent in the Fiench ieseaich aiticles (foi an analysis of inuefinite pionouns in English,
Noiwegian anu ueiman, see }ohansson 2uu2; foi similai finuings on Fiench "%, see ujesual
2uuS, 2uu8). 0ui finuings uemonstiate a veiy vaiieu use of this 'impiecise' anu flexible pionoun.
In uiscouise, the pionoun "% may potentially coiiesponu to all the peisonal pionouns in Fiench.
0f paiticulai inteiest in the KIAP context was the use of "% foi authoi iefeience, i.e.
coiiesponuing to English 1 anu Noiwegian J2+=2+ in single-authoi aiticles oi >2 anu K.
iespectively in aiticles by moie than one authoi, as illustiateu in the following example:

(1) Bans un piemiei temps (paiagiaphe 1), C2 pisenteia une liste ue caiactiistiques,
uans l'ensemble bien connues, qui uistinguent les ueux emplois |...j. (filingu6)
('Fist (i avsnitt 1) vil 9' piesenteie en liste ovei egenskapei, stoit sett ei velkjente i
litteiatuien, som skillei ue to betyuningene |.j')
('Fiist (section 1), we will piesent a list of chaiacteiistics, geneially well known, which
uistinguish the two meanings |...j.')
!" $%&''()* +" ,-.%* /" /" ,01203456 -61 /"7" 89541-%

142

GI@ H$", -82 8D4"-820.0 ") -82 &I02%- &55$20022 -" -82 8D4"-820.0 ") 0-$&-2+.20 )"$ &55$20022
.%K"'K2,2%- .% 0-$#/-#$.%+ &%5 &$+#,2%-&-."% (see Flttum 2uuS, 2uu4a; Kinn 2uuSa, b)

Quantitative seaiches foi linguistic elements explicitly signalling the piesence of the
auuiessee, such as seconu peison pionouns, yielueu few if any iesults, as was expecteu in light
of the fact that the ieseaich aiticle is not tiauitionally consiueieu a uialogic genie. Bowevei,
analyses of vaiious metatextual anu metauiscuisive expiessions pointeu to new potential
manifestations of the textually elusive auuiessee. Expiessions like "In this aiticle we will fiist
consiuei" may inuicate that this is not just a statement by the authoi on how the aiticle is
stiuctuieu anu what it is about. It may also seive as an invitation to the auuiessee to paiticipate
in ueveloping the aiguments anu thus in the co-constiuction of meaning (thiough an inclusive
>2 anu a veib iefeiiing to an action in which seveial people can paiticipate). This becomes even
cleaiei with impeiative constiuctions of the type '2- #0, as in '2- #0 %"> +" I&/: in the following
example:

(2) Baving accounteu foi basic ieconstiuction effects with A-movement, let us now go
back to the uata that motivateu Chomsky's anu Lasnik's conclusion that theie aie no
ieconstiuction effects with A-movement. (engling49)

These analyses yielueu inteiesting iesults with iegaiu to how authoiauuiessee
inteiaction may unfolu.

G/@ H$", -82 8D4"-820.0 ") 2L4'./.- 4$202%/2 ") "-82$ K"./20 -8&% -8&- ") -82 &#-8"$ -" -82 8D4"-820.0
") "-82$ K"./20M .,4'./.- 4$202%/2 -8$"#+8 4"'D48"%./ 2L4$200."%0 (see Flttum 2uuSa, 2u1u, 2u12)

