1 5Te minologija | 2015 | 22
Te ms in Tex s and he Challenge
o Te minology Managemen
KLAUS-DIRK SCHMITZ
CologneUni e si yo AppliedSciences
KEYWORDS: e minology managemen , e m ex ac ion, concep o ien a ion, e m au onomy,
e minological en y, synonymy, ambigui y
1. BASiC THeOReTiCAL FOUNDATiONS
Te minology is de ined as he “se o designa ions belonging o one
special language” (iSO 1087–1:2000). This in e na ional de ini ion e e s
mo e o he ep esen a ion pa o special language communica ion and
igno es he concep ual iew behind he designa ions. A mo e comp e-
hensi e de ini ion is gi en in he Ge man e minology s anda d DiN
2342:2011 whe e “ e minology is he se o in en o y o concep s and
hei ep esen a ions in a speci ic subjec ield”. This de ini ion no only
includes he concep as an impo an aspec o e minology; i also shows
a b oade iew o he concep ep esen a ion side o e minology by no
limi ing i o e ms o o he language- ela ed designa ions and by includ-
ing symbols, icons, ges u es o any o he mul imedia ep esen a ions.
in o de o explain he ela ion be ween concep and e m, e minol-
ogy heo is s and esea che s adap ed he so-called “semio ic iangle”
in oduced by he Ame ican linguis s C. K. Ogden and i. A. Richa ds
(Ogden, Richa ds 1923: 11) o explain he ela ionship be ween concep s
and e ms (Figu e 1). The iangle has unde gone a long his o y o mod-
i ica ions and in e p e a ions, and has also been a acked om se e al
qua e s as an o e -simpli ica ion o mis ep esen a ion o he complex
ela ionships ha exis be ween concep s and e ms. Despi e he c i icisms
ha ha e been le eled agains he iangle, i ’s simplici y makes i an
excellen ool o illus a ing concep - e m ela ionships o people who
a e jus beginning o iden i y e ms in ex s and o c ea e e minological
da a en ies o documen hem. O he and mo e complex models a e
desc ibed in A n z, Pich , Schmi z (2014: 41 .).
1 6 Klaus-Di k Schmi z | Te msinTex sand heChallenge
| o Te minologyManagemen
Figu e 1. Te minological iangle (based on Ogden, Richa ds 1923)
Concep s a e “cogni i e ep esen a i es” (“gedankliche Ve e e ” (Fel-
be , Budin 1989: 69)) o objec s, ha a ise ou o he ac ha humans
ecognize he common cha ac e is ics ha exis in a majo i y o indi-
idual objec s o he same ype, and hen s o e hese cha ac e is ics and
use hem o impose o de on he wo ld o objec s, in o de o achie e
mu ual unde s anding when hey communica e wi h o he people. iSO
1087–1:2000 de ines a concep as a “uni o knowledge c ea ed by a
unique combina ion o cha ac e is ics” and DiN 2342:2011 desc ibes a
concep again mo e explici ly as a “uni o hinking made up o cha ac-
e is ics ha a e de i ed by ca ego izing objec s ha ing a numbe o iden-
ical p ope ies.”
Bo h s anda ds s a e in a no e ha concep s a e no necessa ily bound
o speci ic languages, bu he cul u al, social and echnical backg ound o
he human beings who gene a e he concep s and he en i onmen s in
which he concep s a e used a ec he way hey mani es hemsel es in
any gi en si ua ion. Regional di e ences wi hin a language communi y
(e.g. Ge many and Aus ia) may lead o di e en concep ual o ien a ions
o he same e m, whe eas one cul u al communi y whe e di e en lan-
guages a e spoken (e.g. Swi ze land) may alloca e he iden ical concep
ep esen ed by se e al e ms in di e en languages.
Since concep s a e men al o cogni i e ep esen a ions, we need de ini-
ions o explain and desc ibe concep s. in mos cases, de ini ions e e o
o he concep s (e.g. he supe o dina e concep ) and men ion speci ic cha -
ac e is ics ha a e unique and ypical o he concep o be de ined. On
he basis o de ini ions and concep cha ac e is ics, i is possible o ela e
1 7Te minologija | 2015 | 22
concep s o each o he and o cons uc e minological concep sys ems,
axonomies, on ologies, o o he knowledge o ganiza ion sys ems (SKOS).
These concep sys ems ep esen he knowledge o a domain o sub-domain
in a sys ema ic way.