The explicit piesence of otheis' voices in ieseaich aiticles is manifesteu by uiiect oi
inuiiect iefeience to iesults anu obseivations fiom othei ieseaicheis' woik, often thiough some
foim of iepoiteu speech. 0ui finuings in the KIAP pioject ievealeu that theie aie uiffeiences
between languages anu uisciplines in teims of how othei ieseaicheis anu theii iueas aie
integiateu into a new ieseaich text. Thiough a pilot stuuy eaily in the pioject we iuentifieu
vaiious polyphonic constiuctions as being paiticulaily inteiesting in this iespect. The voices anu
viewpoints of othei ieseaicheis tuineu out to be implicitly integiateu in texts thiough the use of
linguistic featuies such as polemic %"- anu the contiastive anu concessive maikeis I#-. This is a
foim of subtle inteiaction that quantitative analyses in tuin uemonstiateu to be a shaieu anu
fiequent featuie of all the nine subcoipoia of the KIAP coipus. Thiough a closei stuuy of the co-
text of the Noiwegian syntactic maikei of negation .::2, we weie also leu to question the
tiauitional anu absolute uiviue between uesciiptive anu polemical negation. 0ui analyses
inuicateu that theie is a continuum of highei anu lowei uegiees of polemicity, assigning
uiffeient values to the negation, anu ueteimineu in pait by the genie in question (see Flttum
2uu4b).

"#$% & '()*(+,-./0 ./ 1 +.23*(4 +.'350

14S
G5@ H$", -82 +2%2$&' 8D4"-820.0 ") &#-8"$.&' 4$202%/2 &0 ,&%.)20-25 ID ).$0- 42$0"% 4$"%"#%0 -" -82
8D4"-820.0 ") &#-8"$ $"'20 ,&%.)20-25 ID 5.))2$2%- -D420 ") K2$I0 &00"/.&-25 >.-8 -82 42$0"%&'
4$"%"#% (see Flttum 2uu4a; Kinn 2uuSa.)

While items (a)-(c) above iepiesent a fuithei uevelopment of oui initial hypotheses,
item (u) iepiesents a new hypothesis, aiising fiom the concoiuance-baseu analyses enableu by
the coipus seaich function. An impoitant objective foi the quantitative analyses was to examine
oui initial hypothesis of fiist peison pionouns as inuicative of authoiial piesence. Quantitative
concoiuance analyses of fiist peison pionouns uemonstiateu stiiking iegulaiities in the co-text
of this paiticulai featuie. 0bviously, the veib foims associateu with the fiist peison pionouns
offei much moie infoimation on the natuie of authoi manifestation than the pionoun on its
own. An investigation of the veibs involveu inuicateu that at least thiee &#-8"$ $"'20 aie enacteu
in the ieseaich aiticle: the authoi as wiitei (1 >.'' I2+.% >.-8) oi text guiue (1 >.'' $2-#$% -"), as
ieseaichei (1 &%&'D02) anu as 'aiguei', i.e. an actoi positioning him-heiself (1 /'&.,). Analyses
unueitaken by KIAP ieseaichei Toiouu Kinn (Kinn 2uu4, 2uuSa) ievealeu an impoitant
vaiiation between the subcoipoia with iegaiu to the types of veibs that typically co-occui with
fiist peison pionouns.

C7 #+18(2-+ 8-0+.5'0( -/ 1 021** *1/<51<(D E.'F(<-1/ 2(8-+1* 1',-+*(0
+.231'(8 ,. ,?.0( .) .,?(' 8-0+-3*-/(0 1/8 .,?(' *1/<51<(0
In this section we will consider how the different features studied in the KIAP project (both SELF and
OTHER manifestation) have contributed to the characterisation of Norwegian academic discourse. We
in the present context take our point of departure in the discipline of medicine. Medicine represents the
most international discipline in our corpus, and the English-language writing tradition in this field is
also by far the most standardised, both in terms of structure (IMRAD, see e.g. Swales 1990) and the
pervasiveness of the ideal of a non-personal style (see e.g. Gotti & Salager-Meyer 2006).
As already indicated in section 2, the first person pronoun, singular or plural, is the most direct and
explicit indicator of author manifestation (SELF dimension). (Since medical research articles are
typically multi-author articles, the first person singular becomes irrelevant in this discussion.)
However, even if the norm in this discipline clearly favours the use of more impersonal structures, the
Norwegian medical subcorpus did comprise some instances of first person plural vi. Here is an
example of this phenomenon:

(S) N. hai ikke noen sikkei foiklaiing p uenne foiskjellen. (nomeuu2)
('We uo not have any ceitain explanation of this uiffeience.')