The e m is de ined in iSO 1087–1:2000 as a “ e bal designa ion o a
gene al concep in a speci ic subjec ield” o in DiN 2342:2011 as a
“designa ion o a de ined concep in a special language by a linguis ic
exp ession.” The e m se es as he ep esen a ion o he concep and we
can w i e i down, say i ou loud and use i o communica ion. We use
he wo d “designa ion” as a supe o dina e concep when we alk abou
e ms because he e a e also o he ways o ep esen concep s, e.g. ones
ha a en’ necessa ily made up o wo ds, such as symbols, o mulas, pic-
og ams, ges u es, e c.
Some e ms consis o mo e han one wo d. These e ms a e called
mul i-wo d e ms o compounds, e.g. “p in e wi h single-shee eed.”
The way wo ds combine o o m e ms a ies om language o language.
When dealing wi h e ms, he linguis ic side o e minology wo k comes
in o he game.
We ha e seen ha he e m is he e bal ep esen a ion o he concep .
in special o echnical language, i is highly desi able ha his ela ionship
be unambiguous, e en wi hou con ex ual e e ence, which means ha
one e m should be assigned o one concep , c ea ing a condi ion called
uni ocali y. When his condi ion p e ails, he meaning o e ms is com-
ple ely clea , e en i he e m appea s wi hou any explana o y con ex .
O cou se, his ideal si ua ion is di icul o achie e o en o ce. Two p ob-
lems in ol ing e m-concep assignmen ecu equen ly, e en in echni-
cal and scien i ic ex s:
Synonymy exis s i wo o mo e e ms in a gi en language ep esen
he same concep . Thus a synonym is a e m used o designa e he same
concep as ano he e m. e en hough synonymy can comp omise com-
munica ion be ween expe s, i occu s qui e equen ly in p ac ice. This
can happen especially in subjec ields whe e many objec s and concep s
a e s ill unde going de elopmen . in hese kinds o dynamic ields, com-
pe ing e ms a e used in pa allel un il unambiguous e ms a e g adually
es ablished, ei he h ough a na u al selec ion p ocess o by conscious
s anda diza ion. e en when people a e qui e awa e o hese p oblems,
a ian s can emain in use o long pe iods o ime, based on such ac o s
1 8 Klaus-Di k Schmi z | Te msinTex sand heChallenge
| o Te minologyManagemen
as na u al egional a ia ion o , qui e in en ionally, company o p oduc -
speci ic e o s o use e minological di e ences as one means o posi ion-
ing a p oduc in he ma ke .
Homonymy in ol es he opposi e si ua ion om synonymy: he e a
e m o se e al e ms ha ha e he same ex e nal o m e e o se e al
concep s1. iSO 1087:2000 de ines homonymy as a ela ion be ween des-
igna ions and concep s in a gi en language in which one designa ion
ep esen s wo o mo e un ela ed concep s. Homonyms pose huge p ob-
lems o echnical communica ion. As a consequence, expe s a e con-
s an ly s i ing o a oid homonyms in echnical subjec ields. Ne e he-
less, when new concep s e ol e, people like o o m new e ms o hem
by combining amilia exis ing e ms o by adop ing es ablished e ms
om gene al language (e.g. mouse o a compu e inpu de ice) o om
o he ela ed subjec ields (e.g. i us om medicine o malwa e).
2. CONCePT-ORieNTeD TeRMiNOLOGy MANAGeMeNT
The esul s o any kind o e minological wo k ha e o be s o ed oday
in e minological da abases ( e m bases) o e minology managemen
sys ems. Al hough he access o and he e ie al o he con en o he
e m bases will happen in mos cases ia he (sea ch) e m, he o ganiza-
ional p inciple o his e minological know ledge esou ce has o be he
concep .
This is he basic di e ence be ween a lexicog aphical en y in a dic ion-
a y o lexicon and a e minological en y in a e m base o e minology
managemen sys em. While igu e 2 shows he undamen al s uc u e o
a dic iona y en y (e.g. Wik iona y) whe e he wo d ( e m) is he basis
o o ganizing he linguis ic in o ma ion, igu e 2 shows he s uc u e o
a p esc ip i e e minological en y in a e m base (e.g. Wikipedia) wi h
he main ocus on he concep .
Figu e 3 also demons a es he p esc ip i e app oach o e minology
managemen implemen ed by s anda dizing o ganiza ions and companies
o es ablish a consis en and unambiguous co po a e language; only one
e m o each concep is he p e e ed e m and o he e ms o he same
concep a e admi ed o dep eca ed.
1 The di e ence be ween homonymy and polysemy will no be discussed he e; i is i ele an o compu -
e ized e minology managemen and he e o e o his pape .