For the corpus as a whole, our quantitative analyses revealed that medical authors generally use
fewer first person plural subject pronouns than both economists and linguists (for further details, see
Flttum et al. 2006; Flttum 2006). However, in terms of the language variable, we found that
Norwegian medical authors were similar to their English colleagues in the use of this feature: both
groups used more first person plural pronouns than their French peers. In this context, it should be
noted that Norwegian authors in general use vi (we) relatively more frequently than authors writing in
English or French.
!" $%&''()* +" ,-.%* /" /" ,01203456 -61 /"7" 89541-%

144
As mentioned in section 3, indefinite pronouns turned out to play a significant role as
indicators of author presence in academic discourse. It is, however, not easy to compare the use of
these pronouns in the three languages, since the use of one was found to be very limited in the English
corpus texts compared to what was observed for the French and Norwegian texts. For French, our
results supported the traditional view of the use of on as relatively frequent (Gjesdal 2008). For
Norwegian, however, the discipline variable turned out to be important, as the use of the indefinite
pronoun in medical articles turned out to be quite considerable, and more frequent than in Norwegian
economics, but less frequent than in Norwegian linguistics. Here is an example from a Norwegian
medical text:

(4) Man m unng manipulasjon av hjertet fr [...]. (nomed15)
(One must avoid manipulation of the heart before ...)

Another feature serving as a marker of author manifestation is metatext. In the KIAP project we
searched for various expressions referring to (parts of) the text itself. By expressions such as in this
article the author guides the readers through the text (see Dahl 2003, 2004) and explains what will be
done and where (for example In section 4 we analyse ). For methodological reasons, the automated
searches and the quantitative analyses were limited to the following expressions (and their
corresponding expressions in French and Norwegian): article, paper, (sub)section, above, now, below.
Our results indicated very clearly that in medical articles this feature is more or less absent in all the
three languages. In our opinion, this finding may be explained by reference to the highly standardised
text structure in this discipline, making metatext superfluous. For comparative purposes, we may add
that articles in Norwegian economics contain a relatively high number of metatextual expressions.
One reason for integrating polyphonic theory in the analysis of the KIAP corpus was the possibility
this perspective offers to bring in both the SELF and the OTHER perspective. An obvious example of
this is polemic negation where the author implicitly refutes another point of view than his or her own.
The negation expressions studied in our project were not in English, ne pas in French and ikke/ikkje
in Norwegian, all indicating polyphony through the refutation of an underlying positive point of view.
The overall quantitative results showed, roughly, that linguists use more negation than economists,
and economists use it more than medical researchers. As regards our focus discipline in this section,
we note that the language factor is also important, in that Norwegian and English medical authors use
negation more often than French medical authors. Here is a Norwegian example:

(S) |.j, talai alle uesse tie foiholua foi at uen obseiveite auken .::J2 kan tilskiivast
meuiaoppstyiet og pflgjanue seponeiing av pillen Naivelon. (nomeu1u)
('|.j, all these conuitions inuicate that the obseiveu inciease can not be ielateu to |.j')

It may also be noted that Norwegian authors in general use negation more often than authors
writing in English and French. This led us to ask whether authors writing in Norwegian might be more
polemic than their colleagues writing in other languages.
The use of connectives may also be interesting in the perspective of author manifestation.
However, the comparison of connectives in different languages requires particular semantic-pragmatic
"#$% & '()*(+,-./0 ./ 1 +.23*(4 +.'350

14S
considerations. Even if connectives correspond more or less to each other across languages, they often
have particular semantic-pragmatic values in different languages (see e.g. Didriksen 2010). We limited
the automated searches to the adversative connective but and its corresponding connectives in French
(mais) and Norwegian (men).
2
These may also convey different meanings; however, they all have a
concessive meaning in common, and this was the meaning which was particularly interesting in
relation to our research questions. In their concessive capacity, these connectives function as markers
of polyphony as in the following example:

(6) Fleie aibeiuei hai vist at uette gii beuiet piognose, ,2% meuikamentet ei kostbait og
gii kte bluningskomplikasjonei. (nomeuu1)
('Seveial investigations have shown that this gives a bettei piognosis, but the meuication
is expensive anu causes complications |.j.')