1 9Te minologija | 2015 | 22
Figu e 2. S uc u e o a lexicog aphical en y
Figu e 3. S uc u e o a (p esc ip i e) e minological en y
As de ined in iSO 1087-1:2000 and e lec ed in he e minological me-
a model in iSO 16642:2003, a e minological en y has o con ain all
e minological da a ela ed o one concep . The e o e, e minological da-
a modeling has o e lec he p inciple o concep o ien a ion (see
igu e 3), hus allowing o he main enance no only o all concep - ela -
ed in o ma ion bu also o all e ms in all languages wi h all e m- ela ed
in o ma ion wi hin one e minological en y. Te minological en ies de-
signed acco ding o he p inciple o e m o ien a ion (see igu e 2), which
we e y o en ind in bilingual glossa ies o dic iona ies, a e no app op i-
a e o me iculous e minology managemen and will lead e y soon o
inconsis en e minology collec ions ha a e no e y use ul, especially
i mul ilingual e minology managemen is equi ed. i LSP lexicog aph-
ical p oduc s – especially in p in ed o m – ha e o be c ea ed, a e m-
o ien ed alphabe ical iew can be gene a ed om a concep -o ien ed

2 0 Klaus-Di k Schmi z | Te msinTex sand heChallenge
| o Te minologyManagemen
e minological da a base, since only he concep ual o ganiza ion can gua -
an ee an adequa e collec ion, p ocessing, e ision and p epa a ion o he
domain-speci ic e ms.
in addi ion – no in opposi ion – o concep o ien a ion, he second
impo an p inciple o e minological managemen is e m au onomy.
Te m au onomy gua an ees ha all e ms including synonyms, abb e i-
a ed o ms and spelling a ian s can be documen ed wi h all necessa y
e m- ela ed da a ca ego ies such as g amma , s yle, geog aphical es ic-
ion o con ex . This app oach can be ealized by designing he da a
model in a way ha allows he use o c ea e an unlimi ed numbe o
e m sec ions o e m blocks con aining indi idual e ms and all addi-
ional da a ca ego ies desc ibing he e m and i s use.
Te m au onomy is ep esen ed in he e minological me a model (iSO
16642:2003) by he ac ha each e m sec ion is only allowed o ha e
exac ly one e m. Se e al e ms in one language o he same concep
(synonymous designa ions) will be o ganized by using se e al e m sec-
ions each con aining exac ly one e m and addi ional da a ca ego ies
documen ing his e m. in applica ion scena ios whe e he e minology
managemen sys em plays he ole o he e minological knowledge base
o a numbe o applica ions and p og ams, e m au onomy is eally es-
sen ial; e.g. a quali y con ol p og am ha checks he co ec use o e ms
in ex s, has o ha e access o he e m base whe e p e e ed, admi ed
and dep eca ed e ms a e s o ed in he same e minological en y (see
igu e 3), bu in di e en e m sec ions.
The p inciples o concep o ien a ion and e m au onomy a e e lec ed
in he model o a e minological en y shown in igu e 4. Te minology
managemen p og ams designed acco ding o his model a e able o gi e
a wa ning when he use ies o en e in o ma ion ela ed o only one
concep in o di e en e minological en ies (see Schmi z 2011: 242 .).
Concep o ien a ion and e m au onomy a e p e equisi es o domain-
o company-speci ic e minological da a collec ions (see also Schmi z,
S aub 2010: 38 .). Synonymous e ms such as USBs ick, USBmemo y
s ick, USB lashd i e, USBmemo ykey, memo ys ick, pend i e, humbd i e,
o only key as a sho o m, a e s o ed in one e minological en y wi h
only one english de ini ion and concep ela ions o o he concep en ies;
each e m can be documen ed wi h a ibu es speci ying ha e.g. USB
lashd i e is he p e e ed e m o documen a ion and USBkey o
p oduc labelling, and all o he e ms shall no be used wi hin any docu-
2 1Te minologija | 2015 | 22
men s o his company. Because o he concep o ien a ion, any use
consul ing he e m base sea ching o e.g. pend i e will ind he concep
en y (wi h he de ini ion), will see ha pend i e is no he ecommend-
ed e m, and will ind he p e e ed e m USB lashd i e.
On he o he hand, sea ching o key will esul in se e al hi s in he
same da abase, because he e m key is a designa ion o se e al concep s
(homonym o polysemous wo d, e.g. key as pa o he keyboa d) and
he e o e s o ed in se e al en ies wi h di e en de ini ions and concep
ela ions.
3. TeRMS iN TeXTS
Al hough e ms in e minology managemen sys ems a e o ganized in
concep -o ien ed en ies, hey a e used in echnical documen s as elemen s
o he linguis ic su ace o he ex . Depending on he mo phological and
syn ac ical ea u es o a gi en language, e ms appea in in ec ed o m
indica ing e.g. plu al, geni i e, o pas ense. The connec ion om a e m
in a documen o he co esponding concep can only be es ablished by
he eade comp ehending he message o he ex and g asping he mean-
ing behind he e m.
Figu e 4. Model o he e minological en y e lec ing concep o ien a ion and e m au onomy
2 2 Klaus-Di k Schmi z | Te msinTex sand heChallenge
| o Te minologyManagemen
in special o echnical communica ion, i is highly ecommended o use
only one e m o a single concep (no synonyms) and o use only e ms
ha a e no ambiguous and do no ep esen wo o mo e concep s (no
homonyms). Bu his is no always easy o achie e and he wo ld is ull
o documen s ha do no mee he equi emen s o a consis en and
uni ocal usage o e minology. Figu e 5 shows a sample documen wi h
6 e ms, some o hem a e mul i-wo d e ms, ha (may) ep esen he
same concep . Fo unde s anding he message o he ex , he eade has
o cope wi h he ques ion, i all e ms ep esen he same concep o ha e
di e en meanings. Some non-expe s in he domain he documen deals
wi h may be con used, emba assed, o e en o e s ained.
Figu e 5. Sample documen wi h e ms ha may ep esen he same concep
Bu no only human eade s o echnical documen s will ha e p oblems o
deal wi h ex s con aining synonymous and homonymous e ms. Compu e -
assis ed ex p ocessing ools will ha e simila o e en mo e p oblems.
in many o ganiza ional en i onmen s and applica ion scena ios o e -
minology wo k, he ex ac ion o e minology om exis ing ex ual ma-
e ial is ecommended. Typical scena ios a e he p epa a o y e minology
wo k o la ge ansla ion p ojec s wi h se e al ansla o s, be o e he
ansla ion s a s and each ansla o has o do (p obably he same) ad-hoc
e minology wo k, and he ini ial eeding o a new e m base wi h com-
2 3Te minologija | 2015 | 22
pany o subjec speci ic e minology in o de o iden i y he basic neces-
sa y se o concep s and e ms.
Te minology ex ac ion comp ises asks o ex ac ing e minological
in o ma ion, mainly e ms i sel , om ex ual ma e ial. Tex ual ma e ial
can be a se o monolingual documen s, a pai o pa allel ex s ei he
p oduced in bo h languages, a sou ce language ex oge he wi h i s
ansla ion, o a ex co pus wi h a s uc u ed and sys ema ically col-
lec ed se o sen ences.
Human e m ex ac ion by domain expe s o expe ienced e minolo-
gis s is he mos ime consuming and expensi e me hod, bu p obably
leads o he bes esul s. Compu e -assis ed e m ex ac ion p og ams can
handle ex s in (almos ) all languages i hey use only s a is ical me hods.
i hei e m iden i ica ion algo i hm is based on linguis ic me hods, he
esul s o e m ex ac ion will be much be e , especially o mul i-wo d
e ms and ph ases, bu (comme cial) linguis ic-based e m ex ac ion ools
a e only a ailable o “majo ” languages such as english, F ench, Ge man,
Spanish and ew o he s.
Te m ex ac ion ools o e common unc ionali ies known om con-
co dance p og ams (e.g. Wo dSmi h): hey iden i y he wo ds o a ex-
ual documen , c ea e wo d equency s a is ics, display a KWiC index
(Key Wo d in Con ex ), and display he esul s so ed in alphabe ic o de
o by equency. Since wo ds appea in ex s in in lec ed o ms, linguis ic-
based e m ex ac ion ools can educe he ex o m o a wo d o i s
basic canonical o m; his is needed o eal wo d s a is ics and eliable
e m candida e lis s, bu equi es linguis ic knowledge abou he mo phol-
ogy o he espec i e language.
Since e minology is always ela ed o domain-speci ic language, e m
ex ac ion ools should be able o il e ou and igno e unc ion wo ds
(e.g. a icles, conjunc ions, p eposi ions e c.) as well as gene al language
wo ds; o his ea u e, mos o he ools use so-called s op wo d lis s ha
a e language dependen and can be complemen ed by he use . Bu some-
imes i is no so easy o decide i a wo d is a gene al language wo d o
a special language e m.
Al hough e m ex ac ion ools may be e y help ul in speci ic applica-
ion scena ios, he ollowing issues ha e o be aken in o accoun :
• The esul o a e m ex ac ion p ocess is a lis o e m candida es;
his lis mus be checked and “cleaned” by a e minologis .