The author here first makes a concession related to a proposition expressed in the sequence
preceding but (the source of this proposition is not necessarily the author). The author then contrasts
the proposition with a succeeding sequence, presented as the most important (the point of view of the
author here and now). In this way, concessive constructions contribute to author manifestation and
even author positioning.
The quantitative data related to the frequency of the connectives in question revealed that
linguistics is the discipline that displays the highest frequency of concessive constructions, while
medicine is the discipline with the lowest relative frequency. However, Norwegian medical texts
displayed a remarkably high frequency of this feature compared to English and French texts in this
discipline.
Let us now look at the presence of bibliographical references, which are the most obvious markers
of the OTHER dimension. We limit our presentation of this feature here to the main figures related to
the frequency of bibliographical references, without taking the different types of references and their
function in the articles into consideration (see Flttum & Rastier 2003). Without going into detail
about the importance of bibliographical references in all kinds of academic discourse, we just note that
medicine is a discipline with a relatively uniform referencing standard (Gotti & Salager-Meyer 2006).
When analysing the frequencies of bibliographical references (whether introduced by a number or by
author name + year of publication), we find that when considering the three languages together,
medical authors use more references than economists and linguists. However, considering the
languages separately, it turns out that Norwegian medical authors use more references than their
English and French colleagues.
To sum up some of the findings for the KIAP project as a whole: We found that regarding the issue
of cultural identity, discipline has greater influence than language. This means that, for example, there
are more similarities between Norwegian and French medical articles than between Norwegian
medical and linguistics articles.
3
Our quantitative analyses revealed that both discipline and language
have an effect on the frequency of all the main phenomena studied in the project. For most of them,
however, discipline seems to be more important than language (see Table 1).


2
For quantitative findings on a range of French connectives in the French part of the KIAP corpus, see
Didriksen (2004).
S
Foi moie on language anu cultuial uiffeiences, see Flttum 2u12.
!" $%&''()* +" ,-.%* /" /" ,01203456 -61 /"7" 89541-%

146
14?*# .
7))2/-0 ") -82 5.0/.4'.%2 &%5 '&%+#&+2 )&/-"$0 &%5 -82.$ .%-2$&/-."% "% K&$.&%/2 .% -82 <13* /"$4#0

)#4&,$# 3'%@'E*'2# I42J,4J# G2&#$4@&'C2 H#%'8,4*
Netatex S2% 12% 2% S4%
Bibliogiaphical
iefeiences
44% S% 2% 49%
Auveisative
conjunctions
2S% 9% 1% 6S%
Negation 18% 18% u% 64%
Fiist peison subjects 1S% 11% 1% 7S%
Inuefinite pionoun
subjects
9% 49% S% S9%

Another notable general finding is that we have observed clear individual differences between
articles even within one discipline and within one language, indicating that the genre of the research
article to some extent allows academic authors to develop their own voice when writing up their
research results. In a study related to the KIAP-project (Didriksen & Gjesdal 2006), the relationship
between individual linguistic variation and genre conventions was examined through a quantitative,
corpus-based study of the use of the French first person singular pronoun je in research articles. This
study shows that although there is considerable individual variation, authors still tend to comply with
the norms of the genre. Results also indicate that individual language practices vary over time.
As regards medical discourse specifically, our findings indicate, in very general terms, that it is
clearly different from economics and linguistics discourse. The most obvious difference is related to
author presence. Medical authors are not very visible in their texts compared to their economist and
linguist peers. They hide behind passive constructions and impersonal formulations of different
kinds. Scientific observations present themselves to a larger extent than in economics and linguistics
articles. In an overall perspective, we may also conclude that medical articles resemble each other to a
certain extent, whatever language they are written in. However, as already indicated, we have seen that
Norwegian medical authors to some extent exploit the studied features of SELF and OTHER
representation somewhat differently from their colleagues in the other two writing cultures. They are
more clearly and directly present than their English and especially French colleagues as regards the
use of the first person pronoun, the indefinite pronoun, negation and concessive but. Nomed is also the
subcorpus which displays the highest relative frequency of bibliographical references.
At this point it seems fitting to include some reflections on the nature of our various language
subcorpora. While the English texts in all three disciplines are aimed at an international audience, the
French texts attract a somewhat smaller readership. As for the Norwegian texts, their only potential
audiences outside of Norway are found in the other Scandinavian countries. Our basis for selecting the
Norwegian texts was also smaller than for French and much smaller than for English. In fact, studies
from the early 2000s have shown that eight out of ten articles written by Norwegian researchers are
written in English and more than a third of these were published outside of Norway (Schwach 2004,
see also De Smedt et al. 2012). In consequence, the available material was relatively limited for
Norwegian. Only one journal was available for the disciplines of medicine and economics, while there
were two for linguistics. Notably in economics and medicine, the Norwegian articles tended to address
a broader and more diverse audience than just the relevant research community. Practitioners in both
"#$% & '()*(+,-./0 ./ 1 +.23*(4 +.'350

147
fields (e.g. medical doctors outside research hospitals and economists working in government
departments) also published in these journals and were also part of the intended readership. This fact
may also have influenced the findings for the Norwegian subcorpora for these two disciplines.

G7 H-/1* '(21'I0
While KIAP may be denoted as a large and representative corpus of scientific discourse, the rapid
developments in corpus linguistics and the advent of massive data bases and collections imply that the
landscape of corpus studies is changing. However, we believe that carefully composed corpora
developed from philological considerations still offer benefits that massive collections of raw data do
not (Rastier 2011). An important value of the KIAP corpus, which (despite being considered large in
terms of specialised corpora) more generally can be said to be of limited size, is that it is possible to
get a relatively clear picture of the corpus as a whole. It is in fact possible to access each text, get an
overview, and relate observations made there to the entire corpus. There are sometimes objections to
the use of electronic corpora, for instance that you only find what you search for (Flttum 2005b).
However, in the KIAP project, we repeatedly experienced that new hypotheses and questions
developed through the work on the corpus (see above). Furthermore, the possibility to study each
individual article in depth justifies a rejection of such objections.
Finally, we would like to return to the issues we raised at the beginning of the article, i.e. the
status of scientific discourse in smaller language communities in the face of increasing standardization
and pressure to publish in English. The KIAP corpus provides a window into scientific discourse of
the late 90s and early 00s, and is therefore well suited as a basis for diachronic analyses and
comparison with academic discourse today, ten years later. It would be interesting to compare the
analyses of the KIAP corpus with those of a similar corpus of todays academic discourse, in order to
observe which changes if any the genre of the research article has undergone in this period. Has
the trend towards standardisation continued, or has increasing attention to the potential problems
associated with standardisation influenced writing practices? Furthermore, the resource availability for
Norwegian corpus linguistics has also increased significantly over this ten year period, and large
corpora are now available through the Norwegian National Librarys digital language resource
collection, Sprkbanken. Thus, it is now possible to use resources like the Norwegian Newspaper
corpus (Andersen & Hofland 2012) to compare the KIAP corpus of academic discourse with corpora
of general language.



!" $%&''()* +" ,-.%* /" /" ,01203456 -61 /"7" 89541-%

148
H#K#$#2@#%
Authoi, A. A. (Yeai of publication). C.-'2 ") >"$:; O&4.-&' '2--2$ &'0" )"$ 0#I-.-'2. Location:
Publishei.
Beinut, T. }. (2uu2). Fiienuship quality anu social uevelopment. O#$$2%- P.$2/-."%0 .%
*0D/8"'"+./&' 9/.2%/2A QQ, 7-1u.
Anueisen, u., Boflanu, K. (2u12). Builuing a laige coipus baseu on newspapeis on the web. In: u.
Anueisen (eu), 7L4'"$.%+ R2>04&42$ 6&%+#&+2B S0.%+ -82 >2I -" /$2&-2 &%5 .%K20-.+&-2 & '&$+2
/"$4#0 ") ,"52$% R"$>2+.&%. Amsteiuam: }ohn Benjamins, 1-28.
Beige, K. L. (2uuS). The scientific text genies as social actions: Text theoietical ieflections on the
ielations between context anu text in scientific wiiting. In: K. Flttum, F. Rastiei (eus),
3/&52,./ 5.0/"#$02B T#'-.5.0/.4'.%&$D &44$"&/820. 0slo: Novus, 141-1S7.
Beikenkottei, C., Buckin, T. N. (199S). U2%$2 <%">'25+2 .% 5.0/.4'.%&$D /",,#%./&-."%B Billsuale,
N.}.: Lawience Eilbaum.
Bieivega, K. R. (2uuS). N.-0:&4'2+2 &$+#,2%-&0J"%00-$&-2+.&$B 7.% :",4&$&-.K &%&'D02 &K
0#42$0-$#:-#$2''2 :"%).+#$&0J"%&$ . ,25.0.%0:2A 8.0-"$.0:2 "+ 04$V:K.-0:&4'2+2 &$-.:'&$B
9&:4$"0& 8. 0slo: Noisk sakpiosa.
Bahl, T. (2uuS). Netauiscouise in ieseaich aiticles. In: K. Flttum, F. Rastiei (eus), 3/&52,./
5.0/"#$02B T#'-.5.0/.4'.%&$D &44$"&/820B 0slo: Novus, 12u-1S8.
Bahl, T. (2uu4). Textual metauiscouise in ieseaich aiticles: a maikei of national cultuie oi of
acauemic uiscipline. !"#$%&' ") *$&+,&-./0 S6, 18u7-182S.
Be Smeut, K., Lyse, u. I., ujesual, A. N. & u. S. Losnegaaiu (2u12). R"$0: . 52% 5.+.-&'2 -.50&'52$2%.
Bsseluoif: Spiingei.
Biuiiksen, A. A. (2uu4). P"%/ et auties connecteuis aigumentatifs uans les aiticles ue iecheiche
fianais. Nastei thesis. Bepaitment of Romance Stuuies, 0niveisity of Beigen.
Biuiiksen, A. A. (2u1u). La mthoue ues miioiis smantiques : un point ue upait poui
l'iuentification ues instiuctions logico-smantiques u'un connecteui. In: }. Bavu (eu), 3/-20
5# FN112 O"%+$W0 520 $",&%.0-20 0/&%5.%&K20 X 3/-&0 52' FN11 O"%+$20" 52 $",&%.0-&0
20/&%5.%&K"0. Tampeie 0niveisity Piess, 2S8-2S2.
Biuiiksen, A. A., ujesual, A. N. (2uu6). uenie Constiaints anu Inuiviuual Linguistic vaiiation. In:
E. Suomela-Salmi, F. Beivin, F. (eus) Cioss-cultuial anu Cioss-linguistic Peispectives on
Acauemic Biscouise. Tuiku: Bepaitment of Fiench Stuuies, The 0niveisity of Tuiku,
Finlanu, 47-S7.
Flttum, K. (2uuS). Peisonal English, inuefinite Fiench anu pluial Noiwegian scientific authois.
Pionominal authoi manifestation in ieseaich aiticles. R"$0: 6.%+K.0-.0: C.500:$.)- 21, 21-SS.
Flttum, K. (2uu4a). La pisence ue l'auteui uans les aiticles scientifiques: tuue ues pionoms J2,
%"#0 et "%. In: A. Auchlin et al. (eus), 9-$#/-#$20 2- 5.0/"#$02B Qubec: Eu. Nota Bene, 4u1-416.
Flttum, K. (2uu4b). Polyfonisk inteiaksjon via IKKE i vitenskapelig uiskuis. Y82-"$./&
9/&%5.%&K./& S1, 2S-4u.
Flttum, K. (2uuSa). The self anu the otheis - polyphonic visibility in ieseaich aiticles.
1%-2$%&-."%&' !"#$%&' ") 344'.25 6.%+#.0-./0 1S, 29-44.
"#$% & '()*(+,-./0 ./ 1 +.23*(4 +.'350

149
Flttum, K. (2uuSb). Lingvistiske og uiskuisive stuuiei i KIAP-koipuset. 0m utvikling og biuk av
et fleispiklig og fleiuisiplinit koipus av vitenskapelige aitiklei. In: S. Rike (eu), <"$4#0 .
)"$0:%.%+ "+ #%52$K.0%.%+B ZK"$ 0-V$ K. . R"$+2[ Kiistiansanu: Bgskolen i Aguei, 2S-S1.
Flttum, K. (2uu6). En spiklig analyse av noisk meuisinsk piosa. C.500:$.)- )"$ P2% %"$0:2
'\+2)"$2%.%+, 126 (1), 6S-68.
Flttum, K. (2u1u). Linguistically maikeu cultuial iuentity in ieseaich aiticles. In: u. uaizone, }.
Aichibalu (eus), P.0/"#$02A .52%-.-.20 &%5 $"'20 .% 042/.&'.]25 /",,#%./&-."%. Bein: Petei Lang,
267-28u.
Flttum, K. (2u12). vaiiation of stance anu voice acioss cultuies. In: K. Bylanu, C. Sancho uuinua
(eus), 9-&%/2 &%5 N"./2 .% &/&52,./ 5.0/"#$02. Lonuon: Palgiave NacNillan, 218-2S1.
Flttum, K., Rastiei, F. (eus). (2uuS). 3/&52,./ 5.0/"#$02B T#'-.5.0/.4'.%&$D &44$"&/820, 0slo:
Novus.
Flttum, K., Bahl, T., Kinn, T. (2uu6). 3/&52,./ N"./20. AmsteiuamPhilauelphia: }ohn Benjamins.
ujesual, A. N. (2uuS). L'emploi uu pionom "on" uans les aiticles ue iecheiche. 0ne tuue
uiachionique et qualitative. Nastei thesis. 0niveisity of Beigen: Romansk institutt.
ujesual, A. N. (2uu8). ^-#52 0_,&%-.`#2 5# 4$"%", aR 5&%0 #%2 42$042/-.K2 -2L-#2''2 2-
/"%-2L-#2''2B PhB uisseitation. Beigen: 0niveisity of Beigen.
uotti, N., Salagei-Neyei, F. (eus). (2uu6). 35K&%/20 .% T25./&' P.0/"#$02 3%&'D0.0; a$&' &%5
E$.--2% O"%-2L-0. BeinBeilin: Petei Lang.
Bylanu, K. (1998). Z25+.%+ .% 9/.2%-.)./ Y202&$/8 3$-./'20. Amsteiuam: }ohn Benjamins.
Bylanu, K. (2uuu). P.0/.4'.%&$D 5.0/"#$020; 0"/.&' .%-2$&/-."%0 .% &/&52,./ >$.-.%+. Bailow, 0K:
Longman.
}ohansson, S. (199S). T2%0 0&%& .% /"$4"$2 0&%": 0n the Role of Coipoia in Linguistic Reseaich.
C82 7#$"42&% 7%+'.08 T2002%+2$, Iv, 2, 19-2S.
}ohannsson, S. (2uu2). viewing language thiough multilingual coipoia, with special iefeience to
the geneiic peison in English, ueiman anu Noiwegian. In: L.I. Rbaue, S.N. Boval Suaiez
(eus), 9-#5.20 .% O"%-$&0-.K2 6.%+#.0-./0. 0niveisiuaue ue Santiago ue Compostela, S1S-SS4.
Kinn, T. (2uu4). Cognitive ieseaich agents in acauemic piose. 3:&52,.0: *$"0&; 9:$.)-2$ )$& <13*.
Romansk Institutt, 0niveisitetet i Beigen, 1S7-149.
Kinn, T. (2uuSa). "Benne aitikkelen analyseiei ...". Ben tenkjanue foiskaien i noiske
foiskingsaitiklai. TaR9 Qb #-K&'52 &$-.:'&$ )$V 52- -.&%52 Tc-2 ", %"$0: 04$V: . <$.0-.&%0&%5
dbbe. Kiistiansanu: Byskolefoilaget, 18S-194.
Kinn, T. (2uuSb). Tilbou og innbyuingai: impeiativ meu '& i foiskingsaitiklai. R"$0: '.%+K.0-.0:
-.500:$.)- 2S (2), 14S-17S.
Lafuente-Nilln, E. Nui-Bueas, P. Lois-Sanz, R., vzquez-0ita, I. (2u1u). Inteipeisonality in
wiitten acauemic uiscouise: Thiee analytical peispectives. In: R. Lois-Sanz, P. Nui-Bueas,
E. Lafuente-Nilln (eus), O"%0-$#/-.%+ .%-2$42$0"%&'.-D; T#'-.4'2 42$042/-.K20 "% >$.--2%
&/&52,./ +2%$20. Newcastle: Scholais Publishing, 1S-S9.
Nauianen, A. (199S). O#'-#$&' 5.))2$2%/20 .% &/&52,./ $82-"$./. Fiankfuit am Nain: Petei Lang.
!" $%&''()* +" ,-.%* /" /" ,01203456 -61 /"7" 89541-%

1Su
Nauianen, A., Piez-Llantaua C., Swales, }. (2u1u). Acauemic Englishes - A Stanuaiuizeu
knowleuge. In: A. Kiikpatiick (eu), C82 Y"#-'25+2 8&%5I"": ") >"$'5 7%+'.0820. Lonuon:
Routleuge, 6S4-6S2.
Nlke, B., Flttum, K., Noin, C. (2uu4). 9/&*"6.%2B 6& -8_"$.2 0/&%5.%&K2 52 '& 4"'D48"%.2
'.%+#.0-.`#2. Paiis: Kim.
Pouuat, C. (2uu6). ^-#52 /"%-$&0-.K2 52 'f&$-./'2 0/.2%-.).`#2 52 $2K#2 '.%+#.0-.`#2 5&%0 #%2
42$042/-.K2 5f&%&'D02 520 +2%$20. Boctoial uisseitation. 0ilans: 0niveisit u'0ilans.
Pielli, L. }. (1989). 3 Y82-"$./ ") 0/.2%/2; 1%K2%-.%+ 0/.2%-.)./ 5.0/"#$02. 0niveisity of South
Caiolina Piess.
Rastiei, F. (2u11). 6& ,20#$2 2- '2 +$&.% ; 0_,&%-.`#2 52 /"$4#0. Paiis: Bonoi Champion.
Rastiei, F. (2uu1). 3$-0 2- 0/.2%/20 5# -2L-2. Paiis: P0F.
Rinck, F. (2uu6). 6M&$-./'2 52 $2/82$/82 2% 9/.2%/20 5# '&%+&+2 2- 2% 62--$20B H.+#$2 52 'M&#-2#$ 2-
.52%-.-_ 5.0/.4'.%&.$2 5# +2%$2B Boctoial uisseitation uienoble: 0niveisit ue uienoble.
Sanueison, T. (2uu8). O"$4#0A /#'-#$2A 5.0/"#$02. Tbingen: uuntei Naii veilag.
Schwach, v. (2uu4). R"$0: K.-2%0:&4 g 4V 04$V:'.+ I"$-2I&%2[ 7- 4.'"-4$"0J2:- ", 04$V:I$#: I'&%-
)&+0&,)#%%2- &K )"$0:2$2 . R"$+2B NIF0 skiiftseiie, 9.
|http:www.nifu.noNoiwayPublications2uu4skiiftseiie9-2uu4.puf, |accesseu 1u
Septembei 2u12j.
Swales, }. N. (199u). U2%$2 &%&'D0.0; 7%+'.08 .% &/&52,./ &%5 $202&$/8 02--.%+0B Cambiiuge:
Cambiiuge 0niveisity Piess.
vassileva, I. (2uuu). E8" .0 -82 &#-8"$[ 3 /"%-$&0-.K2 &%&'D0.0 ") &#-8"$.&' 4$202%/2 .%
7%+'.08AU2$,&%A H$2%/8A Y#00.&% &%5 h#'+&$.&% &/&52,./ 5.0/"#$02B Sankt Augustin: Asgaiu
veilag.